Methods and Protocols for Electrochemical Energy Storage MaterialsResearchElahe Talaie,† Patrick Bonnick,† Xiaoqi Sun,† Quan Pang,† Xiao Liang,† and Linda F. Nazar*

Department of Chemistry and the Waterloo Institute of Nanotechnology, University of Waterloo, 200 University Avenue, Waterloo,Ontario N2L 3G1, Canada

ABSTRACT: We present an overview of the procedures and methods to prepare and evaluate materials for electrochemical cellsin battery research in our laboratory, including cell fabrication, two- and three-electrode cell studies, and methodology forevaluating diffusion coefficients and impedance measurements. Informative characterization techniques employed to assess newmaterials for batteries are also described, including operando XRD, pair-distribution function analysis, X-ray photoelectronspectroscopy, and operando X-ray absorption spectroscopy. Examples of insightful information that each technique has providedin the research areas of Li-S, Na-ion, and Mg batteries are presented along with excellent references for detailed descriptions ofthe theory, experimental procedures, and various designs, as well as methods for data processing and analysis.

■ INTRODUCTION

Development of high-performance and cost-effective energystorage technologies is an attractive subject of research todaydue to the growing energy demand and the necessity ofproviding it from clean and renewable sources. Efficient use ofelectricity generated from renewable energy sources relies onthe development of sustainable and low-cost energy storagesystems to integrate the energy from intermittent sources, suchas wind and solar energy, into the electrical grid; electro-chemical energy storage systems, and in particular batteries,have emerged as promising candidates to meet this demand.1,2

Lithium-ion batteries are extensively employed in portableelectronic devices, and their application has expanded to theelectric vehicle market, which is expected to increase rapidly.3

Today, various battery technologies beyond Li-ion batteries,such as lithium-sulfur (Li-S),4−6 sodium-ion (Na-ion),7−10 andmagnesium (Mg)11−13 batteries, have gained much attentionfrom the research community due to their advantageoussustainability, cost-effectiveness and/or high capacity, althoughmany challenges have to be overcome for them to competewith state-of-the-art Li-ion batteries. Development of high-performance battery systems requires the exploration of newpromising materials and an in-depth understanding of thescience underlying these devices, such as the redox reactions,the interaction of the electrodes with the electrolyte, and agingmechanisms of each cell component. Characterization of thechemical composition, morphology, crystal structure, and theelectronic structure of electrode materials, and monitoring the

evolution of those properties during the operation of theelectrochemical cell provide a comprehensive evaluation of newelectrode materials.Considering the inevitable importance of advanced character-

ization techniques for battery technology research anddevelopment, this paper presents an introduction to some ofthe most important characterization techniques commonlyemployed for battery evaluation. This paper is not a detailed orcomprehensive description of the characterization techniques ofinterest; instead, the goal of this Review is to present aresourceful starting point for researchers new to the field.Electrode preparation, coin cell assembly, electrochemicalevaluation techniques, operando X-ray diffraction, operandopair distribution function analysis, operando X-ray absorptionspectroscopy, and X-ray photoelectron spectroscopy techniquesare discussed, and an example of the application of eachcharacterization technique in our research is presented.

■ ELECTROCHEMISTRY

Electrode Preparation. Depending on the active materialproperties and the experiment purposes, electrode preparationcan be carried out by either doctor blading or drop casting.

Special Issue: Methods and Protocols in Materials Chemistry

Received: July 4, 2016Revised: August 30, 2016Published: September 4, 2016

Review

pubs.acs.org/cm

© 2016 American Chemical Society 90 DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

Doctor blading requires at least one hundred milligrams ofactive material and gives a more homogeneous coating, whereasdrop casting can be done with 10 times less material and isnormally used for early-stage testing of new materials. Bothprocedures are described in detail below.During either method of electrode preparation, additives

binder are mixed together with the active material in order toincrease the conductivity, elasticity, and viscidity. A commoninitial weight ratio of these three components is active material:carbon:binder = 8:1:1. However, these numbers should bemodified based on the specific properties of the active material;for example, more carbon can be added to aid non-conductiveactive materials. A slurry is made by grinding the above mixturein a solvent that has a high solubility for the binder. N-Methyl-2-pyrrolidone (NMP) is normally used with polyvinylidenefluoride (PVDF) binder, for instance. For the doctor bladingprocedure, the slurry should be relatively “thick”, with a 25−30:1 weight ratio of solvent to the binder as a good startingpoint. The homogeneous mixture is then cast by depositing a“tooth-shaped puddle” onto a large piece of metal or carbon foilthat serves as the current collector, and gliding a doctor bladeover the puddle and down the foil to spread out the slurry.14

The foil is then placed in an oven at, for example, 60 °C for 2 hto evaporate the solvent. Afterward, the cast spread can befurther pressed by passing it through a cold or hot roller pressto achieve better contact.14 Finally, small electrodes are cutfrom the cast spread, usually by using an electrode “punch” topunch out circular electrode disks. The active material loadingof each electrode is determined by weighing “blank” disks of thecurrent collector foil without “electrode material” on them andfinding the average mass. Then this average current collectormass is subtracted from the mass of an electrode to determinethe mass of electrode material present. That mass is thenmultiplied by the percentage of the electrode material that isactive material (80% in our example 8:1:1 mixture) todetermine the active material loading.20

For drop casting, the slurry is added dropwise to theindividual pre-cut current collectors. Thus, the weight of eachcurrent collector can be weighed beforehand to acquire anaccurate loading. Almost all of the slurry can be transferred tothe current collectors during drop casting so that only a smallamount of active material is required. For example, if oneintends to make a batch of 10 electrodes with a loading of ∼5mg/electrode, only ∼50 mg of active material powder isneeded. The slurry is relatively “thin” for drop casting, meaningthat more solvent is needed compared to the doctor bladingmethod. Typically 5−6 drops are required for a 16 mmdiameter current collector, which means ∼80 μL of slurry/electrode. However, this value will fluctuate based on thesolvent and pipet type. The electrodes are then dried in an ovento remove solvent and moisture. Note that it is relativelycomplicated to pass the small electrodes through a roller press,thus this step is normally omitted for drop casting.A very important consideration for handling and storage of

materials and preparing the electrodes is their air stability. Forsome materials, the reactivity with ambient atmosphere impactsvarious properties, including the electrochemical performance.Layered sodium transition metal oxides, in particular, areknown to be unstable in air; insertion of water molecules,uptake of oxide ions, and reactivity with carbon dioxide aredifferent reported mechanisms.15−19 In order to prepare air-protected samples in our Na-ion battery research, the as-prepared powder samples, which are synthesized in air, are

subjected to a heat treatment (at 600 °C for Na0.67[Mn0.5Fe0.5]-O2) under vacuum or an inert atmosphere and then aretransferred to an argon-filled glovebox. All the steps of theelectrode preparation and the cell assembly are performedwithin the glovebox.

Coin Cell Assembly. In addition to their application ascommercial primary batteries, coin cells have been used forresearch on rechargeable Li and Li-ion batteries for the last 30years. Their benefits include being small, fast and easy toassemble, and the feasibility of testing of small amounts ofnovel active materials.20 Therefore, coin cells have been usedfor initial work to identify high-performing materials, inaddition to modifying the compositions, fabrication parameters,and cycling conditions in laboratories.21 There are many typesof coin cells. Their number is based on their dimensions; e.g.,CR2325 represents a cell with a 23 mm diameter and a 2.5 mmthickness. For the purposes of consistency, all the cells used inour laboratory are of the CR2325 type.Figure 1 is a schematic representation of a typical coin cell

assembly in our laboratory. The cell caps and spacers are made

from 430 BA stainless steel (NRC, Canada). Using a non-corrodable stainless steel for the casing and the spring isextremely important. The spring is made from 301 SS,providing loading forces of up to 12 pounds at a deflectionof 1 mm. The separator is either Celgard 3501, made frommicroporous polypropylene monolayer membrane, or glassfiber. The anode material used in our studies includes Li, Na orMg metal, depending on the cathode active materials. Both theanode and cathode are punched to discs with 16 mm or 11 mmin diameter. Precise control of the electrolyte volume isnecessary due to the limited space in the coin cell (theelectrolyte volume should be less than 100 μL according to ourexperience). Particularly for the Li-S battery, the performance ishighly affected by the amount of the electrolyte. The electrolytemeasured by a micropipette is dropped on the separator beforethe anode is placed in the cell. The coin cells are sealed by acrimping machine (MTI Corporation) in an Ar-filled golveboxafter all the components are stacked and aligned as shown inFigure 1. Cells should be conditioned for at least 2 h in a staticstate at room temperature prior to electrochemical inves-tigation, enabling the electrolyte to wet the electrodes andseparator.

Electrochemical Performance Evaluation. Electro-chemically charging and discharging a new electrode materialis the most common and important experiment carried out in

Figure 1. Schematic representation of a coin cell.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

91

the pursuit of better batteries and is generally the firstexperiment that researchers look for when reading a paperabout a new material. Charging an electrode and thendischarging it comprises one cycle, with either the chargestep or discharge step being referred to as a half-cycle. In thelaboratory, cycling is usually carried out using a constantcurrent (i.e., galvanostatically), although other chargingmethods exist that are more appropriate in certain situations.22

Any charge or discharge half-cycle must be terminated after acertain amount of time (or charge) or when the workingelectrode reaches a particular potential. If a time (or charge)limit is used, overcharging or overdischarging commonly occursafter several cycles; over(dis)charging is the continuation of(dis)charge past the point where all of the active material hasbeen (dis)charged, and some other cell component, like theelectrolyte, starts to break down in order to absorb or provideelectrons. When using non-aqueous electrolytes, potentiallimits are usually used to avoid overcharging or overdischargingsince this usually reduces the state of health of the cell (i.e., howmany cycles it is still capable of). In aqueous cells overchargingor overdischarging usually consumes H2O to produce either H2or O2 gas, which can be safely vented, thereby allowing time (orcapacity) limits to be used.20

