j ourna l homepage: www.e lsev ie r.com/ locate /ss i
In-situ neutron diffraction study of the MoS2 anode using a custom-builtLi-ion battery
Neeraj Sharma a,⁎, Guodong Du b, Andrew J. Studer a, Zaiping Guo b, Vanessa K. Peterson a
a The Bragg Institute, Australian Nuclear Science and Technology Organisation, Locked Bag 2001, Kirrawee DC, NSW 2232, Australiab School of Mechanical, Materials and Mechatronic Engineering, University of Wollongong, Wollongong, NSW 2522, Australia
This work presents the first in-situ neutron diffraction study of the MoS2 electrode, undertaken in a custom-built Li-ion battery during discharge. A review of custom-designed cells for in-situ neutron diffractionexperiments is presented along with our optimised cell, which we use to show real-time informationcorresponding to Li-insertion into MoS2 via disappearance of the (103) reflection and increase in the d-spacing of the (002) reflection. The changes in the diffraction patterns begin at the 1.1 V plateau and arecomplete during the 0.5 V plateau. Sequential Rietveld-refinement indicates the presence of an intermediatelithiated phase (LixMoS2) between MoS2 and LiMoS2. We observe the disappearance of all reflections for theMoS2, corresponding to the loss of long-range order, during the 0.5 V plateau and no new diffraction peaksappear with further electrochemical cycling. This result is indicative of a transformation from long-rangeordered MoS2 to short-range ordered LiMoS2, a result that we confirm using ex-situ synchrotron X-ray andneutron diffraction.
Rechargeable Li-ion batteries providing high capacities are widelyused in portable electronic devices [1]. To meet future energydemands, particularly for widespread use in electric vehicles, Li-ionbattery technology needs to provide higher energy densities whilebeing safe and environmentally friendly. Therefore, new materials forcomponents in Li-ion batteries are being developed, characterised, andtested in commercially designed cells [1–4]. In order to direct batteryresearch aimed at improving batteries, scientists and engineers needto understand the processes that occur within a commercial Li-ionbattery as it is being charged and discharged.
Graphite-based anodes are widely used in commercial Li-ionbatteries. The limitations of graphite-based anodes include a lowtheoretical capacity (372 mA h/g) and the formation of a solidelectrolyte interface layer on the anode surface. Si and Sn anodeshave attracted interest due to their high capacities. However, theseanodes have large volume changes as a consequence of alloying andde-alloying with Li during charge/discharge cycles, resulting inlimited cycle life [1,5,6]. Alternatively, layered transition-metaldichalcogenides (MX2, where M=Ti, Nb, Mo, Ta and X=S, Se, Te)represent a class of compounds whose functionality as an anode isachieved through the diffusion of Li+ into the inter-layer space be-tween host layers. In particular, MoS2 is an attractive anode material
+61 2 9717 3606.rma).
11 Published by Elsevier B.V. All rig
due to the relatively weak van der Waals interaction between theMoS2 layers leading to its high theoretical capacity [7,8].
Electrochemical Li+ insertion/extraction, and correspondingstructural changes of, MoS2 have not been thoroughly reconciledwithin the literature [2]. In the first discharge step (Li insertion) thereare electrochemical plateaus at ~1.1 and 0.6 V, with the 1.1 V plateaurelated to the formation of LixMoS2 or Li intercalation into defect sitesof MoS2. Notably, a major reversible plateau is observed uponcharging at ~2.2 V and on subsequent discharging a ~2 V plateauappears, while the ~1.1 and 0.6 V plateaus are less pronounced andshifted to slightly different potentials [7–9]. Some of these plateausdisappear and slightly different behaviour is observed when cycling aMoS2 cell between 3 and 0.3 V when compared with 3–0.01 V.Depending on the MoS2 sample (e.g. synthesis conditions) the firstcycle capacity ranges from 600 to 1100 mA h/g and decreases to andstabilises around 200 mA h/g in subsequent cycles [2,7–9]. This largecapacity decay is widely noted, and strategies to improve and retaincapacities have been put forward, such as the absorption ofcomposites onto the MoS2 layers. Intercalation [9] and structuralchanges [2] are thought to play a critical role in the first cycle.Therefore, it is important to investigate the electrochemical mecha-nisms of charge/discharge in the first cycle, particularly in its earlystages, to determine the cause of capacity degradation. To date, twomechanisms of capacity degradation in MoS2 have been suggested inthe literature. The first mechanism is related to Li insertion into MoS2on tubular sites in nanorods [2,9,10], where the tubular sites trap Li,leading to the capacity loss [10]. The second mechanism describes theformation of Mo and Li2S during the first discharge [7]. To measure
38 N. Sharma et al. / Solid State Ionics 199–200 (2011) 37–43
changes in a MoS2 anode during the initial stages of the first dischargewe undertake in-situ neutron diffraction (ND) experiments.
