Upps al a univ ersit ets l ogot yp
UPTEC K 21010
Examensarbete 30 hp Oktober 2021
Fluorine-free electrolytes for Li-ion batteries
Filippa Wahlfort
Civilingenj örspr ogrammet i k emit eknik
Teknisk-naturvetenskapliga fakulteten Uppsala universitet, Utgivningsort: Uppsala
Handledare: Guiomar Hernández Ämnesgranskare: Andrew Naylor
Upps al a univ ersit ets l ogot yp
Fluorine-free electrolytes for Li-ion batteries
Filippa Wahlfort
Abstract
Lithium-ion batteries are of great importance for today's society. The state-of-the-art batteries that are used today use a fluorinated electrolyte that contains the salt LiPF
6and acts as both a safety hazard and an environmental issue due to its ability to form the toxic gas hydrogen fluoride (HF). This project aims to find a fluorine-free electrolyte that can be used in silicon- based lithium-ion batteries to make them more environmentally friendly without detriment to the electrochemical performance. To do so, an additive that may form a solid electrolyte interphase (SEI) stable enough to allow a fluorine-free electrolyte to replace the ones used today is sought for. The salt of interest is lithium bis(oxalato)borate (LiBOB). Based on previous research electrolytes using LiBOB in either the solvent γ-Butyrolactone (GBL) or a mixture of ethylene carbonate (EC) and ethyl methyl carbonate (EMC) are examined. The additives used are vinylene carbonate (VC) and 1,3,2-dioxathiolane 2,2-dioxide (DTD). Techniques used are cyclic voltammetry, linear sweep voltammetry, galvanostatic charge and discharge, X-ray
photoelectron spectroscopy and scanning electron microscopy. The cells using GBL as solvent have cycled very poorly during this project while LiBOB in EC:EMC + VC shows the most promising results, with highest capacity retention and less amount of degraded LiBOB during the first charge. It is also to be noted that both EC:EMC based electrolytes provide the formation of a passivating solid electrolyte interface (SEI) and are of interest for further investigation based on the results obtained during this project.
Tek nisk-nat urvetensk apliga f ak ulteten, Upps ala universit et . Utgiv nings ort: U ppsal a. Handl edar e: Gui omar Her nández , Äm nesgranskar e: Andr ew N aylor , Exami nator: Pet er Br oqvist
Shortenings used in the report
The shortenings of substances that are used in the report are presented in order of appearance in Table 1 below.
Table 1: Shortenings used in the report
Substance Shortening
Lithium bis(oxalato)borate LiBOB
Ethylene carbonate EC
Ethyl methyl carbonate EMC
Vinylene carbonate VC
Gamma-butyrolactone GBL
1,3,2-dioxathiolane 2,2-dioxide DTD
Dimethyl carbonate DMC
Popul¨ arvetenskaplig sammanfatning
Att batterier spelar en stor roll i dagens moderna samh¨alle ¨ar det f˚a som s¨ager emot. Till exempel l¨agger alla med en mobiltelefon stor vikt vid att dess batteri fungerar som det ska. ¨Aven i ett st¨orre perspektiv ¨ar batterier viktiga. Energilagring ¨ar en stor utmaning f¨or v¨arlden i stort, speciellt vid sol- och vindkraft finns problem med att lagra ¨overfl¨odig energi.
Vidare ¨ar lithiumjonbatterier det mest f¨orekommande batteriet i dag. Deras uppbyggnad best˚ar, n˚agot f¨oren- klat, av tv˚a elektroder som brukar kallas f¨or pluspol respektive minuspol. Vidare kallas ofta pluspolen f¨or anod och minuspolen f¨or katod. Mellan dessa elektroder finns en elektrolyt, som vanligast best˚ar av en v¨atska.
Elektrolyten har som fr¨amsta uppgift att m¨ojligg¨ora att positiva joner kan r¨ora sig mellan katod och anod genom batteriet. Det ¨ar ocks˚a, kanske lite f¨orv˚anande, viktigt att elektrolyten inte ¨ar elektriskt ledande, allts˚a inte leder elektroner. Detta kan f¨orklaras d˚a elektronernas v¨ag mellan anod och katod g˚ar genom en extern led- ning, och det ¨ar elektronernas vandring genom denna som ger str¨om vid anv¨andning. Om elektronerna ist¨allet passerade genom elektrolyten skulle inte bara str¨ommen utebli, utan reaktioner som skulle f¨orst¨ora batteriet skulle ocks˚a snabbt ske.
Dagens typiska litiumjonbatterier anv¨ander sig av en elektrolyt som inneh˚aller saltet LiP F6. Att elektrolyten inneh˚aller ett salt ¨ar en f¨oruts¨attning f¨or att den ska leda joner. Att just detta salt anv¨ands beror p˚a att inget annat hittills har lyckats ge ett lika bra presterande batteri. Problemet ¨ar bara att LiP F6 har sina nackdelar n¨ar det kommer till milj¨o. Vid kontakt med fukt bildar LiP F6 den mycket giftiga och korrosiva gasen v¨ateflu- orid. Detta ¨ar skadligt f¨or b˚ade m¨anniska, milj¨o och sj¨alva batteriet. Bland annat g¨or det batteriet sv˚arare att
˚atervinna.
Alternativ till saltet har d¨arf¨or l¨ange forskats p˚a. Den fr¨amsta kandidaten f¨or att ers¨atta LiP F6 ¨ar lithium bis(oxalato)borate (LiBOB). Anv¨andningen LiBOB kommer med motsatta f¨or- och nackdelar. Det ¨ar milj¨ov¨anligt men har hittills inte gett ett lika bra presterande batteri.
F¨or att f¨orst˚a syftet med projektet m˚aste en till del av batteriet f¨orklaras. N˚agot som ¨ar viktigt b˚ade f¨or prestandan hos ett batteri och detta projekt ¨ar n˚agot som p˚a svenska kan ¨overs¨attas till “fast elektrolytfas”
men som vanligen kallas solid electrolyte interphase (SEI). Detta ¨ar en fast fas som bildas vid b˚ada elektroderna vid batteriets f¨orsta laddning. I detta projekt ligger dock fokus p˚a SEI vid anoden. Denna fas har som fr¨amsta uppgift att skydda batteriet genom se till att r¨att saker passerar genom det.
Ett SEI bildas av degrationsprodukter som bildas n¨ar en reaktion sker i elektrolyten. Dessa reaktioner sker fr¨amst under f¨orsta laddningen, eftersom n¨ar SEI v¨al finns hindras vidare reaktioner att ske. N¨ar LiBOB anv¨ands som salt i elektrolyten visar tidigare forskning p˚a att f¨or mycket LiBOB deltar i bildandet av SEI, vilket g¨or att batteriet presterar s¨amre eftersom mindre salt finns kvar f¨or att f¨orflytta joner. Syftet med detta projekt ¨ar d¨arf¨or dels att hitta en tillsats som reagerar och bildar ett SEI innan LiBOB reduceras, f¨or att f¨orhindra detta. Med andra ord s¨oks det efter ett additiv som till˚ater ett stabilt SEI att bildas f¨or att LiBOB ska kunna ers¨atta LiPF6 utan att prestandan hos batteriet f¨ors¨amras. I f¨orl¨angningen efters¨oks en fluorfri elektrolyt som kan ers¨atta den som vanligen anv¨ands i dag, utan att prestandan f¨ors¨amras.
