Study of Li metal anode surface. interaction with atmospheric gases and impact of impurities in electrochemistry

EUSKAL HERRIKO UNIBERTSITATEA/ UNIVERSIDAD DEL PAÍS VASCO Zientzia eta Teknologia Departamentua/ Facultad de Ciencia y Tecnología Study of Li metal anode surface: interaction with atmospheric gases and impact of impurities in electrochemistry February 2020 Ane Etxebarria Dueñas Thesis Supervisor: Dr. Miguel Ángel Muñoz Márquez EUSKAL HERRIKO UNIBERTSITATEA/ UNIVERSIDAD DEL PAÍS VASCO Zientzia eta Teknologia Departamentua/ Facultad de Ciencia y Tecnología CIC ENERGIGUNE Study of Li metal anode surface: interaction with atmospheric gases and impact of impurities in electrochemistry February 2020 A dissertation submitted to the University of the Basque Country in partial fulfillments of the requirements for the degree of Ph.D. Ane Etxebarria Dueñas Thesis Supervisor: Dr. Miguel Ángel Muñoz-Márquez UPV/EHU Tutor: Dr. Francisco Javier Zúñiga Lagares (cc)2020 ANE ETXEBARRIA DUEÑAS (cc by 4.0) ‖ iii Contents Contents Acknowledgements --------------------------------------------------------------------------- ix Abstract/Resumen/Laburpena ------------------------------------------------------------ xi 1. Introduction ------------------------------------------------------------------------------- 1 Motivation ------------------------------------------------------------------------------ 1 Li-ion technology ---------------------------------------------------------------------- 3 1.2.1 Li-ion technology basics ---------------------------------------------------------- 3 1.2.2 Li-ion battery main component materials ----------------------------------- 5 1.2.3 Solid Electrolyte Interphase (SEI): a key parameter ----------------------- 8 Next generation Li metal batteries (LMB): role of Li metal --------------- 10 1.3.1 Li-sulfur and Li-air batteries --------------------------------------------------- 11 1.3.2 Li metal surface instability ----------------------------------------------------- 13 1.3.3 Artificial solid electrolyte interphases for Li metal anodes ------------ 16 1.3.4 Considerations of Li metal-based energy demand ---------------------- 19 Scope of the thesis ----------------------------------------------------------------- 22 2. Experimental techniques ------------------------------------------------------------- 23 Thin film growth --------------------------------------------------------------------- 23 2.1.1 Thermal evaporation ----------------------------------------------------------- 23 2.1.2 Magnetron sputtering ---------------------------------------------------------- 25 2.1.2.1 Sputtering instrument --------------------------------------------------- 25 Surface modification --------------------------------------------------------------- 26 iv ‖ 2.2.1 Ion bombardment--------------------------------------------------------------- 26 2.2.1.1 Ion source instrument --------------------------------------------------- 27 Surface Characterization --------------------------------------------------------- 27 2.3.1 X-ray photoelectron spectroscopy ------------------------------------------ 27 2.3.1.1 XPS spectra main features ---------------------------------------------- 30 2.3.1.2 Collected intensity and overlayer attenuation -------------------- 32 2.3.1.3 XPS instrument ------------------------------------------------------------ 35 2.3.1.4 Spectra simulation ------------------------------------------------------- 37 2.3.2 Ambient pressure X-ray photoelectron spectroscopy ----------------- 37 2.3.2.1 Synchrotron radiation --------------------------------------------------- 39 2.3.2.2 APXPS instrument -------------------------------------------------------- 41 2.3.3 Ultraviolet photoelectron spectroscopy ----------------------------------- 42 2.3.3.1 UPS instrument ----------------------------------------------------------- 44 2.3.4 Scanning Electron microscopy ----------------------------------------------- 44 2.3.4.1 SEM instrument ----------------------------------------------------------- 45 Electrochemical characterization ----------------------------------------------- 46 2.4.1 Full cell electrochemical characterization --------------------------------- 46 2.4.2 Symmetric cell electrochemical characterization ----------------------- 48 3. Lithium surface interaction with pure atmospheric gases ------------------ 53 Introduction ------------------------------------------------------------------------- 53 3.1.1 Literature review ---------------------------------------------------------------- 53 3.1.2 Work function to monitor lithium surface stability --------------------- 56 Spectra measuring conditions and data analysis guidelines ------------- 58 3.2.1 XPS and UPS data analysis guidelines -------------------------------------- 58 Li foil surface cleaning ------------------------------------------------------------- 60 O2, CO2 and N2 gases effects on clean lithium surfaces ------------------- 62 ‖ v 3.4.1 Oxygen interaction -------------------------------------------------------------- 62 3.4.2 Carbon dioxide interaction ---------------------------------------------------- 68 3.4.3 Nitrogen interaction ------------------------------------------------------------ 75 Conclusions --------------------------------------------------------------------------- 80 4. Study of Li carbonate evolution on Li metal surface ------------------------- 83 Introduction -------------------------------------------------------------------------- 83 Spectra measuring conditions and data analysis guidelines ------------- 84 4.2.1 Data analysis guidelines -------------------------------------------------------- 85 Li foil surface cleaning ------------------------------------------------------------- 87 4.3.1 Characterization of Li foil surface cleaned in Ar atmosphere --------- 87 4.3.2 Characterization of Li foil surface cleaned in UHV ----------------------- 90 Li2CO3 evolution on Li metal surface ------------------------------------------- 95 4.4.1 Evolution of carbon-based compounds ------------------------------------ 95 4.4.2 Li2CO3 growth kinetics -------------------------------------------------------- 102 4.4.3 Depth profiling of lithium-based compounds -------------------------- 104 4.4.4 Insights into the reaction mechanism ------------------------------------ 107 4.4.5 O2 gas effect on Li2CO3 growth --------------------------------------------- 108 Conclusions ------------------------------------------------------------------------- 114 5. Li thin film growth -------------------------------------------------------------------- 117 Introduction ------------------------------------------------------------------------ 117 5.1.1 Structure Development of a thin film ------------------------------------ 118 Experimental procedure -------------------------------------------------------- 119 Study of lithium thin film deposition ----------------------------------------- 121 5.3.1 Li source deposition rate calculation at 8 A ----------------------------- 121 xii ‖ impurities, first step has been to grow a lithium thin film, which has been characterized using a scanning electron microscope. Then, the electrochemical performance and internal resistance of a standard polymer electrolyte system with Li symmetric electrodes has been analyzed. In this study, it has been concluded that avoiding Li foil native surface impurities strongly modifies interfacial properties that determine the electrochemical performance of a system. This emphasizes the need of gaining knowledge about the initial state of metallic lithium surface used in batteries. ‖ xiii RESUMEN Aumentar cuota del mercado de vehículos eléctricos es esencial si se pretenden mitigar las consecuencias del efecto invernadero causadas, entre otros factores, por las emisiones de gases de vehículos de motor térmico. Sin embargo, la tecnología que suministra energía a los vehículos eléctricos, la tecnología de Liion, no es suficientemente competitiva debido a sus limitaciones en carga rápida, autonomía, seguridad y durabilidad. Para que se produzca una completa implementación del transporte eléctrico en el mercado, el desarrollo de la tecnología de Li-ion es vital. Además, su avance asegurará la evolución de los aparatos electrónico portátiles, también alimentados por baterías de Li-ion. De entra las diferentes alternativas existentes para mejorar la densidad energética de las baterías de Li-ion, una de las estrategias más prometedores es el cambio del ánodo actual, grafito, por litio metálico. Esto se debe a la alta capacidad del Li (unas diez veces superior al grafito) y a que posee el menor potencial de reducción conocido (-3.040 V vs electrodo de hidrógeno estándar). Aun así, la gran reactividad de la superficie del litio imposibilita tener una superficie estable entre el ánodo y el electrolito, perdiendo continuamente material activo. Además, la deposición de litio entre los procesos de carga y descarga en el ánodo no es homogénea, y se forman y crecen dendritas. Éstas pueden llegar a alcanzar el cátodo, causando varios problemas de seguridad. En este trabajo de tesis, la superficie del litio ha sido estudiada con el objetivo de adquirir mayor conocimiento sobre su estabilidad. Para ello, primeramente, se ha analizado cómo los gases atmosféricos secos más comunes (O2, CO2 y N2) modifican la composición química y las propiedades electrónicas de la superficie del litio. Este estudio se ha realizado por medio de las técnicas espectroscópicas de fotoemisión de rayos X y de rayos ultravioleta. Se ha concluido que el gas más reactivo es el O2, y que los tres gases reducen la función de trabajo del litio metálico. En el siguiente estudio, la evolución del carbonato de litio en la superficie del litio se ha analizado in situ por medio de la técnica espectroscópicas de fotoemisión de rayos X de presión ambiente. El carbonato de litio incrementa la uniformidad de la deposición del litio metálico cuando está presente en la interfase entre el electrodo y el ánodo. Por ello, es de gran interés evitar xiv ‖ condiciones en las que el crecimiento de este compuesto está favorecido. En este estudio, se ha obtenido información que contribuye al esclarecimiento del mecanismo de reacción, además de proporcionar estudios cinéticos del crecimiento del Li2CO3. Finalmente, se ha analizado el efecto de las impurezas nativas de la superficie de una lámina de litio comercial en un sistema electroquímico. Para ello, con el fin de evitar estas impurezas, el primer paso ha sido crecer una capa fina de litio, la cual ha sido caracterizada por medio de un microscopio electrónico de barrido. A continuación, se ha analizado el rendimiento electroquímico y resistencia interna de un sistema formado por electrodos simétricos de litio y un electrolito polimérico estándar. En este estudio, se ha observado que evitar las impurezas nativas de las láminas de litio comerciales modifica notoriamente las propiedades interfaciales, las cuales determinan la ejecución electroquímica de un sistema. Esto enfatiza la necesidad de adquirir mayor conocimiento sobre el estado inicial de la superficie de litio que se utiliza en las baterías. ‖ xv LABURPENA Energia sortzeko egun erregai fosilekiko dagoen menpekotasunak ondorio zuzenak ditu ingurugiroan, bestak beste CO2 isuriek areagotzen duten berotegi efektua dela eta. Honakoari aurre egiteko, energia berriztagarrien erabilerak eta ibilgailu elektrikoetarako jauziak berebiziko garrantzia daukate. Bi eremu hauetan, energiaren metaketarako gailu eraginkorrak beharrezkoak dira. Iturri berriztagarriek sortzen duten energia baldintza klimatologikoen araberakoa da, ez dute energia denboran iraunkorki sortzen. Hori dela eta, eskaintza eta eskariak bat egin dezaten, sortzen duten energia metatuko duen gailuen menpe daude. Bestetik, ibilgailu elektrikoak lehiakorrak izan daitezen, eskaintzen duten autonomia motore termikoko ibilgailuekiko alderagarria izan behar da. Beraz, hauek ere, energia metaketa gailu eraginkorren beharrean daude. Energia metaketarako gailu desberdinen artean, bateriak dira arruntenetakoak. Bateria bat hainbat zelda elektrokimikoz osatua dago, eta beraietako bakoitzean energia elektrikoa energia kimiko gisa metatzen da erredox erreakzioen bitartez. Zelda elektrokimikoek honako osagai nagusiak dituzte: katodoa edo elektrodo positiboa, anodoa edo elektrodo negatiboa, elektrolitoa eta elektrodo bakoitzeko korronte kolektoreak. Katodo eta anodoen arteko erredukzio potentzial desberdintasuna erredox erreakzioen indar eragilea da. Elektrolitoaren bitartez, elektrodoek ioiak elkar trukatzen dituzten, eta prozesu honen ondorioz elektroiak kanpo zirkuitu baten bidez elektrodo batetik bestera doaz, elektrizitatea sortuz. Anodoari erreparatuz gero, litio metalikoa teorikoko oso aukera aproposa da. Izan ere, kapazitate espezifiko teoriko altua dauka (3860 mAh/g), dentsitate baxua (0.53 g/cm) eta ezagutzen den potentzial elektrokimiko negatibo txikiena (-3.040 V hidrogeno estandarra erreferentziatzat hartuta), azken hau bateriak energia handiagoa ematearen erantzulea delarik. Hala ere, litioaren gainazal ezegonkorrak bere merkaturatzea zaildu du. Elektrolitoarekin etengabe erreakzionatzen du, material aktibo asko galduz, eta ezinezkoa du elektrodo/elektrolito gainazal arte egonkor bat lortu. Honetaz gain, karga eta deskarga artean, litioa ez da era homogeneo batean jalkitzen anodoaren gainazalean, eta dendrita antzeko mikroestrukturak sortzen dira. Hauek hazi xvi ‖ egiten dira eta, katodora helduz gero, zirkuitu laburrak eragin ditzakete, honek dakartzan arriskuekin. Litioaren ezegonkortasunaren arazoari aurre egiteko, 90. hamarkadan anodo bezala Li-ioiak itzulgarriki tartekatu litezkeen matrize bat erabiltzea proposatu zen, non matrizearen eta elektrolitoaren arteko gainazal artea egonkorra izango zen. Material honen aurkikuntzak 1991. urtean Sonyk lehendabizikoz Li-ioi teknologia merkaturatzea ahalbidetu zuen. Bateria haietako anodoa petrolio jatorriko kokea izan zen, katodoa LiCoO2 oxido laminarra eta elektrolitoa disolbatzaile organiko karbonikoetan disolbaturiko Li gatza. Egun, Li-ioi baterietako anodoa petrolio jatorriko kokea izatetik grafitoa izatera pasa da. Teknologia hau sortu zenetik merkatuko lehiakorrena da, energia dentsitate altua eskaintzen duelako era seguru eta eraginkor batean. Hori dela eta, merkatuko ibilgailu elektrikoek Li-ioi teknologian oinarritutako bateriak dituzte. Gailu elektroniko eramangarriek ere, hazkunde etengabean dagoen merkatuak, teknologia mota berdina erabiltzen du baterietan. Hala ere, grafitoaren kapazitate (372 mAh/g), litiorenarekin alderatuz hamar bat aldiz txikiagoa. Beraz, litio metalikoarekiko interesak bizirik jarraitzen du, eta berau egonkortzeko bide desberdinak proposatu dira azken urteetan; hala nola, gainazalaren moldatzea aurre tratamenduen bidez edo elektrolito solidoen erabilera zirkuitu laburrak ekiditeko. Hala ere, oraindik ez da aurkitu litio metalikoa egonkortuko duen epe luzerako konponbidea. Honako tesian litioaren gainazalaren egonkortasuna aztertu da, eta litio komertzialaren berezko ezpurutasunek sistema elektrokimiko batean duten eragina neurtu da. Honetarako, lehendabizi atmosferan ugariak diren O2, CO2 eta N2 gasek litioan zein ondorio dituzten aztertu da 3. kapituluan, eta Li2CO3 konposatuaren bilakaera jarraitu da Li gainazalean 4. kapituluan zehar. Jarraian, ezpurutasunik gabeko litio/elektrolito gainazal artea sortze bidean, litio geruza fina hazi eta karakterizatu da 5. kapituluan. Azkenik, 6. kapituluan, elektrolito polimerikoa duen sistema elektrokimiko batean litio komertzialaren ezpurutasunek elektrokimikan duten eragina ikertu da. Litioaren gainazalean atmosferan aurkitzen diren O2, CO2 eta N2 gasek sortzen dituzten aldaketak aztertzeko fotoigorpen espektroskopia teknikak erabili dira: XPS (X-Ray Photoelectron Spectroscopy) eta UPS (Ultraviolet Photoelectron Spectroscopy). CIC Energiguneko Gainazalen Azterketa Plataforman aurkitzen den teknika anitzeko ekipoan burutu dira bi ‖ xvii espektroskopia hauek. Lehenengo teknikaren bidez gainazalean sortzen diren konposatu kimikoak zehaztu dira. Bigarren teknikaren bidez gainazalen lanfuntzioa (wf, work function) determinatu da. Parametro honek huts mailaren araberako Fermi mailaren posizioa adierazten du, eta elektroi bat gainazaletik ateratzeko beharrezko energia zenbat den adierazten du. Fotoigorpen espektroskopia teknika hauekin litioaren gainazala ex-situ aztertu da; hau da, litioaren gainazala moldatu ostean neurtu da huts altuko egoeran (UHV, Ultra High Vacuum). Gasen eragina ikertu aurretik, lehendabizi argoi atmosferan gordetako litio xafla komertzialaren gainazala aztertu da, baterietarako anodo bezala erabiltzen dena. Xafla honen gainazaleko litio guztia oxidatua dagoela konprobatu da, Li2O eta Li2CO3 konposatuetan bereziki. Hori dela eta, atmosferako gasek beragan duten eragina aztertzeko, litio gainazala Ar ioiekin bonbardatu da UHV egoeran. Metodo hau eraginkorra izan da gainazaleko ezpurutasunak kentzeko: garbituriko gainazalak %(93.6 ± 1.9) Li metalikoz osatuak daude, gainerakoa Li2O delarik. O2, CO2 eta N2 gasen artean, oxigeno gasa da litioarekin bortitzen erreakzionatu duena. 