Contents lists available at ScienceDirect

Solid State Ionics

journal homepage: www.elsevier.com/locate/ssi

Simple and inexpensive coal-tar-pitch derived Si-C anode composite for all-solid-state Li-ion batteries

Nathan Arthur Dunlapa, Seulcham Kima,b, Je Jun Jeongb, Kyu Hwan Ohb, Se-Hee Leea,⁎

a Department of Mechanical Engineering, University of Colorado at Boulder, Boulder, CO 80300-0427, USAbDepartment of Materials Science and Engineering, Seoul National University, Seoul 151-744, Republic of Korea

A B S T R A C T

A simple, inexpensive and scalable method to prepare silicon‑carbon composite particles for use in all-solid-state Li-ion battery anodes is presented. The composite'selectrochemically active soft-carbon matrix is formed through the pyrolysis of coal-tar-pitch, an abundant industrial waste material. Various techniques are used tocharacterize the physical and electrochemical properties of this pitch derived carbon. Optimization of the Si-C composite anode resulted in an all-solid-state Li-ionhalf-cell displaying stable specific capacities of 653.5mAh/g (per mass electrode) and 1089.2 mAh/g (per mass Si-C composite) after 100 discharge-charge cycles.Cross sectional images of cycled electrodes, prepared via FIB milling, were taken in order to investigate the effect of silicon particle expansion on their compositemicrostructures.

1. Introduction

Due to their large volumetric and gravimetric capacities, recharge-able lithium ion batteries (LIBs) are widely regarded as state of the artenergy storage devices. Recently, there has been great interest in thedevelopment of all-solid-state LIBs, which utilize ionically conductiveceramic electrolytes in place of volatile organic liquids. These cells offerincreased safety while enabling high temperature operation without thegrowth of unstable solid-electrolyte-interphase layers [1]. In order toadvance the all-solid-state cell into the competitive battery market, newhigh capacity electrode architectures must be explored. To be deemedreasonable, these new electrode designs must address the significantinterfacial resistances inherent in solid-state cells while utilizing in-expensive and industrially scalable processing techniques and materials[2].

Silicon is an attractive anode material due to its exceptionally largespecific capacity (3579mAh/g, at room temperature), low lithiationpotential and natural abundance. Despite years of intense research in-terest, the practical realization of a reversible silicon anode has beenelusive owing to its extreme volumetric expansion upon lithiation[3–5].

Molina Piper et al. showed that with increased confinement pres-sure, silicon alloys with fewer moles of lithium. This limits the siliconparticles' expansion and improves their cycling stability [6]. We haveexpanded upon this concept by encapsulating silicon particles in mixed-conducting, electrochemically active matrix materials. For example,Yersak et al. showed that confining silicon particles in a

silicon‑titanium‑nickel (STN) matrix greatly improved their cyclingstability in an all-solid-state cell [5,7]. Subsequently, Whitely et al.reported that a simple mixing and cold-pressing process could producea silicon‑tin composite with some of the best cycling characteristics ofany bulk all-solid-state anode to date [3]. These reports demonstratedthat much like the external application of compressive stress, en-capsulation in a robust matrix material suppresses the volumetric ex-pansion and pulverization of silicon particles upon lithiation. Further-more, they concluded that conformal coating by a mixed conductingmatrix prevents active material isolation and reduces resistance in anall-solid-state electrode where conduction pathways are typically lim-ited to small, easily disrupted particle-particle contact points.

In order to ensure commercial viability, inexpensive and sustainablematrix material alternatives should be explored. One such alternative iscoal-tar-pitch (CTP), an abundant byproduct of the coal and steel in-dustries. This black, sooty material is composed of cyclical hydro-carbons with widely ranging molecular weights and configurations [8].In this study, we show that the pyrolysis of CTP produces a mixedconducting amorphous carbon with impressive electrochemical prop-erties. Through a simple and industrially scalable solution coatingprocess, we are able to inexpensively produce Si-C composite particlescapable of being cycled in an all-solid-state cell. While the use of similarSi-C composites has been reported, we are, to our knowledge, the firstto demonstrate this anode material's potential in an all-solid-state cell[9].

https://doi.org/10.1016/j.ssi.2018.07.013Received 17 May 2018; Received in revised form 14 June 2018; Accepted 14 July 2018

⁎ Corresponding author.E-mail address: sehee.lee@colorado.edu (S.-H. Lee).

Solid State Ionics 324 (2018) 207–217

0167-2738/ © 2018 Published by Elsevier B.V.

T

2. Experimental

The composite anode materials presented in this report were pre-pared by first adding 0.5 g coal-tar-pitch (Mitsubishi Chemical) and0.5 g silicon powders (325mesh [≤44 μm], Aldrich, 99%; 1–3 μm, USResearch Nanomaterials, 99.9% or 50 nm, Alfa Aesar, 98%) to a screw-top glass vial containing 2.5 gN,N-Dimethylformamide (DMF) solvent(Fischer). The contents of this vial were mixed overnight with a mag-netic stir bar in order to fully dissolve the pitch powders and evenlydisperse the silicon particles. After 2 h in a sonication bath, this slurrywas evenly spread over copper foil and dried in an oven at 60 °C. Withthe DMF fully evaporated, the silicon-pitch composite could be scrapedaway from the copper foil and heat-treated in a tube furnace for 5 h at900 °C under continuous argon gas flow. The same heat treatmentprocedure was applied to a sample of pure pitch powder so that itsstructure and electrochemical activity could be characterized. A con-sistent pitch mass loss of ~35% was observed after pyrolysis, meaningthat the Si-C composites were roughly 60% silicon and 40% carbonmatrix, by mass.

