FeS 2 ‐Imbedded Mixed Conducting Matrix as a Solid …engineering.snu.ac.kr/pdf/2016/2016_HSS_FeS 2 -Imbedded... · 2017-01-13 · eS F 2-Imbedded Mixed Conducting Matrix as a Solid

FULL P

APER

© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim (1 of 10) 1600495wileyonlinelibrary.com

FeS 2 -Imbedded Mixed Conducting Matrix as a Solid Battery Cathode

Justin M. Whiteley , Simon Hafner , Sang Sub Han , Seul Cham Kim , Kyu Hwan Oh , * and Se-Hee Lee *

J. M. Whiteley, S. Hafner, Prof. S.-H. Lee Department of Mechanical Engineering University of Colorado at Boulder Boulder , CO 80300-0427 , USA E-mail: [email protected] S. S. Han, S. C. Kim, Prof. K. H. Oh Department of Materials Science and Engineering Seoul National University Seoul 151-742 , Republic of Korea E-mail: [email protected]

DOI: 10.1002/aenm.201600495

conductors act as diffusional pathways for both ions and electrons. Perceptively, charge transfer is limited to the triple point of materials. With a liquid, the triple point is virtually guaranteed, whereas if all phases are solid, the presence of the triple point will be far more infrequent. To ensure larger electrolyte/active material boundaries, solid electrolytes exhibiting ductility are preferential. [ 5 ] Unfortunately, obligatory ductility limits the selection of solid electrolytes considerably. Even under high compaction pressures, full density will never be achieved such as permeating an empty space with a liquid. Thus, the total charge transfer sites are limited. The lack of sites leads to “poor rate capability.”

Active materials exhibit volume expan-sions and contractions over the course of

cycling. [ 6 ] In a conventional Li-ion design, the liquid electrolyte can saturate the porous structure and accommodate any expan-sion without ever losing contact, while a solid will most likely not be able to reform to original conditions. If one of the con-ductive additives loses contact with the active material, diffu-sional pathways are severed and principally transport must take place within the active material, often a strenuous process. This is a common problem in solid designs referred to as “active material isolation.”

Finally, the presence of a solid electrolyte powder generates additional variables. Not only can there be chemical instabilities leading to chemical decomposition, [ 7–9 ] but typically solid elec-trolytes are diffi cult to synthesize, [ 10 ] and are unstable in most solvents used in battery preparation. [ 11 ] Variability in the particle sizes of both active material and solid electrolyte adds to “het-erogeneity of the electrode design,” complicating repeatability.

Therefore, we pose the question: can the solid electrode design be simplifi ed?

This study investigates the effect of replacing both the ionic conductor (solid electrolyte) and electronic conductor (carbon) with a single mixed conductor. By doing so, the theoretical interface for charge transfer is enhanced immensely. Any mate-rial contact between active material and mixed conductor will ensure diffusional pathways for both ions and electrons. If the transport of charged particles is high enough, there would be no need for additional electrolyte or carbon.

The concept of pairing a mixed conductor and an insulative active material, while not entirely new, has not been studied in depth. [ 12 ] Recently, studies using a mixed conductor of TiS 2 with

A new concept of pairing an active material and a mixed conductor is explored as a solid-state battery electrode. By imbedding nano-FeS 2 domains into an amorphous LiTiS 2 matrix, a hybrid power-energy system is achieved while additionally improving upon many common solid electrode design fl aws. High-resolution transmission electron microscopy is used to probe the active material/mixed conductor interface over the course of cycling. Arguably the most benefi cial development is enhancement of charge transfer, mani-festing in a signifi cantly increased exchange current as captured in a Tafel analysis. By developing a solution to active material isolation and creating a more homogenous electrode design, cycling at a high rate of C/2 for 500 cycles is obtained. Additionally, the electrode can recover full capacity simply by reducing system rate. Capacity recovery implicates a lack of active material isolation, a common problem in solid-state batteries.

1. Introduction

Numerous designs of all solid batteries have been proposed and developed for superior safety and stability. Whether the construction is the slurry-binder technique, [ 1 ] deposition, [ 2 ] or follows the dynamic compaction strategy, [ 3,4 ] all bulk designs (nonthin fi lm) differ fundamentally from liquid-based systems. The predominant difference between a conventional Li-ion bat-tery electrode and a solid-state electrode is electrolyte perva-sion of empty space. In the solid-state, the electrolyte cannot be added in the fi nal stages to fi ll in the voids. Instead, solid electrolyte powder must be added directly to the original elec-trode synthesis, leading to many of the obstacles encountered in solid batteries.

Charge transfer in essence requires the presence of an ionic conductor, an electronic conductor, and an active mate-rial (assuming the active material is a pure insulator). The

Adv. Energy Mater. 2016, 6, 1600495

www.MaterialsViews.comwww.advenergymat.de

http://doi.wiley.com/10.1002/aenm.201600495

FULL

PAPER

© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim1600495 (2 of 10) wileyonlinelibrary.com

sulfur have been reported [ 13 ] ; however, the purpose of these studies was to protect the active material from electrolytic attack and dissolution, common problems in liquid batteries yet not applicable to solid-state. [ 14 ] Other pairings include CuF 2 with MoO 3 to enhance utilization of the active material. [ 15 ] Yet even these systems still used carbon as a conductive additive and no fundamental charge transfer investigations were conducted. To the best of our knowledge, this is the fi rst report of pairing only a mixed conductor with an active material excluding both ionic and electronic additives in liquid or solid confi guration.

