ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2007 Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 375 Photoelectron Spectroscopy on Atoms, Molecules and Clusters The Geometric and Electronic Structure Studied by Synchrotron Radiation and Lasers TORBJÖRN RANDER ISSN 1651-6214 ISBN 978-91-554-7047-0 urn:nbn:se:uu:diva-8343
82
Embed
Photoelectron Spectroscopy on Atoms, Molecules …171152/FULLTEXT01.pdf · Photoelectron Spectroscopy on Atoms, Molecules and Clusters The Geometric and Electronic Structure Studied
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ACTA
UNIVERSITATIS
UPSALIENSIS
UPPSALA
2007
Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 375
Photoelectron Spectroscopy onAtoms, Molecules and Clusters
The Geometric and Electronic Structure Studied bySynchrotron Radiation and Lasers
“Och kallas därför höga bergen hos oss fjäll, av orsak att där oppå inga trän
eller örter växa, utan de är glatta och bara såsom fjäll uppå en fisk”
– Olof Rudbäck d. ä.
List of Papers
This thesis is based on the following papers, which are referred to in the text
by their Roman numerals.
I Experimental evidence for molecular ultrafast dissociation inO2 clustersT. Rander, M. Lundwall, A. Lindblad, G. Öhrwall, M.
Tchaplyguine, S. Svensson, O. Björneholm
Eur. Phys. J. D 42, 253-257 (2007)
II Suppression of the ultrafast dissociation of bromomethanemolecules in clustersT. Rander, A. Lindblad, A. Rosso, H. Bergersen, G. Öhrwall, S.
Svensson, O. Björneholm
In manuscript
III Core-level electron spectroscopy on the sodium dimer Na 2plevelT. Rander, J. Schulz, M. Huttula, A. Mäkinen, M. Tchaplyguine,
S. Svensson, G. Öhrwall, O. Björneholm, S. Aksela, H. Aksela
Phys. Rev. A 75, 032510 (2007)
IV The diffusion behavior of O2 doped on large Ar clustersT. Rander, A. Lindblad, M. Lundwall, M. Tchaplyguine, G.
Öhrwall, S. Svensson, O. Björneholm
Submitted to J. Chem. Phys.
V The far from equilibrium structure of argon clusters dopedwith krypton or xenonA. Lindblad, H. Bergersen, T. Rander, M. Lundwall, G. Öhrwall,
M. Tchaplyguine, S. Svensson, O. Björneholm
Phys. Chem. Chem. Phys. 8, 1899-1905 (2006)
5
VI The role of molecular polarity in cluster local structurestudied by photoelectron spectroscopyA. Rosso, T. Rander, H. Bergersen, A. Lindblad, M. Lundwall,
S. Svensson, M. Tchaplyguine, G. Öhrwall, L. J. Sæthre, O.
Björneholm
Chem. Phys. Lett 435, 79-83 (2007)
VII Shakedown in core photoelectron spectra from alignedlaser-excited Na atomsJ. Schulz, M. Tchaplyguine, T. Rander, O. Björneholm, S.
Svensson, R. Sankari, S. Heinäsmäki, H. Aksela, S. Aksela, E.
Kukk
Phys. Rev. A 72, 10702-1-4 (2005)
VIII Characterization of weakly excited final states by shakedownspectroscopy of laser-excited potassiumJ. Schulz, S. Heinäsmäki, R. Sankari, T. Rander, A. Lindblad,
H. Bergersen, G. Öhrwall, S. Svensson, E. Kukk, S. Aksela, H.
Aksela
Phys. Rev. A 74, 12705-1-6 (2006)
Reprints were made with permission from the publishers.
6
The following is a list of papers to which I have contributed, but which are not
included in this thesis.
Observation of elastic scattering effects on photoelectron angulardistributions in free Xe clustersG. Öhrwall, M. Tchaplyguine, M. Gisselbrecht, M. Lundwall, R. Feifel,
T. Rander, J. Schulz, R. R. T. Marinho, A. Lindgren, S. L. Sörensen, S.
Svensson, O. Björneholm
J. Phys. B 36, 3937-49 (2003)
Femtosecond Interatomic Coulombic Decay in Free Neon Clusters: LargeLifetime Differences between Surface and BulkG. Öhrwall, M. Tchaplyguine, M. Lundwall, R. Feifel, H. Bergersen, T.