Images of the Si-C composite particles were captured with a fieldemission scanning electron microscope (FESEM, JEOL JSM-7401F)equipped with energy dispersive X-ray spectroscopy (EDS) capabilities.The atomic structure of these materials was characterized with a BrukerAXS D2 Phaser bench-top XRD system using Cu Kα radiation(λ=1.5418).

The following materials preparation and testing procedures wereconducted in the dry atmosphere of an argon filled glove box. All-solid-state cells were constructed and cycled within 13mm diameter poly-etheretherketone (PEEK) lined titanium dies, with two titanium plun-gers acting as current collectors. Titanium was used as the plunger/current collector material because of its mechanical strength and widewindow of electrochemical stability. To construct an all-solid-state half-cell, 150mg of amorphous solid-state electrolyte (SSE) powder was firstpressed into a relatively dense, ionically conductive separator pellet.The 77.5Li2S-22.5P2S5 SSE used in this study was prepared by me-chanochemically reacting stoichiometric amounts of Li2S (Aldrich,99.9%) and P2S5 (Aldrich, 99%) with a planetary ball mill (MTICorporation SFM-1). Two-gram batches of these powders were milledfor 20 h at 400 rpm in airtight 500mL stainless steel jars. These jarscontained three 16mm diameter and twenty 10mm diameter stainlesssteel balls, corresponding to a ball:SSE mass ratio of 60:1. Compositeworking electrodes were prepared by hand mixing the Si-C active ma-terials with SSE using a mortar and pestle. After mixing, 2 mg of thiselectrode composite was evenly spread on one side of the solid elec-trolyte pellet. On the opposing side, 100mg of lithium‑indium (LiIn)alloy was added as a counter electrode. The LiIn alloy, which has apotential of 0.62 V versus Li+/Li, was prepared by vortex mixing(Vortex Genie 2) appropriate amounts of Indium (Alfa Aesar, Puratronic99.999%) and Lithium (FMC Lithium Corp., Lectro Max Powder 100)powders in a glass vial [10]. After depositing both electrodes, a hy-draulic hand press was used to compress the all-solid-state half-cell with5metric tons of force. This final cold pressing densifies the all-solid-state half-cell pellet, improving interfacial contact within and betweenits three layers.

The electrochemical performance of the cells was tested with anArbin BT2000 battery test station using constant current constant vol-tage (CCCV) conditions. All cells were cycled in a voltage window of5mV–1.5 V versus Li+/Li, with one hour voltage holds at both theupper and lower voltage limits. This testing regime was designed tostudy the pitch matrix's ability to confine and withstand the full li-thiation of the various silicon active materials. During testing, all cellswere held at 60 °C under 20MPa compressive clamping force in a dryargon atmosphere. This small external compressive stress was appliedto maintain mechanical contact between the cells' electrodes and thetitanium current collectors. By calculating the approximate siliconcontent in each composite electrode, cycling currents were selected to

achieve theoretical rates of C/20 (1st cycle), C/10 (cycles 2–49) and C/5 (cycles 50–100) assuming 100% silicon utilization throughout testing.Table 1 lists the applied current densities for each cell reported in thisstudy.

A dual-beam focused ion beam (FIB, FEI, Nova Nanolab 200)equipped with energy dispersive X-ray spectroscopy (EDS) was used tocross section and image pristine Si-C composite particles as well ascomposite electrodes after 100 discharge-charge cycles.

3. Results and discussion

A Si-C composite particle prepared with 325meshed (≤44 μm) si-licon particles is displayed in Fig. 1a. Although light colored siliconparticles can be seen protruding through the dark carbon matrix, theirdistribution throughout the composite is unclear. To better understandthe internal microstructure of this composite particle, it was crosssectioned and imaged with a FIB-SEM.

Fig. 1b shows that despite the irregularity in their size and shape,the large silicon particles are evenly distributed and conformally en-capsulated within the carbon matrix. The EDS data in Fig. 1c and dconfirm the composition of the light and dark phases identified by pointscans one and two in Fig. 1b. The small un-labeled peaks in Fig. 1c andd indicate the presence of trace amounts of oxygen and gallium in thesample. These contaminants were likely introduced during the trans-port and FIB milling of the particle. Notice that a high degree of in-terfacial contact has been achieved between the carbon matrix and si-licon particles. This conformal coating of the silicon active materialswill help to prevent their electrochemical isolation with cycling. It willalso facilitate the diffusion of Li-ions and electrons in and out of thesilicon particles, as conduction in all-solid-state electrodes is typicallylimited to small particle-particle contact points rather than conformalinterfaces. Lastly, this large degree of encapsulation may help to confinethe volumetric expansion of the silicon particles upon lithiation, mini-mizing the stresses they experience with cycling.

The variation in size and morphology of the Si-C composite particlesafter heat treatment and hand grinding is shown in Fig. 2a, c and e. Thecomposite particles range in size from tens to hundreds of microns indiameter. These low magnification SEM images show that the heat-treated composites are tough enough to withstand the stresses of handgrinding without crumbling or phase separation. This suggests that thebenefits of the conformal Si-C composite structure will be maintained inan all-solid-state electrode.