Structural studies of electrodes, in particular the anode, aredifficult to investigate by in-situ techniques. Despite these difficulties,in-situ X-ray diffraction (XRD) is routinely used to track structuralchanges of electrodes of Li-ion batteries [11–16]. Modified coin orthin-film pouch cells (e.g. Bellcore's PLiON-type) [17–19] have beenused, which are constructed from components that are virtuallytransparent to X-rays, while the electrode of interest, in many casesthe cathode, is investigated as a function of charge/discharge. As aconsequence of the nuclear scattering mechanism, ND is moresensitive to Li atoms than XRD. This makes ND advantageous in thein-situ studies of Li-ion batteries. Analysis of ND data can providestructural information that is both similar (e.g. lattice parameters) andcomplementary to XRD data. Neutrons are highly penetrating, thusreal-life batteries larger than coin cells can be probed. It is due to theseadvantages of ND for the analysis of Li-ion batteries that there is aneed to develop and test in-situ ND cells that mimic real-life batteries.
The first part of this work reviews Li-ion cell designs for in-situ NDand presents a new cell concept for this application. The second partdetails the first results obtained using this cell, obtained for a MoS2electrode, and discusses the results using in-situND alongwith resultsobtained from ex-situ ND and synchrotron XRD.
2. Electrochemical cell for in-situ neutron diffraction
2.1. Background
In-situ ND experiments on commercial Li-ion batteries haverevealed Li insertion/extraction into cathode and anode materials ofinterest, e.g. LiFePO4 [20], LiCoO2 [21,22], and graphite anodes [20–22]. Tracking Li location and crystal site occupation with charge/discharge is difficult because of the poor signal-to-noise ratios in thecollected ND patterns. The poor signal-to-noise ratios arise from factorssuch as incoherent neutron scattering from hydrogen-containingcomponents in the battery (such as the separator and electrolytesolution) and scattering from liquid components, both ofwhich increasethe background, and geometrical complications arising from the batteryshape. To mitigate these issues, studies have used specially designedcells, tailored to the requirements of ND, to maximise signal-to-noiseratios. These cells are designed to minimise unwanted incoherentscattering from the hydrogen-containing components and the liquidand to maximise the quantity of active electrode material in the
a b
Fig. 1. Specially designed electrochemical cells for in-situ neutron diffraction studies. (a) Cyliseparator soaked in hydrogen-containing electrolyte, D is the stainless steel current collectRef. [24] with permission from Elsevier. (b) Coin cell design [25,26] where 1 is the cell top, 2with glass fibres and Celgard® separator, and 5 is the active material compartment. Reprod
beam, without adversely affecting electrochemical performance.Therefore, in-situ ND studies fall into two categories, those thatinvestigate commercially available batteries (taking advantage of thepenetration depth of neutrons) and those that use cells with a designtailored to ND (enhancing the signal-to-noise ratio by mitigating theunwanted contribution from other scatterers). It is the latter,custom-designed electrochemical cells that may provide sufficientdetail to follow both Li location and occupation as a function of thestate-of-charge of the battery.
Two types of electrochemical cells previously designed for in-situdiffraction studies of Li-ion batteries are shown in Fig. 1a and b. Thefirst cell [23,24] (Fig. 1a), used to investigate LiMn2O4, has a cylindricalgeometry containing approximately 5 g of cathodemixture and uses astainless steel current collector. ND patterns were collected using thiscell at specific voltages during a slow discharge. Although thebackground in the ND patterns is significant, some structural peakswere able to be resolved. Using these data the authors were able tomodel and refine Li occupancies at selected potentials, however, noobservation of two-phase behaviour at high potentials was made, aresult attributed to the slow rate of charge used [24]. The second cell[25,26] (Fig. 1b), used to investigate LiNiO2 and Li4Ti5O12, has a coincell geometry, but the relatively thick electrode used results in anover-potential of approximately 120 mV. Some of the challengesassociated with in-situ experimentation, e.g. effect of separator andelectrolyte solution, are highlighted in this work [25]. Work using thiscell on Li4Ti5O12 was the first anode material to be examined with anin-situ cell custom-built for ND experiments. This work demonstratedthat subtle changes in Li positions can be determined at specificvoltages.
2.2. Cell development
A primary concern in the design of cells for ND is the presence ofhydrogen in components such as the electrolyte and separator.Incoherent scattering from hydrogen is the dominant cause of thereduced signal-to-noise in the diffraction patterns through itscontribution to the background (which arises from the sample andcannot be reduced by increasing measurement times). The effect ofhydrogenated electrolyte in increasing the background is significant;1 mL of hydrogenated electrolyte added to ~1 g of MoS2 powderresults in features corresponding to MoS2 in the diffraction patternbecoming indistinguishable from the background. Even withouthydrogen, the presence of a liquid component adds to the background
ndrical cell [23,24] where A are brass plugs, B is a Pyrex® tube lined with Li foil, C is theor, and E is the active material mixed with carbon black and binder. Reproduced fromis a spring, 3 and 6 are the current collectors, 4 is the cell body comprised of electrodesuced [25] with permission of the International Union of Crystallography.
