Surface reactivity of ultrathin atomic layer deposited Al2O3 on LiNi0.8Mn0.1Co0.1O2

UPTEC K 19036

Examensarbete 30 hp Maj 2020

Surface reactivity of ultrathin

atomic layer deposited Al2O3 on LiNi0.8Mn0.1Co0.1O2

Erik Rozenbeek

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

Abstract

Surface reactivity of ultrathin atomic layer deposited Al2O3 on Li[Ni0.8Mn0.1Co0.1]O2

Erik Rozenbeek

The nickel-rich cathode material Li[Ni0.8Mn0.1Co0.1]O2 is a much sought after material in Li-ion batteries for the future of

electric vehicles due to its high specific capacity. However, it exhibits fast degradation during its use due to the interaction between the electrolyte and the cathode surface leading to capacity loss. In this thesis, the surface interaction of pristine and alumina coated NMC811, and NiO powder with electrolyte was investigated to observe differences in surface reactivity and if hydroxide groups on the surface could be a potential culprit in degradation.

Thermogravimetric analysis (TGA) and Brunauer–Emmett–Teller (BET) analysis were used to measure thermal properties, surface area, and adsorptive properties respectively between pristine and coated powders but no distinct difference was found.

A soaking experiment was performed to induce electrolyte

degradation on the powders by soaking them in the electrolyte LP40 for a duration of two weeks at 50°C. The electrolyte and the soaked powders were thereafter investigated through various spectroscopy methods like Attenuated Total Reflection – Fourier Transform Infrared (ATR-FTIR) and Raman spectroscopy to find potential degradation products but were found insignificant or inconclusive difference.

The electrochemical testing was performed on NMC811 half-cells at 3-4.2V and 3-4.4V with a 0.5C – rate. The coated heat-treated NMC811 was found to have the best cyclability at both potential ranges.

In conclusion, the difference in surface reactivity of the

pristine, coated, and coated heat-treated powders were found to be insignificant. However, the coated heat-treated NMC811 was found to have improved electrochemical performance at both potential ranges but it remains uncertain if hydroxide groups could be a culprit in the degradation.

ISSN: 1650-8297, UPTEC K 19036 Examinator: Peter Broqvist Ämnesgranskare: Erik Lewin

Handledare: Erik J. Berg, Yonas Tesfamhret

Populärvetenskaplig Sammanfattning

Framtidens batterier domineras framförallt av litiumbaserade batterier. Detta beror på att

litiumbatterier bidrar med en högre energikapacitet, lägre underhåll, och lägre självurladdning jämfört med andra uppladdningsbara batterier bestående av bly, nickel-kadmium, eller nickel-metallhydrid.

Trots det är efterfrågan ständigt ökande för batterier med ökad energi- och kraftdensitet, livslängd, bättre säkerhet, och med en lägre kostnad än vad som har åstadkommits hittills. Livslängden är ett ständigt bekymmer hos uppladdningsbara batterier då de ständigt förlorar kapacitet ju längre tid som batteriet används. Förlusten av kapacitet hos ett uppladdningsbart batteri beror på hur delarna interagerar med varandra. Ett uppladdningsbart batteri består essentiellt av tre komponenter, en anod, en katod, och en separator mellan. För att tillåta jontransport mellan anod- och katodmaterialen under upp-och urladdning krävs det ett medium emellan, en så kallad elektrolyt. Elektrolyten är dock normalt sett instabil och degraderar under upp-och urladdning, eller på grund av att den reagerar med vattenmolekyler. När elektrolyten degraderar bildas det biprodukter som attackerar anod- och

katodytan. Biprodukterna från elektrolyten bidrar bland annat med att det bildas ett fast gränsskikt, att övergångsmetaller upplöses från katodmaterialet, att mikrosprickor bildas hos katodmaterialet, med mera. Tillsammans med att elektrolyten förbrukas sådan att anod- och katodmaterialen degraderas sker det en kapacitetsförlust och minskad livslängd hos uppladdningsbara batterier.

Målet med projektet var att undersöka ytreaktiviteten mellan elektrolyt mot katodmaterialet litium- nickel-mangan-kobolt-oxid (NMC) med en ytbeläggning av aluminiumoxid (Al2O3). En

metalloxidbeläggning på anod- eller katodmaterialet används för att agera som en fysisk barriär mot de sidoreaktioner som sker mellan elektrolyten och elektrodytan. Det är framförallt viktigt mot vätefluorid som är känt att formas från fluorbaserade salt i elektrolyten. Däremot, eftersom

hydroxidgrupper lämnas på det yttersta lagret på metalloxidbeläggningen kan det potentiellt reagera med elektrolyten, likt vattenmolekyler, och degradera vidare ned till biprodukter. I detta projekt undersöktes det om en skillnad i degradation mellan ytbelagt aluminiumoxid och obelagt

katodmaterial när det varit i kontakt med en kommersiell elektrolyt. För att potentiellt göra sig av med hydroxidgrupperna på ytan undersöktes det om värmebehandling vid en specifik temperatur skulle minska degraderingen mellan elektrodytan och elektrolyten då hydroxidgrupper potentiellt agerar likt vattenmolekyler. I andra ord, minska ytreaktiviteten.

Eftersom NMC pulver har en låg ytarea användes nickeloxid (NiO) pulver för att öka interaktionen mellan elektrolyt och material. En andel av NMC och NiO pulvren deponerades via ’atomic layer deposition’ (ALD) för att noggrant bestämma tjockleken av aluminiumoxid beläggningen. ALD alternerar flödet mellan två prekursorer för att belägga atomlager efter atomlager med hög kvalitet.

Projektet indelades i tre delar; först undersöktes termiska stabiliteten och ytegenskaper som

materialen NMC och NiO pulvren hade när dem var obelagt eller belagt med aluminiumoxid. Genom termiskgravimetrisk analys (TGA) visades att obelagt och ytbelagt NMC, samt NiO pulvren var stabila upptill 400°C, vilket användes som temperatur för värmebehandling av pulvrena. Via Brunauer–

Emmett–Teller (BET) analys bestämdes ytarean mellan NMC och NiO pulvrena och via vatten – sorption analys visades att obelagd och ytbelagt NMC pulver hade liknande svag interaktion med vatten.

Den andra delen i projektet bestod av ett blötläggningsexperiment där obelagt, belagt och

värmebehandlat NMC samt NiO pulver befann sig i en kommersiell elektrolyt, LP40, under två veckor i 50°C. Målet med experimentet var att observera en skillnad i degradering hos elektrolyten mellan de olika pulvrena och bestämma om ytbeläggningen hade en ökad eller minskad ytreaktivitet, samt om värmebehandlingen hade någon skillnad. Degraderingen av elektrolyt observerades visuellt under experimentets gång, samt efter de två veckorna med hjälp av olika spektroskopimetoder. Elektrolyten

ii

undersöktes efter experimentets gång med en enkel pH-mätning, UV/Vis, ATR-FTIR, och Raman för att respektive bestämma skillnad i surhetsgrad, färgändring, samt kvantifiera ändring i

bindningsstruktur mot potentiella degraderingsprodukter. Pulvren från experimentet separerandes från elektrolyten, sköljdes med organiskt lösningsmedlet dietylkarbonat, och torkades för att sedan

undersökas i ATR-FTIR och XPS. Detta gjordes för att undersöka om degraderingsprodukter som t.ex ett solitt gränsskikt hade formats från elektrolyten och om övergångsmetaller hade upplöst från katodmaterialet under experimentet. Resultaten från både elektrolyt- och pulveranalyserna såg liten till ingen skillnad av degradation av elektrolyten mellan de obelagt och belagda NMC, och NiO-

pulvren. Elektrolyten i experimentet visade tecken på degradation hos NiO-pulvren, men en signifikant skillnad mellan proven kunde inte säkerställas med de spektroskopimetoder som användes.