The term “C-rate” is commonly used in the battery researchcommunity to represent a constant current. The C-rate is theexpected capacity, C, of the working electrode divided by thedesired number of hours to fully (dis)charge that capacity (i.e.,a half-cycle). The expected capacity can be calculated using themass of active material in the electrode and multiplying by thespecific capacity (mAh/g) of the active material, or measuredby cycling the cell once or twice and then declaring the ‘initial’discharge capacity to be the expected capacity (at a specificrate). Researchers can use either, but they must define in theirpublications which method they have used. As an example, aworking electrode with a mass of active material correspondingto 2 mAh that is to be discharged in 4 h would be discharged ata C/4 rate, which is equal to 2 mAh/4 h = 0.5 mA in this case.The cell might finish discharging in only 3 h (conveyinginstantly to the reader that it achieved 75% of its expectedcapacity) but it is still understood to have been discharged at aC/4 rate by the definition chosen in this case. Reportingcurrents in C-rates aids readers by inherently compensating thecurrent for the capacity of the electrode. Note that a C-rate iseffectively the same as mA/gactive material. Currents can also bereported in mA/cm2, which is appropriate when the surfacearea of the electrode is a parameter being varied, or aparticularly large or small surface area is being used.The primary figure that readers are usually interested in is the

discharge/charge curve, which is a plot of potential vs time,specific capacity or a stoichiometric variable, such as x inMgxTi2S4. An example is shown in Figure 2.23 The length ofthis curve is a measure of the charge or discharge capacity and,for a particular material, is dependent on the temperature, rate(/cm2) and state of health of the electrode (i.e., history). TheC-rates in Figure 2 were calculated using the mass of activematerial in the electrodes and the assumption that 1 Mg2+ perTi2S4 unit could be intercalated. Figure 2 shows that highercapacities are achieved at higher temperatures, lower rates, andlower cycle numbers. Reporting the effect of each of these threeparameters on the capacity and shape of the potential curve isrecommended for a new material. The shape of this curve isdetermined by both the thermodynamics and kinetics of solelythe working electrode in a three-electrode cell, or both the

working and counter electrode in a two-electrode cell (seebelow). It is best to know what the potential curve of eachelectrode looks like individually, so at least one three-electrodemeasurement of a new electrode is recommended unless thecounter electrode in the two-electrode cell is highly non-polarizable. In general, a flat plateau in the potential curvesignifies a two phase reaction (with a constant chemicalpotential), whereas a sloped “plateau” (i.e., changing chemicalpotential) signifies a change in solution concentration (e.g., thesolid solution formed via intercalation of Mg2+ in Ti2S4 inFigure 2). A steep decline (or rise) in the potential signifies thekinetic limit at which a reaction is no longer capable ofabsorbing the constant current being forced out of (or into) theelectrode.24 Two types of transitions are common inintercalation materials: ordering25,26 and staging.27,28 Some-times a slight drop in potential between two sloped plateaus,like the drop at about 110 mAh/g in Figure 2a, is due to anorder/disorder transition in an otherwise solid solution.These distinct features in the potential curve are more easily

seen in a differential capacity (dQ/dV) plot, which is theinverse of the derivative of the potential vs specific capacitycurve discussed above, plotted against the potential. Anexample for a LixMn2O4 electrode is shown in Figure 3.29

Plateaus in the V vs Q curve appear as peaks in the dQ/dV plot;the flatter the plateau, the narrower the dQ/dV peak. Each peakfeature represents a different chemical reaction or intercalationenvironment (in the case of ordering and staging) and the areaunder the peak is the capacity of that particular process. Tothose familiar with cyclic voltammograms (CV) this may soundsimilar, and in fact, a CV yields the same information but underdifferent circumstances. The y-axis in a typical CV is current, I,which can be converted to dQ/dV by dividing by the sweep rate(ν = dV/dt):

ν= =I Q t

V tQV

d /dd /d

dd (1)

Figure 2. First and second cycles of spinel MgxTi2S4/APC/Mg coincells at various current rates, and either 60 or 20 °C. The dischargehalf-cycles are the bottom half of each “loop” and begin at the top leftof the plot, progressing down toward the bottom right. Charge half-cycles are the upper half of each “loop” and begin at the bottom right,progressing up toward the top left. Adapted from ref 23.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

92

The difference between the methods is that the differentialcapacity is collected at a constant current, which gives thevarious electrochemical processes more time to reachcompletion, whereas a CV is collected with a constant potentialsweep rate (i.e., μV/s). This forced change in the potential,regardless of the state of charge of the electrode, can increasethe current beyond the kinetic limit of the cell, therebypreventing some capacity from being accessed. Additionally, theIR drop of the cell remains unchanged throughout the constantcurrent scan but varies throughout a CV sweep, which alters thepeak shape. These effects combine to produce the two curves inFigure 3 from the same electrode. Note the superior resolutionof the dQ/dV plot in comparison to the CV. The advantage of aCV is that it contains rate information about the full cell, so thismethod might be preferred by researchers testing electrolytes,separators or electrode architectures, for example. Researchersfocused on electrode materials usually use constant currentdifferential capacity plots. When converting a V vs Q plot into adQ/dV vs V plot, it is common to end up with “noisy” data. Tocollect smoother data, try (1) tightly controlling the temper-ature of the cell, (2) collecting a data point only every few mVas opposed to either every few mV or every few min, and (3)cycling at a very slow rate (e.g., C/100).Another important measurement to perform when character-

izing a new electrode material is the cycle life, which comparescharge and discharge capacities to the cycle number as shownin Figure 4a.30 In cases like this example where the charge anddischarge capacities overlap for the entire plot, it is acceptableto omit the charge data. A cell is considered to be at the end ofits cycle life when its discharge capacity sinks below a criticalthreshold, which is up to the researcher but is usually 80% of itsinitial discharge capacity,31 ignoring the first cycle if a notabledrop in discharge capacity occurs from the first to second cycle.For example, the Li-S cell in Figure 4a has a first dischargecapacity of 1100 mAh/g and a second cycle discharge capacityof 850 mAh/g. The first discharge capacity was ignored, and sothis cell reached the end of its cycle life when it delivered <680mAh/g around cycle 420. In the case of Li-S cells, performingthe first cycle at a very slow rate (C/20 to C/50) improves thecycle life.30 Other cell chemistries like Ni-Zn also benefit fromperforming the first few cycles of the regime differently than therest,20 but this is chemistry-specific and so a complete list ofrecommendations is not possible here. When reporting thecycle life, as in Figure 4, all cycles up to the 80% end of life

threshold should be reported. Some researchers only report thecycles where the cell is healthy and performing well, butreporting all of the cycles collected better reflects thelimitations of the cell design. Two final thoughts on cycle lifemeasurements: (1) Electrolyte loss (evaporation) through theseals of the cell is one of many possible causes of cell failure, butthis one is commonly overlooked and can be identified byweighing cells before and after cycling. (2) The stability andvalue of the temperature at which the cell was cycled can beimportant and should be reported (the temperature range canbe conveyed with a ± as in the caption of Figure 4).Figure 4 also shows the Coulombic efficiency (CE), which is

the discharge capacity divided by the preceding charge capacityor sometimes vice versa (CE = QD/QC, or sometimes QC/QD).

32,33 The CE generally reveals the presence of parasiticreactions that siphon charge away from the intended reaction,i.e., charging the electrode.33 The exact cause of a CE ≠ 100%varies from system to system, but a large departure (|100% −CE| ≳ 2%) is alarming since either the electrode or theelectrolyte is being consumed at a likely unsustainable rate,although cells with aqueous electrolytes can withstand higherdepartures since H2O can be replaced. In the case of Li-ion cellswith a limited supply of Li+ ions, to deliver 3000 cycles beforereaching a 50% end of cycle life limit the CE must be ≥99.98%

Figure 3. (a) Potential curves of LixMn2O4+δ vs Li metal. (b) Differential capacity plot derived from the potential curves in (a) and a cyclicvoltammogram (CV) of the same electrode. Constant current experiment: C/8 = 0.33 mA/cm2. CV: 0.04 mV/S (sweeps up and down in 6.9 h).Adapted from ref 29.

Figure 4. Example cycle life plot featuring a Li-S cell cycled at a C/2rate between 1.8 and 3.0 V, at 20 ± 2 °C. Only every fifth data point isshown for clarity. Adapted from ref 30. Panel b shows the same CEdata as in a, but with the right y-axis magnified to the point ofusefulness.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

93

(CE3000 = 50%), while 500 cycles still require a CE ≥ 99.86%(assuming an exponential capacity loss with cycle number)since Li lost to parasitic reactions permanently reduces thecapacity of the cell.33 As such, a plot of CE vs cycle with a y-axisscale ranging from 0% to 100% (or similar) hides the trueexpected cycle life of the cell since 99.86% cannot bedistinguished from 99.98%, as shown in Figure 4a. Unfortu-nately, most instruments are not accurate enough to measureCE to ± 0.01%.32 In the absence of a high precision charger,researchers should know the accuracy of their instruments andtemperature control, and report it in addition to using a y-axisscale that allows the CE values to be easily determined, asshown in Figure 4b. Researchers can also use symmetric cells toimprove the accuracy of their CE measurements as described inref 34.Two-Electrode versus Three-Electrode Cells. The

electrochemical measurement can be carried out with twoelectrode or three electrode configurations. In the twoelectrode setup, the material of interest is assigned as theworking electrode (WE), while the other electrode serves asboth the auxiliary electrode (AE, also called counter electrode)and the reference electrode (RE). In order to be a good RE inaddition to an AE, the AE should have a small overpotentialand a stable voltage during electrochemical processes. Lithiumand sodium metals in rechargeable Li and Na cells are goodsuch examples. Positive electrode materials are frequentlystudied in two electrode configurations with the correspondingmetal as the AE/RE. Such a configuration also reflects thebehavior of the entire system and provides a more realisticportrayal of the properties to be expected in a commercialversion of the cell. Thus, two electrode cells are widely used forcycle life evaluation when the specific reactions at eitherelectrode are not the major concern. The readily available two-electrode cell designs include coin cells and Swagelok cells. Dueto the easy assembly and low price of two electrode cells, theyare preferred if a separate reference electrode is not necessary.Since the result given in the two electrode setup is a

combination of WE and AE signals, an approximation is takenhere that the electrochemical response at AE is small enough sothat the overall result represents the properties of the WE.However, sometimes the WE signal needs to be well separatedin order to study the mechanisms in detail, or there is nosuitable material to serve as both the AE and RE. In those cases,an individual RE is required and this can be accomplished inthe three electrode setup. A good RE material should provide astable and accurate voltage. Examples include the silver/silverchloride electrode, saturated calomel electrode, and mercury/mercury oxide electrode for aqueous systems, as well as silver/silver nitrate, or ferrocene for non-aqueous systems. A simplesetup is obtained by immersing the three electrodes in a beakercell. There are also other available three-electrode cell designs,such as the Conflat cell,35 T-shaped Swagelok cell,36 andmodified coin cell,37 where the cells are airtight and only a smallamount of electrolyte is required. Note that special care needsto be taken for impedance measurements because the cellgeometry has a significant influence on the results.38,39

One good example for using a three electrode setup is in Mgbattery research. The limitation of available electrolytes thatboth facilitate Mg deposition and stripping, and resist oxidationat high potentials makes it problematic to couple a high voltageoxide-based positive electrode with a Mg metal negativeelectrode in a Mg full cell. As a result, new positive Mgelectrodes are commonly evaluated in cells using high surface

area carbon as the AE. The charge transfer at the AE occursthrough the double layer capacitance mechanism, which resultsin a sloping voltage curve. Therefore, a RE is required tomeasure the voltage response of the positive (working)electrode. A three-electrode Conflat cell was used for testingMnO2 as a new Mg positive electrode.40 A passivation layer isreadily formed on the surface of Mg, so it may experience alarge overpotential (over 2 V) when plating/stripping in theMg(TFSI)2 in the diglyme electrolyte used in this study; thisvalue is largely reduced when extremely low current is involved,nonetheless, making Mg metal a passable RE. The voltageprofile of the MnO2 WE (Figure 5a, blue) exhibits flat plateaus,indicating a two-phase reaction mechanism, whereas thesloping curve of the carbon AE (Figure 5a, red) confirms itscapacitive performance. Without the reference electrode, itwould be impossible to separate the contributions of eitherelectrode from the overall voltage response (Figure 5) or togain insight into the individual electrochemical mechanisms.Another example usage of the three-electrode setup is in a

Mg-Li hybrid cell, with a Prussian blue analogue (PBA) positiveelectrode and Mg metal negative electrode coupling through aMg-Li dual-salt electrolyte.41 In such a system, the Li(de)intercalation into PBA (Figure 5c, blue), Mg plating/stripping at the negative electrode (Figure 5c, red), and theoverall cell performance (Figure 5d) are examined at the sametime. Since all three sets of data are required to understand thedetailed processes of the hybrid cell, the three electrode cell isnecessary to obtain all the information.