The system that we used to test the hypothetical design is an active material of FeS 2 and a mixed conductor of LiTiS 2 . FeS 2 has recently proven to have potential in a solid battery due to its high theoretical capacity (894 mAh g −1 ), [ 16 ] conductive discharge products (iron metal), [ 17 ] and tremendous stability. [ 18 ] LiTiS 2 (LTS) has a diminutive, low-voltage capacity (227 mAh g −1 ) to be useful singularly [ 19 ] ; however, LTS is one of the best lithium mixed conductors known with a Li-ion self-diffusion coeffi cient of ≈1 × 10 −9 cm 2 s −1 at room temperature [ 20 ] and an electronic conductivity in excess of 1 S cm −1 . [ 21 ] The system is advanta-geous as the lithiation voltage of TiS 2 is greater than that of orthorhombic FeS 2 (marcasite) and thus can act as a solid elec-trolyte. The new system is characterized via transmission elec-tron microscopy (TEM) and electron energy loss spectroscopy (EELS) to demonstrate uniform dispersion of nano-FeS 2 in an amorphous LTS matrix. Mass transfer measurements, via elec-trode diffusion coeffi cients, indicate high diffusion capabilities at initial stages of lithiation. Charge transfer is measured using a Tafel analysis revealing that the FeS 2 -LTS system has more than double the exchange current as a standardly prepared solid cathode of the same composition. The result is a hybrid power-energy system where a prominent capacity (567 mAh g −1 of electrode) can be achieved at moderate rates (C/10) and a capacity greater than 400 mAh g −1 of electrode can be achieved at high rates of C/2. The new system persists for over 500 cycles, with 62% capacity retention and 99.8% average coulombic effi -ciency, and is able to recover near theoretical capacity following rate reduction. Comparison of the new mixed conductor/active material design to a standard solid electrode system demon-strates the stark contract in minimal active material isolation. The more homogenized electrode design now allows for better repeatability as the cathode composite can be produced at the mass scale.

2. Results and Discussion

An important aspect of any cathode is maximal contact between active material and conductors. Therefore, the optimal structure of the proposed design will have nanodomains of active mate-rial imbedded in a matrix of the mixed conductor to amplify charge transfer surface area. The continuous mixed conductor matrix will also act as diffusion pathways for both Li ions and electrons. The design is accomplished by concurrently reducing FeS 2 particle size while forming an in situ matrix of LTS through planetary ball-milling of FeS 2 , TiS 2 , and Li 3 N. By main-taining a low weight loading of FeS 2 , particle agglomeration is prevented and FeS 2 domain size is continually reduced. Other studies have investigated the effects of using a solid dispersant

to achieve nano-FeS 2 , therefore LTS acts as a solid dispersant in this design. [ 22 ]

The other important aspect of any electrode is to achieve prominent Li-ion diffusion. LTS naturally forms in the 1T hexagonal layered structure. [ 19 ] Although the Li self-diffusion in 1T-LTS is quite high, [ 23 ] the material suffers from having highly anisotropic diffusion in two dimensions. We discover an interesting effect and a new method to amorphize LTS by ball-milling TiS 2 and Li 3 N for extended periods of time. The addi-tion of Li 3 N reduces TiS 2 particle size through an in situ lithi-ation. [ 24,25 ] Figure 1 is an x-ray diffraction (XRD) pattern of the prepared electrode with amorphous LTS, amorphous LTS sin-gularly, and 1T-LTS. The method for preparing 1T-LTS was pre-viously reported. [ 25 ] It is concluded that LTS is amorphous due to the lack of XRD peaks. The FeS 2 -LTS cathode in Figure 1 (iii) does not contain the crystalline LTS peaks at 33.5° and 42°. Thus, LTS can still amorphize even with inclusion of FeS 2 during the milling process. We are aware of only one study that has been able to achieve an amorphous LTS; both theoreti-cally and experimentally, 3D diffusion was achieved. [ 26 ] Having a more isotropic diffusion without preferential pathways will enhance the active material/mixed conductor interface. As FeS 2 retains cubic peaks after milling with the other precursors, it would appear that FeS 2 does not chemically interact with Li 3 N or TiS 2 . An explanation for this behavior is derived from inves-tigating the theoretical solid-state reaction enthalpies, as com-puted by the Materials Project, [ 27 ] and displayed as Equations ( 1) and ( 2)

+ → + Δ = − −6TiS 2Li N 6LiTiS N H 734kJmol2 3 2 21

( 1)

+ → + Δ = − −3FeS 2Li N 3Li FeS N H 488kJmol2 3 2 2 21

( 2)

While not defi nitive, this gives an indication that the reac-tion of Li 3 N with TiS 2 is more spontaneous and favorable.



Figure 1. XRD pattern of: (i) TiS 2 and Li 3 N (3 to 1 molar ratio) ball-milled 2 h to form crystalline 1T-LiTiS 2 , (ii) TiS 2 and Li 3 N (3 to 1 molar ratio) ball-milled 20 h to form amorphous LiTiS 2 , (iii) FeS 2 , TiS 2 , and Li 3 N (3.1, 3, 1 molar ratio) ball-milled 20 h to form cathode in study. Only cubic FeS 2 peaks are present demonstrating that amorphous LiTiS 2 can also be formed without reacting with FeS 2 .