I detta projekt unders¨oks additiven VC och DTD. VC testas i tv˚a l¨osningsmedel, GBL och en blandning av EC och EMC medan DTD endast testas i EM:EMC. Resultaten ger inte ett tydligt svar p˚a att n˚agot av additiven reduceras innan LiBOB. Vidare fungerar inte cyklingen av batterier som anv¨ander GBL som l¨osningsmedel
¨
over huvud taget. D¨aremot erh˚alls bildandet av SEI f¨or anv¨andandet av b˚ade VC och DTD som additiv n¨ar de anv¨ands i kombination med EC:EMC som l¨osningsmedel. Det m¨arks ocks˚a en skillnad p˚a att reduceringen av LiBOB sker tidigare n¨ar VC anv¨ands j¨amf¨or med n¨ar DTD anv¨ands, samt att mindre LiBOB reduceras i detta fall. Den elektrolyt som presterade b¨ast i detta projekt var LiBOB i EC:EMC + VC.
Contents
Shortenings used in the report iii
Popul¨arvetenskaplig sammanfatning iv
1 Introduction 1
1.1 Aim . . . 1
2 Background 2 2.1 Different types of batteries . . . 2
2.2 Lithium-ion batteries . . . 2
2.3 Silicon based anodes . . . 3
2.4 Solid Electrolyte Interphase . . . 3
2.5 Previous research . . . 4
3 Materials and methods 6 3.1 Electrolyte preparation . . . 6
3.2 Battery preparation and testing . . . 6
3.2.1 Cyclic voltammetry and linear sweep voltammetry . . . 6
3.3 Galvanostatic charge and discharge . . . 7
3.4 Postmortem analysis . . . 7
3.4.1 Scanning electron microscopy . . . 7
3.4.2 X-ray photoelectron spectroscopy . . . 8
4 Results and discussion 9 4.1 Ionic conductivity . . . 9
4.2 Linear sweep and cyclic voltammetry . . . 9
4.3 Galvanostatic cycling . . . 12
4.4 Scanning electron microscopy . . . 18
4.5 X-ray photoelectron spectroscopy . . . 20
5 Conclusion 27
A Appendix 30
1 Introduction
The great importance of batteries for both today’s and tomorrow’s modern society is hard to argue against.
They can be stated to play a great role in the everyday life of every person with a cellphone and for the whole of humanity considering its potential part in a movement towards more sustainable energy, for example by storing redundant energy from green sources. [1]
To take it down a notch, or at least explain the point being made, it is true that the majority of people have access to a cellphone today. It is of course not the only use people have of batteries, and not even the most common or important, but it is a central way in the life of many.
Stating that batteries can save the earth from the climate crisis might seem exaggerated but it plays a greatly important role. One key to move towards more sustainable power production is energy storage. For the moment redundant energy from the sun and wind is lost if it can not be stored. [2]
Further, the batteries that can come with such improvements making society more sustainable, must be envi- ronmentally friendly themselves. The development towards this is ongoing and one large obstacle is still the electrolyte. The state-of-the-art lithium-ion battery used today has a fluorinated electrolyte containing the salt LiP F6. [3] The fluorine present is quite loosely bound to the rest of the molecule. This makes the electrolyte very reactive, and if in contact with moisture it forms the highly corrosive and poisonous gas hydrogen fluoride (HF). [4] This makes recycling of the battery hard. It also acts as a hazard to the battery itself as it cant perform with HF present. [5] Despite its drawbacks, its good properties, including high solubility and ion conductivity, have allowed LiP F6 to remain the used salt, simply because no alternative has yet resulted in batteries with the same level of performance. [5]
The most promising fluorine-free salt that has been researched is lithium bis(oxalato)borate (LiBOB). [5]
The drawbacks are the opposite of those of LiP F6. Both the solubility in a carbonate-based solvent and the ion conductivity is lower. [5] The environmental advantages are however large. LiBOB decomposes to B2O3
and CO2, which are relatively inert and non-toxic. In addition, LiBOB is very thermally stable and does not decompose until above 300◦C. [4]
Batteries that are of interest have anodes that are made up of a silicon-graphite composite. This combi- nation provides a higher capacity than pure graphite, but it comes with the drawback of large volume changes of silicon. [5] This gives a material with low cycling stability. [5] To avoid this, the electrolyte must be able to form a stable solid electrolyte interphase (SEI). This is a passivating and protective layer that is formed during a batterys first cycle due to reduction of the electrolyte. [6] This layer acts as protection for both the anode and the electrolyte during the following cycles. [6] A fluorinated electrolyte is often considered required to form this layer, but previous research has shown that a fluorine-free alternative using for instance LiBOB can form a good SEI. [5] Further, one does not want LiBOB to be the substance forming this layer since this salt is needed when the battery is in use. [5] This project therefore seeks to find an additive that will be reduced before LiBOB, that also forms an SEI that is stable enough to withstand higher cycling numbers and higher current densities.
1.1 Aim
The aim of this project is to find a fluorine-free electrolyte that can be used in silicon-based lithium ion-batteries to make them more environmentally friendly without detriment to electrochemical performance. To do so, an additive that may form an SEI stable enough to allow a fluorine-free electrolyte to replace the ones used today is sought. One way to achieve this is to search for an additive that reduces before LiBOB.
2 Background
The shortenings of substances that are used in the report are presented in order of appearance in Table 2 below.
Table 2: Shortenings used in the report
Substance Shortening
Lithium bis(oxalato)borate LiBOB
Ethylene carbonate EC
Ethyl methyl carbonate EMC
Vinylene carbonate VC
Gamma-butyrolactone GBL
1,3,2-dioxathiolane 2,2-dioxide DTD
Dimethyl carbonate DMC
2.1 Different types of batteries
The history of batteries reaches far back. Findings from centuries before Christ shows what could be something similar to a galvanic cell. [7] The invention that today is known as the Baghdad cell is a jar that contains a rod of iron surrounded by a cylinder made out of copper. [7] The known batteries history is however younger and can be dated to 1782 when Alexander Volta invented a cell that consisted of a stack of silver and zinc disks.
[7] Important to note is that batteries can be divided into primary and secondary cells, depending on whether the process can be reversed or not. [7] Secondary batteries can be charged again, which is not true for primary batteries. [7] The first secondary cell is dated to 1859 when Plant´e created a cell with lead and lead dioxide as electrodes and diluted H2SO4 was used as electrolyte. [7]
Being the oldest secondary system that still is of relevance today, these lead-acid batteries deserves to be explained further. Thanks to the possibility of cheap production, and its remarkable ability to be recycled, it still holds a great share of the market, despite first having been developed over 150 years ago. [8] It does however come with some drawbacks. Lead is a very heavy metal, and due to this, the specific capacity is limited to 259 Ah/kg and the cells specific energy only reaches up to 40 Wh/kg. [8]
One type of secondary cell that has been developed into a cell used today is nickel-based batteries. [8] The nickel-cadmium battery stands out for its duration at low temperatures and can be operated at temperatures as low as -40◦C. [8] This battery consists of a nickel oxide hydroxide cathode and a cadmium anode. The high specific capacity of 477 Ah/kg of cadmium allows as high specific energies as 60 Wh/kg. [8] However regulations regarding cadmium from the European Union, this battery is limited to use in applications with relevance to safety and medicine, as well as to meet requirements of high power. [8]
Lithium metal is widely used as an anode for primary batteries. The main advantages with lithium metal are the high specific charge of 3.862 Ah/kg and the fact that specific energies as high as 600 Wh/kg may be reached, all while being a very light material. [8] Further, the most common cathode is manganese oxide, and examples of applications for this specific setup are batteries used in watches and cameras. [8] However, cells using lithium metal as the anode are typically not seen as secondary, since metallic lithium is highly reactive and therefore a safety hazard. [8] Further lithium’s ability to be deposited on and then stripped from the anodes surface is a part of its redox reaction. There is however a risk that lithium is deposited irregularly and instead of being stripped away starts to form dendrites. [8] This brings the risk of punctuation of the separator, which would result in a short circuit. [8]
2.2 Lithium-ion batteries
First of all, it is important to note that lithium-ion batteries do not contain any lithium metal, unlike its forerunner lithium metal batteries. The lithium in these batteries is found as charged species (Li+) only. [8]
The typical lithium ion-battery is made up of two electrodes, which are connected to one current collector each.