39N. Sharma et al. / Solid State Ionics 199–200 (2011) 37–43
as the scattering from the liquid is not coherent as from a crystallinestructure. Therefore, both hydrogen-containing compounds and theliquid content of the cell were minimised in our design. A minimalamount of electrolyte solution was used, which was deuterated, tominimise the contribution from this component to the background inthe diffraction pattern. The largest remaining hydrogen-containingcomponent is Celgard®, which is used as an insulator of the activecomponents from the vanadium can and as a separator. The insulatingfunction of Celgard® can be replaced by poly-vinyl difluoride (PVDF)film, but no suitable separator film with ion-transport capabilities asefficient as Celgard® were found and Celgard® remained in theneutron beam for these experiments. PVDF, even though it containshydrogen, was used as a binder and insulating film as no significantdiffraction features or loss in the signal-to-noise ratio were observedupon substitution. Replacing PVDF with polytetrafluoroethylene(PFTE) virtually eliminates hydrogen, however, PFTE has a diffractionfeature that overlaps with those of MoS2, and shows a noticeable non-uniform contribution to the background of ND patterns.
Electrochemical cells were constructed usingMoS2 (Sigma Aldrich,99%, particle size b2 μm) mixed with carbon black and PVDF toform a paste and applied to a Cu sheet, followed by drying overnight.Layers of materials were arranged in the following order: Celgard®(insulator), Cu with MoS2 electrode, Celgard® (separator), and Limetal. Cu wires were placed in contact with both electrodes. Theelectrolyte was added drop-wise and was composed of 1 M LiPF6dissolved in a 1:1 vol.% mixture of deuterated ethylene carbonate(CDN, isotopic purity 99%) and deuterated dimethyl carbonate(Cambridge Isotopes, isotopic purity 99%). This assembly was rolled(using the outer Celgard® layer) and inserted into a vanadium can ofinner diameter 9 mm, which was then sealed. These procedures werecarried out in an Ar-filled glovebox. Fig. 2 illustrates this electro-chemical cell for use during in-situND studies, whichwe term the roll-over design. An additional electrochemical cell was constructed usingLi4Ti5O12 in an identical procedure to the MoS2 containing cells.Li4Ti5O12 was synthesised via the protocol available in the literature[27].
The size and geometry of cylindrical batteries is ideal for in-situ NDexperiments andwe use a roll-over cylindrical type cell to collect real-time data during electrochemical cycling, unlike the two previous celldesigns (the cylindrical and the coin cell types) where cells werecharged ex-situ (without continuous data acquisition) and datacollectedwith the cell contents in an equilibrium or quasi-equilibriumstate. As with previous cylindrical cells [23,24] used for in-situ ND, ourcell design mimics the commonly available commercial 18650-type[28] battery.
Seal
Cu wire LiLi MoS2 electrode on Cu
Rolled sheets
Vanadium CanCelgard
Fig. 2. A schematic of a section through the ‘roll-over’ cell for in-situ neutron diffractionstudies. Sheets of Celgard® (white), MoS2 electrode on Cu (red), Celgard®, and Li(blue), are rolled and sealed inside a vanadium can (green). The can is sealed in anargon-filled glovebox.
3. Data collection
In-situ ND data were collected on the high-intensity powderdiffractometer, Wombat, at the Open Pool Australian Light-water(OPAL) reactor facility at the Australian Nuclear Science andTechnology Organisation (ANSTO) [29]. The electrochemical cell wasplaced in a neutron beam of wavelength (λ)=2.406(2) Å andpatterns collected every 5 min for 36 h in the two-theta (2θ) range16≤2θ≤136°. Wombat features an area detector covering 120° in 2θ,enabling diffraction data to be continuously collected rather than a 2θ-step-scan type acquisition. The combination of Wombat's relativelyintense neutron beam and its area detector makes Wombat an idealinstrument for in-situ ND studies of this type. Data correction,reduction, and visualisation were undertaken within the programLAMP [30]. The electrochemical cell was simultaneously electrochem-ically cycled in galvanostatic (constant current) mode with appliedcurrents ranging from ±1 to 30 mA using a Neware battery testingdevice. Rietveld refinements were carried out using the GSAS [31]suite of programs with the EXPGUI [32] interface.