FULL P

APER


Additionally, TiS 2 has diffusion pathways to accommodate lithium. Cubic FeS 2 is known to have a strong activation energy during initial lithiation which helps explain the role of Li 3 N. [ 16 ]

The theoretical structure of the solid cathode is confi rmed using TEM ( Figure 2 ). Using a focused ion beam (FIB) mill to cross-section the uncycled and tenth cycle electrode, [ 28 ] the compacted structure is revealed—a more accurate representa-tion of performance than free-formed powders. Initial FeS 2

domains are detected on the order of 10–100 nm in Figure 2 a. The LTS matrix appears to have disordered domains on the order of a nanometer. The size of these domains would cor-respond to only a few TiS 2 slabs. While not truly amorphous, the severe disordering on this length scale would contribute to 3D diffusion. Therefore references to “amorphous LTS” in this text refer to the x-ray amorphous property. Intimate contact is developed between LTS and FeS 2 . To understand the micro-structural changes that occur during cycling, a cross-section of the tenth cycle FeS 2 -LTS electrode following full charge to 3 V is investigated by high-resolution transmission electron micros-copy (HR-TEM) in Figure 2 b. The large FeS 2 particles pulverize into smaller domains yet remain clustered together. This is in accord with previous studies that only reported this behavior after the fi rst charge. [ 16 ] The TiS 2 matrix now appears to be completely amorphous yet remains in excellent contact with the pulverized FeS 2 .

TEM and EELS elemental mapping of iron and titanium in the uncycled electrode is displayed in Figure 3 a–c, respectively. Due to partial oxidation, it is impossible to confi rm exact 2 to 1 sulfur to iron or sulfur to titanium in both active material and matrix. XRD in Figure 1 along with the distinct domains of FeS 2 would suggest no chemical interaction occurs between FeS 2 and TiS 2 in the prepared state. This is an added benefi t of not using solid electrolyte as solid electrolytes can decom-pose against active materials or even low voltage mixed conduc-tors. [ 8,29 ] EELS elemental mapping is again used to probe iron and titanium following the tenth cycle charge in Figure 3 e,f, respectively. There is a stronger iron signal in this case that could be due to the pulverization of particles. The two mate-rials are still somewhat separated but the FeS 2 appears to form a more continuous matrix rather than localized clusters. Two matrices in contact with one another are still an ideal structure in the solid-state for performance. [ 3 ] While a defi nitive state-ment cannot be made that chemical interaction is not occur-ring, due to the electrochemical response during cycling in the upcoming section, it appears that neither the lithiation of FeS 2 nor conduction capabilities of TiS 2 are being hindered. Infer-ence of replacing a metastable solid electrolyte with a thermo-dynamically stable TiS 2 would suggest less chemical decompo-sition is expected.

Figure 4 displays the basic electrochemical properties via cyclic voltammetry (CV) of the FeS 2 -LTS cathode against a standard cathode using FeS 2 , solid electrolyte, and carbon. The standard cathode is investigated for the rest of this study to pro-vide a stark contrast with the new design. Figure 4 a,b shows CVs of the FeS 2 -LTS and standard systems at various sweep rates, respectively. All values are normalized per gram of com-posite to make accurate comparisons between the two elec-trodes. The most noticeable difference is the shape of the fi rst cathodic peak with increased sweep rates. Since the cathodic peak potential shift is much larger than the corresponding anodic peak potential shift, it can be inferred this is due to a signifi cant activation overpotential in the standard system com-pared to the FeS 2 -LTS system. [ 30 ] This will be explored more in the upcoming sections. For the other main cathodic peak and two main anodic peaks, there is not a substantial difference in peak potential; this is further displayed in Figure 4 c which overlaps the slow sweep rate of 0.1 mV s −1 for both systems.



Figure 2. a) HR-TEM of interface between amorphous LiTiS 2 matrix and FeS 2 nanodomains in uncycled electrode. b) Tenth cycle charge HR-TEM of interface between amorphous TiS 2 and FeS 2 . FeS 2 has pulverized into nanodomains that remain clustered. Intimate contact is preserved between mixed conductor and active material.

FULL

PAPER

© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim1600495 (4 of 10) wileyonlinelibrary.com

An asterisk is used in Figure 4 a,b to denote the anodic peak used for analysis. Figure 4 d plots the peak current of the second anodic peak for both systems versus the square root of the sweep rate. The steeper slope of the FeS 2 -LTS system indicates either a larger diffusion coeffi cient or greater lithium ion con-centration. [ 30 ] Due to the fact that the same active material is being used and greater lithium ion concentration in the solid electrolyte than in the TiS 2 matrix, the difference is most likely due to the larger diffusion coeffi cient now explored more directly.

To analyze mass transfer polarization, the most important term will be Li-ion self-diffusion ( D Li ) apparent in Fick’s second law. We analyze D Li by means of the Galvanostatic Intermittent Titration Technique (GITT) developed by Weppner and Hug-gins [ 31 ] and furthered by Wen et al. [ 32 ] This technique applies a current pulse to the system and measures the relaxation voltage. The calculation is given by Equation ( 3)

πτ= Δ

Δ⎛⎝⎜

⎞⎠⎟

<<4Li

2 s

t

2 2

D LE

Et

L

D

( 3)

where τ is the time of the current pulse, L is the length of the electrode, ΔE s is the change in steady-state voltage as a result of the current pulse, and ΔE t is the total change in voltage during the application of constant current (subtracting out uncompen-sated ohmic effects). [ 32 ]