9 L O2 gas (non 1 L 10-6 Torr presiopean segundo batez eginiko dosifikazioaren baliokidea den) nahikoa dira gainazaleko 8.6 nm-tako litio guztia oxidatua izateko. O2 gasaren presioa 10-4 mbar azpitik denean, erreakzio honetako produktu bakarra Li2O izan da. Hortik gorako presioetan, Li2O2 ere neurtua izan da gainazalean. CO2 gasaren interakzioari dagokionez, Li2O, Li2CO3 eta bestelako karboi oinarridun produktuak identifikatu dira. Erreakzio hau askoz motelagoa da, 8·108 L CO2 gas ere ez dira nahiko gainazaleko 8.6 nm-tako litio metaliko guztia oxidatzeko. Nitrogenoari dagokionez, litioak ez du gas honekin erreakzionatzen 10000 L-etik behera. Eta 10000 L-tan, soilik gainazalen %1.2 dago osatua nitrogeno oinarria duten konposatuekin. Li3N lortzeko modu bakarra litio gainazala nitrogeno ioiekin bonbardatzea izan da. Modu honetan lorturiko gainazala honako konposatuez osatua dago: %68.4 Li0, %19.8 Li3N, %8.1 Li2O eta %3.7 ezpurutasun. Lan funtzioa dagokionez, hiru gasek bere balioaren txikiagotzea dakarte. Li0ren batez besteko lan-funtzioa 3.01 ± 0.08 eV da. 1000 L O2-ren ondorioz, lan funtzioa 2.12 eV-ra txikitzen da, eta 1000L CO2-ren eraginez 2.30 eV-ra murrizten da. Li3N konposatuak era lan funtzioaren txikitzea dakar, 2.49 eV- xviii ‖ ra jaitsiz. Bereziki, gainazala Li2O and Li0 konposatuez osatua badago, lanfuntzioak Li2O kontzentrazioaren araberako erorketa esponentziala jarraitzen duela ondorioztatu da. Beraz, hiru gas hauek moldatuko litio gainazalek litio metilkoak baina erraztasun handiagoaz galduko dute elektroi bat, anodo bezala erabiltzeko ezaugarri kaltegarritzat jo dena. Jarraian, litio karbonatoren garapenaren azterketa egin da litio metalikoaren gainazalean. Izan ere, litio karbonatoa kaltegarritzat hartua dago litio anodo gainazalaren egonkortasunerako. Konposatu honek gainazaleko bestelako konposatu batzuekin alderatuz Li-ioi konduktibitate txikiagoa dauka, eta honek litioaren deposizio ez homogeneoa bultzatzen du. Ikerketa honetarako litio gainazalaren bilakaera neurtu da CO2 atmosferapean APXPS (Ambien Pressure XRay Photoelectron Spectroscopy) teknikaren bidez sinkrotroi bidezko erradiazioa erabiliz. Neurketa hauek ALS (Advanced Light Source) azeleragailuan egin dira, LBNL (Lawrence Berkeley National Laboratory) laborategian. Aurretik erabilitako XPS-rekin alderatuz gero, APXPS teknikaren abantaila nagusia neurketak in-situ egin daitezkeela da; hau da, erreakzioa ematen den bitartean gainazalaren eboluzioa jarraitua izan daiteke. Gainera, sinkrotroiari esker, erradiazioa aldatu daiteke, sakontasun profileko neurketa ez-suntsitzaileak egitea ahalbidetuz. Aurreko kasuan bezala, hemen ere argoi atmosferan gordetako eta garbitutako litio xaflaren hasierako egoera ikertu da, non berriro konprobatu den litioaren gainazal osoa oxidatua dagoela. Sakontasun profileko neurketek bidez Li2CO3 Li2Oren gainean kokatzen dela ikusi da. Kasu honetan, litioaren gainazala garbitzeko bestelako teknika erabili da: gainazala fisikoki urratua izan da UHV egoeran, marraza baten bidez. Litio karbonatoaren eboluzioa aztertzerakoan, berarekin batera beste konposatu baten bilakaera ere neurtua izan da: litio oxalatoa, Li2C2O4. Konposatu hau aurretiaz bitartekari gisa proposatua izan zen Li2CO3 sortzeko, baina ez zegoen bere hazkuntzaren ebidentzia esperimentalik. Beraz, oxalatoaren neurketak karbonatoa sortzeko mekanismoa argitze bidean informazio oso baliagarria eskaintzen du. Karbonatoaren hazkunde motari dagokionez, bi tarte identifikatu dira: erreakzioak kontrolaturikoa eta difusioak kontrolaturiko. Lehenengoak hazkunde lineala dauka, eta bigarrenak parabolikoa. Litioa CO2 gasaren pean egotearen ondorioz, Li2O konposatua ere sortzen da gainazalean. Atal honetan lorturiko informazioarekin erreakzio mekanismo bat proposatu da. CO2 atmosferari O2 gasa gehitzeak dituen ondorioak ere aztertu dira, non ikusi den ‖ xix oxigenoak litio karbonatoaren bilakaera bultzatzen duen, oxalatoa sortzea ekidinez. Azterturiko gainazal guztiek estruktura berdina daukate: Li2O Li metalikoaren gainean kokatzen da, eta Li2CO3 oxidoaren gainean oxalatoarekin batera, baldin eta oxalatoa sortzen bada. Behin litio xafla komertziala aztertuta, litio geruza fina sortu eta karakterizatzeari ekin zaion. Honetarako, baporizazio termiko teknika erabili da. Li iturri komertzial batetik abiatuz, sortutako gainazalak elektroien mikroskopia bidez karakterizatu dira CIC Energiguneko Gainazalen Azterketa Plataforman aurkitzen den SEM (scanning electron microscope) erabilita. Honakoarekin iturri komertzialaren deposizioa abiadura neurtu eta geruzaren hazkuntzaren morfologia behatu dira. Iturritik 8 A-ko korrontea pasatzean, deposizio abiadura 120 – 400 nm/h-koa da eta geruzak mendixkak eta zuloak ditu. Prozesu honetan zehar substratuaren tenperatura 42.3 °C-koa da. Korrontea 10 A denean, berriz, deposizio abiadura 730 – 1400 nm/h-koa da, eta substratuaren tenperatura 51.5 °C-ra igotzen da, zeinak gainazalaren morfologia homogeneizatzen duen. Litioaren hazkundea hainbat substratutan aztertu da: Si monokristalinoa, Ti geruza, altzairu herdoilgaitza, PET (Polyethylene terephthalate) polimeroa eta SrTiO3 monokristalinoa. Hauetatik, Si monokristalinoan ez da lortu litioa geruza moduan haztea. Horren ordez, litioak mikroestruktura ez homogenoa jarraitzen du, dendrita erakoa. Ezin izan da hazkunde mota hau silizioaren propietate jakin batekin erlazionatu. Azkenik, litio xafla komertzialaren ezpurutasunek sistema elektrokimiko batean duten eragina aztertu da. Azterketa honetarako LiTFSI (lithium bis(trifluoromethanesulfonyl)imide) gatza duen PEO (Poly(ethylene oxide)) elektrolito solido polimeriko estandarra sintetizatu da. Polimero honekin Li/PEO:LiTFSI/Li sistema elektrokimikoaren portaera aztertu da bi kasutan, CIC Energiguneko elektrokimika karakterizazio baliabideen bidez. Lehenengo kasuan, Li xafla komertziala erabili da. Aurretik ikusi bezala, xafla honen gainazalak Li2O eta Li2CO3 konposatuak ditu, beraz elektrodo/elektrolito gainazal artean ezpurutasun hauek egongo dira. Bigarren kasuan, litioa zuzenean baporatu da polimeroaren gainean UHV egoeran, gainazal arteko ezpurutasunak minimizatuz. 70 °C-tan, non elektrolito polimerikoak konduktibitate aproposa daukan, litio xaflen ezpurutasunak ekiditeak gainazal artean barneko erresistentzia %26 murriztea dakar. 45 °C-tan, oso tenperatua baxua elektrolitoaren funtzionamendu egokirako, murrizketa hau are eta nabariagoa da, %92-koa. Polarizazio galvanostatikoan ere eragina dauka ezpurutasuna ekiditeak, gain-boltaiaren xx ‖ murriztea baitakar. Emaitza hauekin Li gainazalaren egoerak zelda elektrokimikoaren jardueran erantzukizun zuzena daukala konprobatu da, material honen erreaktibotasun altua ulertze bideko esperimentuen garrantzia azpimarratuz. ‖ xxi 6 ‖ 1. Introduction spinel structure phase transition[23]. In the last decades, the interest in mixed transition metal oxides combining Ni, Mn and Co (NMC) has been growing, due to the synergetic benefits of merging them. These materials can offer a capacity of 200 mAh/g when charging between 2.5 V and 4.5 V[24]. Apart from the above-mentioned cathode materials that rely on the intercalation of Li in layered oxide channels, three-dimensional structures also represent a competitive alternative: e.g. the LiMn2O4 spinel and the LiFePO4 (LFP) olivine structures. The most recent advances are exploring both high voltage lithium-ion cathode materials, as the spinel LiMn1.5Ni0.5O4 which can operate at 4.7-4.8 V, and high capacity cathodes, such as the so-called Li-rich layered oxides; denoted as xLi2MnO3(1-x)LiTO2 (T=Mn, Ni, Co), they can reach capacities higher than 250 mAh/g[22]. Most common standard electrolyte in Li-ion batteries are composed by LiPF6 salt in a mixture of organic carbonate solvents. Generally, the solvent includes ethylene carbonate (EC) and dialkyl carbonates[25]. The advantages of organic liquid electrolytes are the relatively high potential window at which they can operate without degradation (stable until 4.4 V) and the high ionic conductivity. However, these electrolytes are flammable, corrosive and thermally unstable, which could cause explosions and fire accidents when not used properly. Furthermore, LiPF6 salts is highly toxic[26]. Despite water-based electrolytes[27] could be a suitable option to remove organic solvents, main alternative electrolytes to avoid the safety issues of organic liquids are the solid electrolytes and ionic liquids. Solid electrolytes can be divided in two main families: polymer electrolytes and ceramic electrolytes[28]. In order to be competitive, both of them should possess high ionic conductivity (above 10-4 S/cm) at room temperature, have negligible electronic conductivity with high ionic transference number and remain stable in a wide electrochemical window[29]. Polymer based electrolytes are relatively easy to process at room temperature and have a good adhesion, but their conductivity at room temperature is below the desired one[30]. Among the different alternatives, polyethylene oxide-based are the most studied ones. Ceramic electrolytes have a high mechanical rigidity, they are stable at high temperature improving safety and kinetics, and possess a very high Li+ transport number, close to one. However, cracking and delaminating due to high temperature processing constitutes a mayor problem, and still suffer from a lower ionic conductivity than 1.2 Li-ion technology ‖ 7 liquid electrolytes[31]. Some examples of actively researching ceramic electrolytes are NASICON type (Na1+xZr2SixP3-xO12, 0<x<3) Li-ion conductors[31] and garnet type electrolytes, derivatives from the Li3Ln3M2O12 (M = T,W; Ln = Y, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Tm, Yb, Lu) garnet discovered in 1968[32], among others. Currently, hybrid polymer-ceramic electrolytes are under intense study with the aim to find a solid electrolyte that will fulfil all the requirements to be integrated competitively in a Li-ion battery[33]. Ionic liquid electrolytes are also an attractive alternative to organic liquid electrolytes due to their negligible vapor pressure and almost negligible flammability, which enhanced safety of the battery[34]. Furthermore, they also show a wide electrochemical window stability (up to 6 V for certain combinations), high ion density and wide liquidus phase range. These room-temperature molten salts have asymmetrical, large and bulky anions and cations. Typical ionic liquid is comprised of a quaternary ammonium cation, such as imidazolium, pyridinium or pyrrolidinium, combined with an organic or inorganic counter anions such as BF4- , PF6-, [(FSO2)(CF3 SO2)N]- or [(CF3SO2)(CF3CO)N]-, for instance. However, issues as their lower ionic conductivity compared with organic liquid electrolytes, and some incompatibilities with common active materials still represent major challenges for their implementation in Li-ion batteries[35]. The finding of an appropriate anode was the main promoter of the development of Li-ion technology by Sony Energetic of Japan in 1991. Previous attempts of rechargeable batteries used Li metal as anode. Considering its lightness, high theoretical capacity (3860 mAh/g) and lowest reduction potential known (-3.040 V vs standard hydrogen electrode), it was a very attractive anode material. Indeed, in the 1960s, the concept of lithium secondary batteries was presented[36]. In the next decade, first commercial Li metal rechargeable batteries appeared[37]. However, the highly reactive nature of lithium made it impossible to have a stable interface between the anode and the electrolyte, thus metallic lithium anode rechargeable batteries were quickly discarded. More details about the problematic characteristic of cycling a metallic lithium anode are explained later in this chapter (section 1.3.2). In the 90s, the proposed solution to address the instability of lithium was the use of a material where Li ions could intercalate reversible and the interface between the anode and electrolyte could be stable. The finding of a material that met these requirements gave birth to lithium-ion rechargeable batteries. The first chosen material was petroleum coke, a soft 8 ‖ 1. Introduction carbon material with a certain amount of structural disorder that enabled the commercialization of Li-ion batteries[38]. Nowadays, graphitic carbon materials that consist of graphene layers held together by weak van-der-Waals forces[39] are still used as anode due to their outstanding stability[40]. This material can intercalate Li+ at 0.1 V vs Li/Li+, and it has a theoretical capacity of 372 mAh/g when LiC6 is formed. 1.2.3 Solid Electrolyte Interphase (SEI): a key parameter Having a stable solid electrolyte interphase (SEI) over the graphite anode, a layer that was able to stabilize de interface between the anode and the organic liquid electrolyte, was determining for the development of Li-ion battery technology. The SEI forms because the chemical potential of the anode is outside the electrochemical stability window (ESW) of the electrolyte (Figure 1.3a). The organic electrolyte will reduce until the anode electrolyte reaction is blocked by the SEI layer[19], which prevents the anode from further reduction the electrolyte, providing stability to the electrochemical system (Figure 1.3b). Figure 1.3. Relative energies of a liquid electrolyte and the electrodes in an electrochemical cell a) when anode potential lies outside the electrochemical stability window of the electrolyte and b) once SEI layer passivates the anode surface, stabilizing the interface between anode and the electrolyte. (: chemical potential of the anode, : chemical potential of the cathode, LUMO: lowest unoccupied molecular orbital, HOMO: highest occupied molecular orbital, ESW: electrochemical stability window). 1.2 Li-ion technology ‖ 9 This irreversible layer must be both ionic conductive and electronic insulating to avoid the continuous reduction of electrolyte[25], and it also needs to be adhered to the electrode and be insoluble in electrolyte, specially at high temperatures[41]. Furthermore, SEI must be both mechanically stable and flexible enough to expand and contract during cycling without breaking[42]. The composition of the SEI will vary depending on electrolyte and active material compositions. Using classical organic electrolytes and LiPF6 salt, various organic and inorganic components have been identified in the SEI of graphite anodes: Li2CO3, LiOH, LiF, Li2O, ROCO2Li and RCOLi, among others[43]. Same researchers that named the electrode-electrolyte interphase as SEI in 1979 proposed a heteropolymicrophase mosaic type morphology to represent the SEI layer in both graphite and lithium anodes[44] (Figure 1.4). Figure 1.4. Schematic representation of a solid electrolyte interphase (SEI) formed upon graphite or lithium anode and composite solid organic electrolyte. [Reproduction with permission from[44], Copyright (2019) Journal of Electrochemical Society]. It has been proved that this mosaic model, under certain conditions, shows a bilayer structure: an inner layer dominated by inorganic compounds and an outer layer dominated by organic compounds[45]. The inner layer is assumed to be dense. The organic layer, in contrast, is assumed to be porous. However, this bilayer structure interpretation could be more complex in general, so the mosaic of micro phase model is believed to be a more appropriate model[42]. Indeed, modern 10 ‖ 1. Introduction interpretations are still based on this model[46]. The estimated thickness of a standard SEI layer is assumed to be from around 20 Å to several tens of nanometers[45,47]. The SEI is directly influencing the battery performance, irreversible charge, rate capability, cyclability, and safety, as well as preventing graphite exfoliation when using this anode[43]. It has been proved that some additives enhance the battery properties, since they induce the formation of a more stable SEI[41]. A profound understanding of SEI formation, composition and evolution during cycling is essential if want to improve the performance of Li-ion battery. However, despite all the efforts to reach such level of understanding, the elucidation of the formation and evolution of SEI remains elusive[25,41]. 1.3 Next generation Li metal batteries (LMB): role of Li metal Li-ion technology, despite being the best rechargeable energy storage option so far, has an state of the art gravimetric energy density of around 260 Wh/kg[30], one order of magnitude lower than that of petrol. Moving to Li metal anode, Li metal batteries (LMB), is the only possible way to reach very high energy density systems based on Li chemistry. Li metal, as earlier mentioned, has a capacity of 3860 mAh/g, ten times larger than that of actual graphite anode, and it also operates at the lowest reduction potential known. Both parameters will increase the overall energy density of the battery according to equation (1.1). By replacing current anodes of Li-ion batteries with Li metal, this technology will be able to deliver ≈ 440 Wh/kg (Li-LMO batteries in Figure 1.5, where LMO refers to LiMO2, M = Co, Mn, Ni). Besides, emerging technologies postulated as next generation energy storage systems such as Li-air and Li-Sulfur batteries also rely on the use of Li metal as anode. With these technologies, the energy density of batteries could increase up to 650 Wh/kg for Li-S and 950 Wh/Kg for Li-air[15]. The comparison of the gravimetric and volumetric energy density delivered by these technologies is illustrated in Figure 1.5. 1.3 Next generation Li metal batteries (LMB): role of Li metal ‖ 11 Figure 1.5. Comparison of practical specific gravimetric and volumetric energy density obtained by petrol, state of the art Li-ion batteries and next generation Li metal-based (LMB) batteries: Li metal - LMO batteries, Lithium-sulfur batteries and Lithium-air batteries. (LMO: LiMO2, M = Co, Mn, Ni) [Reproduction with permission from[15], Copyright (2019) Nature Nanotechnology]. 1.3.1 Li-sulfur and Li-air batteries Both Li-sulfur and Lithium-air batteries cathodes are not based on intercalation reactions as in Li-ion batteries, but in conversion reactions. In a lithium-sulfur battery (Figure 1.6 for main components schematics of this battery), the reaction from sulfur (S) to lithium sulfide (Li2S) incorporates two electrons per sulfur atom, which results in a cathode capacity up to 1672 mAh/g[48]. Several intermediates are formed during this reaction, summarized in Figure 1.6. One of the biggest challenges of this technology is related to the formation of intermediate polysulfides that are soluble in the liquid electrolyte and can freely move from the cathode to the anode. This so-called shuttle effect results on the passivation of electrodes, loss of active material and self-discharge. Moreover, both Li2S and S are insulating materials, so conductive additives need to be added to the cathodes in order to ensure electron percolation. Another problem of this technology is the volume expansion of about 80% happening when sulfur converts to Li2S[49]. Nowadays, great efforts in research are carried out to overcome all these challenges of Li-Sulfur batteries[50,51]. Some niche technologies that use Li-S batteries can be found, as the Zephyr High Altitude Pseudo-Satellite (HAPS) Aircraft[52]. In this technology, the low cyclability of Li-S batteries is not a matter of 12 ‖ 1. Introduction concern, the interest is to have high capacity during operation. However, for common applications this technology is still not competitive if compared with Liion batteries[53]. Figure 1.6. Schematic representation of a Li-sulfur battery and redox reactions occurring in the cathode during the discharge. [Reproduction with permission from[49], Copyright (2019) Journal of Power Sources]. Li-air batteries are very attractive due to their theoretical high energy density (Figure 1.5) and also due to having a freely available cathode fuel: O2 gas, which although being a convenient gas, its filtering and handling needs to be solved. The typical product of the battery discharge in the cathode is Li2O2, where an oxygen reduction reaction takes place. During the charge, oxygen evolution reaction takes place in the cathode (Figure 1.7). The cathode of this battery consists of a porous material, typically carbon with binder material such as the standard Li-ion batteries binder Polyvinylidene fluoride, or higher stability binder alternatives such as polyethylene[54]. In these cathodes special architectures with an adequate porous structure are necessary to avoid mass transport limitations. Several problems arise in these batteries, mainly related to the parasitic reactions that decompose both carbon electrode and electrolyte: rechargeability becomes poor, charge voltages high, efficiency low, and the cell ends up dying within a few cycles[55]. Finding new cathode designs to overcome these issues is becoming a big challenge due to the unresolved active reaction interface of electrochemical oxidation of lithium peroxide[56,57]. Even with all these problematics, the potential of this batteries is so high that researchers are still putting their efforts to find practical solutions[58]. What is more, the study of how to deal with the air components besides O2 to avoid the purification of the air is being seriously considered in the development of the batteries[59]. 1.3 Next generation Li metal batteries (LMB): role of Li metal ‖ 13 Figure 1.7. Schematic representation of a Li-air battery and the redox reaction happening during discharge and charge in both anode and cathode. [Reproduction with permission from[57], Copyright (2019) Nature Energy]. 