Chemical Diffusion Coefficient Measurements. Forintercalation electrodes (e.g., AxB, where Ay+ is a mobile ionand B is the host lattice), another useful characteristic tomeasure is the chemical diffusion coefficient, D, of the ion, Ay+,traveling through the host lattice, B. There are four differenttechniques generally used to measure D: the galvanostaticintermittent titration technique (GITT),42−44 the potentiostaticintermittent titration technique (PITT),42−44 the charge pulserelaxation (CPR) technique,45 and electrochemical impedancespectroscopy (EIS).46 Here we will limit our discussion toGITT, which provides measurements of several other quantitieslike the equilibrium potential curve and partial conductivity allin one experiment.44 The mathematics behind the theory ofthese experiments were explained by Weppner, Huggins, andWen,42−44 and so only a brief explanation of the method isrepeated here along with a discussion of the experimentalconsiderations. In a GITT experiment, a three-electrode cell isused and a constant current is applied to the working electrodefor a period of time, followed by an open circuit rest periodbefore repeating the current pulse and rest period many times.An example is shown in Figure 6. These pulses are continueduntil the electrode will not accept any more charge (i.e., thepotential plummets/spikes to the point where an undesirableside reaction occurs) or a desired equilibrium potential limit isreached. The current pulses can be short (e.g., 1/500th ofcapacity) or long (e.g., 1/20th of capacity). Short pulses arepreferable when the reference electrode potential does notchange upon activation of the pulse (as it can, for instancewhen using a passivation-prone quasi-reference electrode likeMg foil), and the working electrode potential recovers quicklyduring the rest period, while longer pulses can outlast transitorybehavior of the cell. One of the bonus pieces of informationcollected during a GITT experiment is the equilibrium voltagecurve (V vs Q, but without electrode polarization). Theoret-ically, the equilibrium voltage curves collected during charge

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

94

and discharge would superimpose, but in practice theirproximity to one another depends on how long the rest periodis: the longer it is, the closer the two equilibrium voltage curveswill be, but the longer the experiment will be (e.g., onedischarge in 1 month). Note that very long times (e.g., 2months)47 are required to return the potential to the “true”equilibrium potential.48 Of primary importance to theresearcher attempting to determine D is the pulse V vs tdata. D can be determined using the following equation:43

π= ≪⎜ ⎟ ⎜ ⎟

⎛⎝

⎞⎠

⎡⎣⎢⎛⎝

⎞⎠

⎛⎝⎜

⎞⎠⎟⎤⎦⎥

⎛⎝⎜

⎞⎠⎟D

IVnFS

Vx

dVt

tLD

4 dd d

M2 2 2

(2)

where I is the current, VM is the molar volume of the hostlattice, n is the oxidation state of the mobile ion, F is Faraday’sconstant, S is the electrochemically active surface area, dV/dx isthe slope of the equilibrium voltage curve at the initial value ofx in the present pulse, and dV/d√t is the slope of the V vs √tplot formed from the V vs t data. In theory, a plot of V vs √t,where the initial t value is reset to 0 for any pulse, as shown inFigure 6b, should yield a straight line, as shown in Figure 6c. Inpractice, it is non-linear close to t = 0 due to double layercharging and other transitory effects, and at long times when t≪ L2/D is no longer satisfied. L is the diffusion length, which isbasically half the thickness of the host material particles and canbe estimated using VMnAM/S = L.43 D must be found iteratively(i.e., measure D then go back and check if t ≪ L2/D issatisfied). If the V vs √t plot does not contain any straightsections, the diffusion length might not be long enough. Forthis reason, it is common to make pellet electrodes instead ofcast electrodes and to assume that the pellet acts as one large,thick particle with no pores. Pellet electrodes should ideally be

made without binders, carbon or other additives to maintainelectrode homogeneity, and minimize interfaces. The assump-tion of a pore-free pellet can be rationalized by considering thatdiffusion of active species through the electrolyte down thepores will be slow due to the size and tortuosity of the pores. Assuch, the active species in the pores will become depleted orsaturated relatively quickly when the current is switched on,and so will ultimately only affect the electrode kinetics for ashort time before the activity of the pores is effectively shutdown. Determining the “true” surface area, S, of the sample isalso difficult and made easier by using a pellet and assuming thegeometric surface area is the true surface area. In certainsituations, it is possible to deposit a perfectly flat, pore-free, thinlayer of an active material where the true surface area is thegeometric surface area.49

As an additional check to confirm that the collected signal isindeed reflective of diffusion, pulses with different currents, asshown in Figure 6a, can be collected and used to form a plot ofdV/d√t vs I, as shown in Figure 6d. According to eq 1, aftersome rearranging, this plot should form a straight line. Somecare should be taken to maintain the same x value whenperforming these experiments, which can be accomplished byalternating charge pulses with discharge pulses so that the ionconcentration (x) within the host lattice (B) remains constantbetween pairs of pulses, as shown in Figure 6a. The slope of theresulting dV/d√t vs I line can then be used with eq 1 tocalculate D while inherently compensating for some exper-imental error by using a linear least-squares fit to calculate anaverage D value instead of using individual data points. If theresearcher is interested in determining the migration activationenergy, Em, then the chemical diffusion coefficient, D, must be

Figure 5. (a,b) Voltage profiles of MnO2 vs carbon cycled in Mg(TFSI)2 in diglyme electrolyte at C/20 and the room temperature, using Mg metalas the RE. (c,d) Voltage profiles of PBA vs Mg cycled in Mg-Li dual salt electrolyte at C/10 and the room temperature, using Mg metal as the RE.(blue, red, and green show WE, AE, and WE − AE response, respectively). Adapted from refs 40 and 41.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

95

measured at several different temperatures and then convertedto the so-called self-diffusion coefficient, DA

y+, via

= +D WDAy (3)

where W is a thermodynamic enhancement factor that changesbased on the properties of the host lattice as described byWeppner and Huggins.44 DA

y+ is related to Em through

ν= −+D

af

e E kTA

2/

y m

(4)

where a is the hop distance, ν has the dimension of a frequency,f is the number of possible directions the ion can hop, k isBoltzmann’s constant, and T is the temperature.50 The variablesa and f can be guessed using knowledge of the host structureunit cell; ν can be calculated, but the slope of a plot of ln(DA

y+)vs 1/T will also yield Em and ν.Electrochemical Impedance Spectroscopy (EIS). A high

impedance in an electrochemical cell can cause large over-potentials (the difference between the equilibrium andmeasured potentials), low energy efficiency, and low ratecapability, and can increase the risk of thermal runaway. It isimportant to examine and understand the evolution ofimpedance at each state of discharge/charge and over long-term cycling. Generally, there are three sources of voltage drop:ohmic loss, charge transfer or activation polarization, andconcentration polarization. EIS is a widely used tool todiagnose the impedance of an electrochemical system byrecording the current response to an applied potential atvarying frequencies. Since the response of an electrochemicalsystem, such as a lithium-ion cell, is very non-linear, theimpedance is investigated in a perturbative manner where anAC voltage of 1−10 mV (with frequencies between 1 mHz and1 MHz) on top of the controlled DC voltage (usually at OCV)is applied. Among various ways to present the spectrum, the so-called Nyquist plot is commonly used in the battery field byplotting the negative of the imaginary part of impedance versusthe real part for different frequencies. The Nyquist plot usually

consists of one or more semicircles at high frequencies (whichare charge-transfer processes) followed by a spike at lowfrequencies (which is diffusion toward/away from the surface,i.e., “Warburg” impedance). Note that symmetric axes shouldalways be used for the Nyquist plot (Figure 7a). EIS data arecommonly analyzed by fitting the curve with an equivalentelectrical circuit model to understand the contribution fromeach interface or bulk process. Most of the circuit elements arecommon electrical elements such as resistors, capacitors, andinductors. Each element in the model should have a physicalbasis in the electrochemistry of the system, and the number ofelements should be minimized to be physically meaningful. Anycurve can be fit perfectly by using a large enough number ofelements (variables), but such a fit would likely be meaningless.Practically, there can be a number of interfacial diffusionalprocesses (charge transfer) on both electrodes of a cell. Thiscan cause errors in interpreting different processes as they mayoccur in an overlapping range of frequency resulting in adepressed semicircle; therefore, a three-electrode setup asdiscussed above is preferred. Porous electrodes in particular cangenerate many overlapping charge-transfer processes due totheir large, inhomogeneous surface area. A pelletized electrodewith minimized surface area will be useful in this case.Here we describe the fundamental application of the EIS

technique based on an example of a metallic phase Ti4O7 as asulfur host material for a lithium-sulfur battery.51 Thepolysulfide shuttling remains a major problem that leads torandom precipitation of Li2S on the cathode surface, loss ofactive materials, and long-term capacity fading. We reportedthat by using a bifunctional high surface area Ti4O7 as the sulfurhost, the polysulfides can be effectively adsorbed and readilyreduced/oxidized on the surface over discharge/charge, leadingto a controlled precipitation of Li2S/S and thus lowerimpedance of the cathode. In order to prove that, EIS analysiswas carried out after the first discharge (product: Li2S) and alsoover the first 50 cycles as shown in Figure 7. The spectrum inFigure 7a consists of depressed semicircles with a spike and was

Figure 6. Example procedure for determining the chemical diffusion coefficient D at a particular potential (composition) using a spinel MgxTi2S4pellet in a three-electrode cell. (a) Charge (red), discharge (blue), and open-circuit (black) time periods. (b) Expanded view of left charge curve in(a) with the time reset to start at 0. (c) Data from (b) plotted vs √t and shifted vertically to have a 0 y-intercept for demonstration purposes. Eachline comes from fitting a charge/discharge pulse with a different current from (b). (d) The slopes of the lines in (c) plotted vs the current of thosepulses. The slope of this line can be used to calculate D.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

