November 2014
Insights into the morphological
changes undergone by the anode
in the lithium sulphur battery system
Anurag Yalamanchili
Yalamanchili Satya Lakshmi Devi Yalamanchili Umamaheshwara Rao, R.I.P.
Karnati Jhansi Lakshmi, R.I.P. Karnati Venkateshwarlu, R.I.P.
Teknisk- naturvetenskaplig fakultet UTH-enheten Besöksadress: Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0 Postadress: Box 536 751 21 Uppsala Telefon: 018 – 471 30 03 Telefax: 018 – 471 30 00 Hemsida: http://www.teknat.uu.se/student
Insights into the morphological changes undergone by
the anode in the lithium sulphur battery system
Anurag Yalamanchili
In this thesis, the morphological changes of the anode surface in lithium sulphur cell, during early cycling, were simulated using symmetrical lithium electrode cells with dissolved polysulphides (PS) in the electrolyte. Electron microscopy (SEM) was used as the principal investigation technique to study and record the morphological changes. The resulting images from the SEM were analysed and discussed. The initial surface structure of the lithium anode largely influenced the ensuing morphological changes taking place through lithium dissolution (pits) and lithium deposition (dendrites) during discharge and charge respectively. The rate of lithium dissolution and deposition was found to be linearly proportional to the current density applied to the cell and the effect of cycling on the anode was proportional to the total charge of the cell in general in agreement with the expected reaction. The effect of
self-discharge on the anode was also studied using photoelectron spectroscopy (XPS) in tandem with SEM. The results indicated that self-discharge, occurring in the form of corrosion of the anode SEI by PS reduction, was influenced by the altered
morphology of the cell after cycling.
The findings presented in this project can be understood as a preliminary description for the morphological changes in the anode and their influence in the performance of lithium sulphur battery, which can be further investigated by more advanced methods.
During discharge, three stages of electrochemical reactions take place at the cathode as shown in Figure 2 [34]. During the first stage (~2.3 V), elemental sulphur in solid phase (S8 (S)) is dissolved into the
electrolyte (S8 (L) ) and is then reduced to long chain (S82‐) ions. During the second stage, these sulphur
ions are reduced further to shorter chain ions by disproportionation reactions, where a particular species is simultaneously oxidised and reduced to form two separate products, as shown below:
→ 1
4
This stage typically involves fast reaction kinetics due to the liquid phase reactions witnessing a potential drop from 2.3 – 2.1 V. Further disproportionation of the S62‐ to shorter chain ions like S42‐ takes place
during the third stage at an almost stable potential (~2.1 V). During this stage the viscosity of the electrolyte increases due to the generation of anion species.
It was suggested that short chain anions like S42‐ typically formed polysulphides (PS) through
disproportionation rather than electrochemical reduction given by:
4 → 3
This disproportionation and redox of S42‐ and S62‐ takes place throughout most of this stage eventually
reactions take place both electrochemically as well as chemically. Zhang [36] provides general equations for these reactions as given below: → 1 2 2 (Electrochemical oxidation) 1 2 → (Chemical reduction) These reduced PS diffuse back from the anode to the cathode on the next subsequent charge cycle to be re‐oxidised and a subsequent diffusion to the anode causing a polysulphide “shuttle” between the electrodes. A significant capacity of the Li‐S cell is lost in driving the shuttle mechanism, resulting in underutilization of capacity. Under idle conditions, this mechanism carries on until the PS are reduced to form solid Li2S which can deposit at the anode or elsewhere [9,20,37]. In addition to the shuttle, the PS reduction reactions also cause three main parasitic mechanisms which challenge the use of LiSBs in practical applications: Active consumption of active sulphur and lithium materials to form solid PS through electrochemical reduction, resulting in capacity fade over subsequent cycles (low cycling efficiency) [20,38]. Passive consumption of active lithium material by dissolved PS to from reduced PS species through chemical reduction, resulting in capacity fade when idle (self‐discharge)[20,39]. Corrosion of the passivation layer (SEI) on the anode thereby aiding in the degradation of the anode [40]. The composition of the SEI in presence of sulphur containing electrolytes is discussed further in the forthcoming sections of this chapter.