1.3.2 Li metal surface instability In addition to the abovementioned intrinsic problems of lithium-sulfur and lithium-air batteries, another concern needs to be added to these technologies: having a Li metal anode that will result in handling and stability problems. Getting over the instability hazards arising from the high reactivity of lithium surface that hindered its commercialization back in 1980s is still one of the major drawbacks to achieve a real development of LMB. In contrast to the current graphite anode, solid electrolyte interphase (section 1.2.3) formed on metallic lithium is not stable during the cycling of a cell, directly affecting the performance of the cell[60]. The low reduction potential of metallic lithium will reduce the electrolyte (practically any of them[61]) at the surface of the metal, forming an unstable SEI that will break during plating and stripping process due to volume changes of Li anode[15]. Fresh lithium will be then exposed to the organic electrolyte, forming a new SEI layer. The first SEI model, the mosaic one from Figure 1.4, was proposed both for carbonaceous and lithium metal anode. Analogously to the graphite anode case, for Li metal anodes two layers were also identified in this mosaic SEI: an inner compact layer close to the electrode mainly including inorganic species and an outer porous layer mainly composed by organic 14 ‖ 1. Introduction material[62]. Here also, modern studies based their SEI interpretation in the mosaic model, as show in the Figure 1.8[63]. Significant surface research results summarized in a Li metal SEI review[47] have found that, using several organic solvents and salts, major inorganic compounds are Li2O, Li2S/Li2S2, LiOH, LiF, LiI, Li3N and Li2CO3; whereas the organic ones are ROLi, RCOOLi, ROCOLi, RCOO2Li and ROCO2Li (R = alkyl groups). This complex heterogeneous nature of SEI is rendering very difficult a proper quantitative characterization of SEI chemical composition, structure and mechanical properties. Still, both experimental and theoretical studies keep trying to elucidate the nature of the SEI due to the direct impact of this interphase in the performance of the cell[64,65]. Figure 1.8. Schematic representation of Li plating and stripping effect on lithium metal surfaces based on the mosaic model for SEI interpretation, using organic carbonate liquid electrolyte (LE) with LiNO3 (LNO). [Reproduction with permission from[63], Copyright (2019) Chemistry of Materials]. The other big issue related to the metallic lithium anode is the non-uniform electrodeposition of lithium in the anode during electrochemical cycling. When depositing, it grows forming whisker type structures, named as dendrites (Figure 1.9a). Although the ramified metallic electrodeposition from dilute salt solutions in high electric field was already considered an old subject in 1990[66], the difficult interfacial chemistry of lithium surface makes the explanation of dendrite growth complex. The heterogeneous SEI entails inhomogeneous nucleation due to different ion conductivity of the several compounds, and the cracks in the nonstable SEI increment the non-uniform deposition. In order to explain the selfenhanced nature of the dendritic growth several theories have been proposed. 1.3 Next generation Li metal batteries (LMB): role of Li metal ‖ 15 One of them focused on the higher electric field at the tip of the bulges due to their curvature, which attracts more Li ions and thus forms further protrusions, evolving into dendrites[67]. If the dendrites grow perpendicular to the anode and pierce the separator, thermal runaway and explosion could occur due to the short circuit[68]. Another negative consequence of the dendrites is the loss of active lithium. When the dendrite detaches from the anode, it disconnects electrically. This lithium, surrounded by SEI, becomes inactive, and is usually called dead lithium (Figure 1.9b). Furthermore, the continuous accumulation of the dead Li creates tortuous diffusion pathways that affects the diffusion of Li ions[69]. a) b) Figure 1.9. a) Dendrites formation as a consequence of non-uniform electrodeposition of lithium. b) Inactive dead lithium as a consequence of dendrite detaching from the anode, which decreases the amount of active material of the anode. [Reproduction with permission from[70], Copyright (2019) Cell Press]. From both SEI cracks and dead lithium that cause the loss of active material, main contributing factor to the low Coulombic efficiency is believed to be the dead lithium[71]. This parameter is defined as the ratio of the amount of charge that exits the battery during the discharge and the amount of charge that enters the battery during charge. Normally, in conventional carbonate organic electrolyte, the Coulombic efficiency is lower than 90%[70]. But even when reaching 99% of Coulombic efficiency with advanced electrolytes[72], the inefficiency remains being a problem. The goal for applicability that will allow to have more than 1000 cycles needs a Coulombic efficiency of 99.98%[73]. All these interfacial issues, besides the low coulombic efficiency, have also a direct impact in other parameters that 22 ‖ 1. Introduction 1.4 Scope of the thesis In order to move to next generation lithium metal batteries, the unstable interface between lithium and electrolyte needs to be addressed. Most efforts pursuing this objective are focused on finding an appropriate artificial SEI. However, little attention has been paid so far to the influence lithium native surface exerts in the stability of the interface. The aim of this dissertation is to understand how atmospheric gases modify the surface of metallic lithium and analyze which is the real pristine surface of a battery grade commercial lithium foil to finally see to which extent the preexisting impurities are affecting the interface, which will ultimately drive the electrochemical performance of the cell. In order to do that, the interaction of lithium foil with main pure atmospheric gases (O2, CO2 and N2) is studied using spectroscopic techniques, and the electrochemical performance of an impurities free surface and a standard surface is compared in a symmetric solid electrolyte system. 2.1 Thin film growth ‖ 23 2. Experimental techniques CHAPTER 2 Experimental techniques In this chapter, fundaments of the experimental techniques used to develop the work presented in this thesis are introduced. Chapter is divided in four sections. Firstly, techniques employed to grow thin film are presented. In the second section, the method applied to modify surfaces is explained, and after that surface characterization techniques are detailed. Lastly, a description of the electrochemical characterization techniques can be found. 2.1 Thin film growth Two physical vapor deposition (PVD) processes were used to grow thin films. These techniques are based on moving atoms in gas phase from the target material in solid phase to the growing film, also solid phase. The vaporization of the target atoms in this thesis has been produced either by applying heat (thermal energy) or by cathodic pulverization (sputtering). 2.1.1 Thermal evaporation Vacuum thermal evaporation is the most basic physical vapor deposition process. The element to be evaporated is placed in a metallic crucible, which is heated by passing a current () through it, according to Joule effect. The amount of heat generated is then:  =    (2.1) 24 ‖ 2. Experimental techniques where  is the parallel resistance of the crucible and evaporant combination at the evaporation temperature. Under perfect vacuum conditions and considering a single-component evaporant material, the maximum molar flux of substance from the solid phase to its gaseous form is expressed by the Hertz-Knudsen equation[96]:   ,  =  √ 2  (2.2) where  is the molecular weight of the evaporating compound,  is the universal gas constant,  is the absolute temperature at the evaporant surface and  is the standard vapor pressure of the evaporant, which is a function of the absolute temperature. The relationship between the evaporation flux and maximum evaporation flux is correlated by the evaporating coefficient () according to:   =     ,   (2.3) Most metals have atomic vapors and evaporating coefficient is equal to one. When evaporating an alloy, which is a solid solution or a mixture of solid phases, the evaporated flux will be richer in the more volatile element for any composition, so the melting will continue to deplete in that element as evaporation proceeds. Compounds have a very different evaporation behavior compared to alloys. In contrast with alloys, they have a specific ratio of elements, that is, they have a specific stoichiometry, and during evaporation they can evaporate as molecules, partially dissociated or dissociated completely upon evaporation. This last behavior is very practical when evaporating alkaline metals, because their low sublimation point makes them inappropriate for use in high vacuum evaporators which are usually baked out at temperatures above 100 °C. With an intermetallic compound, the very low sublimation temperature can be significantly increased by high melting intermetallic alkali compound of high enthalpy of formation[97]. One of the main problems of thermal evaporation is the contamination, both crucible material and evaporants release contaminant vapors from their surfaces and from the bulk. Much of the volatile impurity content in the evaporant can be removed before film evaporation, which includes adsorbed gases and dissolved elements of higher vapor pressure than the evaporant. For their removal, crucible is heated at a temperature below evaporation temperature of evaporant, where 2.1 Thin film growth ‖ 25 the dissolved impurities will progressively deplete relative to the evaporant, purging the evaporant. 2.1.2 Magnetron sputtering In this process, represented in Figure 2.1, a solid target is sputtered by energetic ions of inert gases (e.g., argon) from a magnetically enhanced glow discharge. The sputtered material is deposited on the substrate, which is placed opposite the target. A crosswire magnetic field incorporated over the target traps secondary electrons near the target surface. Then, electrons path length is greatly increased before they finally escape to the substrate. When the substrate is electrically insulating, radiofrequency (RF) bias instead of direct current (DC) bias must be used. Figure 2.1. Schematic cross section representation of a magnetron sputtering process. 2.1.2.1 Sputtering instrument The sputtering instrument used in this thesis is a Pfeiffer Classic 500 SP, which consist of a process chamber with 5 magnetron heads: 3 DC power supplies and 2 RF power supplies. The system is part of surface analysis unit of CIC Energigune and it can reach a base pressure of 10-8 mbar. The instrument is equipped with a fast entry chamber that allows keeping good vacuum levels. In addition, the fast entry chamber has a modification that enables the attachment of an air tight transfer system for sensitive samples. 26 ‖ 2. Experimental techniques 2.2 Surface modification In this thesis work, surfaces were modified by ion bombardment for two purposes. On one side, this technique was used to remove impurities from the surface. On the other side, surface chemistry was modified bombarding it with reactive ions. 2.2.1 Ion bombardment The impingement of energetic ions or atoms upon a solid surface produces a variety of effects related to the high efficiency energy transfer of this process. The amount of kinetic energy transferred from the ion to the target atom is defined by:   = 4     (   +   )    =     (2.4) where  is the mass of the impinging particle,  is the mass of the target atom,  is the kinetic energy of the impinging ions and  defines the efficiency of the energy transfer process between the bombardment ions and target atoms. If the masses are within two times each other,  is > 0.9. The ways in which bombarding ions can move surface atoms can be grouped in surface and subsurface processes. Surface processes are usually in the range from few eV to tens of eV. One of the surface displacement process is very useful if the aim is to remove contaminants from the surface of our sample. During this process an inert gas is used to bombard the sample surface, and the adsorbed impurity receives enough vibrational energy to break its bond to the surface and desorbs (Figure 2.2a). Special care has to be taken with the energy of the ions, because if it is too high, the contaminant, instead of being removed, can be implanted on the subsurface (Figure 2.2b). When the ion bombardment energy exceeds a few tens of eV, then particle penetration into the bulk material begins, and one of the most important subsurface phenomena that appears at this point is the ion implantation (Figure 2.2c). When working with an inert gas, implantation is generally undesired, main purpose is the removal of surface contamination. However, it can be used to incorporate a desired dopant or even to form a compound film if the impinging ion is reactive. This last one is called reactive implantation. 2.3 Surface Characterization ‖ 27 In the ion sources, gas ions are produced, focused, accelerated and emitted as a narrow and intense beam towards the sample. In all types of ion sources, the ions are generated by an electric discharge that goes through the gas at low pressure. Ions are produced by electron collision inside ionization cavity forming an electron ion plasma. Figure 2.2. Typical surface and subsurface process generated by ion bombarding. a) adsorbate removal process, b) knock-on implantation of an impurity atom and c) ion implantation. 2.2.1.1 Ion source instrument Ions sources used in this thesis are IQE 11 and IQE12/38, both from SPECS GmbH. Both sources generate and extract ions, but second type also focus and deflect the ion beam using a double lens system and deflection plates. IQE 11 is used for the reaction implantation processes and IQE12/38 to clean surfaces by Ar ion sputtering. 2.3 Surface Characterization Several techniques were used in order to characterize the surfaces studied in this thesis. Chemical composition was determined by X-ray photoelectron spectroscopy. To study the electronic configuration, ultraviolet photoelectron spectroscopy was used. Lastly, information about the morphology and thickness of thin film samples was obtained by scanning electron microscopy. 2.3.1 X-ray photoelectron spectroscopy X-ray photoelectron spectroscopy (XPS) is widely used to determine the chemical composition of the surface. It consists of an X-ray source that irradiates the sample under study with photons that excite electrons from the core levels of the atoms of a solid sample into the vacuum. Thus, XPS probes the electronic structure of matter with elemental sensitivity and chemical state specificity. The photon 28 ‖ 2. Experimental techniques penetrates into the sample surface and is absorbed by a core electron with a binding energy below the energy difference between the photon energy and the vacuum level. Then, electron emerges from the solid with a given kinetic energy as determined by the photoelectric effect. The kinetic energy (KE) of emitted electrons in the sample is determined by:  = ℎ  −  −   ,  (2.5) where ℎ is the photon energy, , the work function of the sample and  the binding energy of the excited electrons. The kinetic energy of these electrons is measured, in our case, by a photoelectron hemispherical analyzer (HAS). Figure 2.3 shows the main parts of the HAS analyzer system. It consists of two metallic hemispherical plates concentrically arranged. A set of electrostatic lenses collects the emitted photoelectrons and focuses them onto the analyzer entrance slit. Electrons are retarded by a potential difference R inside the lens system until they match the electrostatic field of the hemispherical analyzer. This field is called the pass energy and it is applied between the inner and outer hemispheres of the analyzer so that trajectory of the incoming electrons is bent into a curve. A channeltron type electron multiplier (detector) is situated behind the exit slit of the analyzer and counts the emerging electrons. Therefore, by scanning R, spectrum of the photoelectron intensity as a function of kinetic energy can be recorded, the measured kinetic energy being the sum of R and pass energy. One way to increase the energy resolution of the analyzer is decreasing the pass energy; however, the collected intensity will decay. The hemispherical analyzer and transfer lenses can be operated in two modes, namely Fixed Analyzer Transmission (FAT) and Fixed Retard Ratio (FRR). In FAT mode, the pass energy is held constant, transfers lenses are in charge of retard the kinetic energy channel to the one accepted by the analyzer. In FRR, the constant value is the initial electron energy and analyzer pass energy ratio. First one is most used in XPS systems because the energy resolution is constant for the whole spectrum. The HAS is characterized by its own work function. A contact potential exists between the sample and the analyzer when both are electrically connected, if the sample is electrically conducting, the Fermi energies of sample and analyzer are aligned. Consequently, kinetic energy (´) of electrons collected in the analyzer is affected by the contact potential, yielding in:  ´ = ℎ  −  −   ,  +    ,  −   ,   = ℎ  −  −   ,  (2.6) 2.3 Surface Characterization ‖ 29 where , is the work function of the sample. Then, the measured kinetic energy is determined by the photon energy, the binding energy and the work function of the analyzer. A schematic of energies of core level photoelectron spectroscopy is shown in Figure 2.3. Figure 2.3. Energy level diagram of core level photoelectron spectroscopy from a solid. Energetic level terms are explained in the text. A schematic of a hemispherical analyzer is also drawn, showing the path of electrons from the sample to the detector. There is an example of a typical XPS spectrum in the right side of the figure. The measured KE spectrum is a superposition of primary electrons and secondary continuum electrons. The primary electrons result from elastic collisions and are featured as distinct spectra (peaks), these primary electrons are the ones that contain the information about the core levels of the sample elements including their oxidation state. Secondary electrons are primary electrons that undergo inelastic collisions resulting in a reduction of their kinetic energy. They have a more or less continuous energy spectrum. As said previously, the photoemission process relays on the interaction between one X-ray photon and one core level electron. Conventional X-ray sources are based on the bombardment of a solid target with high energy electrons. The 30 ‖ 2. Experimental techniques emission from this target consists of characteristic X-ray emission lines associated with the filling of core holes created by the incident high energy electron beam. Electron energy is generally chosen to create holes of K-shells. The ideal energy photon source should have low background and narrow characteristic line emission, the nearest to a monochromatic source. Another important feature of the target is its capacity to dissipate heat, which will facilitate the cooling down process needed because of the incident electron beam bombarding the target. Most used materials that have appropriate characteristic to fulfill the previous mentioned criteria are Mg and Al. Both of them have a dominant Kα1,2 X-ray emission line, at 1253.6 eV for Mg and 1486.6 eV for Al. These emissions also have other lines associated with doubly and multiply ionized atoms. Table 2.1 shows the energies of main emission lines when using Mg as target to produce X-rays. Table 2.1. X-ray emission line energies for a Mg source. Most pronounced characteristic line is Kα1,2. Relative height of secondary lines is less than 9% of main line. Kα 1,2 (eV) Kα 3 (eV) Kα 4 (eV) Kα 5 (eV) Kα 6 (eV) K β (eV) 1253.6 1262 1263.7 1271.2 1274.2 1302.3 One way to overcome the non-monochromatic nature of targets is adding a monochromator, which is a set of suitable crystals to create single or multiple Bragg reflections. The process leads in having just a part of the dominant Kα1,2; however, this energy resolution improvement suffers from a considerable intensity loss. 2.3.1.1 XPS spectra main features The emitted electrons are recorded according to their kinetic energy. To obtain chemical information, BE (obtained from equation (2.6)) is used to correlated the peaks with tabulated core levels of elements. Since the binding energy of a photoelectron is sensitive to the chemical surroundings of the atom: when changing the chemical environment, there will be a shift in the binding energy, which provides information to identify individual chemical state of an element. These peaks are named as nlj, where n is the principal quantum number, l the orbital momentum quantum number and j the total angular momentum quantum number. s levels (l=0) give rise to a singlet peak (Figure 2.4a for Li 1s), but p, d and f levels (l>0) to a doublet (Figure 2.4b for Cu 2p), which arises from spin-orbit coupling (splitting) effects in the final state. 