Spray-Printed and Self-Assembled Honeycomb Electrodes of Silicon-Decorated Carbon Nanofibers for Li-Ion Batteries

Directional, micron-scale honeycomb pores in Li-ion battery electrodes were fabricated using a layer-by-layer, self-assembly approach based on spray-printing of carbon nanofibers. By controlling the drying behavior of each printed electrode layer through optimization of (i) the volume ratio of fugitive bisolvent carriers in the suspension and (ii) the substrate temperature during printing, self-assembled, honeycomb pore channels through the electrode were created spontaneously and reliably on current collector areas larger than 20 cm × 15 cm. The honeycomb pore structure promoted efficient Li-ion dynamics at high charge/discharge current densities. Incorporating an optimum fraction (2.5 wt %) of high-energy-density Si particulate into the honeycomb electrodes provided a 4-fold increase in deliverable discharge capacity at 8000 mA/g. The spray-printed, honeycomb pore electrodes were then investigated as negative electrodes coupled with similar spray-printed LiFePO4 positive electrodes in a full Li-ion cell configuration, providing an approximately 50% improvement in rate capacity retention over half-cell configurations of identical electrodes at 4000 mA/g.


■ INTRODUCTION
Porous or hollow frameworks and particulate materials are useful in a range of applications, including photovoltaics, 1,2 electronics, 3,4 sensing, 5,6 biologics, 7,8 electrocatalysis, 9,10 and energy storage, 11,12 where maximizing mass transport behavior (e.g., ion mobility) and active surface area is desired. Porous and hollow structures have been studied extensively for rechargeable Li-ion battery (LIB) technologies where interconnected pores within a LIB anode or cathode are essential for effective ion diffusion and "deep" penetration to all active sites within the electrode; 13,14 these relatively open structures may also help to mitigate the mechanical strain of repeated lithiation/delithiation processes. 15,16 The associated high active area may contribute positively to fast electrode charge/discharge kinetics but may also be detrimental where a passivating solid electrolyte interphase (SEI) is formed in the first few charge/discharge cycles, irreversibly consuming Li ions, or especially where a nonpassivating SEI is formed continuously. Although high porosity active materials and engineered porosity at the electrode scale may promote high capacity per unit weight of active materials, they also usually undermine volumetric capacity of resulting cells and battery packs, which is undesirable from a technological point of view.
A high electrode surface area can be achieved by using an active material with fine-scale or even nanoscale porosity. For example, a wide range of high surface area, hollow particles and tube structures at the nanoscale have been proposed and successfully demonstrated in LIB electrodes based on hydro-thermal, sol−gel, and sacrificial template methods. 17−20 Recently, yolk−shell designs for various electrode materials, including TiO 2 , Si and others have been developed with the intention of improving both high-rate capability and charge/ discharge cycle stability in a LIB configuration. 21−24 At the electrode manufacture stage, these high-specific-surface area materials are formed into an LIB electrode using slurry casting, which is the dominant and most productive electrode fabrication method. 25−27 A composite slurry, including carbon-based conductivity enhancers and polymeric binders, in an aqueous or nonaqueous solvent, is cast onto a current collector via a doctor blade or slot-die process and then sequentially dried to produce a robust, well-adhered composite coating over large areas, including using semicontinuous or near-continuous roll-to-roll operations. 28,29 However, the intrinsic electrochemical performance of the as-synthesized materials is usually restricted or inhibited in the slurry cast electrode microstructure because the binder partly obscures some of the active surface, and when reasonable coating thicknesses (50 μm and above) are used to achieve practically usable volumetric capacities, the electrode porosity is so tortuous that Li-ion mobility is restricted at higher charge/ discharge rates and capacity fades.