2.1.3 The sulphur cathode
Elemental sulphur, due to its low conductivity (5 x 10‐30 S/cm at 25 oC), cannot be used as the standalone
PS in the electrolyte. Considering the given requirements for the electrolyte, conventional solvents for other lithium metal batteries based on carbonates, phosphates and esters have been ruled incompatible for LiSBs [52] since they tend to interact with PS anions through redox, nucleophilic and radical reactions [36,51]. Numerous other compatible solvents have been studied for application of to the Li‐S cell. Linear ether glyme based solvents such as 1‐2 dimethoxyethane (DME) and tetraethylene glycol dimethyl ether (TEGDME) have been preferred for LiSBs due to their ability to dissolve larger amounts of PS, low viscosity and faster reaction kinetics to the PS anions [53–55]. In general, a mixture of linear and cyclic ethers (e.g. 1‐3 Dioxalane (DOL)) with a relatively high solubility for PS are used. Solvents such as DME‐DOL [53,56,57], TEGDME‐DOL [45,58,59], and TEGDME‐THF [60] have shown acceptable performance in balancing initial discharge capacity and cycling efficiency. Wang et. al. reported that a ratio of DME‐DOL in either 1:1 or 2:1 yielded optimum results [54]. The typical chemical structures of DME (linear ether) and DOL (cyclic ether) are shown in Figure 3. Figure 3: The chemical structures of the linear and cyclic ether components used in this project 2.1.4.2 Electrolyte salts and additives Lithium salts are added to the solvents in order to obtain a high transference number for lithium and ionic conductivity. However it is important the chosen salt(s) is compatible with the PS dissolved in the electrolyte. Conventional lithium salts used in other lithium metal batteries like LiPF6, LiBF4 and LiBOB
problem during cycling. Xu and co‐workers have presented a comprehensive summary of the dendrite growth models in rechargeable lithium anodes [71]. Chazalviel first proposed a general model for origins and growth of ramified metallic electrodeposits in dilute salt solutions attributing the growth of these deposits to the development of a space charge region due to depletion of anions at the metal electrodes [72]. This model has been adapted by Brissot et. al. for models based on lithium metal batteries using binary electrolyte with lithium salts and a polymer electrolyte such as PEO [73]. DC polarisation of the cell creates a concentration gradient in the cell which is given by the equation: ∂C ∂x μ μ μ Where J is the localized effective current density, e is the elementary charge, D ambipolar diffusion constant and µa and µLi are the ionic mobilities of the anions and Li+ ions respectively. It was proposed
that when > , where CO is the initial salt concentration and L is the inter‐electrode distance, the
Experiments conducted to examine the deposition of lithium at various current densities showed that dendrite nucleation and growth in the shape of elongated dendrites occurred even when current densities were lower that Chazalviel’s limiting current density. The growth of these dendrites was attributed inhomogeneities in the local SEI. Also over multiple cycles, it was observed that dendrite nucleation and growth occurred earlier than Sand’s time due to the breakdown and repair of the SEI from the previous cycle [73–76]. The SEI plays a significant role in any initial morphological changes on the lithium anode. Barton and Brockis [77] proposed an alternate model for dendrite growth applicable in liquid electrolytes suggesting that growth of nucleated dendrites was controlled by the spherical diffusion of cations. Monroe and Newman adopted this model with the thermodynamic reference points of the lithium anode and the conditions associated with ion concentrations proposed by Brissot et. al. [78]. They found that, in liquid electrolytes, dendrite growth rate increases across the surface and is greatly dependent on applied current density. The velocity at which the tip of the dendrite propagates is given by:
Where Jn is the effective current density normal to the hemispherical lithium dendrite tip, Vm is the
