One-pot synthesis of high-capacity silicon-lithium anodes via on-copper growth of a semi-conducting, porous polymer

Silicon-based anodes with lithium ions as charge carriers have the highest predicted charge density of 3579 mA h g -1 (for Li 15 Si 4 ) while being comparatively safe. Contemporary electrodes do not achieve these theoretical values largely because production paradigms remained unchanged since their inception and rely on the mixing of weakly coordinated, multiple components. In this paper, we present the one-pot synthesis of high-performance anodes that reach the theoretical capacity of the fully lithiated state of silicon. Here, a semi-conductive triazine-based graphdiyne polymer network is grown around silicon nanoparticles directly on the current collector, a copper sheet. The current collector (Cu) acts as the catalyst for the formation of a semi-conductive triazine-based graphdiyne polymer network that grows around the inorganic, active material (Si). In comparison to established electrode assemblies, this process (i) omits any steps related to curing, drying, and annealing, (ii) does away with binders and conductivity-enhancing additives that decrease volumetric and gravimetric capacity, and (iii) cancels out the detrimental effects on performance, chemical and physical stability of conventional, three-component anodes (Si, binder, carbon black). This is because, the porous, semi-conducting organic framework (i) adheres to the current collector on which it grows via cooperative van der Waals interactions, (ii) acts effectively as conductor for electrical charges and binder of silicon nanoparticles via conjugated, covalent bonds, and (iii) enables selective transport of mass and charge-carriers (electrolyte and Li-ions) through pores of dened size. As a result, the anode shows extraordinarily high capacity at the theoretical limit of fully lithiated silicon, excellent performances in terms of cycling (exceeding 70% capacity retention after 100 cycles), and high mechanical and thermal stability. These high-performance anodes pave the way for use in exible, wearable electronics and in environmentally demanding applications. as a strong, exible binder that envelops Si NPs with a matrix of covalent bonds that can sustain the dramatic volume changes of silicon in repeated de-/lithiation cycles and prevents detrimental abrasion and reformation of the solid electrolyte interface, (ii) as a facilitator of charge transport along its π-conjugated polymer backbone, and (iii) as a medium for mass transport of lithium ions and electrolyte through its microporous channels. The resulting anodes achieve stable electrochemical cycling performance and an extraordinarily high capacity close to the theoretical limit of electrochemical storage using silicon. The reported process uses raw materials and methods common in industrial electrode manufacture and can be transferred and scaled up with ease. Half-cell electrode assemblies in the off-state retain key performance parameters even after thermal stress.


Introduction
Cheap, high-performance, and safe energy storage solutions are needed to address the increasing demand for portable electronics and the transition to electric mobility. Lithium-ion batteries have replaced conventional secondary battery technology (like nickel-cadmium and nickel-metal hydride batteries) due to their high energy densities, high operating voltages, and stable capacity retention during open-circuit rest. 1 Lithium metal anodes were quickly replaced by lithium-intercalated graphitic materials in order to avoid the formation of dendrites that resulted in short circuiting of the two electrodes. However, while lithium-graphite intercalates are safer and more stable they have only a tenth of the energy density of metallic lithium. 2 Silicon is a good active material for Li-ion anodes because it has a superior theoretical capacity of 3579 mA h g -1 (or 8340 mA h cm -3 ) for Li 15 Si 4 , and -unlike some transition metals -it is not toxic and abundant. Moreover, its alloying reaction with lithium triggered at 0.3 V vs. Li/Li + , prevents the formation of metallic lithium around the anode during discharging -a detrimental process observed for Li-graphite batteries known as "lithium plating" -and allows the use of Si-Li electrodes under harsher conditions. 3 The large number of lithium atoms that silicon can store, however, induces large volume changes during charge/discharge cycles (> 300%). 3 The mechanical stress induced by these drastic volume changes leads to the pulverization of the silicon active material, loss of contact of the electrode lm with the current collector, and loss of overall mechanical integrity of the whole electrode. Repeated cycles of expansion and contraction of silicon expose pristine silicon surfaces and induce the reformation of solid electrolyte interfaces (SEI). This process contributes to the gradual consumption of lithium and electrolyte, and it limits the diffusion of charge carriers through the expanding SEI. [3][4][5][6][7] It is di cult to counter these detrimental mechanical and chemical changes to silicon-based electrodes because conventional methods of electrode assembly rely on the mixing of multiple components that are held together by weak, dispersive forces.