Rander, A. Lindblad, J. Schulz, S. Peredkov, S. Barth, S. Marburger, U.
Hergenhahn, S. Svensson, and O. Björneholm
Phys. Rev. Lett. 93, 173401 (2004)
Final state selection in the 4p photoemission of Rb by combining laserspectroscopy with soft-x-ray photoionizationJ. Schulz, M. Tchaplyguine, T. Rander, H. Bergersen, A. Lindblad, G.
Öhrwall, S. Svensson. S. Heinäsmäki, R. Sankari, S. Osmekhin, S. Aksela,
H. Aksela
Phys. Rev. A 72, 32718-1-4 (2005)
The electronic structure of free water clusters probed by Auger electronspectroscopyG. Öhrwall, R. F. Fink, M. Tchaplyguine, L. Ojamae, M. Lundwall, R. R.
T. Marinho, A. N. de Brito, S. L. Sörensen, M. Gisselbrecht, R. Feifel, T.
Rander, A. Lindblad, J. Schulz, L. J. Sæthre, N. Mårtensson, S. Svensson, O.
Björneholm
J. Chem. Phys 123, 54310-1-10 (2005)
Ioniclike energy structure of neutral core-excited states in free KrclustersS. Peredkov, A. Kivimäki, S. L. Sörensen, J. Schulz, N. Mårtensson, G.
Öhrwall, M. Lundwall, T. Rander, A. Lindblad, H. Bergersen, S. Svensson,
O. Björneholm, M. Tchaplyguine
Phys. Rev. A 72, 21201-1-4 (2005)
7
Postcollision interaction in noble gas clusters: observation of differencesin surface and bulk line shapesA. Lindblad, R. F. Fink, H. Bergersen, M. Lundwall, T. Rander, R. Feifel, G.
Öhrwall, M. Tchaplyguine, U. Hergenhahn, S. Svensson, O. Björneholm
J. Chem. Phys. 123, 211101-1-4 (2005)
Enhanced surface sensitivity in AES relative to XPS observed in free ArclustersM. Lundwall, M. Tchaplyguine, G. Öhrwall, A. Lindblad, S. Peredkov, T.
Rander, S. Svensson, O. Björneholm
Surf. Sci. 594, 12-19 (2005)
Photon energy dependent intensity variations observed in Auger spectraof free argon clustersM. Lundwall, A. Lindblad, H. Bergersen, T. Rander, G. Öhrwall, M.
Tchaplyguine, S. Peredkov, S. Svensson, O. Björneholm
J. Phys. B 39, 3321-33 (2006)
Shell-dependent core-level chemical shifts observed in free xenon clustersM. Lundwall, R. F. Fink, M. Tchaplyguine, A. Lindblad, G. Öhrwall, H.
Bergersen, S. Peredkov, T. Rander, S. Svensson, O. Björneholm
J. Phys. B 39, 5225-35 (2006)
Lineshapes in carbon 1s photoelectron spectra of methanol clustersM. Abu-samha, K. K. Børve, L. J. Sæthre, G. Öhrwall, H. Bergersen, T.
Rander, O. Björneholm, M. Tchaplyguine
Phys. Chem. Chem. Phys 8, 2473-82 (2006)
Preferential site occupancy of krypton atoms on free argon-clustersurfacesM. Lundwall, A. Lindblad, H. Bergersen, T. Rander, G. Öhrwall, M.
Tchaplyguine, S. Svensson, O. Björneholm
J. Chem. Phys 125, 14305-1-7 (2006)
Laser excitation combined with 2p photoionization and Auger decay ofpotassiumK. Jänkälä, R. Sankari, J. Schulz, M. Huttula, A. Calo, S. Heinäsmäki, S.