To address these constraints, new approaches to produce micron-scale porous structures with more efficient integration of active materials, conductivity enhancers, and binders have been proposed. For example, template-assisted methods offer excellent control over pore size and morphology but fabricating and then removing the fine-scale featured sacrificial templates can cause problems for cost and scalability and volumetric capacity tends to be unhelpfully low. 30−32 Lithographic technologies based on ultraviolet light irradiation and/ or ion etching processes allow pore fraction, shape, and alignment to be controlled with micron resolution but require inherently high-cost fabrication apparatus. 33−35 Thus, there remains an opportunity for manufacturing approaches that allow improved and more flexible micron-scale pore engineering in LIB electrodes and that have the potential for costeffective scalability.
In this paper, self-assembled, fibrous honeycomb electrodes were readily manufactured directly onto current collectors via a spray-printing approach. The process is demonstrated for otherwise hard-to-process hollow carbon nanofiber (CNF)based electrodes that were then investigated for LIB applications, first in half-cells and then in full cells using optimized versions of a honeycomb CNF-based anode and a spray-printed LiFePO 4 (LFP)-based cathode. Optimization of the drying behavior of the printed electrode layers, in particular, the formulation of the bisolvent carrier used in the ink suspension and the substrate temperature during sprayprinting, ensured that the honeycomb network of micron-scale through-thickness pore channels formed reliably over large areas. Electrochemical testing showed that the honeycomb pores promoted Li-ion mobility and excellent gravimetric and volumetric capacity at high charge/discharge rate conditions. To boost the specific capacity of the honeycomb CNF electrodes further, an optimum 2.5 wt % fraction of highcapacity Si nanoparticles was grafted onto the CNFs, increasing the discharge capacity at the highest current density of 8000 mA/g by a factor of four and giving a 65% improvement in the sustained discharge capacity after 300 cycles at 400 mA/g. Spray-printed CNF-based anodes were coupled with spray-printed LFP-based cathodes in a full-cell configuration where the honeycomb CNF + 2.5 wt % Si outperformed an identical honeycomb CNF-only equivalent in rate capability, cycle stability, and deliverable discharge capacity. We show that the principal electrochemical benefit of the bottom-up spontaneous self-assembly of the active materials into honeycomb electrodes is to sustain capacity into the high current density region. The honeycomb structure allows the intrinsic electrochemical performance of well-known but difficult-to-process materials to be more readily realized.
■ RESULTS AND DISCUSSION Figure 1 depicts a schematic of the spray-printing of honeycomb electrodes directly onto a large area current collector, typically >20 cm × 15 cm. For the successful selfassembly of the directional pore structure in each sprayed layer and from layer to layer, the suspension containing active materials and inactive components must be optimized for (i) appropriate rheological properties to enable facile pumping, from the reservoir to the atomizing spray head; (ii) stable atomization without pulsing; and (iii) appropriate drying behavior to propagate the honeycomb structure without excessive fugitive liquid buildup and resuspension of predeposited material. In this case, CNF-based active material, conductivity enhancer, and a polymeric binder in a controlled mass ratio were suspended in a fugitive bisolvent liquid carrier of deionized (DI) water and isopropyl alcohol (IPA). The DI/ IPA ratio was optimized to provide both a repeatable suspension without any sedimentation and two-stage drying behavior of the bisolvent to drive the self-assembly of the honeycomb structure. 36 The suspensions were atomized using compressed air in a spray nozzle that moved in a preprogrammed zigzag pattern in the x and y plane with a fixed z spray distance to the current collector for multiple cycles. The heated current collector was set to a temperature higher than the boiling point of the bisolvents to ensure rapid in situ drying and a micron-scale through-thickness honeycomb pore structure, as depicted in the magnified cartoon of Figure 1. Figure 2 presents scanning electron microscopy (SEM) images of a self-assembled, honeycomb porous electrode, fabricated by layer-by-layer spray-printing using CNFs of 100− 200 nm diameter (see Figure S1 in the Supporting Information). Pores of 20 ± 8 μm diameter were separated by interwoven CNF fibrous honeycomb cell "walls" of thickness of 10 ± 5 μm, as shown in the electrode plan view image in Figure 2a. Figure 2b shows a magnified view looking "down" through a pore showing smaller pores (2 ± 1 μm), suggesting that adjacent, smaller pores had combined and merged into the larger pore as the deposition progressed layer by layer. 36,37 The electrode cross section in Figure 2c reveals the intended through-thickness (20 μm) pore channels from the electrode surface to the current collector, as highlighted by the yellow dotted line, with thicknesses from 10 to 80 μm readily available by tuning the number of spray scan cycles. 36 A magnified electrode cross-sectional view in Figure 2d, obtained using focused ion beam (FIB) milling, shows more detail of the interior of the fibrous honeycomb structure, supporting the intent that the micron-scale pores were interconnected throughout the electrode.