In laboratory settings, some strategies were developed to address the inherent aws of these multicomponent assemblies. For example, shaping silicon into hierarchical, nano-sized, or porous structures buffers some of its dramatic volume expansion during lithiation. 7 On the downside, nanostructured silicon is susceptible to restructuring during battery operation, and its larger speci c surface promotes reactions with the electrolyte to form more of electrochemically inactive SEI. In other approaches silicon particles are encapsulated in a carboneous matrix, 8 or they are coated with metal oxides. 9 However, encapsulation of silicon requires supplementary components that do not meaningfully contribute to the volumetric capacity of the electrode and might necessitate the addition of agents that enhance electric conductivity. Such modi cations of the active material prior to electrode assembly have proven too time consuming, low yielding and expensive and, hence, none of these methods have found their way into commercial processes to date.
In this work, we present a departure from the current "blend-and-bake" paradigm of electrode manufacture. In a one-pot process, we embed silicon nanoparticles (Si NPs) in a covalently-linked, porous, semi-conducting polymer matrix whose growth is initiated and templated by the current collector (Cu) itself. The covalent bonds of the organic matrix contribute to a superior mechanical and chemical resistance of our electrode lms. The overall π-conjugated backbone of the triazine-based graphdiyne (TzG) polymer enables the transport of electrons from the active material to the current collector. Since the polymerization is promoted by the reactive metal surface of the current collector, the resulting polymer/silicon composite (TzG/Si) adheres strongly to it. 10 In summary, the covalent polymer matrix acts at the same time as (i) a strong binder, as (ii) an electrical conductor, and as (iii) a semi-permeable membrane that enables transport of ions and electrolyte but prevents the migration of homogeneously dispersed silicon nanoparticles even under harsh conditions. This facile method yields silicon-based anodes (TzG/Si@Cu) of superior performance that suffer little mechanical and electrochemical deterioration from the inherent volume expansion of silicon during lithiation-delithiation cycles and that drastically limit the detrimental loss of lithium and electrolyte at the solid electrolyte interface.
For comparison, we have prepared three different types of electrode systems via the same one-pot method described above but with varying compositions to carefully test the effects of individual components: (i) growing a lm of TzG on Cu, we obtain TzG@Cu, (ii) growing TzG in the presence of Si NPs we get TzG/Si@Cu, and (iii) TzG/Si/CB@Cu is produced by growing TzG around Si NPs and a conventional, conductive additive, carbon black (CB) (Supplementary Information Section S1.6).
Scanning electron microscopy (SEM) images reveal the morphology of pristine TzG/Si@Cu electrodes. The material grown on the copper support adopts a porous, sponge-like, and homogeneous morphology as seen top-down (Fig. 1e) and from cross-sections of the electrode lm (Fig. 1f). Cross-sectional SEM imaging at lower magni cations shows lms of TzG/Si with a thickness of ~ 25 µm that adhere well to the Cu substrate with no apparent gaps ( Figure S4). More detailed transmission electron microscopy (TEM) energy-ltered mapping on TzG/Si lms shows a homogenous distribution of carbon and silicon on the nano-scale (Fig. 1g). On the nanoscale, the electrode lm consists of Si NPs homogenously embedded in an organic polymer matrix of TzG. Residual Cu nanoparticles can be seen withing the polymer matrix that stem from the TzG polymerization process ( Fig. 1h; a comparison of TEM images of TzG@Cu and of pristine Si NPs can be found in Figure S5). 10 Overall, individual Si NPs are enclosed by the conjugated, polymer and held cooperatively as a lm on the current collector.
We have shown previously that neat, unmodi ed triazine-based graphdiyene polymers are narrow-band gap semiconductors (E g,elec = 1.84 eV and conductivity of 1.2 µS cm -1 ) with moderate porosities (N 2 BET surface area of 124 m 2 g -1 at 77 K). 10,23 Hence, the composite TzG/Si on Cu foil (TzG/Si@Cu) has a promising combination of chemical, electrical, and structural features for electrochemical storage.