Ex-situ high-resolution synchrotron XRD data were collected for10 min on the Powder Diffraction beamline (10-BM) [33] at theAustralian Synchrotron with λ=0.82398(1) Å. Ex-situ ND data werecollected on the high-resolution powder diffractometer, Echidna, atthe OPAL reactor facility at ANSTO [34]. Data were collected at bothλ=2.4412(11) and 1.6225(1) Å for 5 h in the 2θ range 5≤2θ≤158°.For ex-situ diffraction patterns, MoS2 was extracted and washed withethylene carbonate and dimethyl carbonate and left to dry overnightunder argon. The electrodes were then sealed in quartz capillaries forex-situ XRD or in vanadium cans for ex-situND. These were carried outin an argon-filled glove bag to minimise reaction with air.
4. Results and discussion
4.1. Electrochemical performance of the cell
Galvanostatic charge/discharge procedures were performed off-line to ensure the electrochemical performance of the roll-overelectrochemical cell was acceptable. From these data we estimate thefirst cycle capacity of our in-situ NDMoS2 cell to be 670 mA h/g whichis within reported capacity ranges for coin cell experiments [2,7–9],but there is a significant decrease in the maximum current rate thatcan be applied for stable battery performance. Hence, the roll-overelectrochemical cell used here required that relatively small currentsbe used during the first discharge (1–10 mA), despite the presence of200–500 mg of active electrode material. Larger currents are able tobe applied during charging, although their application was noted tocause some voltage spikes, presumably due to the higher internalresistance of this electrochemical cell relative to the coin-cell type.Some of these differences in electrochemical performance of the in-situ ND cell compared to a coin cell, may arise due to isotope effects(using deuterated electrolyte solution) or the reduced quantity ofelectrolyte used. The electrodes used for these experiments had amaximum of 500 mg of active material, MoS2, which is less than theprevious in-situ experiments [23,24]. We were able to obtain ND datafrom these small amounts of MoS2, relative to other studies, a resultarising due to the capacity of Wombat to measure smaller samplescombined with the enhanced signal-to-noise ratio obtained from thisspecialist roll-over cell design. The signal-to-noise ratio of the data islimited by the components within the battery itself and our custom-made battery maximises this ratio. ND data were collected for 5 minas no worthwhile gain in neutron signal statistics were obtained withlonger collection times.
ND patterns collected using this cell clearly show powderdiffraction features from MoS2, see Fig. 3a. A second roll-over cellwas constructed using 500 mg of Li4Ti5O12, another anode materialwhich is more crystalline and thus gives a larger diffraction signal
a
2θ (degrees)
Cou
nts
(arb
.)
20 40 60 80 100 120
0
2
4 x
10E
3
LiCu
MoS2
Carbon Black
20 40 60 80 100 1202500
3000
3500
4000
4500
5000
5500
6000 Li4Ti5O12
Cu
2θ (degrees)
Lib
Cou
nts
(arb
.)
(002
)
(103
)
Fig. 3. (a) Amulti-phase Rietveld refinement plot using a 5 minute ND pattern collectedon Wombat with our custom-designed MoS2 anode battery at λ=2.406(2) Å.Structural models for Li, Cu, and MoS2 are used. The calculated pattern is shown inblack, observed data in red, and the difference between the observed and calculated inpurple. Reflection markers are shown as vertical bars. (b) A 5 minute ND patterncollected on Wombat with our custom-designed Li4Ti5O12 anode battery at λ=1.535(1) Å. Intense reflections from Li, Cu, and Li4Ti5O12 are indicated by reflection markersshown as vertical bars.
20.8
22.5
24.2
25.9
(002) of MoS2
(103) of MoS2
62
65
68
a
b
c
d
3.4
3.3
3.2
3.1
12.8
12.6
12.4
12.20 500 1000 1500 2000
1.5
1.0
0.5
0
0
- 3
- 6
- 9
0 500 1000 1500 2000
ac
a la
ttice
par
amet
er
c la
ttice
par
amet
erC
urre
nt (
mA
)
Time (minutes)
Vol
tage
(V
)2θ
(de
gree
s)
2θ (
degr
ees)
Fig. 4. Selected regions (20.8°≤2θ≤25.9° and 62°≤2θ≤68°) of the in-situ ND dataobtained from the MoS2 cell as a function of time showing discharge evolution of:(a) The (002) peak with a reference line (guide to the eye) showing the 2θ shift.(b) Rietveld-refined a (black squares) and c (red circles) lattice parameters of MoS2.(c) The (103) MoS2 peak. (d) Measured voltage (black line) and applied current (red)correlated to the ND data. The reflections intensities are scaled.