Elektrolyten hos NMC-pulvren under experimentet visade nästintill inga tecken på en ökad eller minskad degradation mellan det obelagt och det belagda NMC-pulvren. Det värmebehandlade NMC och NiO-pulvren visade ingen korrelation att hydroxidgrupper hade reducerats på ytan och minskad degradering, det vill säga reaktivitet.

Den tredje delen av projektet bestod av ett elektrokemiskt test av NMC pulvren för att avgöra om ytbeläggningen samt om värmebehandlingen hade en skillnad i cyklingsförmågan.Det elektrokemiska testet bestod utav galvanostatisk cykling genom framställning av halvceller med en NMC811-katod och litiummetallanod. Genom att applicera en högre spänning hos batterier kan mer kraft användas.

Däremot, med en högre spänning kompromissas livslängden och säkerheten hos batterier. En laddning över 4,3V hos litiumbatterier kan leda till att katodmaterialet oxidera, förlorar stabilitet, och producerar koldioxid gas (CO2). Detsamma gäller för upp-och urladdningshastigheten där en högre ström ger mer kraft och lägre laddningstid, men anstränger litiumbatteriet mer och riskerar

litiumplätering på katod- eller anodmaterialet. En så låg ström som möjligt skall därmed helst appliceras sådan att upp-och urladdningstiden ligger respektive över en timme för att undvika dessa degraderingseffekter. NMC811 halvcellerna cyklades mellan 3–4,2V och 3–4,4V med en låg upp-och urladdningshastighet. Det enbart ytbelagda NMC811 visade sämst cyklingsförmåga mellan både 3–

4,2V och 3–4,4V medan det värmebehandlade ytbelagda NMC811-pulvret visade bäst cyklingsförmåga.

Slutsatsen av projektet var att det inte kunde hitta en stor skillnad i degraderingsprodukter med metoderna som användes mellan de obelagt och belagt NMC811, samt NiO-pulvren.

Värmebehandling av ytbelagda NMC811-pulvret visade en tydlig förbättring i cyklingsförmåga, men kunde inte säkerhetsställa om det berodde på reducering av hydroxidgrupper. Därmed behövs en djupare undersökning av vad värmebehandling hade för effekt på ytbelagda NMC811-pulver.

Abbreviations and Definitions

Abbreviations

ALD Atomic Layer Deposition

ATR-FTIR Attenuated Total Reflection - Fourier Transform Infrared

BET Brunauer–Emmett–Teller

EV Electric vehicles

LIB Lithium Ion Batteries

NMC Lithium nickel manganese cobalt oxide SEI Solid Electrolyte Interface

TGA Thermogravimetric Analysis UV/Vis Ultraviolet / Visible

XPS X-ray Photoelectron Spectroscopy

Definitions

Capacity The amount of discharge current a battery can deliver over time under certain conditions measured in ampere-hours (Ah).

Specific capacity To determine the individual material capacity in a battery, ‘Specific capacity’

can be implemented to determine the density of capacity the material can deliver per gram of active material (Ah/g).

C - rate The speed a battery is charged or discharged. At 1C rate, the cell is applied a certain current to fully charge or discharge within one hour.

At a higher or lower C -rate, the current is either increased or decreased respectively, which relates to a faster or slower fully charged or discharged cell.

Cycling The process of repeated charge and discharge in a cell.

Pouch bag A type of battery cell.

iv

Table of contents

Populärvetenskaplig Sammanfattning... i

Abbreviations and Definitions ... iii

Abbreviations ... iii

Definitions ... iii

Table of contents ... iv

List of Figures ... v

List of Tables ... v

1. Introduction ... 1

Scope ... 2

2. Background ... 3

2.1. High-capacity cathode material: NMC ... 3

2.2. Electrolyte degradation ... 3

2.3. Cathode-electrolyte interaction ... 4

2.4. Thin-film coatings on cathode materials ... 5

3. Experimental ... 6

3.1. Material preparation ... 7

3.2. Powder analysis ... 7

3.3. Soaking experiment ... 8

3.4. Electrochemical testing ... 11

4. Results ... 12

4.1. Powder analysis ... 12

4.2. Soaking Experiment ... 14

4.3. Electrochemical testing ... 23

5. Discussion ... 24

5.1. Powder analysis ... 24

5.2. Soaking experiment ... 24

5.3. Electrochemical testing ... 26

6. Conclusion ... 27

7. Future outlook ... 27

8. Acknowledgement ... 28

9. Reference ... 29

List of Figures

Figure 1 Illustration of EC (ethylene carbonate), DEC (Diethyl carbonate), and LiPF6.

Figure 2 Cycling performance of ALD coated Al2O3 on NMC532 electrodes at different thicknesses.

Figure 3 Reference Raman spectroscopy of electrolyte LP40 (EC:DEC v:v 1:1), solvent EC:DEC (v:v 1:1), and DEC.

Figure 4 TGA plot on pristine and coated NMC811, NiO, and alumina powder, between 30°C to 900°C.

Figure 5 BET surface analysis of pristine and coated NMC811, and NiO powder Figure 6 Water sorption measurement of pristine and coated NMC811 powder.

Figure 7 Visual observation of soaking samples during the experiment.

Figure 8 pH-measurement of electrolyte from the soaking samples after 14 days with pH-paper.

Figure 9 UV/Vis spectra of LP40 electrolyte (1M LiPF6, EC:DEC v:v 1:1), EC:DEC solvent (v:v 1:1), and electrolyte from soaking experiment between 200 - 400 nm.

Figure 10 Raman spectra of the electrolyte from NMC811 samples and reference from the soaking experiment.

Figure 11 ATR-FTIR spectra of soaked NMC811 samples electrolyte from the soak experiment.

Figure 12 ATR-FTIR spectra of soaked NiO samples electrolyte from the soak experiment.

Figure 13 ATR-FTIR spectra of the NMC811 powder samples from the soaking experiment.

Figure 14 ATR-FTIR spectra of the NiO powder samples from the soaking experiment.

Figure 15 XPS spectra of A) C 1s, B) O 1s, C) P 2p, D) F 1s, E) Al 2p, and F) Mn 2p from soaked NMC811 powder samples.

Figure 16 Galvanostatic cycling of the NMC811 electrodes.

List of Tables

Table 1 The powder samples used in this project.

Table 2 Corresponding ATR-FTIR vibrational mode peaks of LiPF6, EC, and DEC in the electrolyte LP40.