[8] The electrodes can be divided in cathodes and anodes, with the cathode operating at higher voltages than the anode. [8] The typical setup is that copper is used as current collector at the anode while aluminum is used at the cathode. This is because copper is stable at lower voltages than aluminium. [9] The two current collectors are connected via an outer cable, which allows electrons to move between the two active materials.
[8] Between the two electrodes is an electrolyte and a separator. The electrolyte makes up the medium through which the lithium ions move during the operation of the battery and contains a salt that can conduct said ions.
[8] The separator keeps the two electrodes from getting in contact with each other and prevents a short circuit.
[8] A schematic picture of a lithium-ion battery can be seen in Figure 1 below.
Figure 1: A schematic picture of a typical lithium ion battery. [10]
Cycling a battery can be described in terms of charge and discharge. Charging a battery would for a full cell mean that lithium ions move from the cathode to the anode, and the opposite is true for discharge. [9] At the same time electrons move through the outer connection. [9] It is also common to talk about C-rate. [11] This is a measure of how fast a battery is discharged that is based on how much charge the battery can carry (or in other words, the batterys capacity). A rate at 1C means that the battery is discharged with a current such that it takes one hour to discharge the cell. With the rate 2C, it takes two hours while it at C/2 takes half an hour to discharge a battery completely. [11] However, a more exact terminology would talk about lithiation and delithiation instead. So the process that is commonly called charge for a full cell would instead be described as delithiation of the cathode and lithiation of the anode. Again, the opposite is true for the reversed procedure. [8]
Further, the lithiation process of the anode can take place in different ways depending on the material used. The most common anode material is carbon in the form of graphite, due to its suitable properties. In comparison to other forms of carbon, it has a high theoretical capacity (372 mAh/g) and low lithiation potential. [12]
Graphite is structured so that it consists of a layer of graphene allowing lithium ions to be intercalated between the layers, ideally without changing the structure of the graphite. There are however cases where cracks occur, destroying the ability to store lithium. So although it is widely used, it comes with some drawbacks and other materials have properties that may allow them to compete with carbon. [8]
2.3 Silicon based anodes
One competitor to the conventionally used graphite anodes is a silicon-based one. [5] It is a material of interest due to its many interesting properties. One main is its theoretical capacity of 3579 mAh/g, which compared to graphite is manifold higher. [13] A high energy density is also possible, which may be explained by the low operating potential for silicon vs Li/Li+. It is also stated to be a sustainable material that is present in high amounts in the crust of the earth. [13]
The storage of lithium in silicon alloys is not done by the same mechanism as in graphite. Instead of in- tercalation, the lithium and the silicon form an alloy. [13] This, unfortunately, leads to severe volume changes of the host material, which brings difficulties in forming a stable solid electrolyte interface, which may result in large losses of capacity and cycle life. [14]
The reason why pure silicon is not used as an anode, but has to be combined with graphite, is due to the challenges with volume changes. A small addition of silicon into graphite brings higher capacity compared to pure graphite but makes the volume changes less severe compared to pure silicon. [5]
2.4 Solid Electrolyte Interphase
During a lithium-ion batterys first charge a reduction takes place on the negatively polarized anode, while oxidation takes place on the positively polarized cathode. This leads to degradation of the electrolyte, and the products formed by this form a passivating layer on the anode. [15] This will act as protection of both the electrolyte and the anode during the following cycles of the battery. This layer is known as the Solid
Electrolyte Interface, SEI. [15] The existence of this layer is stated to be highly important for the efficiency, lifetime, and safety of a lithium-ion battery. [15] Two of the most important properties of the SEI is that it must be permeable for lithium ions while being electrically insulating. [6] The conductivity towards lithium ions is needed for the lithium to reach the anode and allowing the battery to store energy and the insulating property will hinder the electrolyte to be reduced in the following cycles, and in other words, inhibit further degradation of the electrolyte. [6] In addition to these fundamental properties, the SEI should also prevent other components than lithium ions to pass and reach the anode. This may be explained by that intercalation of larger particles may crack a layer structure, such as graphite, which anodes typically consists of today. [16]
Further, it is also optimal to have an SEI that is evenly distributed over the anodes surface. [16] When it comes to the composition of the layer, it is preferable to have compact inorganic compounds that are insoluble and stable compared to metastable organic compounds. [16]
2.5 Previous research
This project is based on research previously conducted by Guiomar Hernandez et.al. Their paper ”Elimination of Fluorination: The Influence of Fluorine-Free Electrolytes on the Performance of LiN i1/3M n1/3Co1/3O2/Silicon–
Graphite Li-Ion Battery Cells” [5] investigates the possibility to replace the state of the art electrolyte containing the salt LiP F6with one that uses LiBOB instead. [5] They point out the environmental hazard in the use of fluorinated electrolytes and the main aim is to move towards a more sustainable battery while keeping high efficiency.
The battery structure used was made up of the same electrodes for each battery, with a silicon-graphite anode and a LiN i1/3M n1/3Co1/3O2cathode. The separator Celgard 2325 was also consequently used. Five different electrolytes were used, all with EC/EMC with the volume ratio 3:7, as solvent. The electrolytes used can be seen in Table 3 below.
Table 3: Electrolyte composition and nomenclature used [5]
Electrolyte composition Nomenclature
1 M LiP F6 LP57
0.7 M LiBOB LiBOB
1 M LiP F6+ FEC 10vol% + VC 2 vol% LP57+FEC+VC
0.38 M LiP F6+ 0.5 M LiBOB + FEC 10vol% + VC 2 vol% LiP F6LiBOB + FEC + VC
0.7 M LiBOB + VC 2 vol% LiBOB + VC
The testing and analysis made included galvanostatic cycling at room temperature, electrochemical impedance spectroscopy (EIS), X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM).
Their results show variations in the electrochemical performance that can be derived from the impact of the SEI. The highly fluorinated electrolytes showed to form an SEI that was rich in fluorine while the SEI formed by the fluorine-free was rich in oxygen. It is to be noted that reduction products from the BOB anion was found in the latter. Further, the fluorine-free electrolytes showed to have high Coulombic efficiency and good cycling stability at low currents, which points to the conclusion that an oxygen-rich SEI can provide good enough protection. The effect from the additives used showed to be improved cycle life for all cases.
However it is to be noted that a fluorine-free alternative performs worse at higher currents. When cycled at the rate C/2, the discharge capacity decreases more rapidly for LiBOB + VC than for the fluorinated LP57+FEC+VC and partly fluorinated LiP F6 + LiBOB + FEC + VC electrolytes, which can be seen in Figure 2 below. This may be explained by the fact that LiBOB was reduced before the additives, leaving less salt to take part in the transportation of lithium. It is therefore of interest to find additives that will be reduced and form a stable SEI prior to LiBOB.
Figure 2: Cycle life for fluorinated and non fluorinated electrolytes.