96

fitted using the equivalent circuit in Figure 7e. The circuitconsists of an ohmic resistance, Rs, in series with three R||Qelements. A constant phase element (Q) was used instead ofthe capacitor to represent the non-ideal behavior of the system,as reflected by the depressed semicircles. By referencing to thephysiochemical processes occurring in lithium sulfur cellsreported in the literature,52,53 Rs corresponds to theuncompensated ohmic resistance (mostly the electrolyteresistance). The R1||Q1 at the highest frequencies correspondsto the charge transfer of polysulfide intermediates and R2||Q2 inthe middle frequencies region reflects the precipitation ofresistive Li2S/Li2S2 onto the positive electrode. Lastly, R3||Q3 inthe low frequencies is related to the diffusion of charge carriersin the solids. We observed that the Ti4O7/S cell exhibits acharge transfer resistance that is 2-fold lower than that of VC/S(VC = Vulcan Carbon). This is ascribed to the enhancedinteraction of the LiPS moleculesand subsequently theLi2Swith Ti4O7. In accord, EIS spectra of Ti4O7/S showednegligible changes over 50 cycles, as shown in Figure 7c,d.Operando X-ray Diffraction (XRD) Analysis. Correlation

of the electrochemical performance of electrode materials withtheir structural evolution upon battery operation providesinsights necessary to understand their working mechanism,which is an essential key to the development of high-performance battery materials. In situ and operando X-raypowder diffraction techniques are powerful methods tocharacterize the structural evolution of battery materials duringcharge/ discharge and are widely used today by the researchcommunity.54−62 In an in situ experiment, i.e., on-siteexperiment, the data are collected in the same environmentthat the material is treated/reacted.63 An example of an in situ

experiment is the XRD analysis of an electrode material withinthe electrochemical cell at the specific states of charge, whereasan ex situ experiment involves disassembling of the cell andextraction of the electrode for further analysis, such as XRD.The in situ XRD analysis offers the advantage of monitoring thestructural evolution of the same electrode in the same cellenvironment at different states of charge, whereas the datacollected ex situ from disassembling several cells might beaffected by the imperfect reproducibility of electrodepreparation and cell assembly. Additionally, the state of chargecan be changed due to short circuiting when disassembling thecell. Special care should be taken in the case of air-sensitivesamples for ex situ analyses. On the other hand, ex situ analysisusually results in higher quality data and is required prior toconducting an in situ experiment.The term operando (Latin for “operating”) is used when the

material is characterized in situ, simultaneous with the operationof the reaction, under non-equilibrium and real-time con-ditions;64 for example, the collection of XRD data from anelectrode material when the electrochemical cell is charged/discharged. Research conducted by Rojo et al.65 nicelydemonstrated the unique ability of the operando methodologyto obtain information about non-equilibrium conditions. In thatstudy, a Na0.67[Mn0.8Mg0.2]O2 electrode, with an initialhexagonal phase (P63/mmc), was charged up to 4.0 V vs Na/Na+ and then discharged to 1.5 V, at which point the cell wasdisassembled and the electrode was extracted for an ex situXRD analysis. Two different phases, a small amount ofhexagonal (P63/mmc) and a larger amount of orthorhombic(Cmcm), were detected in the fully discharged electrode.However, it was reported that operando/in situ XRD did not

Figure 7. Use of EIS technique to understand the impedance of a lithium−sulfur cell using Ti4O7 as a sulfur host material. Nyquist plots of (a)Ti4O7/S-60 and (b) VC/S-60 (VC = Vulcan Carbon, S-60 = the sulfur composite with 60 wt% of sulfur) cells at a discharged state on the first cycleat a C/2 rate showing a dramatic difference in the impedance; and Ti4O7/S-60 at (c) discharged and (d) charged status over 50 cycles at a C/2 rateat different cycle status. Fitted resistance values for (a) and (b) correspond to the equivalent circuit in (e) and are shown in table (f). Adapted fromref 51.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

97

detect the orthorhombic (Cmcm) phase when the cell wasdischarged to 1.5 V under the same cycling conditions (thesame voltage range and current rate) under continuousoperation. It should be noted that the electrode analyzed byex situ XRD was disassembled inside the glovebox, washed withthe electrolyte solvent, and protected against air during the datacollection. This observed discrepancy between the results ofthese two methods indicates that the orthorhombic (Cmcm)phase detected by ex situ XRD is formed upon relaxation,whereas operando XRD under continuous galvanostatic cyclingconditions provides real-time information affected by thekinetics of the phase transformations. It should be noted thatthe phase transitions occurring in an electrode material duringcharge and discharge highly depend on the current rate.66

Excellent reviews on various designs of in situ battery cellshave been written by Sharma et al.64 and Brant et al.67 Manydifferent designs based on coin cells and Swagelok cells, aimedfor reflection and transmission XRD data collection geometries,and for using conventional diffractometers and synchrotron X-ray sources are reported.68−77 Generally, cell design formeasurement in transmission geometry is more complicatedcompared to reflection geometry. However, in the case oftransmission geometry, the data are collected from both thepositive and negative electrodes and from all the thickness ofthe electrode materials. For each experiment, the in situ cell hasto be designed based on the goals of the experiment and theavailable experimental setups. An essential feature of a good celldesign is that the electrochemical performance of the studiedmaterial in the in situ cell is equivalent to that of a conventionalcell. A very simple design of an in situ cell is a typical coin cell atwhich a hole is featured in the positive electrode casing for thereflection geometry, or in each stainless steel component of thecell in the case of transmission geometry. Kapton tape or aberyllium window, both of which are transparent to X-rays, iscommonly used to cover the holes. A beryllium windowprovides a hermetic seal and a more rigid and flat surfacecompared to Kapton tape; however, the latter is lower cost andsafer to handle. A glassy carbon film, which is light, safe, a goodelectrical conductor, and has good corrosion resistance, isanother option for the transparent window of an in situ cell.55 Itshould be noted that all the window materials mentioned abovecause an increase in the background at low angles and lowerstack pressure applied on the electrode under the window.64

Chapman et al. demonstrated the impact of cell design and alsoX-ray beam interaction on the electrochemical activity ofelectrodes in a modified coin cell for operando measurements,highlighting the importance of cell design for reliable resultsfrom these experiments.78

Figure 8 shows the results of a study of the structuralevolution of a positive electrode material by the operando XRDtechnique, performed by our team.55 A sample ofNa0.67[Mn0.65Fe0.20Ni0.15]O2 positive electrode material wascycled between 4.3 and 1.5 V vs Na/Na+ in the galvanostaticmode with a current rate of 13 mA/g (C/20 rate) while XRDpatterns were collected continuously. The evolution of theXRD patterns of the electrode during the first cycle revealsthree different phases (Figure 8a): the initial hexagonal phase,P2 (blue), a low-crystalline phase at high voltage, Z (red), andan orthorhombic phase, P′2 (green), at low voltage. Thereversible transition of the hexagonal phase to the high voltagephase occurs through a two-phase region (black). Figure 8bdemonstrates the evolution of the average interlayer distancesof the layered structure Nax[Mn0.65Fe0.20Ni0.15]O2 as a function

of sodium content (x). The lattice parameters were calculatedusing the Le Bail method.79 The interlayer distance at the high-voltage phase (red) decreases upon extraction of sodium ions,whereas the opposite trend is observed for the hexagonal (blue)and orthorhombic (green) phase regions.The interlayer spacing significantly shrinks upon trans-

formation from the hexagonal phase (blue) to the high voltagephase (red). These observations evidently indicate a significantstructural modification at high voltage. Pair distributionfunction analysis (see below) revealed the migration oftransition metals into interlayer space initiating the highvoltage phase transition. In the same study, the deleteriousimpact of the phase transitions on the electrochemicalperformance of the cells was demonstrated.

Pair Distribution Function (PDF) Analysis. ConventionalX-ray and neutron powder diffraction analyses provideinformation about the average structure of crystalline materials.In those techniques, the information is obtained from the Braggpeaks only, and the diffuse scattering is discarded in the form ofbackground. The background beneath and between the Braggpeaks in an XRD pattern includes diffuse scattering, whichcontains information about the local structure. PDF analysisextracts information from both the Bragg reflections and diffusescattering. The broad range of application of PDF analysisincludes characterization of non-crystalline materials, the studyof disorder in crystalline materials, and research of nano-structured materials.80−85

Figure 8. (a) Illustration of the evolution of XRD pattern of P2-Na0.67[Mn0.65Fe0.20Ni0.15]O2 during the galvanostatic cycling at a C/20rate over the voltage range of 4.3−1.5 V (left), along with ademonstration of the operando experiment time vs voltage of the cell(right). (b) Evolution of the average interlayer distance ofNax[Mn0.65Fe0.20Ni0.15]O2 as a function of sodium content, x. Adaptedfrom ref 55.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

98

For an X-ray PDF analysis, X-ray powder diffraction data arecollected similar to a conventional diffraction measurement.The diffracted data are then corrected for Compton scattering,fluorescence, multiple scattering, and scattering from thesample environment (sample holder, air, instrument). Theresulting coherent scattering function, Icoh(Q), is thennormalized, giving the total scattering structure function,S(Q), as follows:86

=⟨ ⟩

+ ⟨ ⟩ − ⟨ ⟩S QN f Q

I Q f Q f Q( )1( )

[ ( ) ( ) ( ) ]2 coh2 2

(5)

where Q is the amplitude of the scattering vector (Q =4π sin(θ)/λ, with θ the Bragg angle and λ the radiationwavelength), f(Q) is the atomic form factor, N is the number ofatoms, and angle brackets stand for an average over atom types.The X-ray atomic form factor, f(Q), decreases with the increaseof Q, meaning that the high-Q data information gains moreweight in the total scattering experiment compared to theconventional XRD. The Fourier transform of the totalscattering structure functions yields the reduced PDF, G(r)80,86

∫π= −

∞G r Q S Q Qr Q( )

2[ ( ) 1] sin( ) d

0 (6)

The PDF gives a measure of the probability of finding twoatoms in a material separated by a distance r, as follows:

∑ ∑ δ π ρ=⟨ ⟩

− −⎡⎣⎢

⎤⎦⎥G r

r

b b

br r r( )

1( ) 4

i j

i jij2 0

(7)

where ρ0 is the average number density of the material; the sumis over all pairs of atoms i and j with an interatomic distance ofrij; bi is Q-independent scattering amplitude of the atom i and isequal to the atomic number for X-rays; ⟨b⟩ is the averagescattering length of the sample. Any two atoms in the materialresults in a peak in its PDF curve in real space (G(r) versus r);the peak positions indicate the distance of atom-pairs; theintegrated intensities of the peaks are related to the number ofneighbors in those coordination shells around the origin atoms;and the shape of the peaks contains information about theprobability distribution of the atom-pairs.An important requirement for PDF analysis measurement is

the collection of data over a wide range of Q, in order to obtainhigh-resolution real space data. Finite Q resolution results in agradual decrease of peak-peak intensities and the appearance oftermination ripples (false oscillations in the data) in thereduced PDF function, G(r). The maximum value of Q in anXRD measurement is proportional to the maximum scatteringangle and is inversely proportional to the X-ray radiationwavelength, which is why a synchrotron source of X-ray or ashort-wavelength X-ray source (e.g., Ag Kα radiation) in aconventional diffractometer is necessary. Good statistics, lowbackground scattering, and stable experimental set up are otherrequirements to ensure high-quality data for PDF analysis. Theexperimental PDF can be quantitatively modeled in real-space.Similar to the Rietveld refinement, an initial structural model isgiven to the software, and structure- and experiment-dependentparameters are refined to obtain the best fit. PDFGetX387 andPDFgui88 are among the available software for reduction ofreciprocal-space data to real-space data and for fitting thestructural models, respectively. Detailed information about thetheory, measurement procedure, and data processing of PDF ispresented in ref 80.