SEM images of cross-sectioned Si-C composite particles are dis-played in Fig. 2b, d and f. The large 325-meshed (≤44 μm) siliconparticles in Fig. 2b are evenly dispersed and conformally encapsulatedin a dark, dense carbon matrix. The smaller 1–3 μm present in Fig. 2dalso clearly show a high degree of interfacial contact with its sur-rounding carbon matrix, although some silicon particle agglomerationcan be observed in this particular cross section. The composite preparedwith 50 nm silicon particles has a slightly different morphology. At thismagnification, this Si-C composite clearly shows porosity not present in

Table 1Cycling current densities of all cells reported in this study.

Cycle 1 (C/20) Cycles 2–49 (C/10)

Cycles 50–100 (C/5)

100% heat-treatedpitch

0.09mA/cm2 0.19mA/cm2 0.38mA/cm2

6:4 Si-C:SSE, 50 nm Si 0.09mA/cm2 0.19mA/cm2 0.38mA/cm2

7:3 Si-C:SSE, 50 nm Si 0.11mA/cm2 0.22mA/cm2 0.44mA/cm2

8:2 Si-C:SSE, 50 nm Si 0.13mA/cm2 0.26mA/cm2 0.52mA/cm2

9:1 Si-C:SSE, 50 nm Si 0.14mA/cm2 0.28mA/cm2 0.56mA/cm2

100% Si-C, 50 nm Si 0.16mA/cm2 0.32mA/cm2 0.64mA/cm2

7:3 Si-C:SSE, 1–3 μm Si 0.11mA/cm2 0.22mA/cm2 0.44mA/cm2

7:3 Si-C:SSE, 325meshSi

0.11mA/cm2 0.22mA/cm2 0.44mA/cm2

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

208

the other cross-sectioned anode particles. This is due to the large sur-face area of the silicon nano-particles, around which the pitch carbonmust coat. The robustness of these composite particles shows that theiractive silicon nano-particles are strongly bound together by theircarbon matrix, despite any visible porosity. This conformal binder likecoating will help to hold the silicon nanoparticles together, linkingthem in a three-dimensional conducting network.

X-ray diffraction (XRD) patterns of the coal-tar-pitch before andafter heat treatment are plotted in Fig. 3a. Before heat-treating, pitch iscomposed of complex arrangements of mostly planar polycyclic aro-matic hydrocarbons with wildly varying molecular weights and con-figurations [11]. The broad, low intensity peaks of the unprocessedpitch are characteristic of the amorphous nature of this extremelycomplex industrial waste material. Upon heating, pitches soften into arelatively low viscosity liquid, making them ideal candidates for theformation of particle-matrix style composite materials. This liquid, ormesophase, allows the pitch's volatile low molecular weight compo-nents to distill while the larger ring compounds polymerize into planararomatic structures [12]. Further heating results in the coalescence andalignment of these planar structures into parallel stacks similar tographite. This ordering of the heat-treated pitch can be verified by theincreased intensity and angle of its diffraction peaks. These peaks, ob-served around 25° and 43°, represent the (002) and (100) diffractionmodes, respectively, and while their increased intensity suggests de-creased d-spacing in the material, their broadness is proof that the heat-treated pitch lacks any long-range repeating order and thus remainsamorphous.

Fig. 3b compiles XRD patterns of the heat-treated Si-C compositepowders along with a baseline scan of pure 1–3 μm silicon powder forreference. The sharp (111), (220) and (311) crystalline peaks of thesilicon powder, shown in green, match the literature [13]. A small peakaround 26° suggests that the silicon particles have at their surface athin, partially crystalline, native oxide layer. This is not uncommon forsilicon stored outside of a dry glovebox [9,14–16]. The similarity of the

silicon powder baseline scan to the profiles of the Si-C compositepowders further demonstrates the amorphous structure of the carbonmatrix material. The absence of any unexpected peaks also shows thatno secondary crystalline phases formed between the silicon and carbon,or any undetected impurities, during the heat-treatment process.

To investigate the electrochemical activity of the heat-treated coal-tar-pitch, an all-solid-state half-cell was prepared with a 2mg heat-treated pitch working electrode. This pitch received the same heattreatment as the other Si-C composite particles presented in this report.The cycling performance of this cell is shown in Fig. 4a, with selectvoltage profiles and dQ/dV plots displayed in Fig. 4b & c, respectively.This pure pitch electrode achieved a large first cycle specific lithiationcapacity of 964.1mAh/g, but only 621.28mAh/g of this capacity wasreversible. This corresponds to a first cycle coulombic efficiency of only64.4%. These capacities are far greater than the theoretical limit ofgraphite intercalation (~372mAh/g), but this is not unprecedented forsoft carbon materials [9,17–20].

Much of this pitch electrode's large first cycle capacity can be at-tributed to lithiation and de-lithiation plateaus centered around 0.25 Vand 1.25 V, respectively. While similar voltage profiles have been re-ported for other pitch based soft carbons, the physical interpretation ofthese large plateaus remains somewhat of a mystery. Larcher et al.suggested that some of the initial discharge capacity could be attributedto irreversible reactions with undetected elemental oxygen and sulfur inthe heat-treated pitch [17,21]. On the other hand, Zheng et al. con-vincingly argued that these high voltage plateaus were the result oflithium atoms semi-reversibly binding in the vicinity of hydrogen atomsin the soft carbon, showing that the presence and magnitude of theselarge voltage plateaus scaled directly with the hydrogen content of thepyrolyzed organic material tested [18–20]. We believe that a combi-nation of these two mechanisms is responsible for this electrode's re-latively large semi-reversible initial capacity.