2.3 Surface Characterization ‖ 31 Figure 2.4. Example of different spectra features collected when measuring XPS using Mg Kα nonmonocromatic source. Satellite from b) corresponds to Mg Kα3 and Mg Kα4 emission lines. When an atom has unpaired electrons in the valence band, emission of an electron from core level arises in multiplet splitting: this is the result of coupling between the unpaired electron in the core with the unpaired electron in the outer shell, creating a number of final states which will be reflected in the measured photoelectron spectrum (Figure 2.4c) Other important features that also appear on the spectrum are X-ray satellites, shake up lines, plasmon loss peaks and Auger electrons. X-ray satellites (Figure 2.4b) are a consequence of irradiation with a non-monochromatic X-ray source, where irradiation has not only the characteristic X-ray but also some minor components at higher photon energies, as shown in Table 2.1 for Mg. Thus, these minor components excite also core level electrons that appears at lower binding energies. Shake up peaks (Figure 2.4b) appear when the outgoing photoelectron interacts with a valence band electron and excites it to a higher energy level changing the kinetic energy of the emitted photoelectron. Shake up peaks have intensities of up to 5-10% of the main peak. The plasmon loss peak (Figure 2.4a) is a typical feature for some metals, where emitted electrons loss a specific amount of energy due to the interaction between the photoelectrons and delocalized electrons in the conduction band that are typically involved in collective oscillations, the so called plasmons. 38 ‖ 2. Experimental techniques which electrons and gas molecules escape. The pressure difference across this aperture depends on the size of the aperture, type of the gas, gas temperature and pumping efficiency, and it is typical of the order of 102 and 104 Torr. Small apertures spaced at large distance improve differential pumping but decrease the effective solid angle of transmitted electrons. In this kind of system, maximum operation pressure in sample environment is about 1 Torr. In order to be able to operate at higher pressures, electrostatic lenses are used to focus electrons through the apertures (Figure 2.9) allowing then the measurements of samples in environments with a pressure up to 10 Torr[105]. Figure 2.9. Schematic of APXPS differential pumping system. Sample is placed in a high-pressure chamber and electrons reach the analyzer under UHV conditions thanks to the differential pumping system between them. The diameter of the aperture (d) defines the optimal distance between surface and aperture (z)[105]. Similar to the attenuation produced by an overlayer mentioned in previous section (section 2.3.1.2), the attenuation of the photoelectron yield in gas environment has also an exponential decay according to:  =   ·    /   (2.13) where  is the photoelectron intensity from the material under study after the gas attenuation,  is the photoelectron intensity that we would observe without the attenuation, z is the distance the electrons travel in gas atmosphere and  is the mean free path of electrons in gas environment, which is defined as:   =   (2.14) 2.3 Surface Characterization ‖ 39  is the Boltzmann constant,  the temperature of the gas,  the pressure of the gas and  the electron scattering cross section. Therefore, a way to decrease the attenuation of ejected photoelectrons is placing the sample near the aperture of differential pumping system. However, there is a minimum distance at which the sample should be kept in order to ensure a homogeneous pressure on the sample surface. This distance is correlated with the aperture dimension of the differential pumping system[105]. If both aperture and distance from sample to aperture are the same, then pressure at sample surface is 95% of chamber pressure and, if the distance is double that from the aperture dimension, the pressure at the surface is 98%. Therefore, the focal distance z at which electrons suffer less attenuation and surface pressure is same as chamber pressure is similar to the aperture dimension (Figure 2.9). Then, the smaller the aperture the more is reduced the path electrons need to travel under gas atmosphere. Standard apertures of APXPS system are less than 1 mm for the front aperture and 2 mm for the rest of apertures between differential pumping stages. With this technique it is also possible to collect the photoelectron signal from the gas phase; this is because X-ray irradiates not only the sample but also the gas. Besides all the possibilities this technique offers to study solid gas interface, nowadays its design is being pushed to study also the solid liquid interface, using instruments that can work at pressures up to 110 Torr[106]. 2.3.2.1 Synchrotron radiation When X-ray spectroscopies use synchrotron radiation instead of laboratory-based X-ray tube as incident X-ray, more information about the surface can be obtained. Brilliance is a parameter that defines the photons generated per second divided by the light source footprint, divergence and bandwidth (BW). When comparing the brilliance obtained from each source, it is around 107photons/(s mm2 mrad2 0.1%BW) for a laboratory X-ray tube, whereas it is around 1022 photons/(s mm2 mrad2 0.1%BW) for a third-generation light source where the generated photon beam is highly collimated. Another important property of synchrotron radiation is its polarization and coherence, the emitted light is linearly polarized in the orbit plane and it can produce detectable wave-like effects. Furthermore, generated Xray covers a wide spectrum, from microwaves to hard X-rays. Most common synchrotron radiation sources are based on storage rings, where a beam of highly energetic electrons is stored and kept traveling on a circular path. 40 ‖ 2. Experimental techniques Relativistic accelerations on the electrons will result on the emission of an electromagnetic field, the so-called synchrotron radiation used as a light source for experiments (Figure 2.10). Figure 2.10. Schematic of a synchrotron radiation facility. Electrons are produced inside the electron source and initially accelerated by a high voltage or radiofrequency field, these electrons are then feeded into the Linac (linear accelerator). The electrons are packaged in bunches and accelerated enough for injection in the booster synchrotron. This is a pre-accelerator where electrons are accelerated to their final energy in the order of GeV before being finally injected into the storage ring. The booster only works when the storage ring has to be refilled. In the storage ring, electrons travel at a constant relativistic speed. In order to recirculate the charged particles along a circular path, a magnetic field perpendicular to the horizontal orbital plane is used. As the electrons travel around the ring, radiation is emitted whenever they are forced to deviate from a straight-line motion. Bending magnets were the first available sources to apply a magnetic field used to deviate electrons and, subsequently, generate synchrotron radiation. A way to increase the intensity of radiation generated by bending magnets is using wigglers, where a series of bending magnets are lineup enhancing the intensity simply by the number of 2.3 Surface Characterization ‖ 41 magnet poles. The spectrum generated by a wiggler is that of a bending magnet but with a higher brilliance, because the individual emissions of each magnet overlap and the intensity adds up. Most modern way to create synchrotron light is using undulators instead of wigglers, these are most powerful generators. They consist of a periodic arrangement of dipole magnets generating an alternating static magnetic field which deflect the electron beam sinusoidally, resulting in radiation with the wavelength of this periodic motion, differing from bending magnets and wiggles spectrum. Figure 2.11 compares the spectra brilliance of a bending magnet, wiggler and undulator of the Spring-8 synchrotron facility. Figure 2.11. Brilliance of the SPring-8 synchrotron bending magnet, wiggler and undulator. The solid curve for the undulator shows the output at a fixed gap between top and bottom poles, the dashed lines the variation in the harmonic peaks as the gap is varied from 25 to 8 mm. Brilliance of sun has also been indicated in the figure. [Modified from[107]]. 2.3.2.2 APXPS instrument APXPS experiments present on this work were carried out using a Scienta R4000 HiPP APXPS system, which is placed at Beamline 9.3.2 Lawrence Berkeley National Laboratory’s (LBNL) Advanced Light Source (ALS). This system is based on a Scienta R4000 with a two-dimensional detector consisting of two multichannel plates coupled to a phosphor screen and charge-coupled device (CCD) camera. It has four pumping stages, and the base pressure of the analyzer is low 10-9 Torr. The separation between the high-pressure chamber and first pumping stage is a removable Ti cone with a 0.425 mm aperture radius on the tip (Figure 2.12). The approximate focal distance of this instrument is 0.8 mm and it can record spectra above 2 Torr. 42 ‖ 2. Experimental techniques The 9.3.2 bending magnet beamline generates soft X-rays with an energy between 250 and 850 eV. A Si3N4 window isolates the UHV X-ray tube from the highpressure chamber. The UHV system has also a preparation chamber which includes an ion gun for ion sputtering processes. An air sensitive transfer tool was used to move samples under argon atmosphere from an argon glove box to the load lock of the UHV system. Figure 2.12. Picture of main chamber of APXPS system placed in beamline 9.3.2 at Advanced Light Source. The sample holder is a Thermionics STLC plate. One of the main advantages of measuring APXPS spectra using synchrotron radiation, apart from the high-resolution spectra, is the capability to tune the energy of the source. When changing the photon energy, kinetic energy of ejected electrons from the same core level is also changed so photoelectrons generated at different depths in the sample surface can be measured and compared. Therefore, a nondestructive depth profile can be measured, which is essential to understand the distribution of the compounds on the surfaces under analysis. Furthermore, it is also useful to measure different core levels at the same kinetic energy for quantification reasons, because in that way we are ensuring that all electrons are coming from same depth. 2.3.3 Ultraviolet photoelectron spectroscopy The basis of ultraviolet (UV) photoelectron spectroscopy (UPS) are the same as for XPS already explained in section 2.3.1, the difference relays on the irradiation 2.3 Surface Characterization ‖ 43 source: instead of using X-rays, photoelectrons are generated after excitation by ultraviolet light. Typical UV source is a He gas discharge line which can be operated to maximize the output of either He I (hν = 21.2 eV) or He II (hν = 40.8 eV). Because of this low energy, only valence levels can be probed, the ones having lower binding energies. These include the occupied band states of a clean solid surface as well as the bonding orbital states of adsorbed molecules. This technique is surface sensitive, but according to the attenuation of the low kinetic energy electrons (Figure 2.6), this attenuation is smaller than that of high kinetic energy electrons. In summary, UPS can probe deeper regions than XPS. Apart from the study of valence band structure, another information that can be obtained by this technique is the value of the materials work function, which stands for the minimum energy required to withdraw an electron from a bound state into the vacuum level. A detailed explanation of the work function and its usefulness to study surface properties can be found in Chapter 3section 3.1.2. The value of the work function corresponds to the difference in the photon energy and the energy of the secondary cut off (estimated with a linear fitting) related to the Fermi edge, as indicated in Figure 2.13. Figure 2.13. Work function (wf) extraction from a UPS He I spectra. The UPS spectra corresponds to a clean lithium metal surface and it was obtained with the UV photons emitted by Helium gas with an energy of 21.8 eV (He I). The figure shows different regions of the spectra and how we can use it to obtain the work function of the surface. The inset is the enlarged spectra in the region of the Fermi edge. During data collection the sample was polarized -12 V to obtain a sharp secondary edge, the binding energy scale is calibrated according this polarization. 44 ‖ 2. Experimental techniques The determination of secondary edge can be tricky because electrons from sample at low kinetic energy overlap with electrons generated on the analyzer itself: these are generated when photoelectrons from the sample hit the internal surface of the analyzer which is typically coated with graphite. The analyzer electrons are not influenced by the contact potential between sample and analyzer, and they form a spectrum superimposed to the secondary edge of the sample spectrum. An easy way to avoid this overlap is applying a potential between the sample and the analyzer. Electrons from sample are going to be accelerated, separating the secondary edges. In Figure 2.13, the binding energy has been corrected, but in order to get a sharp secondary edge sample was polarized -12 V. 2.3.3.1 UPS instrument UPS spectra were taken with a He I emission lamp (hν = 21.2 eV), SPECS UV10/35, and the same photoelectron analyzer used for XPS measurements from the UHV multitechnique surface analysis system at CIC Energigune (Figure 2.7). The helium gas used in the UPS lamp had a purity of 99.99% (Praxair). To increase the purity of the gas, the gas line was guided through a liquid nitrogen trap which acts as a cryopump reducing the amount of impurities in the gas; especially those with a condensation point above the temperature of liquid nitrogen. 2.3.4 Scanning Electron microscopy In a scanning electron microscope (SEM) an electron beam generated by an electron gun is focused using electromagnetic lenses later accelerated onto the sample surface. UHV is needed to avoid interaction of electrons with air. When scanning the beam over the sample, secondary and backscattered electrons ejected by the incoming electron beam are collected in a specific detector for each type of electron, hence obtaining a magnified image of the surface. Secondary electrons are electrons ejected from the sample when the incident beam electrons transfer energy to the atom. Usually, their kinetic energy is lower than 50 eV. The image obtained is a magnification of the surface morphology. Backscattered electrons are electrons from the incident electron beam after interaction with sample atoms. The kinetic energy of backscattered electrons goes from 50 eV to almost the energy of the incident beam electrons[108]. Then, backscattered electrons are coming from deeper regions of the sample than secondary electrons. In contrast to the secondary electrons, backscattered electrons also contain information about the chemical differences of the surface 2.3 Surface Characterization ‖ 45 compounds: heavier elements can deflect incident electrons more strongly, hence those elements appear brighter in the images when compared to light elements. When the sample is not conductive, an overcharging of the surface happens due to electron accumulation that cannot be drain to ground. Before measuring SEM, non-conductive samples are usually sputter coated with a conductive and inert metal, like Au. Enhanced spatial resolution of scanning electron microscope depend on design of the system, but they can typically achieve spatial resolutions below 1 nm owing to the shorter wavelength of electrons if compared to visible light. Hence, SEM allows to obtain higher resolution images than with an optical microscope. 2.3.4.1 SEM instrument FEI Quanta-200FEG microscope from CIC Energigune has been used for the microscopy studies. In the field emission gun (FEG), electrons are emitted from the cathode by applying a high electric field near the filament tip. This technology generates electrons without heating of the gun which can induce problems. An air sensitive transfer tool (Figure 2.14) with a specific coupling for the load lock of this instrument was used to deal with air sensitive samples and to move them from the inert atmosphere of a glove box to the vacuum conditions of the SEM. Figure 2.14. Air tight transfer tool to move samples from an inert atmosphere to the SEM microscope. Another SEM microscope was also used for the measurements presented in this thesis work: Helios NanoLab 450S – FEI, from CIC Nanogune. The particularity of this SEM is that is has a Focused Ion beam (FIB) incorporated. The FIB is used to precisely etch or cut the sample, then the new exposed surface is measured by SEM. An advantage of this microscope is that a clearer cross section can be measured if cutting the sample by other methods presents difficulties. 46 ‖ 2. Experimental techniques 2.4 Electrochemical characterization For the electrochemical measurements, coin cells were assembled using a manual clamper in an argon atmosphere glove box. The type of coin cells used are CR2032 (20 mm diameter and 3.2 mm height). Different parts of a coin cell are specified in Figure 2.15. Figure 2.15. CR2032 type coin cell elements. Case, cap, spring and spacers (current collectors) are made by 316L stainless steel, and the propylene gasket avoid the short circuit of the cell. Note that, when a solid electrolyte is used, there is no need for separator. During this work, two types of CR2032 were assembled. In experiments involving full cells, conventional electrode configuration was used with positive and negative electrodes that deliver an open circuit voltage (OCV) which is the difference between the reduction potential of the electrodes. The second type of CR2032 assembly were symmetric cells. In this case, both electrodes are made of the same material, consequently, OCV of symmetric cells should be zero. All the electrochemical measurements were performed using a Biologic VMP3 potentiostat tester from CIC Energigune. Following the electrochemical characterization methods used in each type of cell are explained. 