For suspension stability, the DI/IPA ratio was critical because although the polymeric binder (carboxymethyl

ACS Applied Materials & Interfaces
Research Article cellulose, CMC) dissolved in DI water only, the CNFs and carbon conductivity enhancer (Super P, SP) were dispersed preferentially in IPA. Figure 3 shows the CNF-based electrode morphology as a function of DI/IPA volume ratios of 100:0, 70:30, 50:50, 30:70, and 0:100, with the bar graphs in each inset of Figure 3a−e indicating the different volume ratios of DI water (blue) and IPA (red). For the suspension formulated with DI water only, the plan view in Figure 3a and the corresponding tilted view in Figure 3f show almost no deposition of the active material, implying a lack of suspension stability. Although the suspension with IPA only (Figure 3e,j) showed a slightly improved deposition yield, the electrode was again too thin and too poorly adhered to the current collector for use, which was attributed to a lack of function of the CMC binder that did not disperse in the IPA-only suspension. For DI/IPA mixtures of 70:30 and 30:70, the suspensions were noticeably more stable without any obvious sedimentation and the deposition efficiency increased markedly. However, no reliably reproducible micron-scale pore structure was formed, as shown in Figures 3b,g, and 3d,i. At a 50:50 DI/IPA volume ratio (Figures 3c,h), micron-scale through-thickness honeycomb pores were spontaneously formed and reproduced readily from experiment to experiment.
The effect of the substrate temperature was investigated using 50:50 DI/IPA suspensions at following conditions: (i) 60°C (lower than the boiling point of both DI water and IPA), (ii) 90°C (higher than the boiling point of IPA at ∼80°C but lower than DI water), and (iii) 120°C (higher than the boiling point of both DI water and IPA), as indicated in the bar graph in Figure 4a. At 60 and 90°C, the resulting electrodes had an excellent deposition efficiency but no micron-scale pores, as shown in the SEM images of the electrode surface in Figure 4b,c. At 120°C, the electrodes spontaneously assembled into a fibrous honeycomb morphology ( Figure  4d), showing that rapid, effective evaporation of the fugitive bisolvent, first the IPA then the DI water, was required. A similar self-assembly of the active component in electrochemical electrodes using a layer-by-layer bisolvent approach has been also achieved for coral-like TiO 2 (B) electrodes in a previous work. 36

Research Article
To investigate any faradaic charge storage benefits of the honeycomb structure in an LIB configuration, the electrochemical performance of spray-printed honeycomb CNF electrodes formulated with the optimum 50:50 DI/IPA ratio was compared with that of nonhoneycomb but otherwise identical CNF electrodes formed using a 70:30 DI/IPA ratio, which is shown in the inset SEM images in Figure 5a, emphasizing the different morphologies of the two electrodes, each 20 μm thick (see Table 1). Figure 5a also shows the deliverable discharge capacity profiles for the two electrodes at charge/discharge current densities of 40−8000 mA/g. In the 1st cycle, the very high discharge capacity of all electrodes was likely associated with side reactions, including SEI formation, which is typical of CNFs. 38,39 In subsequent cycles and at relatively low current density (40−2000 mA/g), the electrodes had almost identical discharge capacities, implying little benefit of the micron-scale pores since there was sufficient time for ion transport to all active sites in both electrodes. However, a difference in deliverable capacity developed on further increase in the current density (≥4000 mA/g), with the honeycomb electrode delivering ∼100 mAh/g at 8000 mA/g that was three times that of the identical electrode without the honeycomb structure, outperforming conventional graphite electrodes (anodes) at similar rates. 40−42 Thus, despite the initial high reactivity and relative low electrode density, the honeycomb structure delivered significant benefits in the high C-rate region.