The large voltage plateaus that greatly contributed to the pitchelectrode's initial capacity and hysteresis had nearly vanished after

Cross Section

1�

2�

�Point 1Si

C

�Point 2

Si

C (d)(c)

(a) (b)

Fig. 1. (a) Low magnification SEM image of a heat-treated Si-C composite particle containing 325-meshed Si-particles. (b) SEM image of the FIB cross sectionedparticle showing location of EDS point scans, (c) point EDS spectrum confirming silicon phase, (d) point EDS spectrum confirming carbon matrix.

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

209

10 cycles, leaving behind a gentle sloping voltage profile with a cou-lombic efficiency>99%. The disappearance of these plateaus waslikely due to both the irreversibility of their underlying reactions andthe development of a large overpotential. This overpotential can beattributed to the disruption of conduction pathways through the all-solid-state electrode. The utilization of a conductive binder or additivesuch as SSE would likely have reduced the magnitude of this over-potential and improved the electrode's capacity retention with cyclingby improving and maintaining interfacial contact between the activeparticles.

Despite containing no SSE, this heat-treated pitch electrode main-tained a stable 100th cycle discharge capacity of over 300mAh/g witha 100% coulombic efficiency at a 0.38mA/cm2 current density. The100th cycle's sloping voltage profile is characteristic of lithium ionspartially charge transferring to the surface of planar aromatic structurespresent in the soft carbon material [22]. A low voltage plateau, re-sulting from the one-hour lower limit voltage hold, suggests that stackintercalation and reversible lithium plating may have also contributedto the electrode's capacity. Not only was this electrode able to achieve a100th cycle specific capacity in the range of the theoretical limit of

graphite, it was able to do so without the addition of any conductingadditives. This result is important because it shows that even at largecurrent densities, heat-treated coal-tar-pitch can cycle on its own bysimultaneously acting as both active material and ionically/electro-nically conductive additive in an all-solid-state electrode. Therefore, bycoating silicon particles in this mixed conducting electrochemicallyactive carbon, we can reduce or even eliminate the need for inactiveadditives such as carbon black in our all-solid-state electrodes, in-creasing their volumetric and gravimetric specific capacities.

In order to investigate the heat-treated pitch's effectiveness as arigid matrix, three different composite anode materials were preparedwith varying sizes of silicon active particles. Much of the literature onsilicon anodes has depended on the superior cycling stability of com-plex nanostructured active materials. By testing a variety of siliconpowders, we hoped to investigate this pitch matrix material's ability toenable the stable cycling of inexpensive and more readily available si-licon micro-particles. To test the effectiveness of these composite anodematerials, three identical all-solid-state half-cells were prepared. Eachcell was prepared in the same fashion, but utilized silicon particles ofdifferent size (50 nm, 1–3 μm or 325meshed). These cells contained

Fig. 2. SEM images of heat-treated Si-C composites containing 325-meshed [≤44 μm] (a, b), 1–3 μm (c, d) and 50 nm (e, f) Si-particles. Images (a), (c), and (e)present a low magnification look at the size and morphology of the Si-C particles after hand grinding. Cross-sectional images (b), (d), and (f) give a highermagnification view into the internal microstructure of these composite particles.

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

210

2mg 7:3 Si-C:SSE working electrodes, corresponding to Si-C compositeand silicon particle mass loadings of approximately 1.4 mg and 0.7mg,respectively. This experiment will be referred to as the particle-sizestudy.

The cycling performance of the three particle-size study cells isplotted in Fig. 5. Note that the capacities are all normalized to the totalelectrode mass, not per gram of active material. From looking at Fig. 5,it is clear that the composite anode made with 50 nm silicon particlesoutperformed the two composites containing micron-sized silicon par-ticles in first cycle capacity, coulombic efficiency and capacity reten-tion. This result was expected, as nano-structured silicon active mate-rials are known to outperform larger silicon particles due to theirdecreased diffusion distances and facile stress relaxation upon lithiation[23,24]. The goal of this experiment was to determine if the mixedconducting coal-tar-pitch derived carbon was robust enough to enablethe deep lithiation of large silicon micro-particles in an all-solid-stateLIB. By encapsulating the silicon active materials in an amorphouscarbon matrix, we hoped to restrict their volumetric expansion and

limit their pulverization with cycling. Unfortunately, this data suggeststhat the amorphous carbon was not robust enough to withstand theextreme volumetric expansion of these large silicon micro-particlesupon deep discharge to 5mV and could not prevent the irreversiblecapacity losses brought on by their electrochemical isolation with cy-cling.

Fig. 6(a–f) shows select voltage profiles and dQ/dV plots for thethree cells included in this particle-size study. The first cycle voltageprofile for all three cells has a long flat lithiation plateau around 0.13 V.Shown as a single sharp cathodic peak in the dQ/dV plots, this lowvoltage lithiation plateau represents the well documented two phaseconversion of crystalline silicon into amorphous LixSi phases [3]. It isinteresting that the first cycle cathodic peak observed around 0.25 V inthe 100% heat-treated pitch electrode cannot be clearly seen in any ofthe cells in Fig. 6, although the pitch delithiation peak ~0.07 V isfaintly present. This makes it difficult to determine the extent to whichthe amorphous carbon matrix material is lithiated and how much itcontributes to the overall capacity of the electrodes with cycling.

Fig. 3. (a) XRD patterns of coal-tar-pitch powders before (blue) and after (red) a 5 hour heat-treatment at 900 °C under continuous argon flow. (b) XRD patterns ofthe heat-treated Si-C composite materials (black) labeled with the silicon particle sizes they contain. The indexed diffraction pattern of pure 1–3 μm silicon powder(green) is added for reference. All peak intensities are left as measured. (For interpretation of the references to color in this figure legend, the reader is referred to theweb version of this article.)