1. Introduction

The initiative to move away from fossil to renewable energy sources has increased immensely during the 21st century. One predominant factor comes from the fact that the transport sector stands for one- fourth of the total CO2 emitted in the world.^1,2 To best utilize renewable energy sources and gradually phase out fossil fuels requires energy storage that is reliable, sustainable, has high capacity, among many other things. The key obstacles for electric vehicle (EV) mass-marketing are range-to-cost, charge time, and lifetime when compared to fossil fuel vehicles. The goal to develop future batteries hence focuses on energy- and power density, lower cost, increased lifetime, safety, scalability, and environmental sustainability. Lithium-based batteries have been the most prominent rechargeable battery for almost three decades. Lithium-ion batteries have been the main component in consumer electronics such as cell phones, laptops, power tools, etc, and are expanding into the automotive industry. The reason that lithium-ion batteries are sought after is that they contribute to a higher energy density, lower maintenance, and lower self-discharge compared to other rechargeable batteries that are lead- or nickel-based.³

Rechargeable batteries consist of an anode, a cathode, a separator, and finally an electrolyte. The anode material in a lithium-ion battery typically consists of graphite together with a polymeric binder, but can also consist of different alloys, composites, or metallic lithium. Metallic lithium has the highest theoretical specific capacity of 3800 mAh/g, but is prone to short circuits due to dendrite growth and is therefore seldom used commercially. Lithium-based cathode materials can, for instance, consists of lithium cobalt oxide (LCO), lithium iron phosphate (LFP), or lithium nickel manganese cobalt oxide (NMC) which have theoretical specific capacities of 274 mAh/g, 170 mAh/g, and 278 mAh/g respectively. The cathode material LCO, consisting of layered cobalt oxide has a high specific capacity and good electrochemical stability but has the main drawbacks of poor thermal stability, and expensive raw material cost. By adding manganese into the structure, it has been shown to improve thermal stability and therefore safety, but on its own in cathode materials like lithium manganese oxide (LiMn2O4) has a low specific capacity and a poor cycle life. A higher nickel content has been shown to provide high capacity and is significantly cheaper than cobalt. A higher nickel content is however not without risk. It has been shown that nickel ions diffuse into the lithium-sites in the crystal structure during charge or discharge, which causes a capacity loss due to fewer lithium ions that can diffuse in or out from the cathode material. Hence, the advantage of NMC cathode material lay in its capability of optimizing the ratio of the transition metals nickel, manganese, and cobalt in the material.

By changing the ratio of these transition metals, NMC powder can lead to a high capacity at a lower cost, higher thermal stability, and electrochemical stability compared to LCO and other cathode materials. Nickel-rich NMC materials are highly sought after for their applications in EV’s but are sensitive to the many problems that occur in rechargeable batteries.

One of the problems that occur in rechargeable batteries is that there is a continuous degradation process during charge/discharge and even when the battery is not in use. The total lifetime of a battery, called calendar life, is related to how long the battery is expected to last during its use. The calendar life of a battery is strongly affected by the electrolyte in the battery, the surrounding temperature, and the state-of-charge. The degradation occurs partly due to the electrolyte being unstable and forming byproducts like hydrofluoric acid (HF), which ultimately attack the battery components like the anode and cathode surfaces. These byproducts from the electrolyte contribute for example to the formation of a solid electrolyte interface (SEI) on the anode and cathode or dissolution of transition metals from the cathode material into the electrolyte. This can, in turn, contribute to a structural change on the cathode material and lead to micro-cracks formation on the surface.⁵ This has a negative effect that influences, for instance, the capacity and lifetime of lithium batteries.

Several different solutions are used to decrease the degradation of cathode materials in rechargeable batteries. The most used solutions to stabilize or protect cathode surfaces against the electrolyte byproducts are for example additives in the electrolyte or coating a surface layer on the

anode/cathode material.⁶By coating with metal oxide, metal fluoride, metal phosphate, or metal hydroxide on the anode or cathode material, a passive layer forms on the surface exposed to the electrolyte. This protective layer has shown to enhance electrochemical performance. From previous studies the primary physical role of the surface coating is to i) protect against side reactions between the electrolyte and electrode surface, ii) act as a scavenger against corrosive acids due to HF is known to form from fluorinated salts.⁷

Scope

This thesis aimed to study the surface reactivity of coated NMC and to better understand the surface reaction mechanisms when in contact with an electrolyte. As NMC has a low surface area, NiO powder, which has a higher surface area than NMC, was used to increase the surface area to achieve a higher surface/electrolyte interaction and simulate NMC. NiO is often spontaneously formed at the surface of NMC during cycling and hence is a suitable substitute. The NMC811 and NiO powders were coated with alumina, Al2O3, through atomic layer deposition. The remaining hydroxide groups on the alumina coated surface could potentially react with the electrolyte. This thesis aims to investigate if there is any difference in electrolyte degradation between a pristine and an aluminacoated material when it was in contact with a common electrolyte. Also, could heat treatment at a specific temperature of the coated Al2O3 reduce surface reactivity by potentially removing hydroxide groups at the surface?

Finally, how would the electrolyte degrade and what byproduct would form?

3

2. Background

The degradation processes in a battery are often related to the surface reactions, and in this instance, the cathode material surface. The interaction between the electrolyte and active material has often been pointed out as a culprit for capacity loss in cathode materials. This chapter details the cathode material NMC, electrolyte degradation, cathode degradation, and role of a thin film coating on the cathode material.

2.1. High-capacity cathode material: NMC

Cathode materials such as LiCoO2 (LCO), LiMn2O4 (LMO), and LiFePO4 (LFP) all have different advantages when it comes to energy capacity, electrochemical and thermal stability, cost, and safety.

However, many of these cathode materials lack in certain areas, making them unsuitable in certain applications. Nickel-rich compounds display higher capacity, while manganese-rich compositions favor cycle life and thermal stability, and cobalt-rich compositions have shown to provide

electrochemical stability.⁶ The nickel-rich compound LiNiO2 (LNO) has a similarly high specific capacity to LCO but has a significantly lower material cost. LNO has however been shown to be unstable in the layered structure, during delithiation, due to Ni²⁺ ions tendency to substitute Li-ion sites. This is due to Ni²⁺ ions having similar ionic radii to Li-ions and hence block the diffusion pathways making LNO both electrochemically and thermally unstable.⁸ To resolve the problem of electrochemical and thermal stability, cost, and safety while maintaining a high energy capacity, the elements nickel, manganese, and cobalt [Ni, Co, and Mn] are integrated into the three-component material known as Li[NixMnyCoz]O2 (NMC).

The typical common composition NMC111 (Li[Ni1/3Mn1/3Co1/3]O2) has superiority when it comes to electrochemical performance, thermal stability, and safety compared to LCO.⁹ However, an increase in nickel content is desired to increase the energy capacity and lower material cost. Due to NMC consisting of several different transition metals, the compound presents different redox couples of Ni²⁺, Ni³⁺, Ni⁴⁺, Co³⁺, Co⁴⁺, and Mn⁴⁺. During initial charge/discharge the Ni²⁺ ions are most involved while towards the end of the charge Co³⁺ becomes active. Mn⁴⁺ ions have been shown to be inactive during the charge/discharge and are associated with a reduced capacity in the material, but

contributes to the structural stability.⁴

2.2. Electrolyte degradation

The electrolyte is an essential part of batteries, providing Li-ion transportation between the anode and cathode. The electrolyte in lithium-ion batteries consists most commonly of a liquid solution of lithium salts dissolved in organic solvents. The electrolyte should have high ionic conductivity, high chemical, and electrochemical stability, and be stable in a wide temperature range. The lithium salt in the electrolyte can consist of LiBF4, LiClO4, LiAsF6, or LiPF6. The salt LiPF6 is common in lithium batteries due to its good solubility, ionic conductivity, and stability. The organic solvents, commonly consisting of carbonates are usually mixed, such as EC (ethylene carbonate) and DEC (Diethyl carbonate) seen in Figure 1.