40 N. Sharma et al. / Solid State Ionics 199–200 (2011) 37–43
than MoS2. Fig. 3b shows ND data for the Li4Ti5O12 battery collectedfor 5 min. The ‘background’ in the data arises from the amorphous,liquid and hydrogen-containing compounds in the battery, whichinclude the electrolyte, carbon black, binder and separator. Thepurpose of the Li4Ti5O12 battery was to allow comparison of thesignal-to-noise ratio relative to other cells used for in-situ ND [26]using the same electrode. Using our roll-over cell design we observecomparable signal-to-noise ratios for Li4Ti5O12 collected for 5 min onWombat to those obtained using a coin cell-type configuration [26]. Akey difference between these data is the collection time, which was5 min on Wombat and up to 9 h for the previous [26] work. Our celldesign was intentionally tailored for use on Wombat, takingadvantage of its fast data collection capability. Furthermore, ourdesign overcomes some electrochemical issues faced in previousdesigns [23–26], in particular, the quantity of electrode material andits contact with the current collector and separator. Previous designsuse thick electrodes, while in this design we spread the electrode overa current collector with a large surface area which is then rolled-over.Therefore, we can satisfy the need for thin electrodes with goodcontact to current collector and separator, and to some extent, controlthe quantity of electrode material by introducing more rolls orreducing the number rolls. The electrodes are thin, so polarisation
effects (differences between the voltage under equilibrium and thatwith a current flow) are minimised and applied currents are higherthan previously applied to in-situ cells [23–26]. We note that thespikes in both current and voltage, as shown in Fig. 4, highlight thatlimits to current density still exist for these custom-built batteries dueto their internal resistance. Therefore, the combination of our celldesign and the use of Wombat permits us to probe structural changesin real-time as a function of electrochemical cycling, possibly at highcurrent rates (e.g. 0.1–1C, where 1C is 1Li+ in 1 h).
Fig. 3a shows themulti-phase Rietveld refinement using a 5 minuteND pattern collected on Wombat of an un-cycled MoS2 cell. For theRietveld refinement the structural models of MoS2 [35], Cu [36], andLi [37] were used, while a broad peak shape approximation was usedfor carbon black centred at 2θ=38°. The refined cell parameters forMoS2 were a=3.149(5), c=12.250(26), the z atomic position for theS atom is 0.618(9), and atomic displacement parameters for Mo andS are 0.08(3) and 0.22(8) Å2, respectively. The figures of merit for therefinement were the profile factor (Rp)=2.36%, weighted profilefactor (wRp)=3.09%, and goodness-of-fit term (χ2)=3.741. Usingλ=2.406(2) Å we are able to better resolve the two MoS2 reflections
10 20 30 40 50 60 70 80
10000
20000
30000
40000
Rel
ativ
e in
tens
ity
2θ (degrees)
2θ (degrees)5 10 15
Fig. 5. Ex-situ synchrotron XRD pattern (10 minute exposure) of the extracted MoS2electrode after discharging in a coin cell.
41N. Sharma et al. / Solid State Ionics 199–200 (2011) 37–43
that are of particular interest, the (002) at 2θ=23.7° and (103) at2θ=65.3°.
4.2. In-situ structure of MoS2
In-situ ND patterns were collected during the first stages ofdischarge of the MoS2 cell. The observed (002) and (103) reflectionsof MoS2 are shown in Fig. 4 and are correlated with the dischargecurve and lattice parameters. Variation in the 2θ value of the (002)reflection (the inter-lattice spacing, d-spacing) corresponds to Liinsertion into MoS2 producing LixMoS2. Li insertion into MoS2 occursduring the 1.1 V plateau highlighted by the gradual shift to lower 2θvalues (increase in the d-spacing of the (002) reflection), a step tohigher d-spacings at 900 min (indicated by the arrowhead in Fig. 4a),followed by further increases in the d-spacing. Overall, the (002)reflection shifts from 2θ=23.7° to 2θ=22.5° and then disappearsduring the 0.5 V plateau. A part of the transition is complete, as shownby the disappearance of the (103) reflection, during the course of the1.1 V plateau, possibly initiating the drop in voltage to the nextplateau region.
This behaviour is quantitatively reflected in the Rietveld refinedcell parameters (Fig. 4b). Sequential Rietveld-refinements wereundertaken using 5 minute datasets, with fixed atomic parameters,where the figures of merit varied from 2.49%≤Rp≤3.22%, 3.14%≤wRp≤4.13%, and 3.39≤χ2≤5.07. We observe an increase in the clattice parameter, followed by a step and further increase until1500 min, as shown by the changes in the 2θ value or d-spacing of the(002) reflection. In the case of the a lattice parameter the step isevident but there is no noticeable fluctuation, as expected, since the(103) reflection disappears. Following the step, the spread in the alattice parameter is larger, a consequence of the fact that the reflectionis disappearing and the reliability of the determination of theparameter diminishes. For both lattice parameters a prominentsecond step appears after 1500 min and indicating that subsequentchanges cannot be determined using the ND data. The (103) reflectiondisappears and the intensity of the (002) reflection begins to decreasefrom its maximum around this point in the discharge.