2.4.1 Full cell electrochemical characterization The electrochemical characterization techniques used in conventional two electrode systems were cyclic voltammetry (CV) and galvanostatic cycling. In a cycling voltammetry experiment, the intensity response of a working electrode 2.4 Electrochemical characterization ‖ 47 (the electrode under study) is measured while applying a voltage sweep using a constant scan rate (Figure 2.16). It provides information about the redox reactions, the voltage at which they occur and their reversibility. Figure 2.16. Example of a cyclic voltammetry experiment of one redox process for element A. a) Applied cyclic potential sweep to the working electrode and b) response of the working electrode resulting in a cyclic voltammetry. Eu indicates the upper limit of the voltage and EL is the lower limit of the voltage, which corresponds to the OCV at discharged state. In a galvanostatic cycling experiment, in contrast to the previous method, the current is controlled and held constant until reaching the upper and lower voltage window limits, and the potential becomes the dependent variable, which is followed as a function of time (Figure 2.17). Within this technique, we can also observe the voltage at which the redox reaction is happening: represented by a plateau in the plot. Figure 2.17. Example of a galvanostatic cycling experiment of one redox process for element A. a) Applied constant intensity until reaching the desired voltage in the working electrode and b) response of the working electrode. Eu is the upper limit of the voltage, and the plateau indicates a redox reaction process. Time needed for charge and discharge is not the same related to irreversible reactions. 54 ‖ 3. Lithium surface interaction with pure atmospheric gases surface[113-141]. In these research works, surface is analyzed by one of the following surface specific techniques: auger electron spectroscopy (AES)[113–122], ultraviolet photoelectron spectroscopy (UPS)[123–127], infrared spectroscopy (IR)[128,129], X-ray photoelectron spectroscopy (XPS)[116,118,134–138,119,120,124,127,130–133], electron energy loss spectroscopy (EELS)[122,134], absorption spectroscopy(XAS)[118], metastable deexcitation spectroscopy[125], ellipsometry[121], surface X-ray diffraction (XRD)[139], density functional theory (DFT) combined with molecular dymanics (MD)[140], selected area electron diffraction[141] and energy filtered transmission electron microscopy[141]. Figure 3.1 shows the distribution of published scientific articles per decade (Figure 3.1a) and per studied gas (Figure 3.1b). We found that, after an interest decay in the first decade of the 21st century, the number of published papers in the last decade (2011-2020) increased (Figure 3.1a), this suggests there are still unsolved questions related to the interaction of metallic lithium with atmospheric gases. In following, we will discuss the effects O2, CO2 and N2 produce on the lithium surface as reported in the works from Figure 3.1 agreed on, in addition, we will also emphasize the controversial issues that entail us to further investigate the gas-lithium interaction. Figure 3.1. Distribution of the number of published articles that analyze the interaction of O2, CO2 and N2 with the surface of metallic lithium; a) per decade and b) per studied gas, from references[113141]. According to our literature review (Figure 3.1b), the most studied reaction is the one between metallic lithium and oxygen gas. All studies, without exception, corroborate that lithium surface is very reactive to oxygen, being Li2O the reaction product. Most of the authors agree that this reaction does not create a stable passivation layer on the lithium surface. Indeed, the oxidation reaction continues into the bulk of the metal. Zavadil et al.[134] explained this phenomenon as a 3.1 Introduction ‖ 55 consequence of a combination of relative thermodynamic stabilities, the solubility of zero valent lithium in its own oxide and the fact that lithium is a highly viscous liquid at room temperature that allows for a continuous structural rearrangement. However, besides thermodynamical and solubility considerations, lithium has a melting point of 180.5 °C and the only metal that is considered liquid at 1 atm and room temperature is mercury[142], then it is more appropriate to say that lithium is a soft metal rather than a highly viscous liquid. An alternative explanation provided some years later attribute the continuous oxidation process to the difference in the atomic density of Li and Li2O: being four times larger for Li2O than for metallic lithium[121]. It was claimed that the density difference produces a contraction of the surface where fresh metallic lithium will be continuously in contact with the atmosphere. By means of ellipsometry, it was concluded that Li2O layer is porous, so it has free pathways for oxygen to reach metallic lithium. In contrast, a recent study suggests that pure oxygen will form a passivation layer if the gas has no traces of moisture and only after certain exposure time[141]; this nm-thick layer blocks the diffusion of oxygen molecules preventing further oxidation of the underlying lithium. The reaction between lithium and CO2 gas was comprehensively studied by Zhuang et al.[137], where the reaction mechanism was investigated by combining XPS, UPS and Ab initio Hartree-Fock self-consistent calculations. These authors concluded that the reaction of CO2 gas with clean lithium leads to a mixture of CO32with O2-. Out of the three interactions, the one with nitrogen gas is of special interest due to the reported strategies based on nitride materials chemistry to stabilize lithium metal anode[143]. Despite its importance, we find some controversial results reported in the literature. Some authors believe nitrogen gas is, together with oxygen and water, the most reactive residual gas for metallic lithium in UHV systems[123]. Indeed, several times, it has been reported the formation of Li3N by direct chemical reaction between the metal and nitrogen gas with the aim of creating a passivation layer that protects lithium upon electrochemical cycling[139,144–146]. In contrast, theoretical studies by Koch et al.[140] reported that direct exposure of N2 to a clean lithium surface does not favor the dissociation of N2 gas. At the same time, some other investigations that analyze the reaction between metallic lithium and nitrogen gas claim that the reaction is not 56 ‖ 3. Lithium surface interaction with pure atmospheric gases spontaneous[131,132], hence contradicting all the studies that confirm Li3N formation. 3.1.2 Work function to monitor lithium surface stability In this chapter lithium surface work function () evolution is monitored in order to evaluate the stability of lithium surface as a result of treatment with different gases. To define the work function, we have to look at the different energetic levels of the surface of a metal as illustrated in Figure 3.2. The Fermi energy  level refers to the energy when the electron occupation probability equals to 0.5 in the electronic energy. As the electron distribution can be represented by a step function, it can be approximately considered that electrons mainly fill the energy levels below Fermi energy level at the finite temperature, while levels above are unoccupied[147]. This term is defined in relation to the average electrostatic potential energy of an electron of the conduction band, , deep inside the metal: (−∞).  becomes constant again at a large enough distance from the surface, (+∞). However, to define the vacuum level we also need to consider the dipole layer defined by the Galvani potential () in which all electrostatic interactions, not included in , are included. The difference in the absence of excess electric charge on the surface is the surface potential :  ( − ∞ ) −  ( + ∞ ) =  (3.1) The chemical potential  of the electron is defined by   =   −   ( + ∞ ) (3.2) And consequently, we obtain the work function   = −   +  (3.3) where  is the charge of an electron. The two terms of equation (3.3) represent the following: one part () describes the electrical work for the electron to go through the intrinsic surface dipole layer and the other (−) is equivalent to the chemical potential. According to this definition, the work function in vacuum corresponds to the minimum work needed to extract one electron from the surface to the vacuum level, being free of excess electric charge. 3.1 Introduction ‖ 57 When the condition of absence of any excess surface charge, i.e. equation (3.1) is fulfilled, the relation between the Fermi energy and the work function is   = −   (3.4) Then, the work function in a metal is equivalent to the position of the Fermi level with respect to the vacuum level[148]. It depends on the surface structure and is affected by the outermost layer of the sample. Figure 3.2. Characteristic electronic energies at the metal/vacuum contact in the absence of excess surface charge. Symbols and terms are explained in the text. Adapted from[148]. In order to correlate the work function of the lithium surface with its stability, in a first approach we could consider the relationship between Fermi energy and the chemical potential of a system. When temperature is zero Kelvin, both terms are equivalent[149]. In turn, open circuit potential () of an electrochemical cell is defined by the chemical potentials of electrodes[19]:   = (   −     ) /  (3.5) Then, at zero Kelvin we should have the following equivalence:   =   =   (3.6) However, considering that our lithium is at room temperature, if the electric potential differences of each interface in an electrochemical system is considered, together with the surface potential of the electrolyte, the cell potential difference 58 ‖ 3. Lithium surface interaction with pure atmospheric gases can be expressed as the difference of absolute electrode potentials () which is related to the electrode work function   (  ) =    + ∆    (3.7) as described by Trasatti[150], where  is the Faraday constant and ∆   is the contact (Volta) potential of the electrode-electrolyte system. With this definition, we observe that work function changes on the lithium surface are indicatives of electrode absolute potential modifications, which will affect its stability against the electrolyte. Then, at a first approximation, an increase in the work function will make the surface less energetically favorable transfer an electron, thus more stable against the electrolyte. In this line, a modified lithium metal anode with a higher work function than bare metallic lithium will help to gain stability in the electrodeelectrolyte interface. 3.2 Spectra measuring conditions and data analysis guidelines Lithium surface reactions with O2, CO2 and N2 gases were characterized with two surface sensitive techniques: XPS and UPS. The first one is used to determine the composition of the Li surface. The second surface characterization technique is used to determine work function, and it also gave information about the valence band structure. Both spectroscopies (XPS and UPS) were carried out in the multitechnique surface analysis system available at CIC Energigune (Figure 2.7), using instruments explained in sections 2.3.1.3 and 2.3.3.1. XPS measurements were recorded with a non-monochromatic Mg Kα photon source (hν = 1253.6 eV). The pass energy was set to 90 eV for survey spectra acquisition and 40 eV for the detailed regions of each element. UPS spectra were taken with a He I emission lamp (hν = 21.2 eV), using a pass energy of 1 eV and polarizing the sample -12 V. 3.2.1 XPS and UPS data analysis guidelines XPS spectra is analyzed with CasaXPS version 2.3.16dev52 (Casa Software Ltd, Teighmouth, UK). The binding energy zero is calibrated in every spectrum prior to 3.2 Spectra measuring conditions and data analysis guidelines ‖ 59 fitting the photoelectron lines for each element. A survey spectrum is recorded for every sample to ensure the surface is free from any contaminants. The binding energy calibration, in the case of the O2 interaction, is done using the metallic lithium component in the Li 1s region and lithium oxide component in the O 1s region. For the CO2 interaction, the binding energy calibration is based on the position of metallic lithium component in the Li 1s region and lithium carbonate component in the C 1s region. For the last gas studied, N2, the binding energy is calibrated with respect to the position of metallic lithium in Li 1s region and position of lithium oxide in the O 1s region. The peak background is simulated by a Shirley function. A Voigt profile (30%-70%, Lorentzian-Gaussian distributions) is used as peak lineshape to fit all components except for metallic lithium. The lineshape of this last one is a pseudo-Voigt function (LF(1.5,2,5,50)) which takes into account the asymmetric tail in the higher binding energy side of the metallic peak; caused by the small kinetic energy losses originated by the interaction of the core level electrons with the conduction band of the metal. This shape is equivalent to the asymptotic form of theoretical Doniach-Sunjic asymmetric lineshape. First two parameters of LF(1.5,2,5,50) define the asymmetry of the lineshape, third one is the Gaussian contribution and fourth the damping parameter to force the tail to reduce towards the limits of the integration limits[151]. The assignment of the compounds has been done based on reported binding energies (BE) in works where the studied system is similar to our case[127,133,137,152]. With these references, and considering a BE uncertainty of ± 0.1 eV, we are able to clearly identify the following compounds: Li0, Li2O, Li2O2, Li2CO3 and Li3N. The maximum FWHM (full width at half maximum) for these compounds is variable depending on the element. Table 3.1 summarizes the BE and FWHM constrains used to fit the data. Any other compound that is not in the table will be discussed in its section. To quantify the surface composition, the concentration of each compound is calculated from equation (2.7). The area of every fitted photoelectron line is corrected with the corresponding sensitivity factor () of each element and orbital based on Scofield cross sections together with a transmission function () specific for this photoelectron analyzer. An exponential factor is also used to correct for the different photoelectron escape depths. This correction is needed because all core levels are measured using same photon energy, so photoelectrons emitted 60 ‖ 3. Lithium surface interaction with pure atmospheric gases from each of these levels will have a different inelastic mean free path (). With these corrections, the atomic concentration () for each compound can be obtained from (2.7). Table 3.1. Fitting parameters used to identify the compounds formed on the lithium metal surface; based on reported BE[127,133,137,152] and experimental evidence. Compound Fitting constrains (eV) Li 1s O 1s C 1s N 1s Li0BE 54.90-55.10 FWHM 1-1.3 Li2O BE 56.30-56.50 531.10-531.30* FWHM 1.6-1.8 1.4-1.6 Li2O2 BE 57.40-57.60 534.05-534.25 FWHM 1.8-2 1.8-2 Li2CO3 BE 57.90-58.10 534.60-534.80 292.60-292.80 FWHM 1.6-1.8 undefined 1.5-1.7 Li3N BE 54.70-54.50 395.20395.40 FWHM 1.4-1.6 1.2-1.4 *the residual amount of oxide we find after cleaning the lithium has a smaller BE, around 530.8 eV, as previously reported[133] and in agreement with suboxide formation due to the ion assisted cleaning process. The work function is calculated from the minimum kinetic energy measured in the photoelectron spectrum (secondary electron cut-off), the maximum kinetic energy measured for a photoelectron emitted from the Fermi level and the photon energy, as explained in chapter 2 section 2.3.3 (Figure 2.13). The secondary electron cut-off is obtained with a linear fitting to the low kinetic energy side of the photoelectron spectrum, whilst the Fermi edge is obtained by fitting a step function that will define the zero for the binding energy. 3.3 Li foil surface cleaning The starting point of this study is a commercial lithium foil (Rockwood Lithium, Battery Grade), which was stored in an argon filled Glove Box (MBRAUN) where O2 and H2O levels were below 0.1 ppm. After being mounted in the photoemission sample holders, the foils were transported to the UHV system with a specific transfer tool (Figure 2.8) that prevented air exposure 3.3 Li foil surface cleaning ‖ 61 This foil has a purity of 99.8%. Even so, the XPS spectra of the Li foil stored in the glove box, represented in Figure 3.3, reveals a completely oxidized lithium surface. Binding energy of its main Li 1s peak is around 57 eV, which can be assigned to a mixture of lithium oxide and lithium carbonate according to Table 3.1. To be able to analyze the interaction of metallic lithium and the selected gases, Ar ion sputtering at 5 keV was performed, at 4·10-7 mbar for at least 5 hours. With this method, previously used in literature[127,133,152,153] we got a surface composed by (93.6 ± 1.9)% of pure metallic lithium, where the rest of the surface is lithium oxide. The Li 1s photoelectron peak of a cleaned foil (Figure 3.3) reveals some plasmon loss structures that correspond to surface plasmons of metallic lithium[154]. These features can be used as an indicative of clean metallic lithium[135]. Another indicative of having a clean lithium surface is the value of the work function measured by UPS, which is 3.01 ± 0.08 eV, in agreement with reported values for metallic lithium surfaces[155]. Figure 3.3.Comparison of Li 1s XSP spectra of a lithium foil stored in argon atmosphere and after cleaning the surface in UHV by Ar+ sputtering. Whit the cleaning procedure, its binding energy of the main photoelectron peak shifts down to the binding energy of metallic lithium, and it also shows plasmon loss structures (highlighted with an arrow), indicative of metallic lithium[135]. 