An excellent example of the application of PDF analysis usedoperando to investigate the local structure evolution of batteryelectrode materials can be found in the study conducted byChapman et al.89 The operando PDF analysis of an ironoxyfluoride conversion electrode, FeII(1−x)Fe

IIIxOxF2−x (x = 0.6),

carried out at a synchrotron, illustrates the variation of atomicdistances and local structural features of the electrode duringthe first discharge/charge (Figure 9a). The initial rutile phase

transforms into body-centered cubic (bcc) Fe nanoparticlesthrough an intermediate rock-salt structure phase during thefirst discharge (Figure 9b). A two-phase nanocompositeelectrode, consisting of the intermediate rock salt phase and apoorly ordered rutile phase (referred to as amorphous rutile), isobtained during the subsequent charge. The evolution of thephase fractions during the first cycle was obtained byrefinement of structural models against the PDF data (Figure9c).

Figure 9. (a) Evolution of the PDF curve of FeO0.6F1.4 during the firstdischarge/charge at 50 °C and a C/30 rate. (b) Simulated PDF curvesof the pristine phase and the phases formed upon cycling. (c)Evolution of the phase distribution. Reproduced with permission fromref 89. Copyright 2013 American Chemical Society.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

99

An example of a conventional diffractometer-based X-rayPDF analysis providing insightful information about adisordered structure can be found in ref 55. In this researchconducted in our laboratory, ex situ PDF analysis along withoperando XRD, Mossbauer spectroscopy, and electrochemicalmeasurements was employed to understand the phasetransition that occurs in P2-Na0.67[Mn0.5Fe0.5]O2 electrodeupon charge. The high-voltage phase, Z-Na0.1[Mn0.5Fe0.5]O2,cannot be characterized by conventional XRD due to the lowcrystallinity of the structure. The PDF data from the pristine P2phase and the high-voltage phase, obtained from the chemicaloxidation of the pristine material, was collected by aPANalytical Empyrean using Ag Kα radiation. Modeling ofthe data showed that the migration of a fraction of transitionmetals from MO2 layer into the interlayer space occurs at thebeginning of the high voltage phase transition.55

Operando X-ray Absorption Spectroscopy (XAS). XASis a widely used technique to determine the electronic structureand atomic geometric/bonding information of matter.90 It isuseful to examine the phase evolution of the electrode materialsduring discharge/charge of the cell. Valuable information canbe obtained by examining the electrodes ex situ by dissemblingthe cells at various states of charge. However, since theelectrochemical reactions always proceed in a non-equilibriumfashion and may not be uniform across the electrodes, ex situmeasurements may not reflect the exact speciation during

galvanostatic experiments. Therefore, operando characteriza-tions are required.91

XAS is performed using synchrotron radiation sources thatare selected by the energy ranges required. The element specificspectrum is obtained by plotting the absorption coefficient as afunction of photon energy (eV) and generally consists of threeregions: pre-edge, X-ray absorption near-edge structure(XANES), and extended X-ray absorption fine structure(EXAFS), as shown in Figure 10. The first two regionscorrespond to the excitation of core electrons to free orbitalsand thus provide information on the electronic states; the lastregion appears due to the scattering of the ejected photo-electrons with the neighboring atoms and thus encapsulatesbonding information. It is worth noting that XAS is a bulk(micrometer size or larger) technique rather than a surfacecharacterization technique, and provides information aboutevolution of the electrode bulk upon the electrochemicalreaction.We have used operando XAS to examine the sulfur speciation

in various electrolyte systems for lithium sulfur cells and theevolution of chemical environments of transition metals inlayered oxides for sodium-host cathodes, in the AdvancedPhoton Source (Chicago) and Canadian Light Source(Saskatoon), respectively. The typical windowed coin cellsetup for the operando experiments in fluorescence mode isshown in Figure 10. Thin aluminized Kapton film was featuredin the cell design for the penetration of X-rays. The electrodes

Figure 10. Scheme and examples for operando XAS experiments. (a) Scheme showing an example XAS spectrum that consists of three regions. (b)Scheme showing the design of the windowed coin cell for operando XAS measurements. (c) 3D plot showing the evolution of sulfur K-edge XANESspectra as a function of the electrochemical cycling of a lithium sulfur cell in ether-based electrolyte; the end of discharge/charge is in red. (d)Evolution of each sulfur species (S8, Li2S6, Li2S4, Li2S) as a function of cycling, as quantified by linear combination fitting with the references. Panels(c) and (d) reproduced with permission from ref 91. Copyright 2013 American Chemical Society.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

100

were coated on porous carbon paper and were placed in the cellfacing the window for maximum beam penetration.91

Here we show the process of operando XAS in our laboratoryusing lithium−sulfur chemistry as an example;91 abundantreports from other research groups can be found else-where.92−94 The beam size was about 450 × 450 μm, andthe cells were cycled under constant helium flow. For samplepreparation, the 200 μm mesoporous carbon spheres infiltratedwith sulfur were used in a diluted concentration in order tominimize self-adsorption (for the S K-edge). Linear combina-tion fitting using synthesized pure (poly)sulfides/sulfurreferences on properly normalized spectra was carried out toquantify the fractions of each species at each stage of cycling.Figure 10c shows the evolution of the XANES spectra as afunction of electrochemical cycling, which nicely suggests theconsumption of sulfur and formation of polysulfide inter-mediates in addition to the final product Li2S. Quantitativeanalysis enables some conclusions on the sulfur speciation inether based electrolytes and the mechanisms that govern thedissolution/deposition of the redox end members, S8 and Li2S.For example, we found evidence that upon discharge the rapiddisproportionation of S8

2‑ by S82−→S6

2− + 1/4 S80 reaction can

lead to the widely reported discharge capacity of only 75% ofthe theoretical value (1675 mAh/g). We have also examinedother electrolyte systems including a polysulfide nonsolvent(acetonitrile based solvate) and a high donor pair solvent(dimethylacetamide), where S3

•− radical was identified as anintermediate product.X-ray Photoelectron Spectroscopy (XPS). The surface

chemistry of the material and the electrode interface influencesthe overall performance of the battery; therefore understandingthe interface phenomenon is obviously necessary.95 Forexample, the solid−electrolyte interphase (SEI) generatedbecause of the electrolyte instability with respect to theanode chemical potential, plays a very important role in the cellperformance.96 Slow and continuous SEI growth leads tocapacity fading and increased cell resistance. XPS is one of themost powerful analytical tools for studying the chemicalcomposition of the material and the oxidation state of theions near the surface. Note that although the X-ray beam usedfor XPS analysis can penetrate to a depth of ∼1 mm, only theelectrons from the top 5−10 nm can escape from the surface.Depth profiling by argon ion sputtering can burrow below thesurface layer; however, the etch rate is typically only 1−2 nm/min.97 In contrast to XRD and XAS techniques, XPS is typicallyoperated under ultrahigh vacuum, therefore, it is difficult tomeasure liquid battery systems in operando or in situ mode.

Near-ambient pressure X-ray photoelectron spectroscopy(NAP-XPS) in the pressure range of millibar can be conductedin operando mode, however,98 as it relies on advanced vacuumand analyzer technologies, but is not a mainstream technique.In our research, we have used XPS, for example, to

characterize the surface chemistry of a wide range of sulfurhost materials and to probe their interactions with polysulfides.Chemisorption of the soluble polysulfides by host materials isone of the promising approaches to enable spatially located Li2Sprecipitation on the host surface, and thus to stabilize thecycling performance of Li-S batteries.4 These chemicalinteractions usually take place on the surface of the sulfurhost materials, accompanied by electron transitions orcoordination, and should be characterizable using XPS. In theresearch studies conducted by our team, modifications of theXPS spectra of the sulfur host materials caused by the reactionwith polysulfides were investigated. The reactions wereconducted in the electrolyte solvent to simulate the cellenvironment. XPS analysis can also be applied to cycledelectrodes. The preparation of the sample is identical to that ofthe ex situ XRD characterization. Note that the sample cleaningprocess is very important as artifacts may be introduced.Solvents that could react with the sample should not be used,obviously. The electrolyte solvent for processing of a cycledelectrode is a good choice. We use CasaXPS software withGaussian−Lorentzian functions and a Shirley-type backgroundto fit the spectrum. The binding energies are calibrated usingthe C 1s peak at 285.0 eV. The energy separation of doubletpairs depends on both the principal and angular momentumquantum numbers of the core level electrons; for example, theS2p3/2 spin−orbit doublet is 1.2 eV lower than the S2p1/2 spin−orbit doublet (Figure 11a,c). This powerful analytical techniqueis widely accepted by the Li-S battery research community toreveal the surface chemistry of the cathode materials. Polarinteractions with shifting binding energies (see Figure11a),99−102 and Lewis acid−base interactions with the newchemical bond formation (Figure 11b)103 were reported toentrap the polysulfides in the cathode. By using XPS, we alsodetected the thiosulfate−polythionate conversion (Wacken-roder’s reaction) phenomena when MnO2 was employed as thesulfur host for a Li-S battery (Figure 11c).104 TheWackenroder’s reaction, discovered in 1845, was based inaqueous solutions,105 but our results show the reaction alsooccurs in an organic solvent.

Figure 11. Examples of XPS spectra to show the interaction between polysulfides and the sulfur host materials. (a) Polar interaction shown by Ti4O7and Li2S4. (b) Lewis acid base interaction shown by Ti2C MXene and Li2S4. (c) Thiosulfate−polythionate conversion shown by MnO2 and Li2S4.Adapted from refs 99, 103, and 104.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

101

■ SUMMARY

This Review presents the procedures and methods to prepareand evaluate electrochemical cells in battery research. Some ofthe most informative characterization techniques employed toassess new materials for batteries are described here, andexamples of insightful information that each technique hasprovided in the research areas of Li-S, Na-ion, and Mg batteriesare presented; excellent references for detailed description ofthe theory, experimental procedures, various designs, and dataprocessing and analysis methods are included.Galvanostatic charge/discharge analysis is the first step in

characterizing the electrochemical performance of a newelectrode material; the delivered capacity and energy, voltage-composition curve, cycle life, and columbic efficiency areamong the most important pieces of information that can beobtained for a new material. Chemical diffusion coefficientmeasurements and EIS are more advanced characterizationtechniques that provide information about the transportproperties and the possible aging mechanisms of electrodematerials. Operando XRD analysis can illustrate the evolution ofcrystal structure of electrode materials during cell operation.Real-time and non-equilibrium data are revealed by an operandomethodology, whereas an ex situ experiment, which is carriedout after a relaxation period, characterizes the stable state of thematerial. PDF analysis is a powerful tool to provide insightsinto the structure of low-crystalline and disordered materials.XAS techniques, which employ synchrotron X-ray radiation, arebroadly used to investigate the electronic state and localstructure of the atoms. Identifying the redox active atoms andmonitoring the evolution of their oxidation states and localenvironment in the electrode materials during the cell charge/discharge processes could be realized by operando XAS. XPS isan effective technique to understand the surface chemistry ofbattery materials, e.g., to study the electrode/electrolyteinterphase.Electron microscopy,106−108 Mossbauer spectroscopy,109,110

nuclear magnetic resonance spectroscopy,111,112 and onlineelectrochemical mass spectrometry113−115 are among manyother advanced characterization techniques for battery researchbut are beyond the scope of this paper. A broad range ofcharacterization of different battery components shapes ourpicture of the reaction mechanisms, transport properties,structural stability, thermal stability, and chemical stability ofeach part, paving the way for the development of high-performance batteries. Last but not least, we note the role oftheoretical studies is gaining increasing importance providingnot only an understanding of how battery materials function atthe atomic level, but also in being a vital predictive tool in thesearch for new materials.116,117 Increasing effort at combiningtheory and experiment is clearly to the benefit of both.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: lfnazar@uwaterloo.ca.