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

211

Because of this, the specific capacities in this report will be normalizedto total electrode mass or total mass of active composite, not per gramof silicon or pitch individually.

With cycling, the particle-size study cells' voltage profiles steadilyretract while becoming increasingly sloping in nature. In the dQ/dVplots, two broad lithiation peaks can be observed in the second cycle ofall three cells. These two peaks, centered around 0.24 V and 0.07 V, arecharacteristic of single-phase transitions between the various amor-phous LixSi phases [3,25]. As cycling progressed, the intensity of thesebroad cathodic peaks slowly decreased as they shifted to slightly lowervoltages. This decrease in peak height represents the electrochemicalisolation of active silicon particles while the shift, or overpotential, is

the result of increasing electrode resistance due to pulverization of itsmixed conducting composite microstructure.

All three cells show a sharp delithiation peak centered ~0.42 V. Thisanodic peak represents a two-phase transition in which the fully lithi-ated Li15Si4 crystal converts back to delithiated amorphous silicon. Thepresence of this sharp anodic peak (a-Li15Si4) is a clear indication thatthe Li15Si4 metastable crystal structure nucleated in the electrodes' si-licon particles during their deep discharge to 5mV. This a-Li15Si4 peakis initially present in Fig. 6d & f, but quickly shrinks into a broad humpmore akin to the delithiation of amorphous silicon phases with con-tinued cycling. The disappearance of the a-Li15Si4 peak shows thatwhile the saturated Li15Si4 crystalline phase initially nucleated in these

0 200 400 600 800 1000

Specific Capacity (mAh/g), Electrode

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

Vo

ltag

e (V

vs. L

i)

Cycle 1

Cycle 2

Cycle 10

Cycle 25

Cycle 50

Cycle 100

0 0.5 1 1.5

Voltage (V vs. Li)

-6000

-5000

-4000

-3000

-2000

-1000

0

1000

2000

dQ

/d

V ((m

Ah

/g

)/V

)

Cycle 1

Cycle 2

Cycle 10

Cycle 25

Cycle 50

Cycle 100

Fig. 4. (a) Cycling performance of a pureheat-treated pitch electrode in an all-solid-state Li-ion cell. This cell was run with noSSE added to its working electrode. Thecoal-tar-pitch powders were heat treated at900 °C for 5 h under continuous argon flow.The voltage profiles (b) and dQ/dV plots (c)for select cycles of this all-solid-state cellhave been included for reference.

Fig. 5. Cycling performance of all-solid-state Li-ion half-cells made with 7:3 Si-C:SSE working electrodes con-taining 50 nm (red) 1–3 μm (blue) and 325mesh (green)silicon particles. All specific capacities are normalized tototal electrode mass. (For interpretation of the refer-ences to color in this figure legend, the reader is referredto the web version of this article.)

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

212

large silicon micro-particles, it became unfavorable with continuedcycling. This can be attributed to an increasing overpotential in thepulverized electrodes shifting the crystallization potential outside of theapplied voltage window. In other words, the electrode's fragmentedmicrostructure introduced so much resistance to the system that theapplied current quickly resulted in polarization and heat loss ratherthan electrochemical work. This polarization led to a slowing of theelectrochemical reaction rate and greatly limited the silicon activematerials' degree of lithiation.

The broad first cycle delithiation curve in Fig. 6b can be interpretedas a combination of the two peaks associated with transitions betweenthe amorphous LixSi phases (0.3 V & 0.49 V) and the sharp anodic peak(a-Li15Si4) representing the two-phase delithiation of the Li15Si4 crystalstructure. The presence of both crystalline and amorphous silicon dQ/

dV peaks shows that despite being potentiostatically held at 5mV, somesilicon nano-particles were unable to nucleate the Li15Si4 crystal. This isproof that the pitch matrix was robust enough to limit the free volumeexpansion of some of its encapsulated silicon nano-particles, preventingthem from fully lithiating and thus suppressing their crystallization. Asmall but consistent overpotential in the cell's initial lithiation peaks isconsistent with this confinement of the silicon particles [6]. Becausethis overpotential remains relatively steady with cycling, we can con-fidently conclude that the continued suppression of the silicon nano-particles' full lithiation is not due to degradation of mechanical contactsin the electrode but is solely related to their confinement in a rigidamorphous carbon matrix. And while similar results have been reportedfor all-solid-state electrodes cycled under large externally appliedcompressive stresses, our data shows that confining forces can be

Fig. 6. Voltage profiles and dQ/dV plots of 7:3 Si-C:SSE all solid state cells prepared with 50 nm (a, b) 1–3 μm (c, d) and 325meshed (e, f) Si-particles.

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

213

achieved in situ by the encapsulation of silicon nano-particles in amixed conducting amorphous carbon matrix [6]. This is highly desir-able because the application of large external compressive stresses isnot practical for future commercialization of the all-solid-state LIB.

Large internal stresses develop at the interface between amorphoussilicon and the saturated Li15Si4 crystal phase. These large stresses canlead to the fracture and electrochemical isolation of silicon active ma-terials during cycling. By confining their expansion below the thresholdfor forming the Li15Si4 crystal, the rigid amorphous carbon matrixhelped to limit the stress in the silicon nano-particles. This limitation ofthe silicon particle expansion results in reduced capacity, but it im-proves cycling stability, an important metric when transitioning to fullcell configurations. While similar lithiation restrictions can be achievedby simply limiting the lower voltage limit of the cell, we were interestedin investigating the pitch matrix's ability to apply confining forces onthe silicon particles in situ. The deep discharge of the active materialswas therefore used as a test of the amorphous carbon matrix material'sability to confine and withstand the expansion of various silicon par-ticles, as any restriction to free volume expansion in this cycling regimewould certainly translate to improved cycling stability at lower degreesof lithiation.