The diffusion coeffi cient will vary as a function of lithium content in the electrode; therefore D Li is measured over the course of a full discharge and charge. The third cycle is chosen

to obtain relevant values due to allotropic transformation in FeS 2 from cubic to orthorhombic during the fi rst cycle. [ 16,33,34 ] Figure 5 a,b plots the diffusion coeffi cient as a function of voltage over the third cycle for the FeS 2 -LTS and standard sys-tems, respectively. Obtained values are in agreement to the only other study for determining D Li in Li x FeS 2 . [ 35 ] As the reduc-tion of the (S–S) 2 2− persulfi de units in marcasite proceeds, a minimum in D Li is present. Lack of diffusion pathways and vast structural changes lead to this occurrence. A maximum in D Li is achieved once the layered γ-phase of Li x FeS 2 is formed and continues for the full intercalation region up to Li 2 FeS 2 is formed, whereby the persulfi de units are fully reduced to S 2− anions. [ 35,36 ] A local minimum is again achieved during the redox of Fe 2+ to Fe 0 . While there are no studies investi-gating diffusion in this region, we postulate a low coeffi cient is the direct result of a two-phase formation as well as crea-tion of resistive Li 2 S. Evidently, there is scarcely a difference in D Li between the two systems; however, there is a dramatic change during initial stages of lithiation. In the fi rst few pulses of current in the FeS 2 -LTS cathode, only the TiS 2 is lithiating due to the higher lithiation voltage. LTS has an extraordinarily high value of D Li , even higher than many sulfi de solid electro-lytes. [ 20,37 ] This process is not apparent on delithiation as the delithiation voltage of FeS 2 and TiS 2 almost overlaps. Interest-ingly, mass transfer is not signifi cantly affected by removing the solid electrolyte. An attempt was made to measure the ionic conductivity directly of the electrodes using DC polarization, however, due to confl icting ionic/electron fl ow, [ 38 ] these values were indeterminate.



Figure 3. a) TEM of FeS 2 (dark) imbedded in LTS (gray) matrix; b) zero loss elemental mapping of iron in uncycled electrode; and c) zero loss elemental mapping of titanium in uncycled electrode. It is clear that FeS 2 and LTS remain separated and not chemically mixed. d) TEM of tenth cycle charge cross-section; e) zero loss elemental mapping of iron; and f) zero loss elemental mapping of titanium. While FeS 2 has agglomerated into larger domain sizes over cycling, the LTS matrix remains in contact without any indication of a chemical decomposition.

FULL P

APER


Perhaps the most important polarization impacting system performance for this design is charge transfer, as previously discussed. In order to measure charge transfer, a Tafel analysis is used. [ 39 ] While this technique may appear unconventional, more common techniques such as electrochemical impedance spectroscopy are severely impacted from poor interfaces such as those found in solid-state. [ 7,40 ] Additionally, Tafel analyses have been used on a variety of cathode and anode materials. [ 41–44 ] A Tafel analysis measures exchange current ( i 0 )—representative of a balanced faradaic activity at equilibrium and is therefore relat-able to the reaction rate and contributing factors. Equation ( 4) is the cathodic Tafel equation

η

α α= −lni lni0

RT

F

RT

F ( 4)

where η is the polarization, R is the gas constant, T is the tem-perature, α is the charge transfer coeffi cient, and F is Faraday’s constant. [ 30 ]

By polarizing a system far enough from equilibrium, the forward or backward reaction will dominate. Extrapolating the linear segment back to equilibrium yields the exchange cur-rent. Using linear sweeps in both the upper plateau (interca-lation) and the lower plateau (conversion) on the third cycle for each system, rearrangement yields overpotential as a func-tion of current. Figure 6 a gives the Tafel plot of both systems at 60 °C. Appropriate Tafel behavior is observed. Due to the reversibility of the system, we replot the measurements in an Allen–Hickling plot in Figure 6 b. Allen–Hickling proposed a rearrangement of the Butler–Volmer equation (Equation ( 5) ) which ignores the reverse reaction so as to isolate the forward reaction [ 45 ]

α η−⎡⎣⎢

⎤⎦⎥ = −

η

ln /(1 ) lni0 0i eF

RT

F

RT

( 5)

From the y -axis intercept in Figure 6 b, the exchange current is determined and displayed in Table 1 . Normalized charge



Figure 4. a) CVs at various sweep rates for the FeS 2 -LTS system. The asterisk denotes anodic peak for analysis in (d). b) CVs at various sweep rates for standard system. Clearly there is a large voltage shift in the fi rst cathodic peak for higher sweep rates. This remains in stark contrast to the FeS 2 -LTS system and highlights the advantage of LTS. c) Overlap of 0.1 mV s −1 sweep rates for the two systems. d) Peak current of second anodic peak (asterisk) versus square root of the sweep rate for the two systems.

FULL

PAPER

© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim1600495 (6 of 10) wileyonlinelibrary.com Adv. Energy Mater. 2016, 6, 1600495


Figure 5. a) GITT of FeS 2 -LTS system over third discharge and third charge. b) GITT of FeS 2 standard system over third discharge and third charge. C/20 current pulses are used for 1-h increments followed by a 6-h relaxation. Both tests are at 60 °C. Squares are relaxed voltages and triangles are calculated diffusion coeffi cients. While both tests yield relatively same diffusion coeffi cients, they differ signifi cantly at initial lithiation as only LTS is utilized during these stages.

Figure 6. a) Tafel plot of both FeS 2 -LTS and standard systems during lithiation in the S redox plateau (intercalation) and Fe redox (conversion).b) Allen–Hickling plot with intercepts on vertical axis yielding exchange current. c) Exchange current as a function of temperature. Vastly superior charge transfer is exhibited in the FeS 2 -LTS system. Computed resistances are in Table 1 .