62 ‖ 3. Lithium surface interaction with pure atmospheric gases 3.4 O2, CO2 and N2 gases effects on clean lithium surfaces We dosed O2 (Praxair, 99.9%), CO2 (Laborgase, 99.995%) and N2 (Praxair, 99.9%) gases in three dose ranges: 1-2-3-4-5-6-7-8-9-10 L as low dose range, 1-10-1001000 L as medium dose range, and higher doses up to the order to 1·108 L. Langmuir (L) unit corresponds to a dose of 10-6 Torr of a given gas during one second. Every dosing sequence was deployed starting from a UHV cleaned lithium. The specific partial pressures we use in each dose are detailed in the analysis of the interaction with each gas. 3.4.1 Oxygen interaction Interaction of a clean lithium surface with oxygen gas was studied at the conditions summarized in Table 3.2. Figure 3.4 shows the evolution of XPS spectra, analyzed with the parameters from Table 3.1. The first compound growing on the lithium metal surface is lithium oxide, Li2O. The oxygen dose that leads to a full coverage of Li surface by lithium oxide has been estimated from the peak area evolution of the Li2O component in the Li 1s photoelectron line. According to the slope change measured in Figure 3.5a, the full surface coverage dose is around 3 – 4 L of O2, which is in agreement with the disappearance of plasmon loss structure and evolution of Li2O energy loss peaks assigned to surface excitons of Li2O[134,156], represented in Figure 3.5b. Table 3.2.Pressures used for each studied dose in the analysis of the interaction of lithium metal surface with O2 gas. Range Dose (L) Pressure range (mbar) Low dose 1,2,3,4,5,6,7,8,9,10 10-8 Medium dose 1 10-8 10 10-7 100 10-6 1000 10-5 High dose 5000,1·10410-4 5·108101 Besides Li2O, there is no other compound evolving on the surface until we get to the high dose range, when a new peak at higher binding energy of O 1s and Li 1s XPS spectra appears, as shown in Figure 3.4 for 104L to 5·108 L O2 doses. Looking 3.4 O2, CO2 and N2 gases effects on metallic lithium surrface ‖ 63 to reported binding energies[133], that peak can be assigned to Li2O2. In order to confirm the assignment of this new compound, we compared the O 1s XPS and O 2p UPS spectra in Figure 3.6. Li2O and Li2O2 positions have been identified in O 2p region according to literature values[127]. The increase of Li2O2 concentration with the oxygen dose is confirmed from both spectra. Figure 3.4. Fitting of the XPS photoelectron peaks from a lithium surface exposed to oxygen gas at selected low (1 L, 5 L, 10 L), medium (100 L, 1000 L) and high (1·104 L to 5·108 L) dose ranges. The compounds that form in the surface are shown by the deconvolution of the 1s photoelectron peaks of oxygen (left panel) and lithium (right panel). In the spectra, the fitted curve (black line) follows experimental data (dots), and background is represented by a dash line. 70 ‖ 3. Lithium surface interaction with pure atmospheric gases To validate which is the most adequate fitting for our data we compare the residual standard error of both situations (calculated by CasaXPS version 2.3.19PR1.0). If we look to the residual standard deviation of the fitted spectra with respect to data, fitting with one or two components will result in very similar values, between 0.85 and 1.35 (Figure 3.10a). However, when comparing the resulting error estimates when using fitting A or fitting B methods, the residual standard deviation of the compounds is notably reduced when two components are used (Figure 3.10b). Figure 3.10. Comparison between the residual standard deviation (RSD) of two types of fitting of CO species in C 1s XPS spectra, where a) represents the RSD of the fitting spectra and b) represents the RSD of each compound. The calculation of the standard deviation of the compounds is based on Monte Carlo analysis where the error estimates are an indicator of how stable a peak model is with respect to noise. One of the advantages of using this error analysis is that it highlights when a quantification parameter is poorly determined by the combination of model and optimization procedure. To be able to apply it, we need to have Poisson noise distribution in the spectra. To validate if we have Poisson noise distribution, a region absent of core-level excitations can be analyzed by a linear regression. If the standard deviation given by CasaXPS is around 1 then we 3.4 O2, CO2 and N2 gases effects on metallic lithium surrface ‖ 71 assume a Poisson noise distribution, which in our case is between 0.8 – 1.15, hence in good agreement with a Poisson noise distribution[161]. Using this approach, we can confirm that at least we have two compounds that form the CO species, which will be referred as (CO)a and (CO)b. Figure 3.11 shows the evolution of O 1s, C 1s and Li 1s spectra when exposing the clean lithium surface to CO2 gas, where C 1s spectra have been analyzed using fitting A (Table 3.4) and rest of compounds have been identified according to the fitting parameters included in Table 3.1. In the figure we observe that compounds created on the surface as a consequence of Li-CO2 interaction are Li2O, Li2CO3 and CO species. If we look to the Li 1s peak evolution at the lowest doses, we detect that Li2O is formed on the Li surface at 1 L CO2, prior to the formation of lithium carbonate. We also see that, for low and medium doses, Li2O is growing the most if compared with the rest of compounds. For this reason, we calculated the CO2 dose needed to cover all the surface metallic lithium by the saturation of Li2O area from O 1s spectra, represented in Figure 3.12a. According to this, the needed dose to reach a monolayer coverage is 8 L, which also agrees with the disappearance of plasmon loss feature represented in Figure 3.12b. In this last figure energy loss peaks corresponding to Li2O are not present, in contrast with the results after O2 dosing, suggesting that carbon-based compounds are growing on top of Li2O and prevent the detection of lithium oxide energy loss features. O 1s spectra from Figure 3.11 cannot be used to identify the contributions from CO species and Li2CO3, which are overlapped above 534 eV. This core level, after forming both CO species and lithium carbonate (Figure 3.11, O 1s spectra after 5 L), presents two main peaks. The one at the lower binding energy (531.20 ± 0.1 eV) corresponds to Li2O as defined in the literature[127,133,137], and its FWHM (1.41.6 eV) does not increase in agreement with the observed FWHM behavior for O2treated surfaces. This suggests that there are no compounds related to carbonbased species in low binding energy side. Therefore, the CO species are going to be somewhere in the high binding energy side, along with lithium carbonate which has a well-defined binding energy at 534.70 ± 0.1 eV[133]. However, as we cannot assign any exact binding energy to the CO species in this peak, we use a broad peak which contains both lithium carbonate and CO species contributions. 72 ‖ 3. Lithium surface interaction with pure atmospheric gases Figure 3.11. Fitting of the XPS photoelectron peaks of a lithium surface expose to carbon dioxide gas at selected low (1 L, 5 L, 10 L), medium (100 L, 1000 L) and high (1·104 L, 8·108 L) dose ranges. The compounds that form the surface are shown by the deconvolution of the peaks of oxygen, carbon and lithium. In the spectras, experimental data (dots) follows the fitted curve (black line) and background is represented by a dashed line. As it can be observed in C 1s spectra evolution from Figure 3.11, binding energy of CH/CC presents a lower value (0.5 eV lower) at the highest dose. This compound, related to adventitious carbon, is widely used as a reference to calibrate the spectra in XPS. However, a recent paper observes that the binding energy of the CH/CC related to adventitious carbon can vary as much as 1.44 eV, and they find a correlation between the changes in the sample work function and 3.4 O2, CO2 and N2 gases effects on metallic lithium surrface ‖ 73 the biding energy of CH/CC[159]. This observation could explain the variation we observe in the binding energy of CH/CC. The evolution of the normalized composition and work function variation in the Li-CO2 system is shown in Figure 3.13. For the normalized surface composition, we considered the Li0 from Li 1s, Li2O from O 1s, and CH/CC, Li2CO3 and (CO)a and (CO)b species from C 1s regions. In this case, the evolution of the work function is only represented for the low dose range and medium dose range, because some technical problems prevented to measure the work function at the high dose range. If the reaction of lithium metal surface with O2 gas and CO2 gas is compared, it is observed that the oxidation process is slower in the case of the CO2 gas, where even after the highest CO2 dose is applied, metallic lithium can still be detected on the surface (Figure 3.11 and Figure 3.13). Then, overlayer thickness should be below 10 nm to allow Li 1s photoelectrons from subsurface Li0 to escape and to be detected. A possible explanation for the slower kinetics of the oxidation reaction is that there is a layer slowing down the lithium oxidation, probably Li2CO3, the predominant one at the highest dose. Figure 3.12. a) Li2O peak area (blue points) measured from O 1s XPS spectra as a function of gas dose, where the intersection, slope change, of both linear fits (dark grey) corresponds to the CO2 dose needed to fully cover the lithium metal surface which is around 8 L. 95% confidence bands of the fitting are shown in light grey. b) the disappearance of metallic plasmon loss structure from Li 1s XPS spectra around 63 eV agrees with the measured slope change in the left panel figure. 74 ‖ 3. Lithium surface interaction with pure atmospheric gases Figure 3.13. Compositional and work function evolution of a clean metallic lithium foil exposed to carbon dioxide gas for a) low dose range b) medium dose range and c) high dose range. In all the cases, the surface has a large amount of metallic lithium. Here, as happens with the O2 gas, the work function decreases because of the reaction of the surface. 3.4 O2, CO2 and N2 gases effects on metallic lithium surrface ‖ 75 The work function evolution for low dose and medium dose of CO2 follows a decreasing trend (Figure 3.13a and b) likewise it happens when dosing with O2. This is also in agreement with the DFT geometry optimization and molecular dynamics calculations performed by Koch et al.[140] up to one monolayer coverage. As earlier mentioned, for the low and medium dose ranges the compound that is growing most on the surface is lithium oxide. Then, it is more than plausible that the work function is going to be dominated by Li2O. Figure 3.14 shows that, in fact,  evolution in the Li-CO2 system adjusts also to the same correlation previously obtained for the oxidation of lithium with O2 from equation (3.8). The deviations of the  exponential decay for O2 and CO2 dosed lithium surfaces can be explained by the effect that carbon-based compounds have on it. Figure 3.14.Correlation of the and the lithium oxide normalized surface percentage. The exponentail fit corresponds to the lithium dosed by O2 gas (blue dots) already shown in Figure 3.8. If we add to this plot the data from the Li-CO2 system in the low and medium dose ranges (orange dots), we observe they follow the same exponential tendency, suggesting that in this range the work function is manly determined by the ammount of lithium oxide on the surface. 3.4.3 Nitrogen interaction The evolution of the lithium metal surface after N2 exposure, according to N2 doses of Table 3.1Table 3.5 is represented in Figure 3.15, where spectra have been analyzed according to fitting parameters shown in Table 3.1. The normalized 76 ‖ 3. Lithium surface interaction with pure atmospheric gases surface composition and work function evolution are shown in Figure 3.16. For the quantification, we consider the Li0 from Li 1s, Li2O and Li2O2 from O 1s and nitrogen-based compounds from N 1s. As occurred for the Li-CO2 system, we were not able to measure the evolution of the work function at high dose range. Table 3.5. Pressures used for each studied dose in the analysis of the interaction of metallic lithium with N2 gas. Range Dose (L) Pressure range (mbar) Low dose 1,2,3,4,5,6,7,8,9,10 10-8 Medium dose 1 10-8 10 10-7 100 10-6 1000 10-5 High dose 1·10410-4 1·108101 For both low and medium dose ranges we do not detect any interaction between metallic lithium and nitrogen gas. The surface composition, and consequently the work function (Figure 3.16), remain almost constant throughout exposure to N2 doses between 1 and 1000 L. When going up to higher doses, we see some nitrogen-based compounds at 1·104 L N2 gas (Figure 3.15). However, this surface is still dominated by metallic lithium, the total amount of nitrogen-based compounds is less than 1.2% (Figure 3.16c). Furthermore, Li3N is just the 0.28% of the surface. Because of this low amount of Li3N, we could not fit a component to account fo it in Li 1s spectra. We name the other nitrogen-based compounds shown in Figure 3.15 at 1·104 L as N1 (binding energy of 397.3 eV) and N2 (binding energy of 399.9 eV). Looking to literature and comparing reported binding energies with ours, we can discard that any of these two compounds is LiN3[162] or LiNO3[163]. Both of them could be related to carbon-based compounds. N1 binding energy corresponds to a poly(aniline)[164,165], and N2 could be pyrrolic-N[166] or carbon nitride[167]. If this would be the case, we should see the corresponding contribution in the C 1s spectra. However, the resolution of the C 1s spectra we have is not enough to determine whether this is the case or not. The relative sensitive factor of C 1s in our system is 1, smaller than that of N 1s (1.77), and we already observe a small amount of N compounds (Figure 3.16c, each N1 and N2 contributions are less than 0.8 % of normalized surface composition). It is fair to mention that possible reaction pathways to produce poly(aniline) or pyrrolic-N just from nitrogen gas is rather unlikely. Also, reported carbon nitride[167] was produced using a magnetron sputtering, then adding more energy to the system 3.4 O2, CO2 and N2 gases effects on metallic lithium surrface ‖ 77 than that we have just with nitrogen gas at room temperature. In order to confirm the assignment of these peaks we would need further studies with reference materials, for this reason we keep naming them as N1 and N2. Figure 3.15. Fitting of the XPS photoelectron peaks of a lithium surface exposed to nitrogen gas at selected low (1 L, 5 L, 10 L), medium (100 L, 1000 L) and high (1·104 L, 1·108 L) dose ranges. The compounds that form the surface are shown by the deconvolution of the peaks of oxygen, nitrogen and lithium 1s. The fitted curve (black line) follows experimental data (dots) and background is represented by a dashed line. 78 ‖ 3. Lithium surface interaction with pure atmospheric gases When treating the surface at the highest dose (1·108 L N2), the surface is completely oxidized, but what we observe is lithium peroxide and none of the expected nitrogen-based compounds. This result at high doses is compatible with the traces of oxygen impurities present in the nitrogen gas line. We also observe here that nitrogen-based compounds formed after a dose of 1·104 L N2 are not strong enough to passivate the lithium surface and prevent its oxidation, and that lithium surface is much more likely to react with oxygen than with nitrogen gas. As mentioned earlier, we could not measure the work function of this high doses because of some technical problems in the UPS system. But we could assume that the work function of the dose of 1·104 L is going to be very similar to clean lithium surface, and the last one is going to have a smaller work function expected from the interaction of lithium and oxygen (section 3.4.1). In order to analyze the effect of lithium nitride formation on the electronic structure of lithium, a different approach based on the work done by Ishitama et al.[132] was used to obtain Li3N: reactive ion implantation, using an ion source that generates a N+ beam with an energy of 0.5 keV at a pressure of 4·10-6 mbar for 5 minutes. This method allowed to prepare a surface mainly composed by lithium metal and lithium nitride, as determined by XPS analysis of Li 1s, O 1s and N 1s shown in Figure 3.17. The normalized surface composition calculated from Figure 3.17 spectra results on a surface composed by 68.4% Li0, 19.8 Li3N, 8.1% Li2O and 3.7% being small amounts of impurities. The formation of Li3N leads to a change on the  value as determined by UPS, from 3.01 eV of Li0 to 2.49 eV in the Li3N-contaninig surface. Considering that the final surface also contains 8.1% of Li2O, one could think that the  decrease is due to lithium oxide formation. However, according to the correlation obtained for O2 and CO2 dosing in equation (3.8), such amount of surface Li2O should result in  = 2.97 eV far above the 2.49 eV measured. So, lithium nitride formation also reduces the work function of the lithium metal surface. 3.4 O2, CO2 and N2 gases effects on metallic lithium surrface ‖ 79 Figure 3.16. Compositional and work function evolution of a clean metallic lithium foil exposed to nitrogen gas for a) low dose range b) medium dose range and c) high dose range. There is no reaction between metallic lithium and nitrogen gas for low and medium dose ranges. For high dose range, a small amount of nitrogen based compunds is detected, represemted by the inset. We could not collect the work function for the high dose range. As mentioned in the introduction, there are several studies where Li3N is obtained just by direct reaction between lithium and nitrogen gas[139,144–146]. All these studies use atmospheric pressures, considering how sensitive is lithium to both 86 ‖ 4. Study of Li carbonate evolution on Li metal surface contribution of the gaussian function and w is a damping parameter to force the tail to reduce towards the limits of the integration limits[151]. This Li0 asymmetric lineshape is, for photon energies of 510 eV, 600 eV and 750 eV, LF(1,2,20,100). For photon energy of 280 eV is LF(1,5,20,100). Other species that are not in the table are explained later in their corresponding section. Three of the studied photon energies (280 eV, 510 eV and 750 eV) are chosen to be able to measure the different core levels of the surface at same kinetic energy. In this way, the only data needed to calculate the atomic concentration of the surface elements is the cross section of the elements at each specific photon energy and the flux of electrons, according to equation (2.7). In this chapter, overlayer attenuation method (section 2.3.1.2) is used to calculate thickness of surface layers, using data from Li 1s core level at different photon energies. Physical parameters needed to calculate both the atomic concentrations and thicknesses calculations are summarized in Table 4.2, Table 4.2 and Table 4.4. Table 4.2. Physical parameters used to quantify surface atomic concentration and to calculate thickness of surface layers later in the chapter. Photon energy (eV) Photon FluxICore level Kinetic energy (eV) Cross sectionII (Mbarn) 280 0.231 Li 1s ~ 222 0.1103 510 0.705 C 1s Li 1s ~ 222 ~ 452 0.2563 0.0199 750 0.47 O 1s Li1s ~ 222 ~ 542 0.2931 0.0063 Inormalized value of photon flux (photons/s mA m), experimental parameter measured in beamline 9.3.2 of Advanced Light Source. II from database[175] Table 4.3. Atomic density of surface compounds used to calculate thickness of surface layers later in the chapter. Compounds Atomic density of Li (10-22 atom/cm3) Li04.6 Li 2 O 8.1 Li 2 CO 3 3.4 Li 2 C 2 O 4 2.5 4.3 Li foil surface cleaning ‖ 87 Table 4.4. Inelastic mean free path of electrons from Li 1s spectra used to calculate thickness of the overlayers. Data obtained from the software QUASES-IMFP calculation by TPP2m formula. PE (eV) Electrons originated in Kinetic energy (eV)  through Li0 layer (Å)  through Li2O layer (Å)  through Li2CO3 layer (Å)  through Li2C2O4 layer (Å) 280 Li0225.0 10.68 Li 2 O 223.6 8.67 Li 2 CO 3 222.0 9.15 Li 2 C 2 O 4 222.2 9.50 510 Li0455.0 18.2 Li 2 O 453.6 13.78 Li 2 CO 3 452.0 14.41 Li 2 C 2 O 4 452.2 14.87 600 Li0545.0 20.96 Li 2 O 543.6 15.69 Li 2 CO 3 542.0 16.38 Li 2 C 2 O 4 542.2 16.90 750 Li0695.0 25.43 Li 2 O 693.6 18.79 Li 2 CO 3 692.0 19.59 Li 2 C 2 O 4 692.2 19.67 4.3 Li foil surface cleaning Lithium foil used in this study is a commercial foil from Alfa Aesar (99.9% purity, metal basis, 1.5 mm thick). This foil has been characterized in two situations: after scraping it in Ar atmosphere and after scraping it in UHV conditions. 4.3.1 Characterization of Li foil surface cleaned in Ar atmosphere Commercial lithium foil was stored in an argon atmosphere glove box, where H2O and O2 gas levels were below 0.1 ppm. Surface was scraped using a UHV cleaned blade in the glove box, a standard procedure in battery community before using the lithium as an anode. Sample was then transferred from the glove box to the load lock of APXPS instrument, preventing surface exposure to atmospheric air. Figure 4.1 represents the APXPS spectra of the Li foil measured with a high photon energy (835 eV) in order to broaden range of measured binding energies. Detected surface elements of this foil are oxygen, carbon and lithium. 