It should be mentioned that no crystallization suppression was ob-served in the silicon micro-particle composite electrodes' initial dQ/dVcharge profiles, as only the sharp a-Li15Si4 peak could be observed. Hadthe lithiation of any of the active silicon micro-particles been limited tolevels below the threshold for forming the Li15Si4 crystal, we wouldexpect to observe a broad anodic dQ/dV profile as in Fig. 6b. Thisshows that the pitch carbon coating was not robust enough to restrainthe free volume expansion of the large silicon particles, pulverizinginstead. Unfortunately, these results suggest that the pitch derivedamorphous carbon matrix is not effective in enabling the long-termutilization of inexpensive silicon micro-particles in a rechargeable all-solid-state LIB. While reduction of silicon mass loading in the compo-site, limiting the lower voltage limit and removal of the low voltagehold would surly improve the cycling stability of these all-solid-stateelectrodes, this optimization is left for future work. The remainder ofthis paper will instead focus on the effect of SSE content on the com-posite electrodes' cycling stability.

In order to demonstrate the full potential of the heat-treated coal-tar-pitch matrix material, a series of all-solid-state half-cells was con-structed utilizing the Si-C active composite prepared with 50 nm siliconparticles. The five cells in this study were assembled with electrodecomposite ratios (nSi-C:SSE) ranging from 6:4 to 100% Si-C (no SSEadded to the working electrode). By varying the amount of SSE in thecells' electrodes, an optimum balance between electrode specific capa-city and capacity retention with cycling was to be discovered. Thisexperiment will be referred to as the electrode-composite study.

The cycling capacities of the electrode-composite study half-cellsare plotted in Fig. 7. Their current densities are listed in Table 1. Noticethat the cells' initial specific lithiation capacity scales directly with themass fraction of active Si-C particles in their composite electrodes. Thisis because these specific capacities are calculated per gram of electrode,not per gram of active material in the electrodes. Fig. 7 shows that thecycling stability of these cells generally increases with the mass fractionof SSE in their working electrodes. For example, while the 100% Si-Ccell achieved the largest first cycle specific lithiation capacity(2082.8 mAh/g) of any cell tested, its capacity quickly decreased to506.3 mAh/g after only 25 discharge-charge cycles. On the other hand,the 6:4 Si-C:SSE composite working electrode, which had the smallestfirst cycle specific lithiation capacity (1115.2 mAh/g), experienced re-latively little capacity loss with cycling and finished 100 discharge-charge cycles with the highest capacity of any cell in this study(653.5 mAh/g at a 0.38mA/cm2 current density). Normalizing thiscell's 1st and 100th cycle capacities to mass of active Si-C compositeresults in specific lithiation capacities of 1858.7 mAh/g and1089.2 mAh/g, respectively. This corresponds to 1st and 100th cycle

areal capacities of 1.66mAh/cm2 and 0.97mAh/cm2. Therefore, notonly does this half-cell outperform all others in this study, it also ex-ceeds the capacity reported for a similar material cycled in a traditionalcoin cell configuration after 100 cycles [9].

All five cells in Fig. 7 displayed first cycle coulombic efficienciesaround 83%, regardless of SSE content in their composite electrodes.Similarly, all of the cells displayed second cycle efficiencies of ~92%.These low coulombic efficiencies can be attributed to initial electro-chemical isolation of active materials and irreversible reactions in theall-solid-state electrodes. With continued cycling, the coulombic effi-ciencies of these five cells diverged and a trend became clear. The cellsthat contained more solid electrolyte achieved and maintained cou-lombic efficiencies> 99% earlier. For example, the 6:4 Si-C:SSE com-posite working electrode achieved a stable> 99% coulombic efficiencyafter< 20 cycles, while it took the 8:2 Si-C:SSE electrode ~50 cyclesand the 100% Si-C electrode nearly 90 cycles to achieve the same. Thisshows that the added SSE improves the reversible utilization of the Si-Cactive materials in an all-solid-state electrode, leading to improvedcapacity retention with cycling. This can be attributed to the additionalSSE improving interfacial contact with the active Si-C composite par-ticles, reducing their likelihood of isolation upon delithiation. It alsoshows that additional SSE does not lead to continual development offragile solid-electrolyte-interphase layers that can limit the coulombicefficiency of conventional cells utilizing liquid based electrolytes.

The voltage profiles and dQ/dV plots for the 100% Si-C and 6:4 Si-C:SSE electrodes are presented in Fig. 8(a–d). These plots display thecharacteristic silicon lithiation plateaus and peaks discussed earlierwith no signs of any unexpected side reactions. The similarity of bothcells' first cycle voltage profiles shows that the Si-C composite materialdoes not require the addition of SSE to cycle in an all-solid-state cell.This is further proof of the facile mixed conductivity of the amorphouscarbon matrix material.

A large overpotential quickly developed in the 100% Si-C electrode.This resulted in the disappearance of the sharp a-Li15Si4 peak after only5 cycles. The 6:4 Si-C:SSE electrode experienced virtually no voltageshift, so rather than disappearing, its a-Li15Si4 peak actually increasedin magnitude over its first 50 cycles. Growth of the a-Li15Si4 peak can beattributed to decreased confinement of the silicon active particlesleading to an increase in their free volume expansion upon lithiation.This reduction of silicon particle confinement may be explained bygradual fracturing of the Si-C composite as well as inelastic deformationof the surrounding SSE matrix with cycling. While a small increase in a-Li15Si4 peak height was observed for this cell, the capacity of theelectrode remained remarkably stable suggesting that continued da-mage to its composite microstructure with cycling was minimal.