Fritsche, T. Rander, S. Svensson, S. Aksela, H. Aksela
Phys. Rev. A 73, 22720-1-8 (2006)
8
Magnetron-based source of neutral metal vapors for photoelectronspectroscopyM. Tchaplyguine, S. Peredkov, H. Svensson, J. Schulz, G. Öhrwall,
M. Lundwall, T. Rander, A. Lindblad, H. Bergersen, S. Svensson, M.
Gisselbrecht, S. L. Sorensen, L. Gridneva, N. Mårtensson, O. Björneholm
Rev. Sci. Instrum. 77, 033106 (2006)
5p photoemission from laser-excited cesium atomsJ. Schulz, M. Määtä, S. Heinäsmäki, M. Huttula, R. Sankari, E. Kukk, T.
Rander, S. Svensson, S. Aksela, H. Aksela
Phys. Rev. A 73, 062721 (2006)
Self-assembled heterogeneous argon/neon core-shell clusters studied byphotoelectron spectroscopyM. Lundwall, W. Pokapanich, H. Bergersen, A. Lindblad, T. Rander, G.
Öhrwall, M. Tchaplyguine, S. Barth, U. Hergenhahn, S. Svensson, O.
Björneholm
J. Chem. Phys. 126, 214706 (2007)
Synchrotron radiation study of chloromethane clusters: Effects ofpolarizability and dipole moment on core level chemical shiftsA. Rosso, A. Lindblad, M. Lundwall, T. Rander, S. Svensson, M.
Tchaplyguine, G. Öhrwall, O. Björneholm
J. Chem. Phys. 127, 024302 (2007)
Free nanoscale sodium clusters studied by core-level photoelectronspectroscopyS. Peredkov, G. Öhrwall, J. Schulz, M. Lundwall, T. Rander, A. Lindblad,
H. Bergersen, A. Rosso, W. Pokapanich, N. Mårtensson, S. Svensson, S. L.
Sorensen, O. Björneholm
Phys. Rev. B 75, 235407 (2007)
Direct observation of the non-supported metal nanoparticle electrondensity of states by X-ray photoelectron spectroscopy M. Tchaplyguine,S. Peredkov, A. Rosso, J. Schulz, G. Öhrwall, M. Lundwall, T. Rander, A.
Lindblad, H. Bergersen, W. Pokapanich, S. Svensson, S. L. Sorensen, N.
Mårtensson, O. Björneholm
Eur. Phys. J. D 45, 295 (2007)
9
Localized versus delocalized excitations just above the 3d threshold inkrypton clusters studied by Auger electron spectroscopyM. Tchaplyguine,A. Kivimäki, S. Peredkov, S. L. Sorensen, G. Öhrwall, J. Schulz, M.
Lundwall, T. Rander, A. Lindblad, A. Rosso, S. Svensson, N. Mårtensson, O.
Working in the field of experimental physics takes a lot of team-work, and
all of the papers in this thesis reflect this fact. Without the fruitful exchange
of ideas and without a helping hand running the experiments, not much of
the work presented here could have been performed. Common to all of the
included articles is that I have been actively involved in the experiments, and
in discussions regarding them. For the papers where I am the first author, I
was the main responsible for data analysis and manuscript preparation.
13
1. Populärvetenskaplig sammanfattning
Världen vi lever i, och allt som finns däri, består av atomer. En atom består av
en atomkärna, som är positivt laddad, och negativt laddade elektroner. Atom-
kärnan består i sin tur av ännu mindre enheter, så kallade neutroner och pro-
toner. De består i sin tur av ännu mindre byggstenar, men detta behöver vi
inte tänka så mycket på i vårt fall eftersom atomkärnan är väldigt liten (c:a
10000 gånger mindre än hela atoms storlek). I våra experiment kan vi där-
för behandla atomkärnan som en laddad punkt. Atomer hittar man ibland
tillsammans med andra atomer, i molekyler, kluster eller fasta material. Ett
kluster är en “liten” klump av atomer eller molekyler (3-50000 st), som sitter
ihop med varandra. Man brukar säga att ett kluster är ett system som länkar
samman förståelsen för egenskaperna hos en enskild atom eller molekyl, och
förståelsen för egenskaperna hos ett fast material. Denna avhandling handlar
om atomer, molekyler och kluster, och vad som kan hända när man blandar
dessa, eller utsätter dem för laserljus eller röntgenstrålning.