In the following (Supplementary Information Section S2.1), we discuss the electric and electrochemical performance of TzG-based electrodes and the effects of the TzG polymer on the formation of the SEI. For the three electrode systems TzG/Si@Cu, TzG/Si/CB@Cu, and TzG@Cu (i) we compared the bulk conductivities of the unlithiated, "as-synthesized" electrodes ( Figure S6), (ii) we recorded cyclic voltammetry (CV) curves ( Figure S7), and (iii) we performed ex-situ XPS measurements probing the electrode surfaces to a depth of approx. 10 nm after a number of de-/lithiation cycles ( Fig. 2a; Figure S8, S9, S10 and S13; Table S1; details of the XPS spectra tting method are described in Supplementary Information section S1.10). During the rst lithiation of the TzG/Si@Cu electrode, the resulting composition of the lm present a content with up to 43.9 mol% of cumulative Li, F. and P elements ( Table 1). Furthermore, after the rst cycle of lithiation and delithiation, we observe that the initial Si 2p signal of the pristine electrode decreases dramatically in intensity from 40.8 mol% to value close to zero ( Table 1). The tting of C 1s, F 1s and Li 1s spectra ( Figure S10) indicate the formation of the SEI during the rst lithiation, resulting in the variations of surface elemental concentration mentioned above. Note though that even after the 100th delithiation event, we detect a signal of Si(0) in the Si 2p region (Fig. 2a, iv, Table 1). This nding suggests that SEI formation is completed at an early stage, and that SEI thickness does not increase dramatically throughout cycling. 24,25 XPS data from the Si 2p region of TzG/Si@Cu electrodes cycled at C/8 (Fig. 2a) shows the emergence of new Si 2p peaks corresponding to Li x Si, SiO x F y , and Li x SiO y after the rst lithiation event. [19][20][21][22] The presence of SiO x F y results from the nonfaradic reaction between the remaining LiPF 6 and the SiO x on the surface of the Si NPs, during the opening the cell for ex-situ measurement. 19 The environments of Li x S and Li x SiO y originate from the electrochemical alloying reaction of silicon and SiO x with the lithium ions in the electrolyte, as further corroborated by emerging signals in the F 1s and Li 1s regions ( Figure S10b and S10c). 20,22 Pristine and delithiated samples of TzG/Si@Cu all show peaks corresponding to Si(0) and SiO x indicating a fully reversible de-/lithiation process.
Ex-situ TEM and SEM imaging of TzG/Si@Cu electrodes at different cycling stages (Fig. 2b and c) reveal that prior to cycling, the TzG polymer acts as a binder that joins Si NPs into a porous TzG/Si composite (i). After the rst lithiation event, individual particles of increased size can be discerned that correspond to lithiated and volumetrically expanded domains of Li x Si all embedded in a fabric of polymer and SEI (ii).
After the rst delithiation, the size of the spherical Si domains decreases (iii), and after the 100th delithiation, all that can be said is that the TzG/Si composite shows no discernible large cracks or deformations (iv) (Figure S11 and S12).
A comprehensive discussion of these ndings is presented in Supplementary Information section S2.2. For now, we conclude that: (i) the alloying reaction between lithium and Si NPs takes place unhindered, hence, the TzG polymer matrix allows lithium to diffuse, (ii) the conductivity enhancing additive in TzG/Si/CB@Cu has no positive effect on the performance of TzG-based electrodes, hence, the initial, modest conductivity of undoped TzG is no impediment for its use in electrodes; and (iii) the TzG polymer does not participate in the detrimental depletion of lithium or electrolyte on its own, and SEI formation occurs exclusively in the presence of Si NPs. It is particularly surprising that TzG-based electrodes perform as well as they do, given their low conductivity in the pristine, unlithiated state. While an in-depth study of the electric conductivity of lithiated TzG/Si@Cu is out of the scope of this work, XPS results for TzG@Cu suggest that the rst lithiation cycle leads to n-doping of the TzG polymer by N 1s signal shifting to higher binding energy ( Figure S8), [26][27][28] analogous to lithiated graphite, 29,30 and, hence, to an increase of its conductivity above and beyond the effect achieved by the conducting additive in TzG/Si/CB@Cu. We subjected the TzG/Si@Cu electrode prepared following method 1 to de-/lithiation cycling at C/8 in a half-cell set-up recording speci c capacity (Fig. 3). In the rst lithiation cycle, we observe the formation of the SEI at a comparatively low CE value of ~ 70%, meaning that the surface of Si NPs is readily accessible for lithium and for the electrolyte via the pore channels of the TzG polymer. Similar Coulombic e ciencies have been reported for nanosized silicon particles with a large external surface area. Microsized silicon particles with a smaller ratio of exposed surfaces show higher initial CE values but suffer from limited discharge capacities and poor cycling performance. 