88 ‖ 4. Study of Li carbonate evolution on Li metal surface Figure 4.1. APXPS survey spectra collected in UHV of a lithium foil scraped in argon atmosphere glove box. Detected surface elements are lithium, oxygen and carbon. Figure 4.2 shows the in-depth distribution of surface compounds. According to it, there is no clear evidence of having metallic lithium on the surface even at the highest photon energy, 750 eV. This photon energy has an estimated probing depth of 8 nm, which corresponds to 3 times the inelastic mean free path () of electrons. Indeed, considering binding energies of Table 4.1, we can easily identify lithium carbonate and lithium oxide. Adventitious carbon contamination is also detected on the surface (CH/CC). A deeper analysis of the nature and binding energy of it can be found later in this chapter. It is also worth mentioning that species such as LiOH and Li2O2 could also be present on the surface. However, aim of this section is to have a general view of the effectiveness of cleaning the lithium in argon atmosphere condition more than conducting a detailed analysis of surface composition, so we are only going to consider the dominant compounds of surface. When comparing O 1s and Li 1s spectra at different photon energies in Figure 4.2, the more surface sensitive (smaller photon energies), the more carbonate there is on the surface, as signal of carbonate increases while signal of Li2O decreases. Then, in a commercial lithium foil cleaned in argon atmosphere, lithium-based compounds in the first 8 nm of the surface are Li2O and Li2CO3, where the carbonate lies on top of the oxide. 4.3 Li foil surface cleaning ‖ 89 Figure 4.2. APXPS spectra showing depth profiling of a commercial lithium foil which has been scraped in argon atmosphere glove box. Looking to the amount of carbonate this Li foil surface presents, one can think the commercial foil has already been exposed to CO2 gas. In order to check it, 400 mTorr CO2 gas were added to the surface. Figure 4.3 shows that surface is not changing after treating it with the gas. However, in the previous chapter we learnt that even a low dose of 10-8 mbar (7.5·10-5 mTorr) of CO2 gas is modifying a clean metallic lithium surface. This difference suggests that, as was speculated, lithium surface stored in an inert gas has already been exposed to an atmosphere that contains CO2 gas, which could be happening in the glove box. Although glove boxes typically have sensors for both O2 and H2O, they neither monitor nor control for potential CO2 gas contamination. 90 ‖ 4. Study of Li carbonate evolution on Li metal surface Figure 4.3. Evolution of C 1s APXPS spectra of a commercial lithium foil measured at a photon energy of 600 eV while dosing 400 mTorr of CO2 gas. 4.3.2 Characterization of Li foil surface cleaned in UHV As in previous chapter, in this one Li foil also needs to be cleaned in UHV conditions in order to be able to study the interaction of Li0 with CO2 gas. The way chosen to clean the surface of lithium foil is different from previous chapter. Here, instead of using argon ion bombardment, Li surface was cleaned by scraping it in UHV conditions. For that, a wobble stick that has both linear and 22o angle motions with a blade at the edge was assembled on a CF port of the load lock of APXPS UHV system. With this tool lithium samples were scraped in UHV conditions, with a base pressure in the range of 1·10-8 Torr. Figure 4.4 shows the difference between scraped and non-scraped surfaces. 4.3 Li foil surface cleaning ‖ 91 Figure 4.4. There is a clear difference in color and shine of a Li foil coming from argon atmosphere between the UHV scraped and non-UHV scraped sides. Figure 4.5a shows the APXPS spectra of the UHV scraped side of the foil, where presence of Li0 in the surface is clear. The three core levels from Figure 4.5a have the same kinetic energy, so the areas of that photoelectron peaks can be used to quantify the concentration of the compounds from the surface, using equation (2.7) and Table 4.2 parameters. This quantification is shown in Figure 4.5b. In Figure 4.5a, Li0 and Li2O have been identified with the constrains from Table 4.1. The presence of Li0 is also corroborated by the plasmon loss structures representative of metallic lithium[135] indicated in Li 1s spectra. About LiOH, it has been identified with the position of the highest binding energy peak of O 1s, which corresponds to that of LiOH according to literature[133]. Both Li2O and LiOH are fitted using just one peak in Li 1s spectra, called Li+. Carbon contamination has contribution from aliphatic carbon C-H/C-C and a higher binding energy carbon that can be correlated to C-O bond[176], both typical from adventitious carbon[159]. A small amount of graphitic carbon (C=C) appears also at lower binding energies than adventitious carbon[177]. According to the surface compounds quantification represented in Figure 4.5b, if 1 oxygen atom is assigned to each C-O species from C contamination, C-O based oxygen only represent 1.1% of total oxygen atoms. Due to this, we neglect its contribution in O 1s spectra. 92 ‖ 4. Study of Li carbonate evolution on Li metal surface Binding energy of aliphatic carbon (around 288 eV, Figure 4.5a) is higher than that from previous chapter (285 eV, Figure 3.11). A study that focuses on the correct assignment of binding energies in lithium foil mentions the existence of two types of CH/CC: one from the bulk around 285 eV, and another one at 3 eV higher than that one, which comes from the surface[178]. Then, CH/CC measured in this chapter will correspond to surface CH/CC, which agrees with the higher surface sensitivity photon energies of this chapter. To confirm that the proposed binding energy based on Li0 position is feasible, we measured the Fermi edge region of this same sample by APXPS. Figure 4.6 shows that the Fermi edge lies at 0 eV, as expected for metallic samples. Then, we can assume calibration based on Li0 is adequate. Figure 4.5. a) APXPS spectra of Li 1s, O 1s, and C 1s core levels of a UHV scraped lithium foil measured at same kinetic energy, which allows to use the areas of the photoelectron peak to quantify the surface composition. In the spectra, the fitted curve (black line) follows experimental data (dots), and background is represented by a dashed line. b) First 3 nm surface composition of a UHV cleaned commercial lithium foil. 4.3 Li foil surface cleaning ‖ 93 Figure 4.6. APXPS spectra of fermi edge region on a UHV scraped lithium foil measured at a photon energy of 280 eV. The position of the fermi edge is at 0 eV, as should be for a metallic sample. According to the surface compounds quantification represented in Figure 4.5b, surface is dominated by Li0 and Li2O. This quantification corresponds to the first 3 nm of the surface, since probing depth can be estimated as 3 times inelastic mean free path () of electrons. In order to obtain the in-depth distribution of Li species in the surface, we measured Li 1s spectrum at several photon energies. Figure 4.7 shows that when surface is measured at the highest photon energies, oxidized lithium intensity decreases compare to that of Li0, meaning oxidized layer is on top of Li0 substrate. In this same figure, probing depth of each photon energy is also indicated. To have an estimation of the thickness of oxidized overlayer, SESSA (NIST Database for the Simulation of Electron Spectra for Surface Analysis) software was used. Considering a photon energy of 600 eV and instrument settings of the spectrometer from beamline 9.3.2 at the Advance Light Source synchrotron, several Li 1s spectra were simulated for the following system: Li2O layer on top of a Li0 substrate, where the variable is the thickness of Li2O overlayer. With these spectra a correlation was obtained between the intensities of Li0 and Li2O measured at a photon energy of 600 eV, from Li 1s core level. This correlation is illustrated in Figure 4.8 and represented in the following equation:     = 59 . 9 − 13 . 1 l n          +      (4.5) 94 ‖ 4. Study of Li carbonate evolution on Li metal surface where  is the thickness of Li2O overlayer (Å) and  are the intensities of Li2O and Li0 peaks from Li 1s measured at a photon energy of 600 eV. With this equation we can directly obtain the thickness of the overlayer, using the intensities of Li 1s Li0 and Li2O measured at 600 eV. Figure 4.7. Li 1s APXPS spectra collected in UHV at different photon energies to illustrate the depth profiling of a clean Li surface. In the spectra, the fitted curve (black line) follows experimental data (dots), and background is represented by a dashed line. The approximated probing depth (3 times the inelastic mean free path) of each measured photon energy and that of Mg source are indicated in the figure. The mean oxide layer thickness present on UHV cleaned lithium surfaces is also specified in the right side of the figure. The oxide layer on top of Li 1s of our UHV cleaned lithium foil is formed by both Li2O and LiOH, as shown in Figure 4.5. If we want to use the above-mentioned equation to calculate the thickness of the overlayer that is attenuating Li0 intensity in Li 1s core level, we are assuming that all Li+ is related to Li2O. However, according to Figure 4.5b, Li2O accounts for the 60% of oxidized lithium. Furthermore, if we compare the  of electrons coming from Li0 measured at a photon energy of 600 eV,  of electrons through Li2O layer is 15.72 Å, and through 4.4 Li2CO3 evolutionon Li metal surface ‖ 95 LiOH layer is 16.28 Å. This similar  makes the attenuation that both compounds produce on Li metal surface electrons comparable. Then, we consider the use of equation (4.5) to calculate the thickness of the oxidized layer on top of lithium metal is adequate. The calculated average thickness of oxidized layer of the several clean surfaces studied on this chapter is 1.6 ± 0.9 nm, which is indicated in Figure 4.7. Figure 4.8. Relation between the Li0 and Li2O compound intensities of Li 1s and the thickness of Li2O overlayer, calculated by SESSA software. Simulation considers a photon energy of 600 eV and instrument settings of beam line 9.3.2 from the Advanced Light Source. 4.4 Li2CO3 evolution on Li metal surface For the study of Li2CO3 growth on Li metal, three CO2 gas (5.0 research purity from Praxair) pressures were considered: 0.1 mTorr, 10 mTorr and 400 mTorr. Gas was dosed for around one hour at each pressure. Every gas pressure dose started with a UHV cleaned Li foil. 4.4.1 Evolution of carbon-based compounds In order to study the evolution of carbon while treating Li with CO2 gas, first of all, the compounds present on C 1s spectrum need to be defined. In this spectrum, binding energy of lithium carbonate is clearly defined at 292.70 eV (Table 4.1). Binding energy of CH/CC from adventitious carbon has also been already found to 102 ‖ 4. Study of Li carbonate evolution on Li metal surface Figure 4.14. APXPS C 1s spectra of a lithium foil after 10 mTorr CO2 gas treatment for about one hour, deconvoluted with two types of fittings. Measurement was done at a photon energy of 600 eV in UHV condition. All carbon-based compounds have been identified. In the spectrum, the fitted curve (black line) follows experimental data (dots), and background is represented by a dashed line. 4.4.2 Li2CO3 growth kinetics When analyzing in more detail the evolution of Li2CO3 represented inFigure 4.15, we can observe that Li2CO3 growth is linear-parabolic, resembling Deal Groove oxide thick growth model[189]. In this type of growth two regimes are distinguished: a reaction limited regime and a diffusion limited regime, where the layer formed in the initial region is responsible for the diffusion limitation. The linear regime is defined by the following growth rate:  −   =    (4.6) And the parabolic regime is:   −    =    (4.7) where  is the thickness of the layer at time ,  the initial thickness of the layer, and  and  are the linear and parabolic reaction rate constant respectively. Both reaction rate constants at each pressure are calculated based on the Li2CO3 area of normalized C 1s spectra measured at a photon energy of 600 eV, which is related to the thickness of the layer. Figure 4.16 shows that, in the linear regime, 4.4 Li2CO3 evolutionon Li metal surface ‖ 103 pressure plays a noticeable role in the reaction rate. However, in the case of parabolic regime, the reaction rate at 10 mTorr CO2 and 400 mT CO2 is very similar. This indicates that diffusion throuhg the layer of Li2CO3 formed at the linear regimes of these pressures is limiting the reaction more than the incoming CO2 molecules. Figure 4.15. In the Li2CO3 area evolution (from C 1s APXPS spectra measured at 600 eV) of a lithium foil dosed by three pressures of CO2 gas, two regimes with different reaction rates can be distinguished. Figure 4.16. Reaction rate constant of Li2CO3 growth on Li metal in linear and parabolic regime at three pressures. Reaction rates have been calculated from the areas of Li2CO3 in C 1s APXPS spectra, measured at a photon energy of 600 eV. 104 ‖ 4. Study of Li carbonate evolution on Li metal surface 4.4.3 Depth profiling of lithium-based compounds An important advantage of performing APXPS measurements in a synchrotron radiation facility is the ability to perform non-destructive depth profiling experiments. To obtain the information of compounds distribution along the surface depth, same surface is measured using several photon energies. Figure 4.17 shows the final surface of Li foil treated at three CO2 gas pressures, measured at three different photon energies. Figure 4.17. APXPS Li 1s spectra showing depth profiling of lithium foil treated with three pressures of CO2 gas, measured at UHV condition. In the spectra, the fitted curve (black line) follows experimental data (dots), and background is represented by a dashed line. 4.4 Li2CO3 evolutionon Li metal surface ‖ 105 According to Table 4.1, Li0, Li2O and Li2CO3 compounds are easily identified in lithium 1s spectra. But, as just concluded analyzing C 1s spectra evolution, Li2C2O4 is also present on the surface. We couldn´t find any XPS binding energy reference for Li2C2O4 in Li 1s spectra. However, reported binding energies of ROCO2Li and Li2C2O4 in C 1s core level spectra are very similar[180,185]. Then, we consider that both Li 1s from Li2C2O4 and ROCO2Li will also have similar binding energies. That of Li 1s in ROCO2Li is 0.2 eV lower than Li2CO3[180]. Because of the small difference between both binding energies, we fit the spectra coupling contributions of Li2CO3 and Li2C2O4 in one peak. Binding energy of this peak lies between 58.1-57.7 eV. FWHM constrains will be 0.2 eV higher than that of Li2CO of Li 1s from Table 4.1 to account for the two types of Li with slightly different binding energies. Looking to Figure 4.17, the higher the photon energy (higher probing depth), the higher the intensity of Li2O. This behavior is the same for the three pressures, suggesting that the layered surface structure is: Li0 on the bottom, an intermediate Li2O layer, and a topmost surface layer with both Li2CO3 and Li2C2O4. In order to check whether actually electrons from Li2O layer are being attenuated by an overlayer, we analyze the attenuation of Li2O photoelectron intensity for the same sample measured at different photon energies. According to equation (2.8), if an overlayer is covering a substrate, intensity should obey an exponential decay when substrate electrons have lower inelastic mean free path, which happens at different photon energies. To be able to compare the intensities measured at different photon energies, according to equation (2.7), intensity has to be corrected by the photon flux (Table 4.2), cross section (Table 4.2) and the inelastic mean free path, (Table 4.4). Figure 4.18 shows Li2O photoelectron intensity is attenuated following an exponential decay at all three pressures, confirming Li2O is covered by an overlayer. The fitting equations for each pressure are the followings: 0 . 1     →      = 3 . 5 · 1 0     .  /  (4.8) 0     →      = 4 . 9 · 10     .  /  (4.9) 400     →      = 4 . 0 · 10     .  /  (4.10) 106 ‖ 4. Study of Li carbonate evolution on Li metal surface where I indicates the corrected intensity of Li2O photoelectron peak form Li 1s core level,  is the inelastic mean free path of Li2O electrons passing through the overlayer of Li2CO3 and Li2C2O4, and 11.1, 16.8 and 20.6 parameters are the overlayer thickness (Å) of Li foil dosed for about one hour at 0.1, 10 and 400 mtorr CO2, respectively. Figure 4.18. Evolution of the corrected intensities of Li 1s Li2O photoelectron peak spectra measured at three photon energies: 280 eV, 510 eV and 750 eV. Electrons have a different inelastic mean free path () in each photon energy. Considering this sequence and the intensities of the compounds from Figure 4.17, thicknesses of surface layers are calculated using equations (2.11) and (2.12) and the parameters from Table 4.3 and Table 4.4. For the carbonaceous layer attenuating Li2O intensity, atomic density and inelastic mean free path are calculated accounting each contribution of Li2CO3 and Li2C2O4. This contribution is given by the area ratios of them in C 1s spectra of dosed surfaces, measured at a photon energy of 600 eV in UHV condition. Figure 4.19 shows layered sketches of final lithium surface, where contribution of each lithium carbonaceous compounds is also indicated. Thickness values are average thickness calculated by each photon energy, and the deviation between the photon energies is indicated by error bars. Obtained Li2CO3/Li2C2O4 layer thicknesses represented in Figure 4.19 are in good agreement with that ones from equations (4.8), (4.9) and (4.10). 4.4 Li2CO3 evolutionon Li metal surface ‖ 107 In Figure 4.19, we observe the higher the gas pressure the larger the Li2CO3 and Li2C2O4 thickness. Li2O also evolves considering initial clean Li surface has an oxidized layer of 16 ± 9 Å thickness, according to Figure 4.7. Li2O as a consequence of this interaction was already observed in previous chapter (section 3.4.2). However, Li2O layer thickness is almost the same for the three pressures, so the gas pressure is not playing a role in the growth of Li2O. Figure 4.19. Layered sketches of lithium surfaces after exposing them to CO2 gas at three pressures for about an hour. Estimated thicknesses of the overlayers and composition of each layer are indicated in the figure for the first 100 Å of the surface. It is worth mentioning that in the layered model we are not considering CH/CC, CO and C=O contributions. These compounds will produce an extra attenuation in Li 1s electrons. But we assume this attenuation to be the same for all Li 1s compounds, so the ratios between Li 1s compounds should not be affected by them. 4.4.4 Insights into the reaction mechanism So far, our analysis reveals that carbonate, oxalate, CH/CC and Li2O evolve on the surface upon Li interacting with CO2 gas. Furthermore, we observe that Li2O lies between Li0 and carbonaceous compounds. In order to explain this surface evolution, and based on proposed reaction mechanisms[137,187], we suggest a possible reaction pathway, represented in Figure 4.20. According to this mechanism, CO2 reaction of Li0 sites leads to Li2O and CO. Depending on the availability, this CO could be adsorbed in both Li0 and Li2O sites. In the first case (pathway A in Figure 4.20), the reaction pathway followed to form carbonate will be the one postulated by Zhuang et al.137, where more Li2O will be created on the surface, and the oxide will then react with CO2 to form carbonate. In the second case (pathway B in Figure 4.20), oxalate will be formed, which will 108 ‖ 4. Study of Li carbonate evolution on Li metal surface act as an intermediate to create carbonate and more CO, as claimed earlier186. This second pathway is a self-recycling process due to the continuous CO evolution. Of the two possible reaction pathways, we think B is favored. Our reasoning relies on the small C increase, compared to that of carbonate, that occurs on the surface, inferred from Figure 4.12 when we compare Li2CO3 and CH/CC evolutions. Another conclusion that we can draw from the reaction mechanism is that pathway B requires metallic lithium to be accessible on the surface to create the required CO for the formation of oxalate. In other words, if the surface were completely oxidized, no oxalate would evolve, and carbonate would dominate the surface. Figure 4.20. Lithium carbonate growth reaction mechanism. 