Egenskaperna hos en atom, en molekyl eller ett kluster bestäms av dess
geometriska struktur (hur atomer eller molekyler sitter ihop i en större en-
het) och av dess elektroniska struktur (hur elektronerna är fördelade kring en
atom, en molekyl eller i ett kluster). Vi har undersökt (med direkta metoder)
denna elektroniska struktur, och kan genom denna dra ganska långtgående
(indirekta) slutsatser också om den geometriska strukturen hos de studerade
systemen i vissa fall.
1.1 De olika experimenten
Grovt sett kan vi dela in denna avhandling i tre delar. De första tre inklud-
erade arbetena behandlar processer i molekyler, som sker när en molekyl be-
strålas med röntgenstrålning av varierande energi. I de två första arbetena har
vi undersökt hur molekylers beteende förändras då de sitter ihop med andra
molekyler av samma sort i ett kluster. Systemen vi har studerat är kluster av
tusentals syremolekyler och kluster av hundratals metylbromidmolekyler. Vi
har även studerat vad som händer när man bestrålar en mer exotisk molekyl,
nämligen natriumdimeren, Na2, utan att den befinner sig i ett kluster. Vi ville
ta reda på om molekylerna går sönder när de bestrålas, och i så fall om de går
sönder “lika mycket” och lika ofta när de befinner sig i kluster som när de
är i en gas. Att studera sådana system är intressant, eftersom det finns många
ställen där kluster och exotiska molekyler förekommer, men där man inte kan
15
komma åt att göra mätningar. Man tror till exempel att mycket av den absorp-
tion av den elektromagnetiska strålning som kommer från rymden görs av små
vattenkluster, och att mycket av den kemi som sker i atmosfären (till exempel
den s.k. Ozon-cykeln) katalyseras på olika sätt genom förekomsten av kluster.
Vi har också studerat den geometriska strukturen hos kluster av metylbro-
mid. Metylbromidmolekylen är vad man kallar en permanent dipol, vilket be-
tyder att många av elektronerna i molekylen finns vid en av dess ändar. Detta
gör att den får en elektriskt negativ laddning i änden där de flesta elektronerna
finns, och en positiv laddning där det finns färre elektroner. Detta fenomen
kallas för polarisation. Strukturen hos kluster som skapats genom att vi låtit
kluster krocka med med atomer och molekyler som då fastnat på eller i klus-
tret (så kallad “dopning”) har även den undersökts. Man kan tänka sig att detta
sker på ungefär samma sätt som om man skulle hälla russin på en stor klump
smör. Som figur 1.1 visar, så finns det två varianter av hur detta kan ske. Vilken
som inträffar beror på smörklumpens temperatur.
Figur 1.1: Den vänstra bilden visar vad som händer med russinen som hälls på smörsom man precis tagit ur kylskåpet. Den högra bilden visar vad som händer med russi-
nen om de hälls på smör som man värmt upp i en kastrull.
Om man precis har tagit ut smörklumpen ur kylskåpet så fastnar som vi alla
vet russinen på smörets yta, men om vi smält smöret innan vi häller russinen
på det så hamnar de inuti det som tidigare var smörklumpen. Ungefär samma
sak händer med klustren när de dopas, eftersom varje krock som klustret ut-
sätts för tillför det lite värme. Vi har mätt hur mängden kollisioner påverkar
strukturen hos dessa kluster.
Sist, men inte minst, har vi också mätt vad som händer med den elektroniska
strukturen hos två olika atomslag (natrium och kalium) som först belyses med
laserljus, och sedan bestrålas med röntgenstrålning. Sådana mätningar kan
ge värdefull information om hur den elektroniska strukturen i en atom eller
molekyl ser ut innan den belyses med laserljus, eftersom det inte alltid är helt
16
ökande våglängd ökande energi
joniserande strålningicke-joniserande strålning
Gamma
IR
Mikrovågor
Radio
Lågfrekvens UV
Synligt
Röntgen
Figur 1.2: Figuren visar olika våglängder av elektromagnetisk strålning, samt vad
dessa brukar kallas.
enkelt att studera detta på grund av diverse effekter som gör att elektronstruk-
turen blir “diffus”; somliga elektroniska tillstånd kan till exempel blanda sig
med varandra i atomens grundtillstånd, så kallad konfigurationsväxelverkan.