LiBOB has been a candidate to replace LiP F6 for several years. This is shown by Kang Zu et. al and their article ”LiBOB: Is it an alternative salt for lithium-ion chemistry?” that was published 2005. [4] Their main purpose in replacing LiP F6 is making the electrolyte more durable to high temperatures, since the state-of- the-art salt is thermally unstable. It is stated that it is difficult to find a candidate good enough due to the many suitable properties this salt possesses. [4]
Their result shows that an electrolyte with LiBOB as salt can form an SEI that stabilizes a graphite elec- trode, but that further improvements are needed to stabilize the silicon-graphite (Si-Gr) anode. Further, it is necessary to find an additive that is reduced before LiBOB. It is however stated that the fluorine-free electrolyte is able to provide cells with good electrochemical performance. [4]
3 Materials and methods
All experiments conducted as well as materials and methods used are described in this section.
3.1 Electrolyte preparation
To determine which amount of LiBOB that were soluble in the different solvents that were to be examined, mixtures with the concentrations of 1 molal salt were prepared. In the cases where this did not dissolve, more solvent was added decreasing the concentration by 0.1 molal per addition until dissolution was achieved.
With the data from the solubility tests in mind, three electrolyte compositions were decided to be of in- terest, and the conductivity of these was tested.
The composition of the electrolytes used for the conductivity test and throughout the project can be seen in Table 4 below. In addition to the electrolytes presented in Table 4 additional electrolytes were prepared during the course of the project in order to investigate some results further. The additional electrolytes can be seen in Table 5
Table 4: Compositions of electrolytes of main interest Conc. of
Electrolyte Solvent LiBOB Conc. Additive additive Nomenclature
(m) (wt./v %) used
LiBOB in GBL + VC GBL 1 VC 2 GBL + VC
LiBOB in EC:EMC + VC EC:EMC 0.9 VC 2 EC:EMC + VC
LiBOB in EC:EMC + DTD EC:EMC 0.9 DTD 2 EC:EMC + DTD
Table 5: Compositions of the additionally examined electrolytes Conc. of
Electrolyte Solvent LiBOB Conc. Additive additive Nomenclature
(m) (wt./v %) used
LiBOB in EC:EMC + VC EC:EMC 0.9 VC 10 EC:EMC + 10% VC
LiBOB in EC:EMC + DTD EC:EMC 0.9 DTD 10 EC:EMC + 10% DTD
LiBOB in EC:EMC EC:EMC 0.7 - - EC:EMC
3.2 Battery preparation and testing
Throughout the project coin cells have been used. They were constructed so that one circularly cut electrode was placed at the bottom of the cell. A separator was then placed on top of it. The electrolyte was then added, and it was made sure that the whole of the separator was covered with the liquid before placing the second electrode. Lastly, the lid was placed on top and the cell was closed using a tool that applies pressure. The whole procedure was done in a glove box.
Compared to the alternative pouch cells, coin cells have the advantage that it is easier to make sure that a constant amount of electrolyte is present. This is an advantage when comparing the performance of elec- trolytes since one wants to make sure it is done in an as constant environment as possible. Further, the amount of electrolyte that has been used in this project was set to 50 µL per cell. Celgard 2325 with a diameter of 17 mm was used as the separator in all cells.
3.2.1 Cyclic voltammetry and linear sweep voltammetry
Cyclic voltammetry (CV) is a widely used method that allows one to study the processes of oxidation and reduction of both inorganic and organic molecular species. [17] In addition to linear sweep voltammetry, it is stated to be the most commonly used method to study these reactions. Its popularity can be explained by the fact that quite a lot of data can be obtained and easily understood without the requirement of expensive equipment. [17]
The main idea is that a potential is applied which is gradually in- or decreased at a set scan rate. The output is a value of which current that responds to which potential. [17] A peak indicates the case of an electron transfer. Characteristically, this technique is not used for scans in only one direction, but both the increase and decrease in volts. It can be used to scan as many cycles as needed. [17]
The main idea behind linear sweep voltammetry (LSV) is rather similar to cyclic voltammetry. It also gives a current as an output to an applied potential, which is decreased or increased linearly. The main difference is that this technique is only used to do one single sweep, up or down, and when the required voltage is reached, it is not swept back in the opposite direction.
The first electrochemical experiments conducted were CV and LSV. These are methods of interest since one can get an idea of at which voltages reactions takes place in the pure electrolyte. It can be seen as a way to disqualify a potential electrolyte since it is of great importance that an electrolyte is stable between the voltages that the battery is operated between when in use. Further, it is also of interest to see when reactions take place at lower voltages to understand what happens when the SEI is formed.
The voltages used for CV ranged from 0 - 5 V vs Li+/Li, and the scan rate used was 1 mV/s. The lim- its set for the voltages were chosen so that reactions of interest both at lower and higher voltages could be observed. For these measurements coin cells with lithium metal, with a diameter of 15 mm, and stainless steel was used. The electrolytes investigated are described in Table 4 and 5.
3.3 Galvanostatic charge and discharge
Galvanostatic charge and discharge differ from CV and LSV such that the electrolytes are in the environment that they will be in also when the battery is in use. One may describe it as charging and discharging a cell repeatably, simulating the battery being used in an application. From the data derived from this, one can create a lot of different plots. This is of interest due to a lot of reasons. For starters, one wants to make sure that the cell works when all components are in place. Compared to CV and LSV, reactions that may take place between the electrodes and the electrolyte will be included in the results from this method, and this may affect how a battery performs.
To look at how well a battery performs, one can look at how capacity varies with cycle number. Another thing that is of interest to know how a battery performs is looking at the Coulombic efficiency. This is a value that describes how much capacity is lost during one cycle.
There is also a lot of information about what takes place in the battery that can be obtained by looking at how voltage and capacity vary during the cycling of a cell. In this project, the formation of the SEI is of interest. This makes the first cycle extra interesting to look at since it is during the first charge that the SEI is typically formed. Looking at how voltage varies with capacity will give information about when a reduction takes place in the system, and graphs displaying this can give an indication about at which potential the SEI is formed. This is of special importance in this project since part of the aim is to find an additive that is reduced prior to LiBOB.
The electrodes used in the cycling of full cells was kindly provided by Varta. The cathode was NMC 6:2:2, which had an areal capacity of 2.2 mAh/cm2. The anode was silicon graphite composite with an areal capacity of 2.4 mAh/cm2. The cycling was done at room temperature and the equipment used was both Arbin BT-2043 battery testing system and Neware. The cycling of the full cells was done so that the cells were precycled at the low rate C/20 (corresponding to 0.146 mA) for two cycles before being cycled at C/2 (corresponding to 1.46 mA) for 500 cycles or until breakdown. The charging process was done at constant current followed by constant voltage, while the discharge was done at constant current only.
3.4 Postmortem analysis
3.4.1 Scanning electron microscopy
This is a kind of microscopy that allows high magnifications when compared to conventional optic microscopy.
Optical microscopy is constrained by the wavelength of visible light which allows only a magnification of about 1000 times. [18] Scanning Electron Spectroscopy (SEM) allows both higher resolution and magnification since electrons are accelerated at high energies, which gives very short wavelengths. Other perks when compared to
optical spectroscopy are several. For example, one reaches a larger depth of field. Information can be obtained about a samples chemical composition, electric properties, and the topography of the surface.