The reduction in reflection intensity is due to the formation ofLiMoS2 [8,38], a material that shows short structural coherence(~50 Å) or lack of long-range structural order, and has been struc-turally characterised by atomic pair distribution techniques [35]. Weobserve the MoS2 to LiMoS2 transformation during the 1.1 V plateau,which is completed during the 0.5 V plateau and corresponds to a lossof long-range order. The (002) remains visible in the 0.5 V plateauregion before disappearing at a time close to 2000 min, highlightingthe transformation of long-range ordered MoS2 to short-rangeordered LiMoS2. The MoS2 (002) reflection is the most intense re-flection of the relatively short-range ordered LiMoS2, determined byatomic pair distribution analysis [35] and the slow disappearance ofthe (002) reflection, relative to the (103) reflection, is consistent withthe transformation to the short-range ordered end member. We notethat once LiMoS2 is formed, shown by the transformation to a productwithout long-range order in the ND data, the crystallinity of the anodematerial does not recover. In-situ ND data were collected on offlinecycled MoS2 cells showing no features corresponding to MoS2 orMoS2-derived compounds. Using in-situ ND data on an un-cycledMoS2 cell we demonstrate the electrochemically driven overalltransformation from crystalline MoS2 to a short-range orderedmaterial, likely LiMoS2.
Fig. 4 reveals a linear increase in the c lattice parameter and a steparound 900 min. The initial insertion of Li into MoS2 results in thelinear increase of the c lattice parameter, as Li inserts betweenadjacent MoS2 layers. Li insertion into MoS2 with some degree oflithiation (i.e. LixMoS2 where xN0) results in a non-linear increase inlattice parameters (a step). This non-linear step in lattice parameterssuggests the formation of an intermediate phase, LixMoS2 with a~3.11
and c~12.43 Å, which is linked to a change in oxidation state of Mo.The structure of this phase is likely to express features similar to theend-member lithiated product, short-range ordered LiMoS2, and thusunable to be precisely determined using this method, which is limitedto the measurement of long-range structural information. In-situ NDusing even custom-designed cells such as presented here presentsignificant difficulties for the determination of short-range order, suchas analysis using the pair-distribution function (PDF), due to theamount of material contributing to the diffraction data and thecomplexity in resolving the contributions arising from the phase ofinterest. Re-design of the electrochemical-cell to allow diffractiondata only from the material of interest, in this case the anode, to becollected, may overcome the major roadblock to in-situ PDF in-vestigations. Our in-situ ND investigation does show the real-timeevolution of lattice parameters and the presence of new intermediatephases.
4.3. Ex-situ measurements of MoS2 electrodes
Ex-situ ND and synchrotron XRD techniques provide a viablemeans to obtain structural information from electrodes of interestwithout the need for often difficult and complex in-situ experimen-tation. Fig. 5 shows an ex-situ synchrotron XRD pattern of the MoS2anode extracted from a battery discharged to 0.1 V. Fig. 6a and c showex-situ ND patterns of the extracted MoS2 electrode at different stagesin the electrochemical cycle (Fig. 6d). Fig. 6a is an ex-situ ND patterncollected at 0.3 V, after the transformation to the short-range orderand for comparison, an in-situND patternwith theMoS2 electrode stillin the battery at 0.3 V is also shown (Fig. 6b). In Fig. 6b only Li, Cu, andC were modelled as no features representative of MoS2 or MoS2-derived compounds were evident. Fig. 6c shows an ex-situ ND patterncollected after discharging to 0.1 V and re-charging to 3 V. The labelsin Fig. 6d show where the MoS2-anodes were extracted for ex-situexperimentation and the region of the electrochemical cycle probedby in-situ ND (shaded).
The ex-situ synchrotron XRD data in particular show broadfeatures and a lack of intense peaks suggesting a loss in long-rangecrystalline order of MoS2. In order to obtain ND data which isstatistically equivalent to synchrotron XRD data, unreasonably longcounting times are required, and so the collected data are statisticallyunsaturated. Ex-situ ND data were collected for 5 h at two incidentneutron wavelengths to probe an appropriate range of d-spacings inthe material. Despite this, the ex-situ ND data clearly show the samelack of diffraction features as the ex-situ synchrotron data. This resultis consistent with the lack of features resembling crystalline MoS2 or
20 40 60 80 100 120 140 160300
400
500
600
700
Discharged to 0.3 V
2θ (degrees)
Cou
nts
(arb
.)
20 40 60 80 100 120
x10E
3 0
2
4
LiCuC
20 40 60 80 100 120 140 160
400
500
600
700Charged to 3V
0 600 1200 1800 2400 3000
0
1
2
3
Vol
tage
(V
)
Time (minutes)
-30
-20
-10
0
10
20
30
Cur
rent
(m
A)
a b
dc
λ = 1.6225 Å Å
Å λ = 2.4412
In-situ cell discharged to 0.3 V
λ = 2.406
(a)
(b)
(c)
2θ (degrees)
2θ (degrees)
Cou
nts
(arb
.)C
ount
s (a
rb.)