The self-assembled honeycomb electrode also had a benefit in terms of volumetric capacity at increasing charge/discharge conditions, as shown in Figure 5b, wherein volumetric capacity was calculated by normalizing the gravimetric capacity by the electrode mass loading and thickness, as given in Table 1. At a relatively low rate of 40 mA/g, the honeycomb CNF electrode had a slightly lower volumetric capacity, implying that the through-thickness pore channels contributed only parasitic, wasted pore volume. However, at 4000 and 8000 mA/g, the honeycomb electrode volumetric performance was significantly better. Figure 5c presents the galvanostatic discharge behavior of both electrodes at a constant charge/discharge current density of 400 mA/g. As repeated charge/discharge cycles progressed to 300 cycles, discharge capacities of both sprayprinted CNF electrodes with and without the honeycomb increased progressively, as previously reported for CNF-based electrodes. 43−45 This behavior was ascribed to the progressive activation of ion storage sites embedded between the interconnected CNFs or the partially reversible formation and dissolution of the passivated SEI layers on the CNFs. After 300 cycles, the honeycomb electrode had a discharge capacity of ∼210 mAh/g compared with ∼196 mAh/g for the identical electrode without the honeycomb. Except for the first few charge/discharge cycles associated with side reactions including SEI formation, all spray-printed CNF electrodes had a coulombic efficiency of almost 100 % up to the 300 th cycle (inset of Figure 5c).
The effective Li-ion diffusion coefficient in each electrode was estimated from cyclic voltammetry (CV) measurements over scan rates in the range of 0.05−1.0 mV/s, which is a common approach to assess overall electrode kinetics at increasing rates. 36,46,47 In the CV profiles for the spray-printed CNF electrodes with and without the honeycomb pore (Figure

ACS Applied Materials & Interfaces
Research Article 6a,b), there were a pair of anodic/cathodic peaks at approximately 0.5 and 0.2 V, respectively, associated with the lithiation/delithiation reaction of graphite. 48,49 As the scan rate increased progressively up to 1.0 mV/s, the anodic/cathodic peak current density increased correspondingly but the relative peak intensity changes were different, implying a difference in the diffusion kinetics between the electrodes. Figure 6c,d shows plots of the anodic peak current density as a function of the square root of the scan rate for the CNF electrodes with and without the honeycomb structure, respectively. In both cases, the peak current density was a good fit to the square root of the scan rate and the Li-ion diffusion coefficient could then be estimated using the Randles−Sevcik equation 36 (Figure 6c,d), the honeycomb electrode was estimated to have a diffusion coefficient of ∼1.0 × 10 −9 that was almost 2 times that of the identical electrode without the honeycomb (∼5.8 × 10 −10 cm 2 /s) (see the comparative bar graph in Figure  6e). Given that the materials are identical in the two electrodes, it is a reasonable assumption that, to a first order approximation, any change in performance is primarily due to a change in the microstructure. This supports that the honeycomb structure led to improved electrode charge/ discharge dynamics through promoting ion mobility, as described in the schematic illustration in Figure 6f, helping to sustain rate performance at fast charge/discharge rates. To improve the reversible charge/discharge performance of the spray-printed honeycomb CNF electrode, especially at high-rate conditions where structural advantages of the honeycomb electrode were more significant, adding a minor fraction of Si, which is an intrinsically high-energy-density Liinsertion material (∼4200 mAh/g), 50,51 was examined. Figure  7a presents a schematic of the intended microstructure comprising bundles of CNFs decorated with Si particles in a honeycomb electrode. Figure 7b shows an SEM image of the corresponding spray-printed CNF-based electrode in which 200−400 nm Si particles were decorated over the CNFs, which maintained the fibrous honeycomb structure of the CNF-only

ACS Applied Materials & Interfaces
Research Article electrode already shown in Figure 2. The inset in Figure 7b shows at higher magnification that the Si particles were entangled and attached to the CNFs. The electrode cross section in Figure 7c again confirmed directional pore channels of approximately 20 μm depth.