The main idea behind the technique is that a sample is bombarded with electrons generated by an electron gun. The electrons pass through electromagnetic lenses and are collected in a beam that is scanned over the sample. The interaction of the sample and the electrons then generate backscattering electrons that are caught by detectors in the microscope to create an image. [18]
3.4.2 X-ray photoelectron spectroscopy
This analysis technique is a very surface-sensitive and precise method. [19] It can be used to analyze almost any sample that can be put under a vacuum. [19] As the name, X-ray photoelectron spectroscopy (XPS), suggests it is based on the use of the emission of photoelectrons from the sample and the use of X-ray radiation. [19] The basic principle is that a sample is bombarded with an X-ray beam. This causes the emission of photoelectrons from the sample. Detectors inside the equipment can then detect the kinetic energy with which the electrons leave the sample. [19] Given a known relation between the energy of the X-ray, the kinetic energy, and the energy needed to induce the emission of an electron, the atom from which the electron has left can be identified.
[19] Further, the energy needed to remove an electron can, a bit simplified, be described as unique for each orbital and each nucleus. [19] This allows one to determine which type of bond was present in the surroundings of a detected electron before it was removed. [19]
The batteries used for both SEM and XPS was prepared as the other full cells, but only cycled the first charge at C/20. After being charged, the batteries were opened inside a glovebox. The anodes were removed and cut into halves and then rinsed using DMC. The rinsing was done so that DMC was dripped three times over the electrode using a dropper while allowing it to dry between each cleaning.
The equipment used for the XPS analysis is Kratos Axis Supra +. The excitation energy used is 1487 eV.
The data has been analyzed using the software ESCApe and Excel. It is also to be mentioned that all spectra have been adjusted, or charge corrected, so that the peak indicating the graphite bond (C-C) is at 284 eV, which makes the results from the different scans comparable. [20]
4 Results and discussion
4.1 Ionic conductivity
As stated in the introduction, the electrolyte can be described as the medium through which the ions move during the operation of a battery. It is therefore important that the electrolyte has a good conductivity for lithium ions. Further, knowing the ionic conductivity for the electrolytes of interest does not only ensure high enough values but will also give more information to the comparison of the electrolytes. The result of the conductivity measurements is given in Table 6 below.
Table 6: Conductivity values of the electrolytes tested Conductivity
Electrolyte (mS/cm)
LiBOB in GBL + VC 7.37
LiBOB in EC:EMC + VC 6.06 LiBOB in EC:EMC + DTD 5.80
The conductivity values presented in Table 6 are all high enough to not disqualify any of the electrolytes. A value higher than 1 mS/cm is enough to allow an electrolyte to be of interest. [21] It can also be seen that LiBOB in GBL + VC has the highest value, while the electrolyte with the lowest conductivity is LiBOB in EC:EMC + DTD. It does however differ with less than 2 mS/cm between the highest and the lowest value.
4.2 Linear sweep and cyclic voltammetry
The electrochemical properties of the electrolytes can be determined by LSV and CV. These analyses can be seen as something that could disqualify a potential electrolyte rather than insure that one performs well. As described in the ”Materials and methods”-section the cells used for CV and LSV are coin cells that uses metallic lithium against stainless steel as electrodes.
The results obtained from linear sweep voltammetry in the negative direction can be seen in Figure 3 be- low.
It can easily be spotted that the electrolytes that contain EC:EMC as solvents shows very similar results while the one electrolyte containing GBL differs. In other words, the additives seem to have little to no impact on the reduction potential of LiBOB and the solvent. This may be explained by the low concentration of additive, which was only 2 wt/v%.
Further it may be stated that the peak at 1.2 V corresponds to the reduction of LiBOB, given that this peak is visible for all four electrolytes. The peak that is yielded at approximately 0.5 V by the electrolytes where EC:EMC is present can be assigned to the reduction of said solvent. It is however uncertain whether it is just one of the components or both that gives rise to this reduction.
Figure 3: Linear sweep voltammetry in negative direction. The cells used are coin cells that uses metallic lithium and stainless steel as electrodes.
Linear sweep voltammetry with increasing voltages yielded the graph shown in Figure 4. The electrolyte with the composition LiBOB in EC:EMC + VC shows quite deviating behavior. First of all, the open-circuit voltage (OCV) is surprisingly low at 2.1 V, compared to the expected value at about 3 V. Further, an oxidation peak appears already at approximately 2.4 V.
Figure 4: Linear sweep voltammetry in positive direction.The cells used are coin cells that uses metallic lithium and stainless steel as electrodes.
To further investigate the behavior of LiBOB in EC:EMC, additional measurements using LSV with increasing voltage were performed. This was done on the EC:EMC based electrolytes, but with the amount of additives increased to 10%. The results from this analysis can be seen in Figure 5 below.
Figure 5: Linear sweep voltammetry in positive direction. The cells used are coin cells that uses metallic lithium and stainless steel as electrodes.
Comparing the results shown in Figure 5 with the results in Figure 4 one can say that the clear peak that is seen at around 5 V for both LiBOB in EC:EMC without additives, and for LiBOB in EC:EMC + DTD is absent for LiBOB in EC:EMC + VC in both cases.
Graphs displaying the first three cycles of cyclic voltammetry measurements of the three electrolytes con- taining additives are seen in Figure 6, and Figure 7 respectively. All three measurements show a peak at just above 1 V when decreasing the voltage. It can therefore be assumed that this peak shows the reduction of LiBOB. Further, all three electrolytes show less reduction in the second and third cycle than the first, which indicates that a passivating layer is formed during the first cycle, which prevents further degradation of the electrolyte in the following.
Comparing the CV measurements of the use of DTD and VC as additives in EC:EMC as the used solvent, that is comparing a) and b) in Figure 7, it can be stated that they seem almost identical with each other. It can however be noted that the reduction peak occurring at just above 1 V reaches a higher absolute value of the current for VC as additive than for DTD.
Figure 6: Cyclic voltammetry for LiBOB in GBL + VC in a coincell that uses metallic lithium and stainless steel as electrodes.
(a) VC (b) DTD
Figure 7: First three cycles of cyclic voltammetry for LiBOB in EC:EMC + 2% VC in a) and for LiBOB in EC:EMC + 2% DTD in b). The cells used are coin cells that uses metallic lithium and stainless steel as electrodes.
Electrolytes with a higher concentration of additives were prepared in order to investigate if clearer differences would be discovered. Graphs comparing the additives VC and DTD with the concentration 10% is displayed in Figure 8.
Figure 8: First cycles of cyclic voltammetry for LiBOB in EC:EMC + 10% VC and LiBOB in EC:EMC + 10%
DTD. The cells used are coin cells that uses metallic lithium and stainless steel as electrodes.
Looking at Figure 8 it can be seen that the differences are very small when it comes to at which voltage reduc- tion takes place. It can however be stated that the reduction of LiBOB takes place before in the electrolyte containing VC than the reduction in the DTD-containing electrolyte.
When looking at how the electrolytes perform in the cyclic voltammetry analyses, one of the most impor- tant aspects is whether the electrolyte is stable in the voltage window used when full cells are cycled. In
other words, it is important that no electrochemical reactions takes place between 3 and 4.2 V at the working electrode. This is of interest since reactions within this interval would lead to constant degradation of the electrolyte on the working electrode. Looking at the result from the cyclic voltammetry for batteries using 2%
additive displayed in Figure 6 and Figure 7 it can be stated that no peaks are present between 3 and 4.2 V.
4.3 Galvanostatic cycling
The cells used are coin cells and NMC 6:2:2 is used as cathode and silicon graphite composite is used as anode.