Fig. 6. Ex-situ ND pattern (5 hour exposure) of the MoS2 cell (a) discharged to 0.3 V and (b) the multi-phase Rietveld refinement using a ND pattern of the battery containing theMoS2 electrode at the same point in the discharge curve, with the calculated pattern in black, observed data in red, and the difference between the observed and calculated in purple.Ex-situ ND pattern of (c) the MoS2 cell charged to 3 V. The discharge/charge curve (d) of the custom-designed MoS2 cell with measured voltage (black line) and applied current (redline). Labels in (d) indicate where the ex-situ samples shown in this figure were extracted from and the shaded region represents the section of the discharge/charge cycle probed byin-situ ND.
42 N. Sharma et al. / Solid State Ionics 199–200 (2011) 37–43
MoS2-derived compounds in the in-situ ND data collected at a similarstate-of-charge, see Fig. 6b at 0.3 V. Note that exposure to air canresult in similar ex-situ data via degradation of the samplecrystallinity, however, given the similar results obtained from thein-situ ND data (Figs. 4 and 6b) and the care taken during theextraction method to exclude air we believe these data represent lossof crystallinity as a result of Li-insertion processes. Comparatively, a5 hour dataset with an ex-situ discharged and extracted MoS2electrode on the high resolution neutron diffractometer providessimilar signal-to-noise ratios as a 5 minute collection on a new MoS2battery on the high-intensity diffractometer.
For an anode material, lack of crystallinity is not necessarily alimitation for electrochemical performance, for example, somegraphiticanodes used in Li-ion batteries are mostly amorphous, yet showreasonable performance [3]. If the MoS2 electrode reduces in crystal-linity or in particle size in the first cycle this may cause anomalouselectrochemical behaviour of these electrodes in the first cycle relativeto later cycles. In this case ex-situ experimentation is not suitable andtime-dependent in-situ experimentation is required to link thestructural behaviour of MoS2 to its electrochemical performance.
5. Conclusions
A custom-designed electrochemical cell was successfully used toextract real-time information using in-situ neutron diffraction during
the first stages of discharge (Li insertion) into MoS2. We demonstratethat structural information of anode materials such as MoS2 can becorrelated to electrochemical behaviour. The non-uniform variation inlattice parameter indicates the presence of an intermediate phase,LixMoS2, that is formed during the transition from MoS2 to LiMoS2,with lattice parameters a~3.11 and c~12.43 Å. We observe a loss ofcrystallinity resulting from the Li-insertion process into MoS2,characteristic of the formation of LiMoS2, a material with knownshort-range order only. Ex-situ neutron and synchrotron diffractiondata show only a static picture of the electrochemical process, and thissnapshot can be contaminated, e.g. degradation of componentcrystallinity via reaction with air. Real-time in-situ neutron diffractiondata shows that the first 1.1 V and part of the 0.5 V plateau in theelectrochemical discharge curve arise primarily as a result of theMoS2to LiMoS2 transition, which is accompanied by the loss of long-rangestructural order in these anode compounds. Whilst the loss ofcrystallinity does not negatively impact on the anodic performancein Li-ion batteries, it limits the detail that can be extracted from eitherex-situ or in-situ traditional powder diffraction.
We show that our cell design is practical for in-situ neutron dif-fraction studies, with only a slight perturbation in the electrochem-istry relative to the more commonly-used coin-type cell. Our celldesign is sensitive to weakly diffracting materials and we antici-pate its use to further the understanding of the mechanisms of Liinsertion/extraction from a variety of electrode materials and to
43N. Sharma et al. / Solid State Ionics 199–200 (2011) 37–43
characterise the phases and their transitions with electrochemicalcycling.
Acknowledgements
Part of this project was supported by the Australian ResearchCouncil (ARC) through an ARC Discovery project (DP0878611). Thisresearch was undertaken on the Powder Diffraction beamline at theAustralian Synchrotron, Victoria, Australia and wewould like to thankDr. Justin Kimpton for his assistance.
References
[1] J.-M. Tarascon, M. Armand, Nature 414 (2001) 359–367.[2] R. Dominko, D. Arcon, A. Mrzel, A. Zorko, P. Cevc, P. Venturini, M. Gaberscek, M.
Remskar, D. Mihailovic, Adv. Mater. 14 (2002) 1531–1534.[3] M. Winter, J.O. Besenhard, M.E. Spahr, P. Novak, Adv. Mater. 10 (1998) 725–763.[4] J.B. Goodenough, Y. Kim, Chem. Mater. 22 (2009) 587–603.[5] M. Winter, J.O. Besenhard, Electrochim. Acta 45 (1999) 31–50.[6] A.N. Dey, J. Electrochem. Soc. 118 (1971) 1547–1549.[7] J. Xiao, D. Choi, L. Cosimbescu, P. Koech, J. Liu, J.P. Lemmon, Chem. Mater. 22
(2010) 4522–4524.[8] G. Du, Z. Guo, S. Wang, R. Zeng, Z. Chen, H. Liu, Chem. Commun. 46 (2010)
1106–1108.[9] C. Feng, J. Ma, H. Li, R. Zeng, Z.P. Guo, H. Liu, Mater. Res. Bull. 44 (2009) 1811–1815.