Li2O and organic lithium salts such as lithium alkoxy species: ROLi, RCOLi, RCO2Li and salt anions and contaminants [82]. Surface films dominated by ionic species cannot accommodate the volume changes of the anode due to their inelastic physical properties, eventually breaking down under deposition [83]. A study by Morigaki and Ohta advocated that inorganic lithium compounds such as Li2O tend to form along the ridge lines and grain boundaries of the lithium anode, suggesting an SEI dominated locally by ionic species [84]. The study of the surface films formed in DOL showed that similar surface species dominated the SEI only with the addition of oligomer products of polydioxolane [85–87]. These oligomers are insoluble and stick to the lithium anode thereby leading to the formation of elastomers, which provide flexibility to the SEI allowing it to accommodate the volume changes of the anode during cycling. The effect of DOL and ethers forming alkoxy species such as DME is shown in Figure 8 [83]. Figure 8: The role of various surface species and the SEI species in morphological changes (Reprinted from [83] Copyright 2000, with permission from Elsevier) A major influence on the SEI composition in a Li‐S cell is the presence of dissolved PS in the electrolyte. Aurbach et. al. conducted a study examining the chemical composition of the SEI using on the lithium metal etched in a PS dissolved electrolyte solution with and without LiNO3 additives [26]. The x‐ray photoelectron spectroscopy (XPS) study of the SEI indicated that a lithium strip etched in the electrolyte without the additive exhibited the formation of a small quantity of Li2S formed after 5 hours of etching and a large quantity of Li2S formed after 9 days. This was not observed in the lithium strip etched in the
LiNO3 contained solution. Instead, various species of LixSOy compounds were noted in greater quantities
than the insoluble Li2S. This has been attributed to the reaction of LixNOy species in the SEI with PS to
3.1.1.1 Energy dispersive X‐ray Spectrometry (EDX) X‐rays are also emitted from the sample surface in the SEM. When high energy electrons impact the atoms of the surface, the electrons from the atoms at various states are ejected, creating vacancies. These vacancies are filled by other electrons e.g. valence electrons. As a result, x‐rays are emitted with energy equal to the difference in states, characteristic to the element from which the x‐rays were emitted. The x‐rays are detected by an EDX detector, in this project, an X‐MaxN detector from Oxford Instruments. It must be noted that the EDX detector is an approximate method of elemental analysis and cannot be calibrated for lithium samples as the detection of characteristic x‐rays in the EDX detector is limited to elements with atomic numbers (Z) greater than 4 in principle due to the low energy of the emitted characteristic x‐ray photons. And in practice, the accuracy of the readings for EDX is low for elements with Z < 10. 3.1.1.2 Error possibilities on SEM image formation The atmosphere of the SEM chamber under venting during introduction and removal of the samples was nitrogen gas which ideally does not react with the highly reactive lithium metal. However, given the reactive nature of lithium metal and inevitable presence of atmospheric species that could contaminate the samples during initial transfer of samples, it not possible to get an entirely pristine image of the lithium surfaces as they are when extracted from cycled cells. However precautions and measures have been taken to ensure the images used in this project can be the closest approximation to the possible surface morphologies. Charging can also occur locally in case of insulated samples, which manifests in the SEM images in the form of very high or very low brightness, making it difficult to identify the surface topography of the sample. But this phenomenon was identified easily during studies.
3.1.2 X‐ray photoelectron spectroscopy (XPS)
X‐ray photoelectron spectroscopy (XPS) is a method of characterizing the chemical composition XPS is based on the Photoelectric Effect discovered by Hertz in 1887 [89] and described in its present form by Einstein in 1905 [90]. Electrons emitted by the photoelectric effect have a kinetic energy (Ekinetic) equal tothe photon energy minus its binding energy (Ebinding) and the work function of the sample ( ). In
3.2 Anode sample preparation
In this project, two types of cells, Li‐S cells and symmetrical Li‐Li, were assembled and cycled up to various states of charge (SoCs). All the cells were assembled in an argon (Ar) filled glove box (O2: < 1
ppm; H2O: < 8 ppm) to avoid contamination of the cells and electrodes by O2 and H2O present in the air.
The constituent cell materials such as electrolytes and the lithium anode strips were prepared and stored in the glove box under a similar atmosphere.