The exceptional stability of the 6:4 Si-C:SSE electrode's dQ/dV plotssuggests that the growing overpotential in the 100% Si-C electrode canbe attributed to the fragmentation and disruption of interfacial contactsbetween its composite particles. The presence of SSE in the electrodeclearly improves the cycling stability of the Si-C composites. This isbecause the SSE behaves as a secondary ionically conductive matrix,improving interfacial contact between the active Si-C composite parti-cles much in the same way that the pitch derived amorphous carbonencapsulates and confines the silicon nano-particles. Increasing themass loading of SSE in the composite electrode allows for more con-formal coating of the electrochemically active Si-C particles. This en-sures improved interfaces and connectivity in an electrode that wouldotherwise be limited to particle-particle contact points between rela-tively brittle active materials.

No clear peaks corresponding to the lithiation or delithiation of thecoal-tar-pitch derived amorphous carbon matrix can be observed ineither cells' dQ/dV plots. Because of this, it is difficult to make anyconfident conclusions on the extent of soft carbon utilization in thesecells. While these electrodes' capacities could theoretically be attributedto their silicon particles alone, we believe it is highly unlikely that theamorphous carbon matrix remained inactive. For example, the 6:4 Si-

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

214

C:SSE electrode had a total mass of 1.98mg, corresponding to about1.188mg of Si-C composite particles or around 0.7128mg of silicon and0.4752mg of amorphous carbon. Assigning conservative specific ca-pacity values of 3000mAh/g and 300mAh/g to the silicon and carbonphases would result in first cycle specific capacities of ~1152mAh/g(electrode) and 1920mAh/g (active material). These values vary fromthose actually observed for the cell by only 3%. When applied to the

100% Si-C electrode, this calculation predicts a first cycle specific li-thiation capacity 7% smaller than what was observed experimentally.This shows that the Si-C composite materials were able to achieve highdegrees of silicon and carbon utilization without the aid of the ionicallyconductive additives, and that the addition of SSE primarily helpedwith maintaining interfacial contact throughout the working electrode.

After 100 cycles, select electrodes were cross-sectioned and imaged

Fig. 7. Specific lithiation and de-lithiation capacities ofall-solid-state Li-ion half-cells containing the nSi-C com-posite anode material. The Si-C:SSE ratio in these cells'working electrodes, 6:4 (blue), 7:3 (red), 8:2 (green), 9:1(orange) and 100% Si-C (black), was varied in order tooptimize the cycling performance of this composite ma-terial. All specific capacities are normalized to totalelectrode mass. (For interpretation of the references tocolor in this figure legend, the reader is referred to theweb version of this article.)

Fig. 8. Select voltage profiles and dQ/dV plots for two all-solid-state half-cells included in the electrode-composite study. Both cells contained 50 nm silicon particlesin their working electrodes. Plots a & b represent the 100% nSi-C working electrode. Plots c & d represent the 6:4 nSi-C:SSE working electrode.

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

215

with a FIB-SEM system. Fig. 9a shows the internal microstructure of the6:4 Si-C:SSE electrode prepared with 50 nm silicon particles. Notice thatthis electrode is almost totally dense with very little cracking or por-osity. This cell maintained a stable capacity throughout cycling becauseit contained enough malleable SSE to extensively link its electro-chemically active Si-C composite particles in an ionically conductivematrix. Maintaining a high degree of interfacial contact between thecomposite particles is critical to preventing active material isolation inall-solid-state cells where conduction pathways are limited to small,particle-particle contact points. These bottleneck points restrict themovement of Li-ions through the cell and can be easily disrupted duringcycling leading to irreversible capacity losses.

Compare Fig. 9a to the cross-section of the 7:3 Si-C:SSE electrodeprepared with the same 50 nm silicon composite anode material(Fig. 9b). Large voids can clearly be seen permeating through the 7:3 Si-C:SSE electrode. Since the same Si-C composite material was used inboth cells, the presence of these large voids must therefore be attributedto the lower mass loading of SSE in the electrode. These voids or se-parated interfaces are detrimental to the electrochemical performanceof the cell because they act as roadblocks to electrons and Li-ions tra-veling through the electrode, increasing resistance. Furthermore, thesevoids allow for the free volume expansion of surrounding active ma-terials, enabling their fracture and isolation with cycling.

The relationship between electrode porosity, pulverization and ca-pacity fade with cycling is further highlighted by the cross-sectioned7:3 Si-C:SSE electrode presented in Fig. 9c. This cell, which was pre-pared with larger 1–3 μm silicon particles, displayed the poorest cyclingperformance of any tested in this study (Fig. 4). The cross sectionalimage of this electrode shows a high degree of cracking and delami-nation at the interfaces between the SSE matrix and the Si-C compositeparticles. This delamination was likely exacerbated by the free volu-metric expansion of the large silicon micro-particles. Therefore, byfailing to maintain interfacial contact with the active composite mate-rials and restrain their volumetric expansion, these voids acceleratedthe electrode's irreversible capacity losses with cycling. In the future,we will investigate alternate processing methods to further improveinterfacial connectivity in these all-solid-state electrodes, increasing thecycling stability and utilization of the Si-C composite anode materials.