[10] G.X. Wang, S. Bewlay, J. Yao, H.K. Liu, S.X. Dou, Electrochem. Solid-State Lett. 7(2004) A321–A323.
[11] J.N. Reimers, J.R. Dahn, J. Electrochem. Soc. 139 (1992) 2091–2096.[12] S.T. Myung, S. Komaba, N. Hirosaki, N. Kumagai, K. Arai, R. Kodama, I. Nakai,
J. Electrochem. Soc. 150 (2003) A1560–A1568.[13] C. Baehtz, Th. Buhrmester, N.N. Bramnik, K. Nikolowski, H. Ehrenberg, Solid State
Ionics 176 (2005) 1647–1652.[14] M.A. Rodriguez, M.R. Keenan, G. Nagasubramanian, J. Appl. Crystallogr. 40 (2007)
1097–1104.[15] K.W. Nam, W.S. Yoon, H. Shin, K.Y. Chung, S. Choi, X.Q. Yang, J. Power Sources 192
(2009) 652–659.
[16] M.R. Palacin, F. Le Cras, L. Seguin, M. Anne, Y. Chabre, J.-M. Tarascon, G. Amatucci,G. Vaughan, P. Strobel, J. Solid State Chem. 144 (1999) 361–371.
[17] C. Lampe-Onnerud, J.O. Thomas, M. Hardgrave, S. Yde-Anderse, J. Electrochem.Soc. 142 (1995) 3648–3651.
[18] M.R. Palacin, G.G. Amatucci, M. Anne, Y. Chabre, L. Seguin, P. Strobel, J.M. Tarascon,G. Vaughan, J. Power Sources 81–82 (1999) 627–631.
[19] B. Gerand, A. Blyr, A. Du Pasquier, J.B. Leriche, L. Seguin, J. Power Sources 81–82(1999) 922–924.
[20] M.A. Rodriguez, M.H. Van Benthem, D. Ingersoll, S.C. Vogel, H.M. Reiche, PowderDiffr. 25 (2010) 143–148.
[21] M.A. Rodriguez, D. Ingersoll, S.C. Vogel, D.J. Williams, Electrochem. Solid-StateLett. 7 (2004) A8–A10.
[22] N. Sharma, V.K. Peterson, M.M. Elcombe, M. Avdeev, A.J. Studer, N. Blagojevic, R.Yusoff, N. Kamarulzaman, J. Power Sources 195 (2010) 8258–8266.
[23] O. Bergstom, A.M. Andersson, K. Edstrom, T. Gustafsson, J. Appl. Cryst. 31 (1998)823–825.
[24] H. Berg, H. Rundlov, J.O. Thomas, Solid State Ionics 144 (2001) 65–69.[25] F. Rosciano, M. Holzapfel, W. Scheifele, P. Novak, J. Appl. Cryst. 41 (2008) 690–694.[26] J.-F. Colin, V. Godbole, P. Novák, Electrochem. Commun. 12 (2010) 804–807.[27] Y. Qi, Y. Huang, D. Jia, S.-J. Bao, Z.P. Guo, Electrochim. Acta 54 (2009) 4772–4776.[28] IEC 61960–2:2001 Secondary lithium cells and batteries for portable applications,
International Electrotechnical Commission, 2001.[29] A.J. Studer, M.E. Hagen, T.J. Noakes, Physica B 385–386 (2006) 1013–1015.[30] D. Richard, M. Ferrand, G.J. Kearley, J. Neutron Res. 4 (1996) 33–39.[31] A.C. Larson, R.B. Von Dreele, Los Alamos National Laboratory Report LAUR, 1994,
pp. 86–748.[32] B.H. Toby, J. Appl. Cryst. 34 (2001) 210–213.[33] K.S. Wallwork, B.J. Kennedy, D. Wang, AIP Conference Proceedings, 2007, p. 879.[34] K.-D. Liss, B.A. Hunter, M.E. Hagen, T.J. Noakes, S.J. Kennedy, Physica B 385–386
(2006) 1010–1012.[35] V. Petkov, S.J.L. Billinge, P. Larson, S.D. Mahanti, T. Vogt, K.K. Rangan, M.G.
Kanatzidis, Phys. Rev. B 65 (2002) 0921051–0921054.[36] W.L. Bragg, Phys. Z. 23 (1922) 114–120.[37] R. Berliner, O. Fajen, H.G. Smith, R.L. Hitterman, Phys. Rev. B 40 (1989)
12086–12097.[38] Y. Sharma, N. Sharma, G.V. Subba Rao, B.V.R. Chowdari, Adv. Funct. Mater. 17