As shown in the inset bar graph of Figure 8a, as the Si fraction increased from 0 to 10 wt %, the CNF fraction decreased correspondingly, from 80 to 70 wt %, so the total CNF + Si active fraction was fixed at 80 wt % (refer Table 2). Figure 8b shows the deliverable discharge capacity profiles as a function of Si fraction at charge/discharge current densities of 40−8000 mA/g. At relatively low charge/discharge current densities of 40 and 80 mA/g, as the Si fraction increased from 0 to 10 wt %, the electrode discharge capacities increased progressively, showing the effectiveness of even only minor fractions of high-capacity Si. However, with increasing charge/ discharge current density up to 8000 mA/g, the contribution of the Si became less marked and the lower Si fractions retained capacity better. For example, discharge capacities of a CNF + 10 wt % Si electrode, which had the highest discharge capacity at 40 and 80 mA/g, had the lowest discharge capacity at 8000 mA/g and even lower than that of a CNF-only electrode, suggesting that the Si had become fragmented and electrically isolated within the electrode due to its large volume expansion and pulverization at ultrafast lithiation/delithiation conditions, then contributing only to parasitic mass, as reported often in other similar Si-incorporating composite electrodes. 52,53 The CNF + 2.5 wt % Si electrode retained the highest discharge capacity in the range of 800−8000 mA/g, for example, ∼170 mAh/g at 4000 mA/g and ∼120 mAh/g at 8000 mA/g. The galvanostatic charge/discharge profiles in the inset of Figure 8b show rate performance behavior at 8000 mA/g: the CNF + 2.5 wt % Si electrode had the highest discharge capacity, whereas the CNF + 10 wt % Si electrode had the lowest. The performance of a spray-printed Si-only electrode was investigated at increasing current densities of 40−8000 mA/g, as shown in Table S1 and Figure S4

ACS Applied Materials & Interfaces
Research Article electrode had discharge capacities higher than 1000 mAh/g but capacity catastrophically collapsed as C-rates increased progressively up to 8000 mA/g. Eventually, the capacity reduced almost to zero at 8000 mA/g. By contrast, at 8000 mA/g, the CNF + 2.5 wt % Si electrode had a higher discharge capacity than that of the CNF-only electrode, suggesting that mechanical compliance in the CNF structure and the multiscale pores were effective in mitigating volume expansion effects and Si pulverization even at fast charge/discharge rates. In the galvanostatic discharge plots at 400 mA/g in Figure  8c, a fibrous honeycomb CNF + 2.5 wt % Si electrode sustained the highest discharge capacity of ∼290 mAh/g after 300 cycles, showing an approximately 40 % improvement in deliverable discharge capacity over a CNF-only honeycomb electrode without Si. CNF electrodes with 5 and 10 wt % Si had lower discharge capacities of ∼240 and ∼200 mAh/g, respectively. The Si particles in the honeycomb CNF + 2.5 wt % Si electrode were estimated to have ∼340 % volume expansion after 300 cycles, as shown in Figure S5 in the Supporting Information. In the inset of Figure 8c, the CNF + 10 wt % Si electrode showed the lowest coulombic efficiency of 45 % in the 1 st charge/discharge cycle, which, as previously described, was likely associated with a loss of irreversible capacities from SEI formation on Si and CNF surfaces; however, excluding the first few cycles, all electrodes recovered and sustained a coulombic efficiency of almost 100 % up to the 300th cycle.