Figure 1: Illustration of a) EC (ethylene carbonate), b) DEC (Diethyl carbonate), and c) LiPF6

The mixing of cyclic and linear carbonates is done to optimize ionic conductivity due to their different physicochemical properties such as dielectric constant and viscosity. This property affects battery performance. EC has a high dielectric constant, which gives high salt solubility and mobility, but has high viscosity while DEC has a low dielectric constant and low viscosity.

The drawback of the LiPF6 salt is that it has poor thermal stability and fluorinated salts are well known to form reactive and corrosive species such as HF acid. Water content is ideally not present in a lithium cell, but moisture and hydroxide traces from undried components are common, contributing to salt degradation. The LiPF6 saltcan decompose into HF in several ways as seen in reaction (1), and (3):^10,11

LiPF6 → LiF + PF5 (1)

PF5 + H2O → 2HF + POF3 (2)

LiPF6+ 4H2O → 5HF + LiF + H3PO4 (3)

The HF harms the battery performance not only by causing loss of electrolyte but by attacking the anode and cathode surfaces. The SEI formation on the anode and cathode can be attributed to contact with reactive species such as HF and PF5 but also from thermal or electrochemical degradation. The solvent EC reacting with PF5 species in the electrolyte often leads to its ring-

opening, which undergoes polymerization into polyethylene and CO2 gas formation. The DEC reacting with HF and PF5 leads to further HF formation and CO2 gas formation exemplified in reaction (4) and (5):¹²

C2H5OCOOC2H5 + PF5 → C2H5OCOOPF4 + HF + C2H4 (4) C2H5OCOOPF4 + HF → PF4OH + C2H5F + CO2 (5)

A variety of solvent and salt decomposition products can be found on the anode and cathode surfaces including LiF, Li2CO3, R-O-Li, M-CO3, R-O-CO2-M, MF2 (R = organic chain, M = Ni, Mn, Co) along with different polymeric compounds.¹³ The different species at the anode or cathode surface can cause a vicious cycle of HF producing small amounts of water in the electrolyte. An example is seen in reaction (6) with lithium carbonate, often originating after synthesis on the anode or cathode surface:^4,14

2HF + Li2CO3 (s) → 2LiF + H2O + CO2 (6)

2.3. Cathode-electrolyte interaction

nickel-rich NMC compounds are prone to a wide range of different kinds of degradation influenced by surface effects, structural effects, and morphological effects.⁹ The cathode-electrolyte interface has been shown to be vulnerable to side-reactions from the electrolyte when it decomposes into corrosive species and forming SEI layers, leading to transition metal dissolution and an insulating layer blocking electron transport.

The dissolution of the transition metals, where manganese is especially susceptible, is often caused by HF acid. The dissolution causes loss of active material in the cathode material and can precipitate into metal fluoride species on the anode and cathode surface, inducing further SEI formation on the surface.¹¹

The parasitic surface reactions are accelerated when cycling at higher potentials beyond 4.2 V vs Li⁺/Li, where destabilization in the surface crystal structure has been observed. At higher potentials, the layered NMC structure may convert into a NiO-type rock-salt structure due to cation mixing when Ni²⁺ ions migrate into vacant Li sites along with the release of oxygen at the surface.^7,15,16 Due to the

5

structural change within the cathode material and oxygen release, it induces mechanical strain in the particles which form microcracks near the surface, exposing the inner parts of the NMC particles to electrolyte degradation.

2.4. Thin-film coatings on cathode materials

To protect the surface from electrochemical side-reactions and parasitic reactions induced by the electrolyte, introducing a surface coating to the cathode has shown to improve stability. In previous studies, different surface coatings such as metal oxides, metal fluorides, metal hydroxides, etc have shown significant improvements in cycling performance and structural stability of the cathode material.⁷ The role of the surface coating can be summarized as a protective layer between the electrolyte and cathode surface, protecting the cathode surface from HF attacks and suppressing transition metal dissolution into the electrolyte. Metal oxide coatings like Al2O3, MgO, ZnO, and ZrO2

will act as acid scavengers in electrolytes based on partially decomposing fluorinated salts, thus transforming the surface into a metal fluoride layer. For instance, Al2O3 coating in contact with HF will gradually form a passive aluminium fluoride layer (AlF3), seen in reaction (7):^7,17

Al2O3 (s) + 6HF → 2AlF3 (s) + 3H2O (7)

The layer of AlF3 is stable against any other HF interaction. In a previous study, it has been proposed that the beneficial role of Al2O3 comes directly when in contact with LiPF6 salt. The LiPF6 reacting with Al2O3 produces small amounts of lithium difluorophosphate (LiPO2F2), an electrolyte additive

commonly used to improve cycling stability and lifetime in lithium batteries. The reaction can be seen in reaction (8):

2Al2O3 (s) + 3PF6- → 4AlF3 (s) + 3PO2F2- (8)

Some amount of LiPO2F2 has been shown to form on uncoated materials but significantly more on Al2O3 coated materials.¹⁸

The coating methods commonly used are atomic layer deposition (ALD) or a wet-chemical process.

The ALD method is more expensive compared to the wet-chemical process, however, the ALD method provides a better film quality and control of the coating thickness. Common precursors used for ALD coating of Al2O3 are trimethylaluminium (TMA) and H2O were the ALD process has two self- terminating reactions involved in (9) and (10):

AlOH* (s) + Al(CH3)3 (g) → Al–O–Al–(CH3)2* (s) + CH4 (g) (9)

Al–CH3* (s) + H2O (g) → AlOH* (s) + CH4 (g) (10)

The active surface sites are noted as *. This inevitably leaves hydroxide species at the surface which could potentially react further with the electrolyte.

In previous studies, controlling the film thickness has shown improved cycling performance when ALD coated Al2O3 was compared with various thicknesses as shown in Figure 2.^7,19–21

Figure 2: Cycling performance of ALD coated Al2O3 on NMC532 electrodes from literature where the thicknesses corresponding to 0 (pristine electrode), 2 (2-ALD electrode), 5 (5-ALD electrode), and 8 (8-ALD electrode) coating cycles.

Galvanostatic cycling in 3.0–4.5 V vs. Li⁺/Li at a cycling rate of 0.5 C for 100 cycles.²¹

The thin ALD coating allows lithium-ion diffusion through with little interfacial resistivity as Al2O3 has insulating properties.