4.4.5 O2 gas effect on Li2CO3 growth To further explore the system, CO2 gas was codosed with O2 gas (5.0 research purity from Praxair). Starting from a UHV cleaned Li foil, we first added 0.1 mTorr CO2 gas and after 10 minutes, O2 gas was also added to the system for one hour. When doing so, Li2CO3 growth is promoted, changing the growth rate of the reaction if it is compared with a pure CO2 reduction, as shown in Figure 4.21. To further study the effect of the oxygen, we repeated the codose experiment but inversing the order of adding the gases. In this second codose, first we added O2 gas, and then CO2 gas. If we compare the final C 1s spectra and the evolution of Li2C2O4 for both co doses (Figure 4.22), when O2 gas is added first, no Li2C2O4 is evolved, and the final surface is pure carbonate without even adventitious carbon. In Figure 4.22 we also clearly see how the addition of O2 gas is changing the mechanism to form Li2CO3 in the first co-dose studied, where Li2C2O4 stops evolving. 4.4 Li2CO3 evolutionon Li metal surface ‖ 109 Figure 4.21. Evolution of raw area of Li2CO3 from C 1s APXPS spectra, measured at a photon energy of 600 eV and at two different dose conditions. Figure 4.22. Evolution of raw areas of Li2C2O4 from C 1s spectra measured at a photon energy of 600 eV, at two CO2 and O2 gas codoses experiments. In the first codose CO2 gas is added first (orange) and in the second co dose O2 gas is added first (green). Final C 1s spectra are also indicated in the right side of the figure for each co dose. In the spectra, the fitted curve (black line) follows experimental data (dots), and background is represented by a dashed line. 110 ‖ 4. Study of Li carbonate evolution on Li metal surface This behavior indicates that the reaction mechanism to create lithium carbonate is bypassed when the atmosphere is comprised of both CO2 and O2; this has been schematically represented in Figure 4.23. Due to the interaction of oxygen with lithium, lithium oxide is created on the surface, and there is no metallic lithium accessible to react with CO2, thus no oxalate evolves on the surface, further supporting the proposed reaction mechanism of Figure 4.20. Figure 4.23. Effect of O2 gas in lithium carbonate growth reaction mechanism. Another consequence of adding the gases in a different order is the rate of lithium oxidation, as can be inferred from Figure 4.24 by the fast disappearance of surface Li when adding first O2. It is worth mentioning that oxalate contribution has not been taken into account considering its small contribution compare to that of Li2CO3 (Figure 4.22). The faster oxidation of the surface with O2 agrees with the conclusion obtained in previous chapter: O2 gas oxidation rate is higher than that of CO2 gas (section 3.4). This behavior implies that, if lithium is pretreated with CO2 and then O2 gas is added, oxidation is slower and more metallic lithium will be available near the surface region, as shows Figure 4.24. We calculate the evolution of the thickness of Li2O and Li2CO3 for the first co dose, the one started with CO2 gas. For that, Li 1s 600 eV function of time spectra compounds intensities, equations (2.11) and (2.12) and parameters from Table 4.3 and Table 4.4 are used. Results are summarized in Figure 4.25. There, we observe that, when CO2 gas contacts Li0 surface, both Li2CO3 and Li2O are formed with a high reaction rate. When adding O2 gas to the system (green region in Figure 4.25), Li2O keeps constant, meaning all O2 is used to form Li2CO3, further affirming bypassed reaction represented in Figure 4.23. 4.4 Li2CO3 evolutionon Li metal surface ‖ 111 Figure 4.24. APXPS Li 1s spectra evolution while codosing a lithium surface with CO2 and O2 gases, changing the order of adding the gases. Spectra is measured at a photon energy of 600 eV. In the spectra, the fitted curve (black line) follows experimental data (dots), and background is represented by a dashed line. 118 ‖ 5. Li thin film growth The process of growing a thin film by deposition has six steps: first the arriving atoms have to adsorb on the surface, then they diffuse some distance, after that a reaction of the adsorbed species with each other and with the surface occurs to form the bonds of the film. The fourth step is the nucleation, the initial aggregation of the film material, and then the structure develops. Finally, diffusional interactions occur with the bulk of the film and with the substrate[193]. Following, we are going to summarize the possible structural development morphologies, which will be useful to compare with the morphologies obtained experimentally later in the chapter. 5.1.1 Structure Development of a thin film There are three basic structural zones that depends on the ratio between the substrate temperature (Ts) and the melting point of the film (Tm), all of them illustrated in Figure 5.1. Z1 occurs when Ts/Tm is so low that surface diffusion is negligible. Columns of Z1 have poor or none crystallinity and are separated by voids. In Z2, when Ts/Tm is higher than Z1, surface diffusion is significant and the structure consist of columns having tight grain boundaries between them. Crystalline columns are less defected than Z1 and are often facetted at surface. In Z3, due to the higher temperature of the substrate compared to previous zones, we can consider bulk annealing of the film is taking place during deposition. This is characterized by more isotropic or equiaxed crystallite shape. There is an extra zone between Z1 and Z2 called the transitional zone (ZT), which contains similar columns to those of Z1 but voids and domes are absent, and is usually associated with energy enhanced processes as sputter deposition. Sometimes, anomalous structure forms occur, in particular the whiskers, illustrated also in Figure 5.1. Figure 5.1 a) Characteristic cross section of the three basic structure zones when developing a thin film. Ratio of substrate temperature to film melting option increases from left to right. b) Whiskers anomalous structure formation. Adapted from[193]. 5.2 Experimental procedure ‖ 119 5.2 Experimental procedure Figure 5.2 shows the UHV chamber system designed to evaporate Li. Load lock of the system is compatible with the air sensitive portable transfer arm from Figure 2.8, thus air exposure of the samples is prevented. Base pressure of the system is low 10-8 – high 10-9 mbar. Figure 5.2. a) Front view and b) side view of the UHV chamber system where Li evaporations were performed. Li sources used in this study are commercially available chromate-free metal vapor sources (alvasources, from alvatec[97]). They contain an intermetallic compound in argon atmosphere inside a small stainless-steel tube, sealed with indium, as shown in Figure 5.3. Capacity of the sources is 190 mg and diameter of the tube 5 mm. Lithium metal is thermally evaporated from the intermetallic compound when passing a current through the contacting flaps. To do that, the contacting flaps were welded to two conductive rods connected to the feedthrough of Figure 5.2b. Fundaments of thermal evaporation are explained in chapter 2 section 2.1.1 section. In the activation of the sources, indium seal is removed in UHV conditions. According to the information provided by the supplier, indium melts from 1.5 A to 4 A, which causes the release of the argon, increasing the pressure of the chamber. Experimentally, we observed this release at an intensity of 5.3 A. Besides, in order 120 ‖ 5. Li thin film growth to remove all the In from the source, we kept it at a higher intensity for about 3 hours. Once the source was activated, we never exposed it to air atmosphere. Figure 5.3. Configuration of a typical alkaline source from alvatec. Adapted from[97]. Steps followed to perform an evaporation were: 1) Substrate cleaning step. This step was performed in an ultrasonic bath, with the following sequence: first acetone, then ethanol and finally water, 10 minutes in each one. After that, substrate was dried in an oven at 80 °C overnight. 2) Substrate loading step. Substrate was load in the load lock and left in vacuum at least 12 hours. During this step, a titanium sublimation pump was run to help recover the base pressure of the chamber. 3) Removal of impurities step. In this step, explained in section 2.1.1, volatile impurities were removed. For that, sample was isolated from the Li source closing the corresponding gate valve, and intensity was increased up to 3 A and kept it for half an hour. 4) Li source stability step. Intensity passing through the source was increased up to the evaporating value, which will be higher than 5 A for our commercial sources according to supplier’s manual[97]. We kept that intensity until pressure was stable. Pressure should be in the range of 10-7–10-8 mbar. 5) Evaporation step. Substrate was moved to the front of the lithium source, opening the corresponding gate valve. Sample was kept there the desired evaporation time. 5.3 Study of lithium thin film deposition ‖ 121 5.3 Study of lithium thin film deposition 5.3.1 Li source deposition rate calculation at 8 A The deposition rate of the Li thin film is defined by the intensity applied to the Li source, so each intensity will have a specific deposition rate. Aim of this section is to determine the deposition rate at an intensity of 8 A. Thickness of thin films were estimated by scanning electron microscope (SEM), using the instrument detailed in section 2.3.4.1, and measuring secondary electrons at 30 kV. Evaporated samples were cut in the glove box using a diamond scribe and transferred to the SEM with the air sensitive transfer arm from the SEM system (Figure 2.14). There, measuring a cross section, thickness of the deposited layer was obtained, and the top view images provided information about the morphology of the lithium. In order to measure the thickness from a cross section image, a substrate which will be easy to cut and will produce a sharp edge is needed, as silicon wafer is, common substrate for thin film depositions. We then started the study with a silicon monocrystalline wafer (<100>, Bo additive, ρ > 1 Ω·cm, Virginia semiconductors). When characterizing the evaporation of lithium in Si wafer some singularities were found, which led us to study the evaporation process in several layer sequences, all of them indicated in Table 5.1. Following we are going to explain the conclusions obtained in each sequence. Table 5.1. Sequences used to study the evaporation and growth morphology of lithium. Lithium layers are evaporated at the current and times specified in the table and Ti is sputtered using a magnetron sputtering. Sequence Substrate Layer 1 Layer 2 Layer 3 A Si waferILi 8 A, 12 h B Si waferILi 8 A, 6 h C Si waferILi Ti 8 A, 23 h 340 nm D Stainless steelII Li Ti 8 A, 23 h 340 nm E Si waferITi Li Ti 255 nm 8 A, 24 hours 680 nm I Si wafer <100>, Bo additive, ρ > 1 Ω·cm II Stainless steel from the sample holder 122 ‖ 5. Li thin film growth In order to make sure that the source is evaporating lithium, first evaporation (sequence A from Table 5.1) was characterized using X-ray photoelectron spectroscopy instrument explained in section 2.3.1.3. Figure 5.4 shows that, after an evaporation of 8 A for 12 hours in a Si wafer, the surface is covered by a layer that contains mainly lithium and oxygen. Figure 5.4. XPS survey spectra measured by Mg source of a silicon wafer after lithium evaporation, following sequence A from Table 5.2. The normalized surface atomic concentration is indicated also in the figure. The inset represents Li 1s region measured at higher resolution with same photon source. Apart from the expected adventitious carbon impurity, there is also a small amount of fluorine and sulfur on the surface. These elements are a cross contamination because of some impurity we had at that time in the XPS instrument, and both represents around 5.5% of the normalized atomic concentration from the surface. For the calculation of this concentration we considered the areas of the peak of the elements from the survey of Figure 5.4. Areas were corrected with the corresponding relative sensitive factor of each element based on Scofield cross sections and a transmission function correction 5.3 Study of lithium thin film deposition ‖ 123 specific for this equipment. An exponential factor was also used to correct the difference in the inelastic mean free path of electrons. With this measurement, it was confirmed that source evaporated just Li. Figure 5.5 shows the cross section and top view of this sample, where the expected thin film is not observed. Indeed, the surface is full of microstructured features, as large as 3 micrometers. Figure 5.5. Air sensitive SEM a) cross section and b) top view of a lithium evaporation on a monocrystalline Si wafer following sequence A from Table 5.1. Looking to the weird Li deposition obtained on Si, the state of Si wafer used in this study was checked. Figure 5.6a shows that the cut made to measure cross section using a diamond scribe was not responsible of the microstructured features. In Figure 5.6b it can be observed that initial Si wafer surface was flat, so Li morphology was not induced by an inhomogeneous substrate. Figure 5.6. SEM images of the a) cross section and b) top view of the monocrystalline silicon wafer used to evaporate lithium. Then, to have a better understanding of how the lithium was growing on the silicon wafer, a new Li evaporation was performed, shorten the evaporation time 124 ‖ 5. Li thin film growth (sequence B from Table 5.1). Surface of sequence B, represented in Figure 5.7, shows the same non-uniform surface got in previous evaporation. In addition, when trying to get images of the lithium surface structures at higher magnification than the ones from Figure 5.5, Li was degraded. This degradation, a consequence of the interaction of Li metal with SEM electron gun, took place when measuring with magnifications that show a scale equal or smaller to 2 m (magnification of x50000). Figure 5.7b clearly shows how the area where SEM images are taken is completely different from the rest of the surface. Figure 5.7c also confirms this behavior, showing the disappearance of a lithium feature from the surface just by trying to focus the area with SEM microscope. Time between each image is the one needed to save the picture, around 5 seconds. Figure 5.7. Air sensitive SEM a) cross section and b) top view of a lithium evaporation on a monocrystalline Si wafer following sequence B from Table 5.1. c) Evolution of a surface Li feature from the cross section exposed to SEM electron gun. Time between each picture is around 5 seconds, the one needed to save the images. To avoid the problem of lithium degradation, in the following deposition (sequence C from Table 5.1) evaporated Li was covered with a Ti layer, deposited by magnetron sputtering. Details about the technique and equipment can be found in chapter 2 section 2.1.2. Optimal sputtering conditions and deposition 5.3 Study of lithium thin film deposition ‖ 125 rates of Ti were obtained by other researchers of the Advanced Interface Analysis group from CIC Energigune. The same air sensitive transfer arm used to move samples from Li evaporation UHV chamber system is compatible with the sputtering system, preventing sample air exposure. Li evaporation time in the sequence C was higher (24 hours) than previous cases in order to have better sense of the structural development of the lithium. When measuring the cross section and top view of sequence C, represented in Figure 5.8, we clearly observe that lithium was not depositing as a uniform layer. Figure 5.8. Air sensitive SEM a) cross section, b) top view and c) tilted view of a lithium evaporation following sequence C from Table 5.1, where a Ti layer is added to the lithium evaporated on Si wafer. This type of structural development does not correspond to any of the basic zones explained in Figure 5.1a. In fact, this is an anomalous type growth, similar to the whiskers of Figure 5.1b, that sometimes happens when trying to obtain a thin film[194]. About the effect of covering Li with Ti to prevent its degradation, it was been effective. In this sample we were able to go to high magnifications without damaging the surface, which can be seen if Figure 5.7b and Figure 5.8b are compared 126 ‖ 5. Li thin film growth Because of the configuration of the evaporation system, we were also evaporating lithium and titanium on the holder, not just on the Si substrate. This holder is made of stainless steel, so we have a new deposition sequence here, the one called sequence D in Table 5.1.Figure 5.9 shows the comparisons of the surface morphology of lithium deposited on the Si and Li deposited on the stainless-steel (SS) holder. There, it can be concluded that whisker type growth is related to the nature of the substrate. For this reason, in next evaporation a Ti layer was added between the Li and Si wafer, a metal considered a good bonding material[193]. This deposition, called sequence E, is summarized in Table 5.1. Figure 5.9. Air sensitive top view SEM of a) sequence C and b) sequence D from Table 5.1, where the difference between the sequences relies on the substrate, Si wafer and stainless steel (SS) respectively. Figure 5.10 shows the surface morphology of several steps from sequence E. The morphology of the deposited Ti (Figure 5.10a, b) is very similar to that one from literature for similar deposition conditions[195]. About the morphology of the evaporated lithium into Ti surface (Figure 5.10c), it is more similar to the evaporated lithium into stainless steel (Figure 5.9b) than to the evaporated lithium into Si wafer (Figure 5.9a). Here, according to Figure 5.10e, we also observe that the evaporation of Ti on top of lithium is not changing the morphology of lithium. When analyzing the morphology of the evaporated Li from Figure 5.10b, we can observe some voids on the surface. This kind of morphology corresponds to a Z1 structural development, explained in the introduction section and schematically represented in Figure 5.1a. In this type of development surface diffusion is neglected. 5.3 Study of lithium thin film deposition ‖ 127 In the cross section images of sequence E, represented in Figure 5.11, the layer of lithium and the two Ti layers can easily be identified. With this information, the deposition rate of Li source at 8 A was estimated to be in the range of 120 – 400 nm/h. Figure 5.10. Air sensitive top view SEM images and pictures of the surfaces made while following sequence E: a) sputtered Ti on Si wafer, b) same as a) with higher magnification, c) evaporated lithium on Ti, d) photo of evaporated lithium on Ti, e) sputtered Ti on the evaporated Li and f) photo of sputtered Ti on the evaporated Li.