The honeycomb negative electrode with 2.5 wt % Si was further investigated in full-cell configurations with a nonhoneycomb spray-printed LFP positive electrode, to investigate

ACS Applied Materials & Interfaces
Research Article any electrochemical benefits of the honeycomb structure in the CNF (refer to the half-cell performance of the spray-printed LFP electrode in Figure S6 and Table S2 in the Supporting Information). The CNF-LFP LIB configuration was constructed with a balanced 1:1 anode/cathode capacity ratio (see Table 3) that delivers the best performance in energy density and power density. 54 A Si-containing honeycomb negative electrode was also compared with the equivalent electrode without Si in full LIB arrangements (see Figure S8 in the Supporting Information). Figure 9a shows the galvanostatic charge/discharge profiles of the CNF + Si/LFP and CNF-only/LFP full cells at charge/ discharge current densities of 40−4000 mA/g. The CNF + 2.5 wt % Si electrode had much higher charge/discharge capacities than those of the CNF-only electrode at all current densities, with discharge capacities of ∼320 mAh/g at 40 mA/g, ∼220 mAh/g at 400 mA/g, and ∼150 mAh/g at 4000 mA/g (also see comparative performance of spray-printed CNF + 2.5 wt % Si electrodes with and without the honeycomb in Figure S8 in the Supporting Information). Figure 9b presents the Ragone plots of the spray-printed CNF + Si/LFP and CNF-only/LFP full cells at charge/discharge current densities of 40−4000 mA/g. CNF + Si/LFP had an areal energy density of ∼1.3 mWh/cm 2 at 40 mA/g and an areal power density of ∼7.1 mW/cm 2 at 4000 mA/g, whereas CNF-only/LFP had an energy density of ∼0.9 mWh/cm 2 and a power density of ∼4.6 mW/cm 2 for identical conditions (also see the massnormalized Ragone plots in Figure S7 of the Supporting Information). Note that the positive electrode was not optimized to have the same honeycomb pore structure or CNF network as the negative CNF-based electrode, and thus further performance gains at high charge/discharge conditions might be realizable using both honeycomb positive and negative electrodes.
The inset in Figure 9b shows a photograph of an A5-scale (approximately 20 cm × 15 cm) spray-printed honeycomb CNF + 2.5 wt % Si electrode (larger electrode areas are readily available using a larger hot plate) and a magnified SEM image of a portion of the spray-printed A5 electrode, confirming the self-assembled honeycomb pore pattern over large areas.
It is worth noting that the benefits of the honeycomb pore structure were realized more significantly and more effectively in the full-cell arrangement, where Li-ion supply is finite, rather than in the half-cell arrangement with Li foil where Li-ion availability is essentially unrestricted. Figure 9c,d shows the rate capacity profiles of half-cell and full-cell configurations at different current densities. Here, the rate capacity retention is defined as the ratio of the discharge capacity obtained at 40,80,200,400,800,2000, and 4000 mA/g to the discharge capacity at 40 mA/g. Both CNF-only and CNF + Si in full cells with LFP outperformed identical half-cells with Li foil at all current densities (Figure 9c,d), with 30 and 45 % improvements in the capacity retention at 4000 mA/g, respectively. Slightly lower capacity but better capacity retention arose because there was, in general, a lower Li-ion concentration in the full cell and therefore less intercalation and less electrode damage and any volumetric changes were better tolerated, especially in the more mechanically compliant, higher porosity honeycomb electrode. Figure 9e shows the galvanostatic discharge capacity profiles of CNF + Si/LFP and CNF-only/ LFP full cells at 200 mA/g. The CNF + Si and CNF-only electrodes sustained a discharge capacity of ∼240 and ∼130 mAh/g after 500 cycles, respectively. Both the electrodes had a coulombic efficiency of almost 100 % after 500 cycles excluding the few initial cycles (the inset in Figure 9e).