Heat treatment of the coated cathode material is often used after a wet-chemical process because the method initially forms a loose and non-uniform coating on the cathode materials. When annealing an Al2O3 coated cathode material that has been coated by a wet-chemical process at 400°C for 8h, the coating becomes denser but retains its morphology.²² However, increasing the annealing temperature higher than 400°C the surface coating becomes smoother and hard-shell-like, but leads to

morphological change and has been recorded to lead to insertion of Al into the bulk material. The diffusion of Al ions into the bulk material from the coating layer may lead to that the coating no longer forms a protective layer around the cathode material to prevent surface degradation, which in turn leads to lower capacity and cyclability. Previous studies have shown that heat treatment of Al2O3

coated NMC with low manganese content results in the diffusion of aluminium cations into the NMC bulk.^22,23 Meanwhile, NMC with higher manganese content showed that the manganese cations had a blocking effect on the aluminium cation insertion.

3. Experimental

The scope of this thesis was to determine if there was a difference in surface reactivity between pristine and coated samples. The project was divided into three parts; powder analysis, soaking experiment, and electrochemical performance. The aim of the powder analysis was to find a difference in the surface area and adsorption properties between pristine and coated powders. The powder analysis also included finding the thermal stability so a specific temperature for heat-treating the coating could be found. The aim of the soaking experiment was to see a difference in pristine, coated, and heat-treated powders when in contact with an electrolyte for a set period. Finally, the aim of the electrochemical performance test was to find which of the pristine, coated, or heat-treated cathode electrodes had the best cyclability.

This chapter provides a brief description of the experimental procedures used to prepare and analyze the samples in this project.

7

3.1. Material preparation

The base materials used were NMC811 powder and NiO nanopowder (<50 nm particle size), which was commercially purchased from Helium and Sigma Aldrich, respectively. NiO powder was used because of the reported low surface area of the NMC811 powder. The higher surface area of the NiO powder was anticipated to show an increased electrolyte-surface interaction. The NMC811 and NiO powders were coated with 5 ALD cycles of Al2O3, using trimethylaluminum (TMA) and H2O as precursors. The ALD was performed in a PICOSUN™ R-200 Standard under vacuum condition, controlled at 6-10 hPa in the reaction chamber, with a substrate temperature of 120°C. The

sequenced program consisted of H2O exposure time of 2 s, purge for 100 s, TMA exposure time of 2 s, and final purge for 100 s, per cycle. Nitrogen gas was used as a purge gas with a flow rate of 100 sccm. The expected growth per cycle of alumina taken from the literature is approximately 0.13 nm/cycle hence the expected thickness of the alumina layer after 5 cycles were 0.65 nm thick.²⁴ Half of the coated NMC811 and NiO powders were separated and heat-treated in a glass tube furnace at a specific temperature chosen from the TGA measurements for 8h under 50 sccm nitrogen gas flow using an alumina crucible. Additionally, the pristine NMC811 powder sample was heat- treated similarly at the same temperature for 8h, to investigate if the heat treatment affected uncoated samples as well. The powders were after the heat treatment quickly taken back into a glove box and labeled as “LowTemp”. All the powders were stored in a glove box (O2 ≤ 5ppm, H2O ≤5 ppm) when not used in experiment or analysis. The NMC811 and NiO powder samples used during this thesis can be seen in table 1.

Table 1: The powder samples used in this project. The coated samples marked as “5 cycles ALD Al2O3”, while the heat treatment samples marked as “LowTemp”.

# Material

1 NMC811 Pristine

2 NMC811 + 5 cycles ALD Al2O3

3 NMC811 + 5 cycles ALD Al2O3 + LowTemp 4 NiO Pristine

5 NiO + 5 cycles ALD Al2O3

6 NiO + 5 cycles ALD Al2O3 + LowTemp 7 NMC811 Pristine + LowTemp

3.2. Powder analysis

Thermogravimetric Analysis (TGA)

The first objective was to find the thermal stability of the pristine and coated powders. Thermal Gravimetric Analysis (TGA) was used to investigate where the NMC811 and NiO powders would potentially show water or hydroxide reduction and determine the heat treatment temperature.

The powder samples were spread out to a thin layer on an alumina crucible and weighed. Pristine and coated NMC811 and NiO powders were measured on a TGA Q500 between the temperature 30°C to 900°C with a temperature gradient of 5°C/min. Al2O3 powder was measured as a reference if the coated NMC811 and NiO powders didn’t show any reduction from the coating. The TGA

measurements were performed under nitrogen gas with a balance flow rate of 40 ml/min and a sample flow rate of 60 ml/min.

Brunauer–Emmett–Teller (BET) & Water - sorption

To investigate the surface area and adsorptive properties between pristine and coated NMC811 and NiO powder, a BET measurement with nitrogen gas was performed using a Micromeritics ASAP 2020.

The degassing was set to ramp at a temperature of 5°C/min with a hold temperature of 400°C for 240 min.

Water sorption was used to see a difference between the pristine and the coated NMC811 powders ability to absorb and desorb moisture. The water sorption measurement was performed using a similar Micromeritics ASAP 2020 instrument.

3.3. Soaking experiment

The main goal of the soaking experiment was to induce an interfacial reaction without electrochemical processes between the electrolyte and the active materials for two weeks. The interfacial reactions were accelerated by storing the samples at an elevated temperature.

The soaking experiment consisted of mixing the powders listed in table 1 with a commercially

purchased electrolyte LP40 (1M LiPF6, EC: DEC v:v 1:1) from Solvonics. The powders were dried in a vacuum oven at 120°C for 12h inside a glove box before soaking in the electrolyte. The ratio of 125 mg powder / per ml electrolyte was used to simulate the same ratio used during cell assembly. An amount of 375mg active powder with 3ml of electrolyte was chosen to have enough electrolyte sample volume for all measurements. The soaked materials were stored in UV/Vis protected clear glass vials, sealed with a polypropylene screw-top and parafilm fitted around. The soaked samples were then put in a glass jar, sealed with a lid together with a parafilm fitted around to ensure minimum oxygen and water exposure. A reference sample with the only electrolyte was filled into a glass vial and sealed. The samples were then taken outside of the glove box to be stored in a 50°C oven for two weeks. The colour change of the electrolyte inside the soaked samples was regularly checked during the experiment.

After the experiment duration, the acidity was measured by opening the sealed samples and pipetting a small amount of the electrolyte onto a pH paper. The soaked electrolyte was at first analyzed and afterward, the soaked powders were separated from the electrolyte. The separation of the soaked powder material was performed inside the glove box and began by individual filtering using filter paper (grade 3), separating the electrolyte from the powder, and pouring DEC solvent to remove excess electrolyte. The powder samples were thereafter put in glass vials and dried in a <50°C Buchi oven for 1-2h under vacuum. The analysis methods used in this project are described in the following section.

3.3.1. Spectroscopy

Spectroscopy was used mainly to see a difference in binding and formation of complexes that had formed during the soaking experiment. The electrolyte from the soaking experiment was measured using ultraviolet/visible (UV/Vis), Raman spectroscopy, and Attenuated Total Reflection - Fourier Transform Infrared (ATR-FTIR) spectroscopy, while the soaked powders were measured in ATR- FTIR andX-ray Photoelectron Spectroscopy (XPS) after the soaking experiment.

Ultraviolet/Visible Spectroscopy

The ultraviolet/visible (UV/Vis) spectroscopy was used to quantify the colour difference from the electrolyte degradation in the UV/Vis region. Electrolyte from the soaking experiment was filled into cuvettes and an absorbance measurement was performed between 100 and 1100 nm on a HP 8453 UV/Vis spectrometer.