Author Contributions†The paper was written with the active contribution of all co-authors except for L.F.N., who primarily oversaw theconception of the Review and content, and its final stageediting.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

L.F.N. and her team gratefully acknowledge past and currentresearch support from BASF (SE and Canada) through theirAcademic Battery and Electrochemistry International Network;the Joint Centre for Energy Storage Research, a Hub funded bythe Department of Energy, USA; NSERC through theirDiscovery, Research Chair programs and student fellowshipprograms; and Natural Resources Canada. The WaterlooInstitute for Nanotechnology and the Ontario Governmentare thanked for scholarships to some co-authors of this work.

■ REFERENCES(1) Dunn, B.; Kamath, H.; Tarascon, J. M. Electrical Energy Storagefor the Grid: A Battery of Choices. Science 2011, 334, 928−935.(2) Yang, Z.; Zhang, J.; Kintner-Meyer, M. C. W.; Lu, X.; Choi, D.;Lemmon, J. P.; Liu, J. Electrochemical Energy Storage for Green Grid.Chem. Rev. 2011, 111, 3577−3613.(3) Andre, D.; Kim, S.-J.; Lamp, P.; Lux, S. F.; Maglia, F.; Paschos,O.; Stiaszny, B. Future generations of cathode materials: anautomotive industry perspective. J. Mater. Chem. A 2015, 3, 6709−6732.(4) Pang, Q.; Liang, X.; Kwok, C. Y.; Nazar, L. F. Review-TheImportance of Chemical Interactions between Sulfur Host Materialsand Lithium Polysulfides for Advanced Lithium-Sulfur Batteries. J.Electrochem. Soc. 2015, 162, A2567−A2576.(5) Xu, R.; Lu, J.; Amine, K. Progress in Mechanistic Understandingand Characterization Techniques of Li-S Batteries. Adv. Energy Mater.2015, 5, 1500408.(6) Rosenman, A.; Markevich, E.; Salitra, G.; Aurbach, D.; Garsuch,A.; Chesneau, F. F. Review on Li-Sulfur Battery Systems: an IntegralPerspective. Adv. Energy Mater. 2015, 5, 1500212.(7) Yabuuchi, N.; Kubota, K.; Dahbi, M.; Komaba, S. ResearchDevelopment on Sodium-Ion Batteries. Chem. Rev. 2014, 114, 11636−11682.(8) Kundu, D.; Talaie, E.; Duffort, V.; Nazar, L. F. The EmergingChemistry of Sodium Ion Batteries for Electrochemical EnergyStorage. Angew. Chem., Int. Ed. 2015, 54, 3431−3448.(9) Fang, C.; Huang, Y.; Zhang, W.; Han, J.; Deng, Z.; Cao, Y.; Yang,H. Routes to High Energy Cathodes of Sodium-Ion Batteries. Adv.Energy Mater. 2016, 6, 1501727.(10) Luo, W.; Shen, F.; Bommier, C.; Zhu, H.; Ji, X.; Hu, L. Na-ionBattery Anodes: Materials and Electrochemistry. Acc. Chem. Res. 2016,49, 231−240.(11) Muldoon, J.; Bucur, C. B.; Gregory, T. Quest for NonaqueousMultivalent Secondary Batteries: Magnesium and Beyond. Chem. Rev.2014, 114, 11683−11720.(12) Yoo, H. D.; Shterenberg, I.; Gofer, Y.; Gershinsky, G.; Pour, N.;Aurbach, D. Mg rechargeable batteries: an on-going challenge. EnergyEnviron. Sci. 2013, 6, 2265−2279.(13) Muldoon, J.; Bucur, C. B.; Oliver, A. G.; Sugimoto, T.; Matsui,M.; Kim, H. S.; Allred, G. D.; Zajicek, J.; Kotani, Y. Electrolyteroadblocks to a magnesium rechargeable battery. Energy Environ. Sci.2012, 5, 5941−5950.(14) Marks, T.; Trussler, S.; Smith, A. J.; Xiong, D.; Dahn, J. R. AGuide to Li-Ion Coin-Cell Electrode Making for AcademicResearchers. J. Electrochem. Soc. 2011, 158, A51−A57.(15) Lu, Z.; Dahn, J. R. Intercalation of Water in P2, T2 and O2Structure Az[CoxNi1/3‑xMn2/3]O2. Chem. Mater. 2001, 13, 1252.(16) Buchholz, d.; Chagas, L. g.; Vaalma, C.; Wu, L.; Passerini, S.Water sensitivity of layered P2/P3-NaxNi0.22Co0.11Mn0.66O2 cathodematerial. J. Mater. Chem. A 2014, 2, 13415.(17) Stoyanova, R.; Carlier, R. D.; Sendova-Vassileva, M.; Yoncheva,M.; Zhecheva, E.; Nihtianova, D.; Delmas, C. Stabilization of over-stoichiometric Mn4+ in layered Na2/3MnO2. J. Solid State Chem. 2010,183, 1372.(18) Sathiya, M.; Hemalatha, K.; Ramesha, K.; Tarascon, J. M.;Prakash, A. S. Synthesis, structure, and electrochemical properties of

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

102

the layered sodium insertion cathode material: NaNi1/3Mn1/3Co1/3O2.Chem. Mater. 2012, 24, 1846.(19) Duffort, V.; Talaie, E.; Black, R.; Nazar, L. F. Chem. Mater. 2015,27, 2515.(20) Bonnick, P.; Dahn, J. R. A Simple Coin Cell Design for TestingRechargeable Zinc-Air or Alkaline Battery Systems. J. Electrochem. Soc.2012, 159, A981−A989.(21) Trask, S. E.; Li, Y.; Kubal, J. J.; Bettge, M.; Polzin, B. J.; Zhu, Y.;Jansen, A. N.; Abraham, D. P. From coin cells to 400 mAh pouch cells:Enhancing performance of high-capacity lithium-ion cells viamodifications in electrode constitution and fabrication. J. PowerSources 2014, 259, 233−244.(22) Lemaire-Potteau, E.; Perrin, M.; Genies, S.Charging Methods.In Encyclopedia of Electrochemical Power Sources; Garche, J., Ed.;Elsevier: Amsterdam, 2009; pp 413−423.(23) Sun, X.; Bonnick, P.; Duffort, V.; Liu, M.; Rong, Z.; Persson, K.A.; Ceder, G.; Nazar, L. F. A high capacity thiospinel cathode for Mgbatteries. Energy Environ. Sci. 20169227310.1039/C6EE00724D.(24) Emly, A.; Van der Ven, A. Mg Intercalation in Layered andSpinel Host Crystal Structures for Mg Batteries. Inorg. Chem. 2015, 54,4394−4402.(25) Dahn, J. R.; McKinnon, W. R. Lithium intercalation in 2H-LixTaS2. J. Phys. C: Solid State Phys. 1984, 17, 4231−4243.(26) Reimers, J. N.; Dahn, J. R. Lattice gas and spin ordering onstacked triangular lattices. J. Phys.: Condens. Matter 1992, 4, 8105−8118.(27) Dahn, J. R. Phase diagram of LixC6. Phys. Rev. B: Condens. MatterMater. Phys. 1991, 44, 9170−9177.(28) Chen, Z.; Lu, Z.; Dahn, J. R. Staging Phase Transitions inLixCoO2. J. Electrochem. Soc. 2002, 149, A1604−A1609.(29) Potyrailo, R. A.; Maier, W. F. Combinatorial and High-Throughput Discovery and Optimization of Catalysts and Materials;CRC Press: Boca Raton, FL, 2006.(30) Pang, Q.; Nazar, L. F. Long-Life and High-Areal-Capacity Li-SBatteries Enabled by a Light-Weight Polar Host with IntrinsicPolysulfide Adsorption. ACS Nano 2016, 10, 4111−4118.(31) Wenzl, H. Batteries and fuel cells-lifetime. Encyclopedia ofElectrochemical Power Sources; Elsevier: Amsterdam, 2009; pp 552−558.(32) Smith, A. J.; Burns, J. C.; Trussler, S.; Dahn, J. R. PrecisionMeasurements of the Coulombic Efficiency of Lithium-Ion Batteriesand of Electrode Materials for Lithium-Ion Batteries. J. Electrochem.Soc. 2010, 157, A196−A202.(33) Smith, A. J.; Burns, J. C.; Xiong, D.; Dahn, J. R. InterpretingHigh Precision Coulometry Results on Li-Ion Cells. J. Electrochem. Soc.2011, 158, A1136−A1142.(34) Burns, J. C.; Krause, L. J.; Le, D.-B.; Jensen, L. D.; Smith, A. J.;Xiong, D.; Dahn, J. R. Introducing Symmetric Li-Ion Cells as a Tool toStudy Cell Degradation Mechanisms. J. Electrochem. Soc. 2011, 158,A1417−A1422.(35) Periyapperuma, K.; Tran, T. T.; Trussler, S.; Ioboni, D.;Obrovac, M. N. Conflat Two and Three Electrode ElectrochemicalCells. J. Electrochem. Soc. 2014, 161, A2182−A2187.(36) Blyr, A.; Sigala, C.; Amatucci, G.; Guyomard, D.; Chabre, Y.;Tarascon, J.-M. Self-Discharge of LiMn2O4/C Li-Ion Cells in TheirDischarged State. J. Electrochem. Soc. 1998, 145, 194−209.(37) Delacourt, C.; Ridgway, P. L.; Srinivasan, V.; Battaglia, V.Measurements and Simulations of Electrochemical Impedance Spec-troscopy of a Three-Electrode Coin Cell Design for Li-Ion CellTesting. J. Electrochem. Soc. 2014, 161, A1253−A1260.(38) Ender, M.; Weber, A.; Ivers-Tiffee, E. Analysis of Three-Electrode Setups for AC-Impedance Measurements on Lithium-IonCells by FEM Simulations. J. Electrochem. Soc. 2012, 159, A128−A136.(39) Bunzli, C.; Kaiser, H.; Novak, P. Important Aspects for ReliableElectrochemical Impedance Spectroscopy Measurements of Li-IonBattery Electrodes. J. Electrochem. Soc. 2015, 162, A218−A222.(40) Sun, X.; Duffort, V.; Mehdi, B. L.; Browning, N. D.; Nazar, L. F.Investigation of the Mechanism of Mg Insertion in Birnessite in