FULL P

APER


transfer resistance is calculated using the following relationship while additionally dividing by the mass of the electrode [ 39 ]

=CT

0

RRT

Fi

( 6)

In Figure 6 b, the relative measurements between the two systems have similar slopes; this outcome is expected as this represents the FeS 2 reaction barrier and should be analogous for both systems. To the best of our knowledge, we have not found any Tafel analyses performed on an FeS 2 material or a conversion material for that matter. However, exchange cur-rents have been measured between a solid electrolyte and metallic interface. [ 41,42 ] These reported exchange currents are higher than what is measured in this study, but rationalized due to known facile kinetics for metallic stripping and deposi-tion. Exchange currents measured for the intercalation cathode material, LiFePO 4 , are very similar to this study however. [ 44 ]

Exchange current and thus charge transfer resistance meas-urements were taken for temperatures between 60 and 90 °C and are plotted in Figure 6 c. Plots using an Arrhenius rela-tionship between temperature and inverse resistance have been used previously to extract a charge transfer activation energy. [ 46,47 ] Tabulated activation energies are displayed in Table 1 . In Table 1 , the most drastic difference between the two systems is that the FeS 2 -LTS system is over half as resistive as the standard system. In both cases, the charge transfer resist-ance is more profound for the conversion region of the system. The conversion region generally demonstrates more sluggish kinetics than the intercalation portion and therefore should decrease the exchange current. Although the precise reason for the smaller charge transfer resistance in the FeS 2 -LTS is diffi cult to specifi cally assign, we postulate that this is due to the enhanced contact area for charge transfer to occur. As indi-cated by Figure 2 , the FeS 2 is almost completely encapsulated in LTS promoting near theoretical charge transfer whereas the standard system will not be able to achieve this connection. The charge transfer activation energies are greater for the conver-sion regions as compared to the intercalation region. It should be noted that the experimentally derived activation energies are quite a bit smaller than those measured for liquid sys-tems. [ 41,46 ] In a solid system, there is no solvation and desolva-tion of the ion, nor specifi c adsorption. Thus a smaller energy barrier is incurred and charge transfer resistance is much less than a solvated interface, chronicling the benefi ts of solid–solid interfaces. [ 47 ] High charge transfer resistance in the standard system at end of lithiation could be attributed to differences

in concentration gradients developed in LTS versus the solid electrolyte. Future work will have to be pursued in a thin-fi lm format, where near theoretical interfaces can be formed and more accurate measurements of charge transfer can occur. [ 48 ]

Figure 7 a is a rate study of the FeS 2 -LTS and standard sys-tems. Rate is symmetrically increased every three cycles from C/10 to 4C with C/10 recovery cycles allowed between rate changes. Corresponding voltage profi les are displayed in Figure 7 c,d along with differential capacity plots for the FeS 2 -LTS and standard systems, respectively. The FeS 2 -LTS cathode achieves near theoretical capacity (567 mAh g −1 electrode) at a rate of C/10. Additionally, no electrolyte activation is notice-able in the LTS system, whereas in the standard system, it has been well documented that electrolyte activation adds addi-tional capacity manifesting in a gradual capacity increase. [ 49 ] The LTS system achieves greater capacity with better stability at higher rates—the voltage profi les provide feasibility evidence. At higher rates, the superior diffusion coeffi cient in LTS acts to “absorb” much of the current shock to the system, expressed in a smoothing effect at immediate application of current. In the standard system, a sharp ohmic drop is perceived followed by a recovery. The smoothing effect is a product or both the lower activation overpotential (as probed by CV) and higher diffu-sion coeffi cient (from GITT). Larger capacities arise from the presence of the iron redox (conversion) reaction even at higher rates. The conversion plateau is present in the LTS system even at a rate of 2C. Lower charge transfer resistance exhibited in this region, as detailed in the Tafel analysis, provides an expla-nation. One way to visualize the drastic differences between the two systems is to subtract out the C/2 voltage profi le from the C/10 voltage profi le for each system. This is displayed in Figure 7 b. In doing so, the relative polarization between high and low rates is observed. The LTS system has a smaller ohmic overpotential (initial capacity) and despite having a small over-potential during the sulfur redox (intercalation) region, full utilization actually occurs as the overpotential drops to zero. Finally, only a gradual rise in polarization occurs at the fi nal stages of lithiation in the LTS system due to a smaller charge transfer resistance. Contrastingly, the standard system has a steep tail at fi nal stages of lithiation due to a greater charge transfer resistance.

Long-term cycling of the two systems at a high rate of C/2 is presented in Figure 8 a. The standard system undergoes a rapid capacity fade, common to the FeS 2 active material, and is attributed to discharge product isolation and poor recombina-tion. [ 34,40,49,50 ] The FeS 2 -LTS system, on the other hand, is able to maintain 62% capacity over 500 cycles—an unprecedented value for a solid-state, high capacity electrode at high rates. Figure 8 b,c shows the evolving voltage profi les of the FeS 2 -LTS and standard systems, respectively. The LTS system maintains the conversion region throughout the course of cycling. The standard system loses the lower plateau and then gradually loses capacity from the upper plateau. This results from active material isolation of FeS 2 due to electrode heterogeneity, impli-cating that FeS 2 cannot reform upon charging and the majority of retainable capacity transitions to a sulfur active material. [ 49 ] The delithiation profi le of the FeS 2 -LTS system regains a higher proportion of capacity during the 3 V hold than the standard system—a behavior explained by results in Figure 5 . The



Table 1. Results of Tafel analysis on FeS 2 -LTS and standard systems. Exchange current at 60 °C, charge transfer resistance, and charge transfer activation energy are tabulated.

i o,60C [mA mg −1 ]

R ct [Ω mg −1 ]

E a [eV atom −1 ]

FeS2-LTS intercalation 0.197 145 25.0

FeS2-LTS conversion 0.162 172 30.7

Standard intercalation 0.085 367 18.0

Standard conversion 0.079 427 20.7

FULL

PAPER



diffusion coeffi cient is the same on the charging cycles as the FeS 2 delithiation voltage shifts to higher potentials than the del-ithiation of TiS 2 . Thus, higher voltages are necessary to com-pletely delithiate the system.