The same electrolytes as for CV and LSV are examined but as explained in the materials and method section, one important difference to keep in mind when looking at electrolyte performance is that the environment in the full cell is the same as the one the electrolyte is in when a battery is used in an application. This means that the cycling of full cells differs from the measurements using CV and LSV, not only by what equipment and cell set up is used, but the properties of electrolyte may differ as well since the reactions that may occur due to interaction with the electrodes are present in these cells.
Something similar between the cycling of full cells and CV and LSV is however that an indication of when reduction occurs in the electrolyte can be obtained. When cycling full cells, a first idea of when that is can be obtained by looking at the voltage profile, or more precise, how voltage varies with capacity at the beginning of the first charge. In these types of plots, a reduction is indicated by a small initial plateau in the curve. As LiBOB in EC:EMC with different additives showed similar behavior, these are depicted in the same graph and can be seen in Figure 9 below. The voltage profile of a cell containing an electrolyte using GBL as the solvent can be seen in Figure 10.
Figure 9: Voltage vs Capacity for LiBOB in EC:EMC + 2% VC/DTD.
Figure 10: Voltage vs Capacity for LiBOB in GBL + VC.
2.0 V for the electrolyte containing DTD. This might seem like a small difference, but it may however be stated that the reaction happens at a lower voltage when VC is used as the additive than when DTD is used. Being the first charge, this means that the reduction of LiBOB in EC:EMC + VC happens before LiBOB in EC:EMC + DTD in the full cell. Comparing this with the results from the CV measurement presented in Figure 8 it can be stated that the reduction occurs earlier when VC is used as the additive than when DTD is used both when the active material is present and when its not. Another thing that is interesting to note is that the plateau lasts longer in the DTD case. This suggests that more LiBOB is being reduced when DTD is present than when VC is.
The voltage profile for LiBOB in GBL + VC differs from the voltage profiles of the electrolytes using EC:EMC as the solvent, with no clear initial plateau visible. There is a first change in the slope of the curve at around 2.9 V, followed by another just above 4.0 V. This behavior suggest that although a reduction seems to have taken place to some extent, it has not been done fully.
In addition to the electrolytes with 2% additives, a couple of batteries were prepared with electrolytes using EC:EMC as solvent and the amount of additive enhanced to being 10%. These electrolytes were initially prepared for the purpose of getting more clear results on the voltammetry experiments, but since some were left available it was decided to investigate whether the same results would be seen for 10% as for 2% additives in full cells. Plots of the voltage profile for the electrolytes with 10% additive can be see in Figure 11 below.
Figure 11: Voltage vs Capacity for LiBOB in EC:EMC + VC/DTD with 10% additives.
Looking at Figure 11 it can be stated that the small early plateau occurs at lower voltages for VC than for DTD, which confirms the results seen when 2% additive was used in Figure 9. Figure 11 also confirms that this plateau is longer for when DTD is used than for VC.
From the voltage profiles, plots of how dQ/dV varies with voltage during the first charge may be derived. Or, in other words, graphs showing how the slope of the voltage profile varies. This gives another way of telling when degradation of the electrolyte occurs. Where in the voltage profile, this information was given by a plateau, it is indicated by a peak in dQ/dV for differential capacity. The plots displaying differential capacity only look at the first charge since this is when degradation of the electrolyte takes place and gives the formation of the SEI.
In the following charges, it is expected and preferred that the SEI hinders further reduction of the electrolyte.
In Figure 12 a graph of how dQ/dV varies with voltage for LiBOB in EC:EMC + 2w/v% additives is shown.
Figure 12: dQ/dV vs Voltage for LiBOB in EC:EMC + VC and LiBOB in EC:EMC + DTD. The concentration of the additive is 2% in both cases.
Comparing the behavior in differential capacity between the two cells shown in Figure 12 it can again be seen that the peak indicating that degradation of the electrolyte takes place occurs at a lower voltage for VC as additive than for DTD.
Plots of how dQ/dV varies with voltage have also been derived from the voltage profile of the electrolytes containing 10% additive. These are compared with the data yielded from the electrolytes containing 2% in Figure 13 below.
(a) DTD (b) VC
Figure 13: A comparison between how dQ/dV varies with voltage for LiBOB in EC:EMC + 2 and 10 % additives. DTD is displayed in a) and VC in b).
As seen in Figure 13 it can be stated that a higher amount of additive seems to result in the degradation of the electrolyte to take place in higher voltages. It may also be seen that a quite higher peak is visible for 10%
DTD. This might indicate that more LiBOB is being degraded in this case.
The differential capacity for LiBOB in GBL + VC can be seen in Figure 14. Comparing this to the behavior for the EC:EMC based electrolytes displayed in Figure 12 the first peak does not occur until 2.9 V.
Figure 14: dQ/dV vs Voltage for LiBOB in GBL + VC.
Several identical batteries have been made and analyzed to investigate the reproducibility of the cells as well as getting more certain results. However, when analyzing and comparing the results between the different electrolytes the main focus is kept on one battery of each kind. This is simply to make the results more com- prehensible and easy to follow. The batteries that are chosen for this are batteries that showed good general performance compared to the others, and that were prepared at the same time and analyzed with the same equipment. Plots for each individual cell can be found in Figure A.1 - Figure A.16 in the appendix. As ex- plained earlier, two out of four batteries were prepared a few months before the other two for both electrolytes using EC:EMC as solvent. Since no new cells containing GBL were prepared, this comparison focuses on the cells with EC:EMC as electrolyte solvent.
Looking at the reproducibility within the experiments analyzed so far one may start by looking at the graphs displayed in Figure A.1 - Figure A.4 showing the electrolytes containing 2% DTD a clear peak can be seen at just about 2 V for FV201104 and FV201107, and the corresponding peak occurs at around 2.2 V for FV210103 and FV210104. As the two first batteries were prepared around the same time, and the two latter were pre- pared together a few months later, this may be explained by differences in the preparation of the batteries and experimental differences. One big difference between the cells using DTD as additive is that the cells prepared at the later occasion had a freshly prepared electrolyte. Further, the diameter of the anodes used in the latter produced cells was 13 mm compared to 14 mm for the earlier prepared cells, while the cathode had a diameter of 13 mm for all cells. Adding to that, the smaller anodes were a lot more bent after drying than the larger ones. Both of these factors made good alignment of the electrodes harder to achieve during the making of the cells with the smaller area. It may be noted that it in general is preferable that the anode and the cathode are the same sizes. Any part of an electrode that is not aligned with its counter electrode might not be fully utilized, which is not desirable. The profits of same area electrodes do however only make a difference if the electrodes are perfectly aligned which was hard to achieve when the cells studied in this project were prepared.
Other factors that may have affected the performance of the cells are the environment in the glovebox during preparation and the temperature during cycling.
Looking at LiBOB in EC:EMC + VC a similar behavior can be seen, where the reduction peak corresponding to BOB occurs at slightly higher voltages for the batteries prepared at the later occasion than for the batteries prepared earlier. In the VC case, the same electrolyte was used in all batteries, which makes it possible that the electrolyte used in FV210101 and FV210102 might have degraded more before cycling than it had when FV201105 and 201108 were tested.
Looking at how the capacity varies with each cycle will give a general idea of how well the battery performs.
An as high and steady value as possible is required, since this means that the retention of the capacity is high.