31 In the third cycle, the CE rises above 97% and steadily improves up to 99.5% which means that SEI formation is completed early on. The speci c capacity observed during the rst lithiation exceeds the theoretical value for fully lithiated silicon (3579 mA h g -1 ) due to the formation of the SEI. [32][33][34] Starting with the second cycle, no excess capacity is recorded and, hence, the newly formed SEI does not contribute to the measured discharge capacity. The recorded speci c lithiation capacity in the second cycle is ~ 3500 mA h g -1 , and it remains at a record high in subsequent cycles (e.g. ~3450 mA h g -1 in the fth cycle). In comparison, the TzG/Si/CB@Cu electrode prepared together with carbon black does not perform better than the additivefree electrode ( Figure S14). Taking the speci c lithiation capacity in the second cycle as a baseline, TzG/Si@Cu retains 81.0% of its speci c capacity after 50 cycles (2862 mA h g -1 ) and 69.1% after 100 cycles (2443 mA h g -1 ) Conversions of these values to areal and volumetric capacities can be found in Figure S15. The remarkable performance parameters of TzG/Si@Cu anodes stem in parts from the increased Si mass loading enabled by our method of preparation (Supplementary Information Table S2 and S3). For comparison, we prepared two electrodes using conventional binders and CB additive, PVdF/Si/CB@Cu and PAA/Si/CB@Cu ( Figure S16). 21,35 These conventional formulations allow a total Si mass loading of up to 60% without compromising the performance of the electrode, compared to 75% of Si in the TzG/Si@Cu system. Overall, the speci c capacity and retention of TzG/Si@Cu anodes exceeds that of the best-performing multi-component systems with and without conductivity-enhancing additives that are reported to date (Table S4). [36][37][38][39][40][41][42][43][44][45] Simpler methods have been implemented to prepare the TzG/Si@Cu electrodes, presenting improved performances (Supplementary information, Section S2.4, Figure S17-S23 and Table S5). It is also noteworthy to mention that our TzG/Si@Cu anodes can withstand high temperature without showing any diminution in their performances, as shown in Figure  S24. Table 1 Relative molar concentrations of elements as determined by peak integration using X-ray photoelectron spectroscopy (XPS) data recorded at 1486.6 eV for TzG/Si@Cu electrodes at various stages of de-/lithiation cycled at a constant current of 0.32 mA cm -2 (i.e. C/8 for the capacity of Si, within 0.01-1.2 V vs. Li /Li +

Conclusions
We present here a one-pot synthetic protocol that yields high-performance silicon-lithium anodes within one hour of reaction time. These anodes consist of silicon nanoparticles that are fully encapsulated by a semi-conducting, porous triazine-based graphdiyne (TzG) polymer that grows directly on the Cu current collector. Cu foil plays three roles in this paradigm-changing method of anode fabrication: it acts (i) as a source of Cu species for a Glaser-type oxidative coupling polymerization, (ii) as templating substrate for the polymer lm, and (iii) as the current collector of the electrode. The porous, semi-conducting TzG polymer acts (i) as a strong, exible binder that envelops Si NPs with a matrix of covalent bonds that can sustain the dramatic volume changes of silicon in repeated de-/lithiation cycles and prevents detrimental abrasion and reformation of the solid electrolyte interface, (ii) as a facilitator of charge transport along its π-conjugated polymer backbone, and (iii) as a medium for mass transport of lithium ions and electrolyte through its microporous channels. The resulting anodes achieve stable electrochemical cycling performance and an extraordinarily high capacity close to the theoretical limit of electrochemical storage using silicon. The reported process uses raw materials and methods common in industrial electrode manufacture and can be transferred and scaled up with ease. Half-cell electrode assemblies in the offstate retain key performance parameters even after thermal stress.  Characterization of TzG/Si@Cu electrodes prepared following method 1 using a, X-ray photoelectron spectroscopy (XPS) data from the Si 2p region, b, transmission electron microscopy (TEM), and c, scanning electron microscopy (SEM). Data is presented for TzG/Si@Cu electrodes in the pristine state (i), after the rst lithiation (ii), after the rst delithiation (iii) and after the 100th delithiation (iv). Cycling was performed at a constant current of 0.32 mA cm-2 (i.e. C/8 for the capacity of Si, within 0.01-1.2 V vs.