3.2.1 Materials
Sulphur powder, Carbon black Super P Li (Timcal Graphite), poly (ethylene oxide) (PEO, MW = 4,000,000; Aldrich), polyvinylpyrrolidone (PVP, MW = 360,000; Aldrich), 1,2‐dimethoxyethane (DME; Novolyte), 1,3‐ dioxolane (DOL; anhydrous; Aldrich) were used as received. Lithium nitrate (LiNO3; Aldrich), lithiumperchlorate (LiClO4) and lithium sulphide (Li2S; Aldrich) were dried under vacuum at 120 °C prior to use.
3.2.2 Cathode preparation
For the Li‐S cells a carbon‐sulphur composite cathode was prepared by coating cathode ink to an aluminium substrate. 0.5 g of sulphur, 0.4 g of Super P carbon black, 0.08 g of PEO and 0.02 g of PVP were dispersed in 8 ml of water and mixed by ball milling for 2 hours to prepare the cathode ink. The ink was bar‐coated on to an Al foil and left to dry at room temperature for 48 hours. Circular cathodes of a diameter of 20 mm were punched out and dried under vacuum in the glove box at 298 K for a period of 16 hours.3.2.3 Electrolyte preparation
For both, the Li‐S cells and Li‐Li cells, the electrolytes were prepared in the glove box. In case of Li‐S cells, 5 ml of electrolyte consisted of 1 M (0.532 g) of LiClO4, 0.25 M (0.086 g) of LiNO3, and 2.5 ml of DOL and2.5 ml of DME.
In case of Li‐Li cells, 5 ml of electrolyte consisted of 1 M (0.1064 g) of LiClO4, 0.25 M (0.172 g) of LiNO3,
and 2.5 ml of DOL and 2.5 ml of DME, 0.1 M (0.046 g) of Li2S and 0.1 M (0.257 g) of S8, to give a
polysulphide‐saturated electrolyte with an average PS stoichiometry of Li2S9 (Since the cell did not
4 Experimentation
This chapter presents and discusses the results from various experiments conducted in the project and is divided in four experiments investigating the lithium metal anode morphology under various conditions of cycling.4.1 Comparison of Li‐S cells with Li‐Li symmetrical cells
In this experiment, 4 Li‐S cells and 4 Li‐Li cells were assembled and cycled at various states of charge (SoCs) within 1 cycle. The anode samples were then examined by the SEM. The resulting images were then observed and analysed. The morphologies of Li‐S cells and Li‐Li cells are compared and contrasted. The aim of the experiment was to examine the effects of operating the cells on the lithium metal anodes at a constant current density. The cells were cycled at 200 µA/cm2. The list of Li‐S cells that were cycled and the Li‐Li cells that were used for electrochemical dissolution and deposition of lithium are listed in Table 1. Table 1: List of cells cycled/simulated to their respective SOCs Li‐S cellsSoC (after cycling) Active Sulphur mass (mg) Cycling Procedure
0 % 3.545 Discharge
6.25 % 3.515 Discharge + Charge
50 % 3.315 Discharge + Charge
100 % 2.645 Discharge + Charge
Li‐Li cells
SoC (after cycling) Duration of charge Cycling procedure
0 % 6 hours Discharge
6.25 % 6 hours + 23 mins Discharge + Charge
50 % 6 hours + 3 hours Discharge + Charge
Table 2: Dimensional statistics of the anode samples from Li‐Li cells and Li‐S cells SOC (after cycling) Range of pit diameters (µm) Dendrite widths (µm) Mean dendrite area % within cycled area with pits Li‐S cells Li‐Li cells Li‐S cells Li‐Li cells Li‐S cells Li‐Li cells
Table 3: Li‐Li cells at their respective simulated SOC cycled at varying current densities
Current Density (J)
(μA/cm2) Cycling Procedure Duration
probably be slowed down by increasing tLi, the transference number of lithium ions in the electrolyte, as proposed by the Chazalviel model. This can be increased by increasing the concentration of lithium salts in the electrolyte, thereby mitigating dendrite growth to an extent, if not totally prevent it. The typical values for tLi in dilute electrolytes is approximately 0.3 and studies have shown that this can be increased to 0.6‐0.7 in concentrated electrolytes [96].