Nonaqueous and Aqueous Rechargeable Mg-Ion Batteries. Chem.Mater. 2016, 28, 534−542.(41) Sun, X.; Duffort, V.; Nazar, L. F. Prussian Blue Mg-Li HybridBatteries. Adv. Sci. 2016, 3, 1600044.(42) Weppner, W.; Huggins, R. A. Electrochemical Methods forDetermining Kinetic Properties of Solids. Annu. Rev. Mater. Sci. 1978,8, 269−311.(43) Wen, C. J.; Boukamp, B. A.; Huggins, R. A.; Weppner, W.Thermodynamic and Mass Transport Properties of LiAl. J. Electrochem.Soc. 1979, 126, 2258−2266.(44) Weppner, W.; Huggins, R. A. Determination of the KineticParameters of Mixed-Conducting Electrodes and Application to theSystem Li3Sb. J. Electrochem. Soc. 1977, 124, 1569−1578.(45) Yamamoto, T.; Kikkawa, S.; Koizumi, M. Chemical diffusivity oflithium in LiyTi1+xS2. Solid State Ionics 1985, 17, 63−66.(46) Wang, L.; Zhao, J.; He, X.; Gao, J.; Li, J.; Wan, C.; Jiang, C.Electrochemical Impedance Spectroscopy (EIS) Study of Li-Ni1/3Co1/3Mn1/3O2 for Li-ion batteries. Int. J. Electrochem. Sci. 2012,7, 345−353.(47) Novak, P. CuO cathode in lithium cells − II. Reductionmechanism of CuO. Electrochim. Acta 1985, 30, 1687−1692.(48) Poizot, P.; Laruelle, S.; Grugeon, S.; Tarascon, J.-M.Rationalization of the Low-Potential Reactivity of 3d-Metal-BasedInorganic Compounds toward Li. J. Electrochem. Soc. 2002, 149,A1212−A1217.(49) Xia, H.; Lu, L.; Ceder, G. Li diffusion in LiCoO2 thin filmsprepared by pulsed laser deposition. J. Power Sources 2006, 159, 1422−1427.(50) Van der Ven, A.; Ceder, G.; Asta, M.; Tepesch, P. D. First-principles theory of ionic diffusion with nondilute carriers. Phys. Rev. B:Condens. Matter Mater. Phys. 2001, 64, 184307.(51) Pang, Q.; Kundu, D.; Cuisinier, M.; Nazar, L. F. Surface-enhanced redox chemistry of polysulphides on a metallic and polarhost for lithium-sulphur batteries. Nat. Commun. 2014, 5, 4759.(52) Deng, Z.; Zhang, Z.; Lai, Y.; Liu, J.; Li, J.; Liu, Y.Electrochemical Impedance Spectroscopy Study of a Lithium/SulfurBattery: Modeling and Analysis of Capacity Fading. J. Electrochem. Soc.2013, 160, A553−A558.(53) Canas, N. A.; Hirose, K.; Pascucci, B.; Wagner, N.; Friedrich, K.A.; Hiesgen, R. Investigations of lithium−sulfur batteries usingelectrochemical impedance spectroscopy. Electrochim. Acta 2013, 97,42−51.(54) Mortemard de Boisse, B.; Carlier, D.; Guignard, M.; Bourgeois,L.; Delmas, C. P2-NaxMn1/2Fe1/2O2 Phase Used as Positive Electrodein Na Batteries: Structural Changes Induced by the Electrochemical(De)intercalation Process. Inorg. Chem. 2014, 53, 11197−11205.(55) Talaie, E.; Duffort, V.; Smith, H. L.; Fultz, B.; Nazar, L. F.Structure of the high voltage phase of layered P2-Na2/3−z[Mn1/2Fe1/2]-O2 and the positive effect of Ni substitution on its stability. EnergyEnviron. Sci. 2015, 8, 2512−2523.(56) Han, M. H.; Gonzalo, E.; Sharma, N.; Lopez del Amo, J. M.;Armand, M.; Avdeev, M.; Saiz Garitaonandia, J. J.; Rojo, T. High-Performance P2-Phase Na2/3Mn0.8Fe0.1Ti0.1O2 Cathode Material forAmbient-Temperature Sodium-Ion Batteries. Chem. Mater. 2016, 28,106−116.(57) Sharma, N.; Gonzalo, N.; Pramudita, J. C.; Han, M. H.; Brand,H. E. A.; Hart, J. N.; Pang, W. K.; Guo, Z.; Rojo, T. The UniqueStructural Evolution of the O3-Phase Na2/3Fe2/3Mn1/3O2 during HighRate Charge/Discharge: A Sodium-Centred Perspective. Adv. Funct.Mater. 2015, 25, 4994−5005.(58) Pinedo, R.; Weber, D. A.; Bergner, B.; Schroder, D.; Adelhelm,P.; Janek, J. Insights into the Chemical Nature and FormationMechanisms of Discharge Products in Na−O2 Batteries by Means ofOperando X-ray Diffraction. J. Phys. Chem. C 2016, 120, 8472−8481.(59) Bleith, P.; Kaiser, H.; Novak, P.; Villevieille, C. In situ X-raydiffraction characterisation of Fe0.5TiOPO4 and Cu0.5TiOPO4 aselectrode material for sodium-ion batteries. Electrochim. Acta 2015,176, 18−21.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

103

(60) Takahashi, I.; Mori, T.; Yoshinari, T.; Orikasa, Y.; Koyama, Y.;Murayama, H.; Fukuda, K.; Hatano, M.; Arai, H.; Uchimoto, Y.; Terai,T. Irreversible phase transition between LiFePO4 and FePO4 duringhigh-rate charge-discharge reaction by operando X-ray diffraction. J.Power Sources 2016, 309, 122−126.(61) Lim, L. Y.; Fan, S.; Hng, H. H.; Toney, M. F. Operando X-rayStudies of Crystalline Ge Anodes with Different Conductive Additives.J. Phys. Chem. C 2015, 119, 22772−22777.(62) Permien, S.; Indris, S.; Schurmann, U.; Kienle, L.; Zander, S.;Doyle, S.; Bensch. What Happens Structurally and Electronicallyduring the Li Conversion Reaction of CoFe2O4 Nanoparticles: AnOperando XAS and XRD Investigation. W. Chem. Mater. 2016, 28,434−444.(63) Banares, M. A. Operando methodology: combination of in situspectroscopy and simultaneous activity measurements under catalyticreaction conditions. Catal. Today 2005, 100, 71−77.(64) Sharma, N.; Pang, W. K.; Guo, Z.; Peterson, V. K. In SituPowder Diffraction Studies of Electrode Materials in RechargeableBatteries. ChemSusChem 2015, 8, 2826−2853.(65) Sharma, N.; Tapia-Ruiz, N.; Singh, G.; Armstrong, A. R.;Pramudita, J. C.; Brand, H. E. A.; Billaud, J.; Bruce, P. G.; Rojo, T. RateDependent Performance Related to Crystal Structure Evolution ofNa0.67Mn0.8Mg0.2O2 in a Sodium-Ion Battery. Chem. Mater. 2015, 27,6976−6986.(66) Sharma, N.; Han, M. H.; Pramudita, J. C.; Gonzalo, E.; Brand,H. E. A.; Rojo, T. A comprehensive picture of the current ratedependence of the structural evolution of P2-Na2/3Fe2/3Mn1/3O2. J.Mater. Chem. A 2015, 3, 21023−21038.(67) Brant, W. R.; Schmid, S.; Du, G.; Gu, Q.; Sharma, N. A simpleelectrochemical cell for in-situ fundamental structural analysis usingsynchrotron X-ray powder diffraction. J. Power Sources 2013, 244,109−114.(68) Chianelli, R. R.; Scanlon, J. C.; Rao, B. M. L. Dynamic X-RayDiffraction. J. Electrochem. Soc. 1978, 125, 1563−1566.(69) Boyanov, S.; Zitoun, D.; Menetrier, M.; Jumas, J. C.; Womes,M.; Monconduit, L. Comparison of the Electrochemical Lithiation/Delitiation Mechanisms of FePx (x = 1, 2, 4) Based Electrodes in Li-Ion Batteries. J. Phys. Chem. C 2009, 113, 21441−21452.(70) Richard, M. N.; Koetschau, I.; Dahn, J. R. A Cell for In Situ X-Ray Diffraction Based on Coin Cell Hardware and Bellcore PlasticElectrode Technology. J. Electrochem. Soc. 1997, 144, 554−557.(71) Morcrette, M.; Chabre, Y.; Vaughan, G.; Amatucci, G.; Leriche,J.-B.; Patoux, S.; Masquelier, C.; Tarascon, J. M. In situ X-raydiffraction techniques as a powerful tool to study battery electrodematerials. Electrochim. Acta 2002, 47, 3137−3149.(72) Gustafsson, T.; Thomas, J. O.; Koksbang, R.; Farrington, G. C.The polymer battery as an environment for in situ X-ray diffractionstudies of solid-state electrochemical processes. Electrochim. Acta 1992,37, 1639−1643.(73) Leriche, J. B.; Hamelet, S.; Shu, J.; Morcrette, M.; Masquelier,C.; Ouvrard, G.; Zerrouki, M.; Soudan, P.; Belin, S.; Elkaim, E.;Baudelet, F. An Electrochemical Cell for Operando Study of LithiumBatteries Using Synchrotron Radiation. J. Electrochem. Soc. 2010, 157,A606−A610.(74) Gross, T.; Buhrmester, T.; Bramnik, K. G.; Bramnik, N. N.;Nikolowski, K.; Baehtz, C.; Ehrenberg, H.; Fuess, H. Structure-intercalation relationships in LiNiyCo1‑yO2. Solid State Ionics 2005, 176,1193−1199.(75) Rhodes, K.; Meisner, R.; Kim, Y.-G.; Dudney, N.; Daniel, C.Evolution of Phase Transformation Behavior in Li(Mn1.5Ni0.5)O4

Cathodes Studied By In Situ XRD. J. Electrochem. Soc. 2011, 158,A890−A897.(76) Borkiewicz, O. J.; Shyam, B.; Wiaderek, K. M.; Kurtz, C.;Chupas, P. J.; Chapman, K. W. The AMPIX electrochemical cell: aversatile apparatus for in situ X-ray scattering and spectroscopicmeasurements. J. Appl. Crystallogr. 2012, 45, 1261−1269.(77) Kimpton, J.; Gu, Q. Synchrotron X-ray Powder Diffraction forFast, In-Situ Insights. Synch. Rad. News 2014, 27, 18−20.