■ CONCLUSIONS
Fibrous honeycomb electrodes of CNFs with micron-scale through-thickness pores were manufactured directly on current collectors using a layer-by-layer spray-printing process based on bisolvent fugitive carriers containing DI water and IPA. The conditions for the formation of the honeycomb structure were identified to depend on the volume ratio of the DI/IPA bisolvent carrier and the current collector temperature. The self-assembly of honeycomb pores occurred only at a current collector temperature of 120°C, high enough to evaporate the liquid bisolvent carrier that was formulated in an optimum volume ratio of 50:50 DI/IPA. At low current densities, CNFbased electrode gravimetric and volumetric capacities were low. However, through-thickness pore channels promoted efficient Li-ion dynamics throughout the electrode as current density increased, leading to marked overall improvements in capability and cycling performance.
Incorporating an optimum fraction of 2.5 wt % Si onto CNFs boosted the honeycomb CNF electrode performance in the fast charge/discharge 800−8000 mA/g region further, which was realized most markedly in a full-cell configuration where the CNF-based electrode served as the negative electrode and a spray-printed LFP electrode was used as the positive electrode. The material combination used here is a model system to demonstrate the generic possibilities of how large area, directional, and beneficial micron-scale honeycomb pore structures can be spray-printed directly on current collectors to better realize the intrinsic properties of energy storage materials at high charge/discharge rates.

■ EXPERIMENTAL SECTION
Materials. CNFs were obtained from Sigma-Aldrich, U. K.; Si particles were purchased from Alfa Aesar, U. K.; and LFP particles were obtained from Hydro-Quebec, Canada.
Spray-Printing Process. Prior to spray-printing, active materials (CNF or CNF + Si), SP conductivity enhancers, and CMC binders in a controlled mass ratio of 80:10:10 (these relatively high concentrations of inactive components are typical of CNF-based electrodes where an open, porous structure is preferred, although inactive fractions have been successfully minimized to 5% or less in other spray-printed electrodes 54 ) were suspended into a mixture of DI water and IPA using ultrasonication for 30 min. A current collector was fixed on a vacuum chuck of a hot plate that was maintained at temperatures in the range of 60−120°C. The CNF-based suspensions were then atomized through a spray nozzle at an air pressure of 0.4 bar (see Figure 1). The entire spray process was carried out in a wellventilated fume cupboard.
Electrochemical Testing. The electrochemical performance of the spray-printed electrodes was examined using coin cells (CR2032) assembled from thoroughly dried electrodes in an Ar-filled glovebox (H 2 O < 0.1 ppm, O 2 < 0.1 ppm). For half-cells, the working electrode was assembled with pure Li chips (99.9 % trace metals basis, MTI Corporation) as the counter/reference electrode. In the manufacture of full LIBs, spray-printed CNF-based anodes (e.g., CNF-only or CNF + 2.5 wt % Si) were coupled with spray-printed LFP cathodes that were formulated with a 80:10:10 mass ratio of LFP/SP/CMC for an optimum capacity balance between anodes and cathodes. The anodes and cathodes were electrically separated by a polyethylene separator (Cellgard 2400) soaked into 1 M LiPF 6 electrolyte solution in a 1:1 mixture (by volume) of ethylene carbonate and dimethyl carbonate (Merck). Galvanostatic charge/discharge tests for both half-cells and full LIBs were carried out at room temperature using an Arbin battery cycler (Arbin BT2000). Although charge/discharge capacity values of

ACS Applied Materials & Interfaces
Research Article half-cells were calculated using the mass of active materials (e.g., CNF, CNF + Si, or LFP), full LIB capacities were estimated on the basis of the weight of the active materials in the anode (e.g., CNF or CNF + Si). The theoretical capacity of CNF and LFP was assumed to be ∼370 mAh/g 48,49 and ∼170 mAh/g, 55