4.3 Effect of self‐discharge of Li‐S cells on the lithium anode
This experiment aims to chemically characterize the surface films formed on the lithium anode from three distinct conditions of cycling in order to understand the self‐discharge, caused by chemical PS reduction, on the anode. This experiment used XPS for chemical characterization combined with SEM for study of Li anode morphologies at various cases. The XPS studies were conducted by Julia Maibach at the Ångström Advanced Battery Centre at Uppsala University. A current density of 400 µA/cm2 was applied to Li‐Li cells which were cycled. A pristine Li was studied in the XPS by Dr. Maibach as a reference for XPS studies of other samples which were soaked in PS containing electrolyte. The samples studied in this experiment and their experimental conditions are listed in Table 5. The samples which are underlined were studied by the SEM for morphologies. The self‐ discharge condition in the Li‐Li cell was simulated by storing the cell in idle open circuit conditions for a period of 10 days after the 1st cycle. Table 5: List of samples studied in the experimentSample Idle duration after assembly Idle duration after 1st cycle
Pristine Li ‐ ‐
Soaked Li 12 hours ‐
Charged Li 0 hours 0 hours
Self‐discharged Li 0 hours 10 days
4.3.1 Results
A low intensity peak was noted as at 161 – 162 eV, which was indicative of sulphide species such as Li2Sx, where the oxidation state of S is negative. This indicated that LixSyOz compounds are present in higher concentration than Li2Sx at the surface. The sample also exhibited a peak for sulphur compounds of intermediate oxidation states (163 – 165 eV) as well the presence of other forms of sulphur on the SEI. SEM studies of the soaked Li sample showed no presence of pits or dendrites (Figure 24) indicating that no electrochemical reactions took place in the Li‐Li cell without cycling. While the XPS results indicated that some chemical reactions occurred at the anode surface, these were not visible by the SEM due to the spatial resolution of the technique. Charged Li XPS studies of the charged Li sample showed no peaks for F1s. Sulphur spectrum, S2p showed 2 peaks of similar intensities for Li2Sx (161 – 162 eV) and LixSyOz (166 – 170 eV), suggesting that both kinds of
4.3.2 Discussion
The XPS study of the samples has provided distinguishing insights about the composition of the SEI on each sample. The decreasing concentrations of fluorine based compounds with the increasing exposure to the electrolyte and the electrochemical reactions can be explained by the formation of over‐layers of other compounds from the electrolyte, resulting in a thicker SEI. Due to the limited probing depth of XPS, it is possible that the signal for fluorine was suppressed in the deeper regions of the SEI. When soaked in the electrolyte, the dissolved PS molecules rapidly react with the lithium metal in the presence of LiNO3 resulting in the formation of oxidised sulphur species on the soaked sample as seen in Figure 22. LiNO3 in the electrolyte prevents the formation of a high concentration of sulphide compounds on the lithium metal surface by readily reacting with the lithium surface to form a passivating layer consisting of stable compounds. When a current is applied to the cell, the constituent compounds in the electrolyte instantly reduce at the lithium surface to form further passivating compounds on the surface of the charged Li sample as seen with oxidised sulphur species (SOx) (Figure 20). The peak for sulphide compounds (Li2Sx) can be explained by the electrochemical redox reaction in the Li‐S cell (refer Chapter 2.1.2) resulting in the deposition of solid PS on the anode. The self‐discharged sample has exhibited a significantly larger quantity of Li2S with respect to LixSyOz species that the charged sample. This indicates that the chemical PS redox reaction takes effect when the cell was left idle for 10 days. The dissolved PS in the electrolyte corroded the lithium anode forming the precipitates Li2Sx on the SEI. Previous literature on the phases of lithium sulphide have pointed out that Li2S is the only thermodynamically compound that could exist in the operating window of the Li‐Scell [97]. While other PS species such as Li2S2 and Li2S4 may have precipitated initially, it is possible that