An interesting aspect of this new electrode design is the ability to recover capacity. Since active material isolation does not occur, as well as minimal chemical decompositions, once the FeS 2 -LTS system is slowed back down to C/10 after 500 cycles (Figure 8 d), nearly all capacity is regained. The standard system can only recover a small portion of this capacity. The ability to recover capacity is of monumental impact. Essentially, this gives the possibility of a “recyclable” battery in the sense that simply by slowing down the system, an original electrode design can be recovered. This will be studied in the future as this possibility could provide a paradigm shift in discovering signifi cantly more sustainable batteries.

3. Conclusion

In this study, we present a new solid-state electrode design. By using a mixed-conducting matrix, it is possible to remove the solid electrolyte and conductive additive yielding many benefi ts. The combination of FeS 2 and LiTiS 2 has the capabilities of both a high power and high-energy system. The LiTiS 2 provides high diffusivity during initial stages of lithiation. Charge transfer is greatly enhanced moving from a three-material interface to a two-material interface. Through a Tafel analysis, the exchange

current of the system was shown to effectively double. By devel-oping a possible solution to active material isolation and cre-ating a more homogenous electrode design, cycling at a high rate of C/2 for 500 cycles is attainable. Additionally, the elec-trode can recover full capacity simply by reducing system rate. Capacity recovery implicates a lack of active material isolation, a common problem in solid-state batteries.

4. Experimental Section All procedures outlined were conducted in a dry argon environment. The 77.5 Li 2 S-22.5 P 2 S 5 solid electrolyte (denoted as a77.5) was prepared by planetary ball-milling Li 2 S (Alfa, 99.9%) and P 2 S 5 (Sigma, 99%) in a 77.5 to 22.5 molar ratio for 20 h in a 500 mL stainless steel jar. The standard cathode was prepared by hand-mixing FeS 2 (Washington Mills, SULFEX Red), a77.5, and C65 (Timcal Super C65) in a 5:5:1 weight ratio. A large enough batch was prepared to use for all experiments. The FeS 2 -LTS cathode was prepared by planetary ball-milling FeS 2 , TiS 2 (Sigma, 99.9%), and Li 3 N (Alfa, 99.4%) in a 3.1:3:1 molar ratio for 20 h in a 100 mL agate jar. The theoretical capacity was therefore 51% weight FeS 2 (894 mAh g −1 ) and 49% weight LiTiS 2 (227 mAh g −1 ) yielding 567 mAh g −1 .

A reinforced cell die was used for all electrochemical tests. [ 51 ] 150 mg of a77.5 was pressed at 75 MPa as a separator. Either the standard or FeS 2 -LTS cathode was pressed onto one side of the separator at 375 MPa (3.75 mg cm −2 for cycling and rate study; 2.25 mg cm −2 for GITT; 0.75 mg cm −2 for CV and Tafel). Concurrently, lithium metal foil (CV, GITT, Tafel) or an indium-lithium alloy (In + Li x In, 0< x <1) [rate study and cycling] was pressed onto the other side of the separator as a counter electrode. Cyclic voltammetry and Tafel analysis were carried

Figure 7. a) Rate study of FeS 2 -LTS and standard systems. FeS 2 -LTS achieves theoretical capacity at a C/10 rate. The system also shows high stability at extreme rates whereas the standard system shows poor stability at rates greater than C/5. b) Polarization of C/2 with respect to C/10 voltage profi les for both systems. The standard system has a greater ohmic polarization at low capacities and charge transfer polarization at high capacities. c) Discharge voltage profi les of the FeS 2 -LTS system with respective d Q /d V values (C/10–C/2) on the right-hand side. Not only sharp plateaus are seen at all rates, but the Fe redox plateau (lower) is present even at a 2C rate. d) Discharge voltage profi les of the FeS 2 standard system with respective d Q /d V values (C/10–C/2) on the right-hand side. Fe redox plateau disappears at rates greater than C/2.

FULL P

APER

© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim (9 of 10) 1600495wileyonlinelibrary.comAdv. Energy Mater. 2016, 6, 1600495


out on a Solartron 1280C Electrochemical Test System. Tafel analysis, more specifi cally, took place on the third cycle discharge by holding the cell at either 2.17 V (intercalation) or 1.53 V (conversion) for 1 h; then a linear sweep was performed at 1 mV s −1 between ±150 mV around open circuit voltage. GITT and cycling were performed using an Arbin

BT2000 Battery Test Station. GITT was performed by using a rate of C/20 for 1 h followed by a 6 h relaxation. An aerial current density of 0.21 mA cm −2 corresponded to a rate of C/10. The rate study and long-term cycling used an operating voltage window of 3–1 V (vs Li + /Li). All electrochemical tests were performed at 60 °C unless otherwise noted; cells were allowed 3 h of temperature acclimating before tests were performed.

XRD patterns were collected with Cu K α radiation ( λ = 1.5418 Å) in the 2 θ range of 25°−50°, using a Bruker AXS D2 Phaser benchtop XRD system operated at 30 kV and 30 mA. A Lynxeye XE 1D detector with a step size of 0.02° and collection time of 1 s per step was employed. Sample displacement was corrected by using a pure corundum internal standard.