9

Raman Spectroscopy

Raman spectroscopic measurements were done on the electrolyte to identify a change in the C-C, C- O, C=O, P-F, etc bonds from the salt LiPF6 or organic solvents, EC and DEC, between the soaked samples. Samples with strong colour change would be problematic due to Raman spectroscopy is susceptible to background fluorescence interference, which makes measurements incomprehensible or small deviations in the vibrational distribution harder to detect.²⁵

The Raman spectroscopy was conducted on a Renishaw inVia™ Qontor. The measurement used a 785 nm laser with a 1200 nm grid line, 1% intensity, 50x aperture, exposure time of 20 seconds, with 5 acquisitions. Droplets of the electrolyte sample were put on a stainless-steel sample holder with a glass window for each measurement. The Raman spectra were measured between wavenumber 100 cm^-1 to 3200 cm^-1.

Reference Raman peaks of the electrolyte LP40, the solvent EC: DEC, and DEC can be seen in Figure 3.

Figure 3: Reference Raman spectroscopy of electrolyte LP40 (EC:DEC v:v 1:1), solvent EC:DEC (v:v 1:1), and DEC. Source:

Nataliia Mozhzhukhina, Uppsala university

Attenuated Total Reflection - Fourier Transform Infrared Spectroscopy (ATR - FTIR) Attenuated Total Reflection - Fourier Transform Infrared (ATR-FTIR) measurements were similarly performed to identify a change in C-C, C-O, C=O, P-F, etc bonds. The ATR is an accessory to the FTIR spectroscopy which allows non-destructive analysis on liquids, semi-solids, polymers, and powders with little sample preparation needed and easy to use. The ATR works by letting an infrared

beam at a set angle through a ZnSe or diamond crystal. When the infrared beam reaches the surface of the crystal, which has a higher reflective index than the sample, the reflection creates an

evanescent wave that extends into the sample that is held in close contact with the crystal. Some of the energy from the evanescent wave is absorbed by the sample and the reflected wave will be altered before it returns to the crystal. The infrared beam reflects several times with the sample until it reaches the end of the crystal and exits to a detector. This makes the ATR accessory suitable for surface characterization of materials that are either too thick or too strong absorbing. The main advantage with ATR-FTIR spectroscopy over Raman spectroscopy is that it is not susceptible to fluorescence interference.

The ATR-FTIR measurements were performed using a PerkinElmer Spectrum One. From the soaking experiment, both the electrolyte and soaked powder samples were measured in the ATR-FTIR by placing droplets of sample electrolyte or by applying a thin film of the powder sample respectively on top of the diamond/ZnSe analyzing area. A borosilicate glass cover was used in the electrolyte measurements to prevent the electrolyte from evaporating. The ATR-FTIR spectra were measured between 450 nm to 4000 nm with 20 acquisitions per sample with a resolution of 4 cm^-1. The background was fitted with the borosilicate glass cover in the electrolyte sample measurements.

The expected peaks from the electrolyte LP40 can be found in table 2.

Table 2: Corresponding ATR-FTIR vibrational modes peaks of LiPF6, EC, and DEC in the electrolyte LP40.^26,28

Wavenumber (cm

^-1

) Compound Vibrational mode

557 LiPF6 F-P-F bending

716 EC Ring bending

728 EC Ring bending Li⁺ shifted

774 EC Out-of-plane ring –C=O bending

792 DEC CH2 wagging

838 PF6-

904 EC Ring breathing

972 EC Ring stretch

1016 DEC C–C–O asymmetric stretch

1071 EC O-C-O asymmetric stretch

1085 EC O-C-O asymmetric stretch Li⁺ shifted

1158 EC CH2 twist

1196 EC CH2 twist Li⁺ shifted

1259 DEC C-H bend

1300 EC CH2 wagging

1376 DEC C-H bend

1390 EC CH2 wagging

1407 EC C-H bending in adsorbed or solvated EC

1482 EC CH2 scissor

1720 DEC C=O stretch Li⁺ shifted

1740 DEC C=O stretch

1770 EC C=O stretch

1802 EC C=O stretch

2938 DEC C-H stretch of the -CH3 groups 2986 DEC C-H stretch of the -CH3 groups

11

From degradation of the electrolyte, the EC peaks at 716, 728, 972, 1158, 1196, and 1390 cm^-1 were expected to vanish in the soaking experiment. If a SEI of EC had formed the peaks at 1770 and 1802 cm^-1 were expected to decrease in the electrolyte.²⁶ In the LiPF6 salt degradation, the peaks 557 and 838 cm^-1 were expected to decrease.

X-ray Photoelectron Spectroscopy (XPS)

X-ray Photoelectron Spectroscopy (XPS)was used to investigate transition metal dissolution and surface species formed during the soaking experiment. The XPS powder samples were prepared inside a glove box (O2 ≤ 5ppm, H2O ≤5 ppm), where a thin film of the powder samples was pressed on individual indium foil bits, situated on a sample holder with copper tape.

The XPS measurements were conducted on a Perkin Elmer Phi Model 5600 MultiTechnique system with Al K-∝ source, which has a 1486.6 eV photon energy.²⁶ The carbon C 1s, oxygen O 1s,

phosphorous P 2p, and fluoride F 1s peaks were analyzed to find if a potential SEI layer had formed and what potential surface species could be on the soaked powder samples. The manganese Mn 2p peaks were measured to analyze any metal dissolution. As previously mentioned in the Cathode- electrolyte interaction the ratio of Mn⁴⁺ ions will decrease if manganese metal dissolution would have occurred. The C 1s, O 1s, P 2p, F 1s, and Mn 2p spectra were plotted in the software CasaXPS with the intensity in arbitrary units and the binding energy in eV.

3.4. Electrochemical testing

To determine whether the pristine, coated, or coated heat-treated NMC811 powder had better electrochemical performance compared to each other, NMC811 half-cells in pouch-bag format were made and galvanostatically cycled. The cathodes were prepared by making a slurry containing 90 wt% of active material, 5 wt% C65 Carbon, and 5 wt% polyvinylidene difluoride (PVDF) (Kynar FLEX 2801) binder in N-methyl-2-pyrrolidone (NMP). The slurry was ball milled in a Retsch MM400 shaking mill mixer, on 25 Hz for 60 min. The slurry was then spread out on an aluminum foil using a gap bar coater with a gap size of 200 μm with transparent tape under the legs, resulting in a gap of 248 μm.

The coated film was pre-dried in a vacuum oven at 70°C for 3h. The electrodes were punched out with a diameter of 10 mm resulting in a mass loading of 12-14 mg/cm². The punched-out electrodes were transferred to a Buchi oven inside a glove box and fully dried in vacuum 120°C for 12h.

In cell assembly, pouch-bags were used with copper and aluminum current collectors for the anode and the cathode respectively. The anode consisted of a commercially purchased lithium - chip (Ø15mm). During cell assembly 80 μL of the electrolyte LP40 (1M LiPF6, EC: DEC v:v 1:1) was used with Celgard 2325 as a separator. After vacuum sealing, the cells were left for 12h to ensure wetting.