How the batteries performed in this regard is presented in Figure 15 and Figure 16 below. Looking at these graphs it can be stated that LiBOB in EC:EMC + VC maintains the highest capacity, while LiBOB in GBL + VC does not cycle at all. This is a fact that has raised a lot of questions. First of all, the electrolyte has quite good conductivity. In addition to this, the electrochemical characterization of the pure electrolyte showed promising results with a passivating layer formed during the first cycle, and less reduction of LiBOB compared to the EC:EMC based electrolytes. Other papers such as ”Electrochemical performances of a novel high-voltage electrolyte based upon sulfolane and γ-butyrolactone” [22] published in 2013 has managed to cycle batteries using GBL as electrolyte-solvent. Further, batteries using LiBOB in EC:EMC + DTD as electrolyte starts at a reasonably high capacity but the value decreases rapidly. It may be noted that these batteries does not break down, but continue to cycle at a very low capacity. This is however not that relevant since the capacity is too low to be used in commercial cells.
The comparison of how the discharge capacity varies with cycle number between full cells with different elec- trolytes can be seen in Figure 15 and Figure 16. Further, the mean value of the capacity for the two EC:EMC based electrolytes are shown in Figure 17.
(a) VC (b) DTD
Figure 15: Capacity vs cycle number for LiBOB in EC:EMC with 2% VC in a) and 2% DTD in b).
Figure 16: Capacity vs cycle number for LiBOB in GBL + VC.
Figure 17: Mean value of discharge capacity vs cycle number for the additives VC and DTD in EC:EMC.
It can be seen in Figure 15 that the batteries prepared at the earlier occasion, that is FV201104, FV201105, FV201107, and FV201108, has a higher capacity and lasts for more cycles than the ones prepared later. This may, as in the case for differential capacity, indicate that some experimental differences affect the result. One thing that might be seen as surprising in this case is that a fresher electrolyte can be expected to give a better performing battery. This was as seen in Figure 15 b) not the case.
By looking at both Figure 15 and Figure 17 it can be stated that the electrolytes using EC:EMC as the solvent and containing VC as additive performed better than the ones containing DTD.
The coulombic efficiency gives information about how much irreversible capacity is lost during a cycle. How the coulombic efficiency varies with each cycle can be seen in Figure 18. The cycle that is mainly of interest is the first one since this is when the SEI is expected to be formed. Looking at the coulombic efficency for the first cycle it can be stated that it is very similar between the use of 2% DTD and 2% VC.
(a) DTD (b) VC
Figure 18: Coulombic efficency for LiBOB in EC:EMC + VC/DTD 2%.
4.4 Scanning electron microscopy
Scanning electron microscopy, (SEM) was performed on anodes after one charge in the electrolytes using EC:EMC as the solvent, both with 2% additive. This means that the anodes are in a lithiated state when analyzed. No anode cycled in LiBOB in GBL + VC was analyzed in SEM due to the poor cycling of these cells.
In addition to the cycled anodes, pictures of the pristine electrode, consisting of silicon graphite composite, are included in Figure 19 below.
(a) Magnification: 5K (b) Magnification: 10K
Figure 19: Pristine anode with a smaller magnification in a) and a larger magnification in b).
Looking at Figure 19 one gets a quite good picture of the morphology of the anode. Larger, darker formations can be seen, which are pieces of graphite. On top of these, smaller and brighter particles are present, which consists of silicon.
Pictures of the anode that was cycled one charge in LiBOB in EC:EMC + 2% VC can be seen in Figure 20 and 21.
Figure 20: Anode cycled in LiBOB in EC:EMC + 2%VC.
(a) Magnification: 10K (b) Magnification: 10K
Figure 21: Anode that was cycled one charge in LiBOB in EC:EMC + 2% VC.
Comparing the pictures of the anode cycled in LiBOB in EC:EMC + 2% VC above with the pristine one displayed in Figure 19 no big difference is immediately seen, and the morphology of the anodes surface does not seem to change after one charge. This is however no longer true when looking at Figure 22 below. This shows another part of the anode cycled in LiBOB in EC:EMC + 2% VC but another position of the sample.
Figure 22: Anode cycled in LiBOB in EC:EMC + 2% VC.
The ”spikes” seen in this picture are believed to be salt still present from the electrolyte. When this suspicion was raised, the solubility of LiBOB in the solvent used for washing, DMC, was tested. This showed that the salt did not dissolve, suggesting that the electrodes might not have been properly cleaned when analyzed in SEM.
The anode that was cycled in LiBOB in EC:EMC + 2% DTD are displayed in Figure 23 and 24.
Figure 23: Anode cycled in LiBOB in EC:EMC + 2% DTD.
(a) Magnification: 10K (b) Magnification: 10K
Figure 24: Anode that was cycled in LiBOB in EC:EMC + 2% DTD.
As in the case for the anode cycled in the electrolyte using VC as additive, the anode cycled with DTD present did also present some areas with a lot of ”spikes” was present. Looking at the morphology besides that, no great changes can be seen in Figure 23 and 24 when comparing with the pristine anode showed in Figure 19.
It might however be suggested that the larger graphite particles are a bit more covered, which might suggest that a thicker SEI has been formed when DTD is used as additive.
4.5 X-ray photoelectron spectroscopy
The results obtained from the XPS can be seen in Figure 25 - Figure 29. In the cases when the pristine anode gives rise to a clear peak, a red line has been inserted at its position to make a comparison easier. If a peak from the pristine anode is missing, the peak from LiBOB in EC:EMC + VC is used. A quick look at the overall results suggests that the anode cycled in LiBOB in GBL + VC shows quite similar results to the pristine anode.
This makes a lot of sense since the cell containing GBL did not cycle properly, and the process of lithiating the battery was not made fully. The results from the anodes cycled in the presence of LiBOB in EC:EMC + VC/DTD differs from the other results.
Looking at the results element by element, the C1s spectra is seen in Figure 25. All four samples have a peak indicating that graphite is present. The pristine anode also has peaks at 285.5 and 289 eV. Comparing this with reference values collected from previous research it is probable that the peak at 285.5 eV comes from either C-O or C-O-C and that peak at 289 eV comes from C=O. [5] [20] Looking at what is present in the pristine anode, there is silicon-graphite, conductive carbon, and the binder which is lithium polyacrylate. This means that the only possible source of the C-O and C=O bonds in the pristine anode is the binder. Further, the C-C bond may be found in the binder, the conductive carbon and the graphite.
Moving on to look at the anode lithiated in LiBOB in GBL + VC, peaks are seen at roughly the same values (285.5 and 290 eV). The peak at 290 eV can be assumed to be carbonate. [23] Since the cell containing GBL did not cycle properly, the source of these peaks is less certain. The main hypothesis is that the anode was not rinsed properly before the analysis, which left LiBOB on the anodes surface, or that some degradation of the electrolyte occurred before the cycling ended. It is also possible that a combination of these cases has taken place. In addition to this, some contribution to the results from the anode cycled in LiBOB in GBL + VC could still come from the anode itself, since both scenarios would not necessarily cover the anodes surface completely. It is also to be noted that the peaks obtained from the pristine sample show a ratio where the C-C peak has a higher intensity than the peaks that come from oxygen bonded to carbon. This strengthens the hypothesis that something is present on the anode cycled in GBL. Whether it is remaining non-degraded LiBOB or degradation products from the electrolyte is not possible to say from looking at only carbon.
The anodes charged in EC:EMC based electrolytes show a similar shape to each other, but a small peak at 286 eV is more present for DTD than for VC. This peak is, despite having a little higher binding energy than for GBL and the pristine anode, believed to come from either C-O or C-O-C. The fact that the peak is a little shifted in binding energy can be explained by the fact that the environment differs when a full charge has taken place. For example, the rings in the structure of pure LiBOB can be expected to have opened up when the reduction of said compound took place. It can be stated that the shape of the spectra from both EC:EMC based electrolytes differs significantly from the pristine anode, with the C-C peak being smaller than the peaks
yielded from carbon bonded to oxygen. This suggests that the contribution from the anode is very little and that an SEI has been formed in both cases. Given this conclusion, it can further be stated that the SEI is rich in both carbon and oxygen.