(78) Borkiewicz, O. J.; Wiaderek, K. M.; Chupas, P. J.; Chapman, k.W. Best Practices for Operando Battery Experiments: Influences of X-ray Experiment Design on Observed Electrochemical Reactivity. J.Phys. Chem. Lett. 2015, 6, 2081−2085.(79) Le Bail, A. Whole powder pattern decomposition methods andapplications: A retrospection. Powder Diffr. 2005, 20, 316−326.(80) Egami, T.; Billinge, S. J. L. Underneath the Bragg peaks: structuralanalysis of complex materials; Pergamon Press/Elsevier: Oxford,England, 2003.(81) Thakral, S.; Terban, M. W.; Thakral, N. K.; Suryanarayanan, R.Recent advances in the characterization of amorphous pharmaceuticalsby X-ray diffractometry. Adv. Drug Delivery Rev. 2016, 100, 183−193.(82) Zeng, D.; Cabana, J.; Breger, J.; Yoon, W.-S.; Grey, C. P. CationOrdering in Li[NixMnxCo(1−2x)]O2-Layered Cathode Materials: ANuclear Magnetic Resonance (NMR), Pair Distribution Function, X-ray Absorption Spectroscopy, and Electrochemical Study. Chem.Mater. 2007, 19, 6277−6289.(83) Breger, J.; Dupre, N.; Chupas, P. J.; Lee, P. L.; Proffen, T.;Parise, J. B.; Grey, C. P. Short- and Long-Range Order in the PositiveElectrode Material, Li(NiMn)0.5O2: A Joint X-ray and NeutronDiffraction, Pair Distribution Function Analysis and NMR Study. J.Am. Chem. Soc. 2005, 127, 7529−7537.(84) Breger, J.; Kang, K.; Cabana, J.; Ceder, G.; Grey, C. P. NMR,PDF and RMC study of the positive electrode material Li(Ni0.5Mn0.5)-O2 synthesized by ion-exchange methods. J. Mater. Chem. 2007, 17,3167−3174.(85) Hu, Y. Y.; Liu, Z.; Nam, K.-W.; Borkiewicz, O. J.; Cheng, J.;Hua, X.; Dunstan, M. T.; Yu, X.; Wiaderek, K. M.; Du, L. -S.;Chapman, k. W.; Chupas, P. J.; Yang, X. Q.; Grey, C. P. Origin ofadditional capacities in metal oxide lithium-ion battery electrodes. Nat.Mater. 2013, 12, 1130−1136.(86) Dinnebier, R. E.; Billinge, S. J. L. Powder Diffraction: Theory andPractice; The Royal Society of Chemistry: Cambridge, UK, 2008.(87) Juhas, P.; Davis, T.; Farrow, C. L.; Billinge, S. J. L. PDFgetX3: arapid and highly automatable program for processing powderdiffraction data into total scattering pair distribution functions. J.Appl. Crystallogr. 2013, 46, 560−566.(88) Farrow, C. L.; Juhas, P.; Liu, J. W.; Bryndin, D.; Bozin, E. S.;Bloch, J.; Proffen, T.; Billinge, S. J. L. PDFfit2 and PDFgui: computerprograms for studying nanostructure in crystals. J. Phys.: Condens.Matter 2007, 19, 335219.(89) Wiaderek, K. M.; Borkiewicz, O. J.; Castillo-Martínez, E.;Robert, R.; Pereira, N.; Amatucci, G. G.; Grey, C. P.; Chupas, P. J.;Chapman, K. W. Comprehensive Insights into the Structural andChemical Changes in Mixed-Anion FeOF Electrodes by UsingOperando PDF and NMR Spectroscopy. J. Am. Chem. Soc. 2013,135, 4070−4078.(90) Ortega, R.; Carmona, A.; Llorens, I.; Solari, P. L. X-rayabsorption spectroscopy of biological samples. A tutorial. J. Anal. At.Spectrom. 2012, 27, 2054−2065.(91) Cuisinier, M.; Cabelguen, P. E.; Evers, S.; He, G.; Kolbeck, M.;Garsuch, A.; Bolin, T.; Balasubramanian, M.; Nazar, L. F. Sulfurspeciation in Li−S batteries determined by operando X-ray absorptionspectroscopy. J. Phys. Chem. Lett. 2013, 4, 3227−3232.(92) Patel, M.U. M.; Arcon, I.; Aquilanti, G.; Stievano, L.; Mali, G.;Dominko, R. X-ray Absorption Near-Edge Structure and NuclearMagnetic Resonance Study of the Lithium−Sulfur Battery and itsComponents. ChemPhysChem 2014, 15, 894−904.(93) Gorlin, Y.; Patel, M. U. M.; Freiberg, A.; He, Q.; Piana, M.;Tromp, M.; Gasteiger, H. A. Understanding the Charging Mechanismof Lithium-Sulfur Batteries Using Spatially Resolved Operando X-RayAbsorption Spectroscopy. J. Electrochem. Soc. 2016, 163, A930−A939.(94) Wujcik, K. H.; Pascal, T. A.; Pemmaraju, C. D.; Devaux, D.;Stolte, W. C.; Balsara, N. P.; Prendergast, D. Characterization ofPolysulfide Radicals Present in an Ether-Based Electrolyte of aLithium−Sulfur Battery During Initial Discharge Using In Situ X-RayAbsorption Spectroscopy Experiments and First-Principles Calcu-lations. Adv. Energy Mater. 2015, 5, 1500285.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

104

(95) Goodenough, J. B.; Kim, Y. Challenges for Rechargeable LiBatteries. Chem. Mater. 2010, 22, 587−603.(96) Owejan, J. E.; Owejan, J. P.; DeCaluwe, S. C.; Dura, A. SolidElectrolyte Interphase in Li-Ion Batteries: Evolving StructuresMeasured In Situ by Neutron Reflectometry. Chem. Mater. 2012, 24,2133−2140.(97) Kanamura, K.; Tamura, H.; Shiraishi, S.; Takehara, Z. XPSAnalysis of Lithium Surfaces Following Immersion in Various SolventsContaining LiBF4. J. Electrochem. Soc. 1995, 142, 340.(98) Salmeron, M.; Schlogl, R. Ambient Pressure PhotoelectronSpectroscopy: A New Tool for Surface Science and Nanotechnology.Surf. Sci. Rep. 2008, 63, 169.(99) Pang, Q.; Kundu, D.; Cuisinier, M.; Nazar, L. F. Surface-enhanced redox chemistry of polysulphides on a metallic and polarhost for lithium-sulphur batteries. Nat. Commun. 2014, 5, 4759.(100) Zhou, G.; Zhao, Y.; Manthiram, A. Dual-Confined FlexibleSulfur Cathodes Encapsulated in Nitrogen-Doped Double-ShelledHollow Carbon Spheres and Wrapped with Graphene for Li−SBatteries. Adv. Energy Mater. 2015, 5, 1402263.(101) Pang, Q.; Tang, J.; Huang, H.; Liang, X.; Hart, C.; Tam, K. C.;Nazar, L. F. A Nitrogen and Sulfur Sual-Soped Carbon Derived fromPolyrhodanine@ Cellulose for Advanced Lithium−Sulfur Batteries.Adv. Mater. 2015, 27, 6021−6028.(102) Wang, Z.; Dong, Y.; Li, H.; Zhao, Z.; Wu, H. B.; Hao, C.; Liu,S.; Qiu, J.; Lou, X. W. Enhancing lithium−sulphur battery performanceby strongly binding the discharge products on amino-functionalizedreduced graphene oxide. Nat. Commun. 2014, 5, 5002.(103) Liang, X.; Garsuch, A.; Nazar, L. F. Sulfur cathodes based onconductive MXene nanosheets for high-performance lithium−sulfurbatteries. Angew. Chem., Int. Ed. 2015, 54, 3907−3911.(104) Liang, X.; Hart, C.; Pang, Q.; Garsuch, A.; Weiss, T.; Nazar, L.F. A highly efficient polysulfide mediator for lithium−sulfur batteries.Nat. Commun. 2015, 6, 5682.(105) Wiberg, N.; Holleman, A. F.; Wiberg, E. Holleman-Wiberg’sInorganic Chemistry; Academic Press: San Diego, CA, 2001.(106) Qian, D.; Ma, C.; More, K. L.; Meng, Y. S.; Chi, M. Advancedanalytical electron microscopy for lithium-ion batteries. NPG AsiaMater. 2015, 7, e193.(107) Wang, C. M. In situ transmission electron microscopy andspectroscopy studies of rechargeable batteries under dynamicoperating conditions: A retrospective and perspective view. J. Mater.Res. 2015, 30, 326−339.(108) Chen, D.; Indris, S.; Schulz, M.; Gamer, B.; Monig, R. In situscanning electron microscopy on lithium-ion battery electrodes usingan ionic liquid. J. Power Sources 2011, 196, 6382−6387.(109) Baggetto, L.; Hah, H. Y.; Jumas, J. C.; Johnson, C. E.; Johnson,J. A.; Keum, J. K.; Bridges, C. A.; Veith, G. M. The reaction mechanismof SnSb and Sb thin film anodes for Na-ion batteries studied by X-raydiffraction, 119Sn and 121Sb Mossbauer spectroscopies. J. Power Sources2014, 267, 329−336.(110) Chen, R.; Heinzmann, R.; Mangold, S.; Chakravadhanula, V. S.K.; Hahn, H.; Indris, S. J. Phys. Chem. C 2013, 117, 884−893.(111) Ogata, K.; Salager, E.; Kerr, C. J.; Fraser, A. E.; Ducati, C.;Morris, A. J.; Hofmann, S.; Grey, C. P. Revealing lithium−silicidephase transformations in nano-structured silicon-based lithium ionbatteries via in situ NMR spectroscopy. Nat. Commun. 2014, 5, 3217.(112) Gotoh, K.; Ishikawa, T.; Shimadzu, S.; Yabuuchi, N.; Komaba,S.; Takeda, K.; Goto, A.; Deguchi, K.; Ohki, S.; Hashi, K.; Shimizu, T.;Ishida, H. J. Power Sources 2013, 225, 137−140.(113) Berkes, B. B.; Jozwiuk, A.; Vracar, M.; Sommer, H.;Brezesinski, T.; Janek, J. On-line continuous flow differentialelectrochemical mass spectrometry with a realistic battery setup forhigh-precision, long-term cycling tests. Anal. Chem. 2015, 87, 5878−5883.(114) Berkes, B. B.; Jozwiuk, A.; Sommer, H.; Brezesinski, T.; Janek,J. Simultaneous acquisition of differential electrochemical massspectrometry and infrared spectroscopy data for in situ character-ization of gas evolution reactions in lithium-ion batteries. Electrochem.Commun. 2015, 60, 64−69.

(115) Castel, E.; Berg, E. J.; El Kazzi, M.; Novak, P.; Villevieille, C.Differential Electrochemical Mass Spectrometry Study of the Interfaceof xLi2MnO3·(1−x)LiMO2 (M = Ni, Co, and Mn) Material as aPositive Electrode in Li-Ion Batteries. Chem. Mater. 2014, 26, 5051−5057.(116) Jain, A.; Persson, K. A.; Ceder, G. Research Update: theMaterials Genome Initiative: Data Sharing and the Impact ofCollaborative Ab Initio Databases. APL Mater. 2016, 4, 053102.(117) Jain, A.; Shin, Y.; Persson, K. A. Computational Predictions ofEnergy Materials Using Density Functional Theory. Nature ReviewsMaterials 2016, 1, 15004.

Chemistry of Materials Review

DOI: 10.1021/acs.chemmater.6b02726Chem. Mater. 2017, 29, 90−105

105