The electrochemical cells were later cycled in an Arbin BT2000 between 3-4.2V and 3-4.4V with a rate of 0.1C for the first three cycles and 0.5C for 98 cycles. The reason for comparing the NMC811 electrodes at higher potentials above 4,2V vs Li⁺/Li was discussed in the chapter Cathode-electrolyte interaction. The reason was that previous studies had shown destabilization of the surface crystal structure on the NMC material leading to microcracks. The coated NMC811 electrodes should be protected from the harmful oxidative environment and hence show better cyclability compared to the pristine NMC811 electrodes.

4. Results

4.1. Powder analysis

Thermogravimetric Analysis (TGA)

TGA measurements were performed on pristine and coated NMC811, NiO, and alumina powder. To ensure reproducibility, measurements on each sample were taken at least twice and an average value was plotted for each sample. The TGA plot from the pristine and coated NMC811, NiO, and alumina powder can be seen in Figure 4.

Figure 4: TGA plot on pristine and coated NMC811 and NiO, and alumina powder between 30°C to 900°C with a temperature gradient of 5°C/min.

The TGA measurements in Figure 4 of the pristine and coated NMC811 powders can be seen to be thermally stable up until 600°C. The weight of pristine NiO powder can be seen to reduce until it reaches two plateaus, one between 250 to 400°C and the second one at 550 to 900°C. The coated NiO can be seen to have its first plateau between 250 to 350°C and similarly the second one between 550 to 900°C. The alumina powder can be seen to have an exponential weight decrease up until 900°C indicating the alumina powder loses more than absorbed water and organic compounds. From the TGA measurements and data from previous studies, the temperature for heat treatment was chosen to be 400°C for the coated NMC811 and NiO powders. This was to ensure the reduction of water and other compounds on the NiO powders and stay off the second plateau.

Brunauer–Emmett–Teller (BET) & water - sorption

The pristine and coated NMC811 and NiO powders were dried before analysis in a Buchi oven inside a glove box at 120°C for >12h. Due to the low surface area of NMC811 powders, an amount of ~7 grams of the NMC811 powders were used in the BET measurements. For the BET analysis of the pristine and coated NiO powders, ~2g of each powder was used. The results from the BET analysis can be seen in Figure 5.

13

Figure 5: BET surface analysis of a) pristine and coated NMC811 powder, b) pristine, and coated NiO powder.

The BET surface area of the pristine and coated NMC811 powders was estimated to be 0.6107 m²/g and 0.7266 m²/g, respectively. The NMC811 powders can be seen in Figure 5 a) to have weak interaction between the surface and nitrogen gas as there appeared to be incredibly low nitrogen gas quantity adsorbed at the surface even at relatively high pressures. This indicated that the nitrogen gas adsorbent interacted with itself rather than the sample surface until it reached the saturation pressure, Po, forcing capillary condensation on the powder samples surface. The hysteresis from the desorption curve indicates that certain nitrogen adsorbent was obtained into the pores of the powder.

The BET surface area of the pristine and coated NiO powders were measured to 75.57 m²/g and 73.97 m²/g respectively. The NiO powders can be seen to have a stronger interaction with nitrogen gas compared to the NMC811 powders. At low pressures, nitrogen gas can be seen to fill the micropores, while at higher pressures the monolayer formation of absorbed nitrogen starts to form multilayers. At the highest pressures, capillary condensation occurs at the powder samples surface.

The result from the pristine and coated NMC811 powders analyzed in the water-sorption measurement can be seen in Figure 6.

a)

b)

Figure 6: Water sorption measurement of pristine and coated NMC811 powder.

The water-sorption measurement of the NMC811 powders showed similar physisorption properties with nitrogen gas of weak interaction with the water-absorbent gas. From low to high pressure the water vapour have weak interaction with the surface on both the coated and pristine NMC811 powders. It’s not until at higher pressure when capillary condensation occurs on the powder samples where the water is absorbed on the surface.

4.2. Soaking Experiment

The soaking samples were prepared inside the glove box, sealed and put in a sealed glass jar, and finally put outside the glove box in a 50°C ventilated oven.

During the first 7 days, after 24h the sample 5 (containing NiO + 5 cycles ALD Al2O3, see Table 1) began to show red colour change, followed by a green colour change in sample 4. Sample 6 showed a green/yellow colour change after 3 days. No observable colour change could be seen in the NMC811 samples, however, after 5 days a layer could be seen forming on all powder samples. The surface layer was most noticeable on the NMC811 powder samples, indicating some form of electrolyte degradation on the surface had occurred.

After 7 days, the reference sample (containing electrolyte LP40 only) began to turn yellow/red. The colour change in the NiO powders continued as it had previously, with the strongest change shown in sample 5 > 4 > 6. As the experiment reached its end after 14 days, the NMC811 powder samples 2 and 3 began to show a slight colour change, with the strongest change shown for sample 2. No observable colour change could be seen in the pristine NMC samples, 1 and 7, but a layer could be seen to have formed on top of the powders, similarly in samples 2 and 3. The colour change from day 0 to 7 days and 14 days can be seen in Figure 7.

15

Figure 7: Visual observation of soaking samples during the experiment. The top Figure represents from day 0, middle Figure after 7 days, and the bottom Figure after 14 days.

After the soaking duration, the samples were opened in air under a fume hood and pH measurement was performed on the electrolyte. The result of the pH measurement can be seen in Figure 8.

Figure 8: pH-measurement of electrolyte from the soaking samples after 14 days with pH-paper.

The NMC811 samples in samples 1, 2, 3, and 7 had a pH-level of ~4. Meanwhile the NiO and reference samples in the 4, 5, 6, and ref. had a pH-level ≤1.

UV/Vis Spectroscopy

Because of too much interference from the electrolyte in the NiO samples from the colour change when measuring in the UV/Vis range, no conclusive spectrum could not be obtained. Of the full UV/Vis spectra that was observed between 100 and 1100 nm the only significant peaks that could be observed were below 400 nm in the UV/Vis measurement, hence only UV spectra’s was be plotted.

The UV spectra of the fresh electrolyte LP40, the solvent EC: DEC, and electrolyte from the reference and soaked NMC811 samples from the soaking experiment can be seen in Figure 9.

Figure 9: UV spectra of LP40 electrolyte (1M LiPF6, EC:DEC v:v 1:1), EC:DEC solvent (v:v 1:1), and electrolyte from the soaking experiment between 200 - 400 nm. The absorbance signal is normalized into arbitrary units (a.u).

In Figure 9, the electrolyte LP40 showed to have a strong absorption below 220 nm, a plateau from 220 to 240 nm and gradual descent in absorption until 290 nm until a faint broad peak could be seen at 350 nm. The EC: DEC solvent showed a decrease in absorption where a minimum could be seen at 223 nm with a peak at 232 nm could be observed, beyond 260 nm no other absorption peaks could be observed. The reference sample showed similar characteristics to the EC: DEC solvent with a minimum seen at 238 nm with a peak at 266 nm, with however a broad absorption tail above 300 nm.

The NMC811 samples showed little difference between UV/Vis spectra’s and were hence unable to distinguish a difference in electrolyte degradation.

17 Raman Spectroscopy

The electrolyte from the coated NiO samples was too fluorescent to get an adequate measurement and hence electrolyte in the NiO soaked samples could not be shown in the report. The results of the Raman spectroscopy of the electrolyte from the soaking experiment can be seen in Figure 10.