For the detailed observation of uncycled and tenth cycle composite electrodes, assembled cells were extruded from the pressure die. Samples cross-sections were prepared using an FIB (FEI NOVA200 dual beam system) machine equipped with air-lock system. The air-lock system enabled the composite electrode to remain in a vacuum state while samples were loaded from the glove box to the FIB chamber. HR-TEM (JEOL 3000F) was then performed on the fi ne-milled cross-sections. These techniques have been detailed in previous studies. [ 52 ]

Acknowledgements This work was supported by the National Science Foundation (NSF, CHE-1231048).

Received: March 3, 2016 Revised: April 7, 2016

Published online: June 6, 2016

[1] a) T. Inada , K. Takada , A. Kajiyama , H. Sasaki , S. Kondo , M. Watanabe , M. Murayama , R. Kanno , J. Power Sources 2003 , 119 , 948 ; b) T. Inada , T. Kobayashi , N. Sonoyama , A. Yamada , S. Kondo , M. Nagao , R. Kanno , J. Power Sources 2009 , 194 , 1085 .

[2] a) M. Ogawa , R. Kanda , K. Yoshida , T. Uemura , K. Harada , J. Power Sources 2012 , 205 , 487 ; b) T. Uemura , K. Goto , M. Ogawa , K. Harada , J. Power Sources 2013 , 240 , 510 .

[3] J. M. Whiteley , J. W. Kim , C. S. Kang , J. S. Cho , K. H. Oh , S.-H. Lee , J. Electrochem. Soc. 2015 , 162 , A711 .

[4] M. J. G. Jak , E. M. Kelder , M. Stuivinga , J. Schoonman , Solid State Ionics 1996 , 86–8 , 897 .

[5] Z. Q. Wang , M. S. Wu , G. Liu , X. L. Lei , B. Xu , C. Y. Ouyang , Int. J. Electrochem. Sci. 2014 , 9 , 562 .

[6] M. N. Obrovac , L. Christensen , Electrochem. Solid State Lett. 2004 , 7 , A93 .

[7] J. M. Whiteley , J. H. Woo , E. Hu , K.-W. Nam , S.-H. Lee , J. Electrochem. Soc. 2014 , 161 , A1812 .

[8] F. Han , T. Gao , Y. Zhu , K. J. Gaskell , C. Wang , Adv. Mater. 2015 , 27 , 3473 .

[9] S. P. Ong , Y. F. Mo , W. D. Richards , L. Miara , H. S. Lee , G. Ceder , Energy Environ. Sci. 2013 , 6 , 148 .

[10] A. Kuhn , O. Gerbig , C. Zhu , F. Falkenberg , J. Maier , B. V. Lotsch , Phys. Chem. Chem. Phys. 2014 , 16 , 14669 .

[11] T. Inada , K. Takada , A. Kajiyama , M. Kouguchi , H. Sasaki , S. Kondo , M. Watanabe , M. Murayama , R. Kanno , Solid State Ionics 2003 , 158 , 275 .

[12] a) B. D. Katz , L. C. De Jonghe , M.-Y. Chu , S. J. Visco ( Polyplus Battery Company, Inc. ), US6200704 B1 , 2001 ; b) L. Bai , C. Li , A. A. Anani , G. Thomas , H. Wu , K. K. Lian , F. R. Denton Iii , J. N. Howard ( Motorola, Inc. ), US5744258 A , 1998 .

[13] Z. W. Seh , J. H. Yu , W. Y. Li , P. C. Hsu , H. T. Wang , Y. M. Sun , H. B. Yao , Q. F. Zhang , Y. Cui , Nat. Commun. 2014 , 5 , 5017 .

Figure 8. a) Cycling at C/2 (1.05 mA cm −2 ) for FeS 2 -LTS and standard cathodes. b) Voltage profi les for FeS 2 -LTS electrode over cycling. c) Voltage profi les for standard electrode over cycling. d) Fifth cycle C/10 voltage profi les and recovered cycle at C/10 following long-term cycling at C/2 for both electrodes.

FULL

PAPER



[14] D. Marmorstein , T. H. Yu , K. A. Striebel , F. R. McLarnon , J. Hou , E. J. Cairns , J. Power Sources 2000 , 89 , 219 .

[15] F. Badway , A. N. Mansour , N. Pereira , J. F. Al-Sharab , F. Cosandey , I. Plitz , G. G. Amatucci , Chem. Mater. 2007 , 19 , 4129 .

[16] T. A. Yersak , H. A. Macpherson , S. C. Kim , V. D. Le , C. S. Kang , S. B. Son , Y. H. Kim , J. E. Trevey , K. H. Oh , C. Stoldt , S. H. Lee , Adv. Energy Mater. 2013 , 3 , 120 .

[17] R. Fong , J. R. Dahn , C. H. W. Jones , J. Electrochem. Soc. 1989 , 136 , 3206 .

[18] V. Pele , F. Flamary , L. Bourgeois , B. Pecquenard , F. Le Cras , Electrochem. Commun. 2015 , 51 , 81 .