4. Conclusions

Silicon-Carbon composites derived from the industrial waste pro-duct coal-tar-pitch were, for the first time, utilized as anode materials inan all-solid-state Li-ion cell. We demonstrated how a simple and in-dustrially scalable solution coating process could be used to encapsulatesilicon particles of various shapes and sizes in an amorphous carbonmatrix. On its own, this soft carbon material displayed excellent Li-ioncapacity and mixed conducting capabilities. Various silicon particlesizes and Si-C:SSE electrode composite ratios were investigated in thisreport. While it was concluded that the amorphous carbon matrixmaterial was not robust enough to enable the long term utilization oflarge silicon micro-particles, the optimization of the nano‑siliconcomposite electrode resulted in a half-cell with a 100th cycle specificcapacity of 653.5 mAh/g (mass electrode) and 1089.2 mAh/g (mass Si-C) with a coulombic efficiency>99%.

Acknowledgements

This material is based on research sponsored by Air Force ResearchLaboratory under agreement number FA9453-15-1-0304. The U.S.

Fig. 9. FIB cross section view of Si-C:SSE composite electrodes after 100 charge-discharge cycles. (a) 6:4 nSi-C:SSE composite electrode. (b) 7:3 nSi-C:SSEcomposite electrode. (c) 7:3 Si-C:SSE composite electrode containing 1–3 μmsilicon particles.

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

216

Government is authorized to reproduce and distribute reprints forGovernmental purposes notwithstanding any copyright notationthereon.

References

[1] J.W. Fergus, J. Power Sources 195 (2010) 4554.[2] T.A. Yersak, T. Evans, J.M. Whiteley, S.B. Son, B. Francisco, K.H. Oh, S.H. Lee, J.

Electrochem. Soc. 161 (2014) A663.[3] J.M. Whiteley, J.W. Kim, D.M. Piper, S.H. Lee, J. Electrochem. Soc. 163 (2) (2016)

A251–A254.[4] J.E. Trevey, K. W, C.R. Stoldt Rason, S.H. Lee, Electrochem. Solid-State Lett. 13

(2010) A154.[5] S.B. Son, S.C. Kim, C.S. Kang, T.A. Yersak, Y.C. Kim, C.G. Lee, S.H. Moon, J.S. Cho,

J.T. Moon, K.H. Oh, S.H. Lee, Adv. Energy Mater. 2 (2012) 1226.[6] D.M. Piper, T.A. Yersak, S.H. Lee, J. Electrochem. Soc. 160 (2013) A77.[7] T.A. Yersak, S.B. Son, J.S. Cho, S.S. Suh, Y.U. Kim, J.T. Moon, K.H. Oh, S.H. Lee, J.

Electrochem. Soc. 160 (2013) A1497.[8] H. Tillmans, Fuel 64 (1985) 1197–1203.[9] Y.X. Wang, S.L. Chou, J.H. Kim, H.K. Liu, S.X. Dou, Electrochim. Acta 93 (2013)

213–221.[10] K. Takada, N. Aotani, K. Iwamoto, S. Kondo, Solid State Ionics 86–8 (1996) 877.[11] H. Preiss, K. Szulzewsky, P. Kölsch, Fuel 72 (1993) 1301–1303.

[12] I.C. Lewis, Fuel 66 (1987) 1527–1531.[13] C.R. Hubbard, H.E. Swanson, F.A. Mauer, J. Appl. Crystallogr. 8 (1975) 45.[14] D.Q. Yang, J.N. Gillet, M. Meunier, E. Sacher, J. Appl. Phys. 97 (2005) 024303.[15] M. Morita, T. Ohmi, E. Hasegawa, M. Kawakami, M. Ohwada, J. Appl. Phys. 68

(1990) 1272.[16] C.N. Barnakov, G.P. Khokhlova, A.N. Popova, S.A. Sozinov, Z.R. Ismagilov, Eurasian

Chem. Technol. J. 17 (2015) 87–93.[17] D. Larcher, C. Mudalige, A.E. George, V. Porter, M. Gharghouri, J.R. Dahn, Solid

State Ionics 122 (1999) 71–83;J.M. Whiteley, J.W. Kim, C.S. Kang, J.S. Cho, K.H. Oh, S.H. Lee, J. Electrochem. Soc.162 (4) (2015) A711–A715.

[18] T. Zheng, J.S. Xue, J.R. Dahn, Chem. Mater. 8 (1996) 389–393;A. Hayashi, S. Hama, T. Minami, M. Tatsumisago, Electrochem. Commun. 5 (2)(2003) 111–114.

[19] T. Zheng, W. Xing, J.R. Dahn, Carbon 34 (1996) 1501–1507.[20] T. Zheng, Y. Liu, E.W. Fuller, S. Tseng, U. von Sacken, J.R. Dahn, J. Electrochem.

Soc. 8 (1995) 142.[21] D. Larcher, C. Mudalige, M. Gharghouri, J.R. Dahn, Electrochim. Acta 44 (23)

(1999) 4069–4072.[22] I. Mochida, C.H. Ku, Y. Korai, Carbon 39 (2001) 399–410.[23] R. Teki, M.K. Datta, R. Krishnan, T.C. Parker, T.M. Lu, P.N. Kumta, N. Koratkar,

Small 5 (2009) 2236–2242.[24] J.R. Szczech, S. Jin, Energy Environ. Sci. 4 (2011) 56.[25] W.J. Zhang, J. Power Sources 196 (2011) 877.

N.A. Dunlap et al. Solid State Ionics 324 (2018) 207–217

217