Figure 10: Raman spectra of the electrolyte from NMC811 samples and reference from the soaking experiment. The top Figure;

a) 300-1400 cm^-1, bottom Figure; b) 1400-3200 cm^-1, right Figures; c) enhancement of the peak between 940-870cm^-1, d) enhancement of the peak between 760-700cm^-1. The reference spectra were increased by a magnitude of x5 due to low signal.

Note that the offset Y-value differs between the plots.

By comparing with the reference Raman spectra seen in Figure 3 the Raman measurement of the NMC811 samples seen in Figure 10 showed no indication of solvent degradation of EC or DEC. The Raman measurement did show a change in the overlapping peak of DEC (O-C-O) and solvated EC+Li⁺ at 904 cm^-1 and PF6 peak at 743 cm^-1. The peak at 904 cm^-1 from the electrolyte in the pristine and pristine heat-treated NMC samples was not visible and showed a lower intensity of the PF6 peak at 743 cm^-1.

ATR-FTIR Spectroscopy

Electrolyte analysis

The ATR-FTIR measurements on the electrolyte from the NMC811 and NiO soaking samples can be seen in Figure 11 and Figure 12 respectively.

Figure 11: Normalized ATR-FTIR spectra of soaked NMC811 samples electrolyte from the soak experiment between 0 to 100 transmission (%). Top Figure a) 550-1600 cm^-1, middle Figure b) 1600-2300 cm^-1, bottom Figure c) 2300-4000 cm^-1.

19

Figure 12: Normalized ATR-FTIR spectra of soaked NiO samples electrolyte from the soak experiment between 0 to 100 transmission (%). Top Figure 550-1600 cm^-1, middle Figure 1600-1900 cm^-1, bottom Figure 2900-3100 cm^-1.

The peaks in Figure 11 and Figure 12 identified with Table 2 showed no clear difference between the NMC811 and NiO samples. The peaks that were expected to decrease from electrolyte degradation, as discussed in the Experimental section, showed no clear difference from the fresh LP40 or

reference sample in the ATR-FTIR spectra. The EC peaks from the EC ring at 716, 728, and 972 cm^-1 showed no significant decrease in the signal as seen in Figure 11 and Figure 12. The CH2 bonds from EC at 1158, 1196, and 1390 cm^-1, and the C=O bonds at 1770 and 1802 cm^-1 showed no significant drop in signal. The LiPF6 salt peaks at 557 and 838 cm^-1 showed similarly no clear difference between the NMC811 and NiO samples.

Powder analysis

The powder from the soaking experiment was analyzed in ATR-FTIR after separation from the electrolyte and drying in a vacuum. A thin film of the powder sample was put on the ATR-FTIR diamond detector to cover the whole detection area. The ATR-FTIR spectrums from the NMC811 and NiO powder samples in the soaking experiment can be seen in Figures 13 and 14.

Figure 13: ATR-FTIR spectra of the NMC811 powder samples from the soaking experiment. The transmittance signal was normalized between 0 and 100.

21

Figure 14: ATR-FTIR spectra of the NiO powder samples from the soaking experiment. The transmittance signal was normalized between 0 and 100.

The spectra’s in Figure 13 and Figure 14 showed that surface species could be found on the surface of the NMC811 and NiO samples. The pristine and as-coated NMC811 powder can be seen to have a strong signal from EC while the coated heat-treated NMC811 has a stronger signal from DEC. A strong signal from the LiPF6 salt can be observed on the coated and pristine heat-treated NMC811 powder, but not on the pristine and coated heat-treated NMC811 powder.

The NiO powders shown in Figure 14 does not follow the same trend as the NMC811 powders. The pristine and coated heat-treated NiO powder show similarity to the NMC811 powders but the coated NiO shows a strong signal from both EC and DEC. All the NiO powders show a weak signal to the LiPF6 salt. This could indicate that the ATR-FTIR measurements on the soaked NMC811 and NiO powder could have been sensitive to the sample preparation.

X-ray photoelectron spectroscopy (XPS)

The soaked powder samples were prepared and analyzed in the in-house XPS. The F1s, O1s, P 2p C 1s, Al 2p, and Mn 2p can be seen in Figure 15.

Figure 15: XPS spectra of soaked NMC811 powders after separation and drying. The elements analyzed were A) C 1s, B) O 1s, C) P 2p, D) F 1s, E) Al 2p, and F) Mn 2p.

In Figure 15, the spectra from the XPS measurements showed that various chemical compounds exist on the soaked NMC811 powders. In the C 1s spectra, carbon-based species relating to C-C, C-H, C- O, and more can be observed between 284-294 eV. The coated and coated heat-treated NMC811 can be seen to have carbon-fluoride species of CF3 and CH2CF2 respectively, while the pristine samples lack any such peaks. The coated heat-treated NMC811 lacks any C-C peak at 284.8 eV which questions if any carbon species are on the surface.

In the O 1s spectra, the metal oxide, M-O, can only be seen in the pristine heat-treated NMC811 sample indicating that there may have been a thick layer of organic species on the pristine and coated samples.

The phosphorus peak in P 2p and fluoride peaks in F 1s indicate the presence of LiPF6 and LiF on the pristine samples. The coated samples showed, however, no such indication, and in the F 1s spectra

23

showed unknown peaks between 691-696 eV which could not be determined. The peaks may arise from various polymeric compounds as indicated by the (-CF2-CF2-)n at 690 eV but may arise elsewhere.

The Al 2p spectra in Figure 15 E) shows that the aluminium signal was absent. This could be because of a thick surface layer formed from the electrolyte degradation products blocking the signal to the alumina coating.

The Mn 2p spectra showed a slight indication of metal dissolution. It could be argued that the pristine heat-treated NMC811 has a higher Mn⁴⁺ signal compared to the pristine and coated heat-treated NMC811 indicating less Mn²⁺ ions were dissolved. The coated NMC811 which had almost no manganese signal indicating most probable that a thick surface layer was blocking the manganese signal. An Indium peak could be seen at 665.8 eV.

The XPS measurements showed no consistency in surface layer formation and manganese metal dissolution between the pristine and coated NMC811 samples.

4.3. Electrochemical testing

The pristine, coated, and coated heat-treated NMC811 electrodes were prepared and cycled. The results of the galvanostatic cycling at a rate of 0.5C of the pristine, coated and coated heat-treated NMC811 electrodes between 3-4.2V and 3-4.4V can be seen in Figure 16. To ensure reproducibility, the different electrodes were cycled at least 2 times.

Figure 16: Galvanostatic cycling of the NMC811 electrodes. The cycling was performed between a) 3-4.2V and b) 3-4.4V with the first 3 cycles at a 0.1C rate and thereafter 98 cycles of 0.5C rate. The NMC811 electrodes consisted of pristine, coated, and coated heat-treated active material.

The discharge capacity retention of the pristine, coated, and coated heat-treated NMC811 electrodes when cycled with a 0.5C - rate at 3-4.2V was 73.2%, 68.8%, and 84.8%, while cycled at 3-4.4V was 67.5%, 60.9%, and 76% respectively over 100 cycles. The low steep slope in both cases of the coated heat-treated NMC811 electrodes in Figure 16 indicated an improved cyclability in addition to overall higher capacity retention after 100 cycles compared to the pristine and as-coated NMC811 electrodes.