[19] M. S. Whittingham , Science 1976 , 192 , 1126 . [20] A. J. Vaccaro , T. Palanisamy , R. L. Kerr , J. T. Maloy , Solid State Ionics

1981 , 2 , 337 . [21] C. W. Lin , X. J. Zhu , J. Feng , C. Z. Wu , S. L. Hu , J. Peng , Y. Q. Guo ,

L. L. Peng , J. Y. Zhao , J. L. Huang , J. L. Yang , Y. Xie , J. Am. Chem. Soc. 2013 , 135 , 5144 .

[22] P. P. Chin , J. Ding , J. B. Yi , B. H. Liu , J. Alloys Compd. 2005 , 390 , 255 . [23] a) B. G. Silbernagel , Solid State Commun. 1975 , 17 , 361 ;

b) M. S. Whittingham , Prog. Solid State Chem. 1978 , 12 , 41 . [24] T. A. Yersak , J. E. Trevey , S. H. Lee , J. Power Sources 2011 , 196 , 9830 . [25] T. A. Yersak , Y. F. Yan , C. Stoldt , S. H. Lee , ECS Electrochem. Lett.

2012 , 1 , A21 . [26] R. Winter , P. Heitjans , J. Phys. Chem. B 2001 , 105 , 6108 . [27] A. Jain , G. Hautier , C. J. Moore , S. P. Ong , C. C. Fischer , T. Mueller ,

K. A. Persson , G. Ceder , Comput. Mater. Sci. 2011 , 50 , 2295 . [28] 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. 2012 , 2 , 1226 .

[29] A. J. Bard , L. R. Faulkner , Electrochemical Methods , Wiley , Hoboken, NJ, USA 2001 .

[30] W. Weppner , R. A. Huggins , J. Electrochem. Soc. 1977 , 124 , 1569 . [31] C. J. Wen , B. A. Boukamp , R. A. Huggins , W. Weppner ,

J. Electrochem. Soc. 1979 , 126 , 2258 . [32] S. B. Son , T. A. Yersak , D. M. Piper , S. C. Kim , C. S. Kang , J. S. Cho ,

S. S. Suh , Y. U. Kim , K. H. Oh , S. H. Lee , Adv. Energy Mater. 2014 , 4 , 1300961 .

[33] B. R. Shin , Y. J. Nam , J. W. Kim , Y.-G. Lee , Y. S. Jung , Sci. Rep. 2014 , 4 , 5572 .

[34] T. Evans , D. M. Piper , S. C. Kim , S. S. Han , V. Bhat , K. H. Oh , S.-H. Lee , Adv. Mater. 2014 , 26 , 7386 .

[35] E. Kendrick , J. Barker , J. Bao , A. Swiatek , J. Power Sources 2011 , 196 , 6929 .

[36] R. Brec , A. Dugast , A. Lemehaute , Mater. Res. Bull. 1980 , 15 , 619 .

[37] a) P. Bron , S. Johansson , K. Zick , J. S. auf der Gunne , S. Dehnen , B. Roling , J. Am. Chem. Soc. 2013 , 135 , 15694 ; b) A. Van der Ven , J. Bhattacharya , A. A. Belak , Acc. Chem. Res. 2013 , 46 , 1216 .

[38] H. I. Yoo , J. H. Lee , M. Martin , J. Janek , H. Schmalzried , Solid State Ionics 1994 , 67 , 317 .

[39] J. Tafel , Z. Physik. Chem. 1905 , 50A , 641 . [40] T. A. Yersak , T. Evans , J. M. Whiteley , S. B. Son , B. Francisco ,

K. H. Oh , S. H. Lee , J. Electrochem. Soc. 2014 , 161 , A663 . [41] M. Chiku , W. Tsujiwaki , E. Higuchi , H. Inoue , Electrochemistry 2012 ,

80 , 740 . [42] M. Chiku , W. Tsujiwaki , E. Higuchi , H. Inoue , J. Power Sources 2013 ,

244 , 675 . [43] P. Bai , M. Z. Bazant , Nat. Commun. 2014 , 5 , 3585 . [44] H. Munakata , B. Takemura , T. Saito , K. Kanamura , J. Power Sources

2012 , 217 , 444 . [45] P. L. Allen , A. Hickling , Trans. Faraday Soc. 1957 , 53 , 1626 . [46] T. Abe , H. Fukuda , Y. Iriyama , Z. Ogumi , J. Electrochem. Soc. 2004 ,

151 , A1120 . [47] T. Abe , F. Sagane , M. Ohtsuka , Y. Iriyama , Z. Ogumi , J. Electrochem.

Soc. 2005 , 152 , A2151 . [48] M. Gellert , K. I. Gries , J. Zakel , S. Kranz , S. Bradler , E. Hornberger ,

S. Muller , C. Yada , F. Rosciano , K. Volz , B. Boling , J. Electrochem. Soc. 2015 , 162 , A754 .

[49] T. A. Yersak , C. Stoldt , S. H. Lee , J. Electrochem. Soc. 2013 , 160 , A1009 .

[50] J. M. Whiteley , P. Taynton , W. Zhang , S.-H. Lee , Adv. Mater. 2015 , 27 , 6922 .

[51] D. M. Piper , T. A. Yersak , S. H. Lee , J. Electrochem. Soc. 2013 , 160 , A77 .

[52] S. B. Son , J. E. Trevey , H. Roh , S. H. Kim , K. B. Kim , J. S. Cho , J. T. Moon , C. M. DeLuca , K. K. Maute , M. L. Dunn , H. N. Han , K. H. Oh , S. H. Lee , Adv. Energy Mater. 2011 , 1 , 1199 .

Documents

FeS 2 ‐Imbedded Mixed Conducting Matrix as a Solid …engineering.snu.ac.kr/pdf/2016/2016_HSS_FeS 2 -Imbedded... · 2017-01-13 · eS F 2-Imbedded Mixed Conducting Matrix as a Solid