Figure 25: The carbon spectra obtained from XPS on anodes cycled one charge in the electrolytes described in Table 4 as well as a pristine anode.
Moving on to examine the O1s spectra found in Figure 26 , it can be stated that one single peak is seen in all cases. The red line indicates the position of the peak from the pristine anode (532 eV). As described above, the only contribution of oxygen bonded to carbon to this sample is the binder, so that peak can be believed to come from lithium polyacrylate. In addition to this, silicon dioxide can be assumed to be present on the silicons surface. It is therefore lightly that SiO2contributes to this peak as well. [20] When it comes to the anode cycled in LiBOB in GBL + VC, the peak is shifted to higher binding energy (533.5 eV). This further strengthens the theory that something is present on this anodes surface. Looking at the results from the EC:EMC based electrolytes, both peaks are found at 532.5 eV. Despite being a bit shifted, the peaks yielded from the pristine anode and the anodes cycled in EC:EMC based electrolytes are believed to be either SiO2 or carbon bonded to oxygen. It is possible that both types of bonds contribute to the peak but are not possible to distinguish.
In the case of GBL, where the peak is found at 533.5 the peak is believed to be yielded by organic C-O. [20]
Figure 26: The oxygen spectra obtained from XPS on anodes cycled one charge in the electrolytes described in Table 4 as well as a pristine anode.
Looking at the Li1s spectra seen in Figure 27 there is again only one peak yielded by each sample, and the red line indicates the placement of the peak from the pristine anode (56 eV). Further, this peak can with certainty be assigned to come from the lithium in the binder. The GBL peak is found at slightly higher energy (56.5 eV) than the one from the pristine anode. This can be explained by that some of the lithium giving rise to this peak can come from either remaining pure LiBOB or LiBOB that has been reduced to some extent. It is again possible that a contribution from both cases gives the shift in peak, as well as it is likely that some come from the anode itself. The peaks from the anodes charged in EC:EMC based electrolytes are both found at 55 eV. The fact that these peaks are both shifted compared to the pristine anode is not surprising, due to the fact that the lithium in this case can be expected to come from the SEI or the lithiated active material, and not the binder in the anode. It is also to be noted that both the SEI and the lithiated active material could give rise to these peaks.
Figure 27: The lithium spectra obtained from XPS on anodes cycled one charge in the electrolytes described in Table 4 as well as a pristine anode.
Figure 28 shows the B1s spectra. As expected, there is no peak from the pristine anode, and the red line here indicates the position of the peak from LiBOB in EC:EMC + VC (192.8 eV). The fact that a peak is yielded by the anode cycled in GBL further strengthens the hypothesis that although the charging process was very short for this anode, it is not identical to the pristine one. The boron spectra also bring a bit more information about what the contribution from LiBOB is. The fact that the peak is shifted when comparing to the anodes cycled in EC:EMC based electrolytes points to the conclusion that something else than what is formed on the anodes cycled in EC:EMC based electrolytes. Whether it is pure LiBOB or a different degrada- tion product does however remain uncertain. It is known from previous results that the LiBOB does degrade when a full charge is achieved, which it is for the cells using an EC:EMC based electrolyte. As stated above, one probable degradation product of LiBOB includes a form where the rings in the structure of pure LiBOB is opened. This would create a different environment for the bonds present and could explain the small shift in binding energy when comparing the anode cycled in GBL with the ones cycled in EC:EMC. This would however not disqualify the idea that degradation takes place to a small extend, and is not enough proof to indi- cate that pure LiBOB is present due to a lack of knowledge about the form of an unknown degradation product.
Looking at the anodes cycled in LiBOB in EC:EMC + VC and LiBOB in EC:EMC + DTD, the peaks seen are found at very similar energies. One can see a small difference in approximately 0.3 eV, with the peak for DTD found at 192.5 eV and the peak for VC found at 192.8 eV, but this is such a small difference that it is almost not worth mentioning. It is safe to say that the main source of the boron present is degraded LiBOB. The problem with improperly washed anodes can be expected to affect these results as well, but given the difference in binding energy compared to the not fully charged cell, this possible contribution can be regarded as small compared to the boron present in the degraded LiBOB. In addition to this, degraded LiBOB is expected to be a part of the SEI, given the previous results. When it comes to determining exactly what the degradation product is, and how much of each compound of the electrolyte contributes to the SEI, it gets a bit more complicated.
The boron peaks in this case could be believed to come from BCO2, when comparing the binding energy to reference values. [24] This would be chemically possible since all components are found in the electrolyte.
Figure 28: The boron spectra obtained from XPS on anodes cycled one charge in the electrolytes described in Table 4 as well as a pristine anode.
Looking at the Si2p spectra seen in Figure 29 it can immediately be stated that the result from the pristine anode is very similar to the results from GBL, and that the results from the anodes cycled in LiBOB in EC:EMC + VC is similar to when DTD is present. The peak found at 99 eV for both the pristine anode and GBL is believed to be pure silicon. [5] This is likely to come from the silicon-graphite composite anode. The fact that the results from the GBL anode are so similar strengthens the hypothesis that no thick SEI is covering this anode.
The peaks seen for the anodes lithiated in EC:EMC based electrolytes differs from the other anodes. No pure silicon is present in these results. This can on one hand indicate that a thick SEI has been formed. There is however some silicon present. The anodes analyzed in this analysis comes from charged cells, meaning that the anodes are lithiated. This would create a very different environment for the silicon in the anode, given that it when charged is part of a lithium-silicon alloy. Therefore and the peak seen at 101 eV can be expected to come from lithiated silicon. This fact does not contradict that a thick SEI has indeed been formed.
Figure 29: The silicon spectra obtained from XPS on anodes cycled one charge in the electrolytes described in Table 4 as well as a pristine anode.
Summarizing all the results given by the XPS, it can with certainty be stated that a thick and both carbon and oxygen-rich SEI has been formed on the anodes after one charge in both EC:EMC based electrolytes. Only small differences are found in the composition of these passivating layers when comparing the use of the two different additives VC and DTD.
5 Conclusion
During this project, the performance of several fluorine-free electrolytes has been investigated. Fluorine-free alternatives using LiBOB as salt still shows promising results. The cells using LiBOB in GBL + VC did not cycle well during this project. The reason for this would need to be further investigated. It is also to be noted that these result does not disqualify the idea that GBL could be a useful compound in fluorine-free batteries.
In contrast to the electrolyte using GBL as solvent both EC:EMC based electrolytes provide the formation of a passivating SEI and are of interest for further investigation. The results from the XPS suggests that the SEI formed on the anodes cycled in EC:EMC based electrolytes are rich in both carbon and oxygen. Looking at the results from the SEM, it is suggested that the use of DTD as additive might yield a bit thicker SEI than the use of VC. These results are however a bit uncertain since the anodes are suspected to not have been rinsed proparly.
Comparing the usage of VC and DTD as additives to LiBOB in EC:EMC, the LiBOB present in EC:EMC + VC appears to be reduced before the LiBOB present in EC:EMC + DTD. The usage of VC also seems to decrease the amount of LiBOB being reduced in the full cell, which could indicate better performance for these batteries. Further, neither of the additives are found to be reduced prior to LiBOB based on the results presented in this report, but as stated above the additives affect the reduction potential of LiBOB, and the amount degraded.