1.2 Experimentella metoder
En mängd kompletterande experimentella tekniker har använts för att genom-
föra de studier som presenteras i denna avhandling. De flesta av dessa kan
klassificeras som s.k. elektronspektroskopier, men även andra tekniker, så som
mass-spektroskopi och fluorescensmätningar har varit till hjälp under det ex-
perimentella arbetet.
1.2.1 Elektromagnetisk strålning
Egentligen är ljus och röntgenstrålning samma sak, och kallas tillsammans
med många andra former av strålning (ex. vis. radiovågor, gammastrålning
och mikrovågor) för elektromagnetisk strålning. Det som skiljer de två först-
nämnda åt är våglängden, alltså energi-innehållet hos var och en av dem. Rönt-
genstrålning innehåller mycket mer energi än vanligt ljus gör. Figur 1.2 visar
en bild av de olika våglängdsområdena för elektromagnetisk strålning.
I våra experiment används dels laserljus och dels röntgenstrålning för olika
ändamål. Laserljuset främsta användningsområde var att preparera exciterade
tillstånd i metallatomerna, medans röntgenstrålningens främsta uppgift var att
jonisera våra prov.
1.2.2 Elektronspektroskopi
Efter det att vi joniserat vårt prov med röntgenstrålning kan vi mäta en mängd
olika saker. En möjlighet är att mäta massan hos den kvarvarande jonen, för att
på så sätt få reda på om systemet vi joniserat har gått sönder i processen eller
17
ej. En annan möjlig mätning är att mäta energin hos den elektron som frigörs
från molekylen genom jonisationen. Det är detta, som kallas för fotoelektron-
spektroskopi, som vi oftast utför i våra experiment. Figur 1.3 visar en schema-
tisk bild över den elektroniska strukturen hos en atom. De energinivåer där det
finns, eller potentiellt kan finnas elektroner kallas för orbitaler.
E E
} Fylldavalensorbitaler
Tom valensorbital
Core-orbital
Grundtillstånd Exciterat tillstånd
Laserljus
Jon
isation
sgrän
s
Figur 1.3: En schematisk bild över elektronstrukturen hos en atom i sitt grundtillstånd,och i ett exciterat tillstånd.
Man kan även mäta så kallade Augerelektroner. Vid en röntgenjonisation
bildas en vakans i elektronstrukturen. Detta gör att elektroner i den yttre elek-
tronstrukturen känner en positiv laddning. Elektronerna kommer, eftersom de
själva har negativ laddning, attraheras mot denna vakans, och rusar dit för att
fylla den. Den energi som frigörs i en sådan process ges till en annan elektron
– Augerelektronen – som frigörs från systemet. Sluttillståndet i detta fall är ett
dubbelladdat tillstånd.
Det finns även en variation på samma tema, så kallad resonant Augerspek-
troskopi. Här föregås det s.k. Augersönderfallet inte av en jonisation, utan av
en excitation av en elektron till en tidigare tom elektronplats, som befinner
sig mycket långt ut i elektronstrukturen. Detta har ungefär samma effekt som
en jonisation, med skillnaden att den exciterade elektronen kan deltaga i det
påföljande Augersönderfallet. Man får alltså en jon med ett enkelladdat slut-
tillstånd. Resonant excitation försätter många molekyler i dissociativa till-
stånd, vilket innebär att molekylen går sönder. Detta kan ske mycket snabbt,
ofta inom ett fåtal femtosekunder. Om en molekyl går sönder så fort så hinner
inte Augerjonisationen hända innan molekylen har blivit till två fragment. Ef-
fekten av detta blir att man kan mäta Augersönderfallet från en av de två bitar
av molekylen som skapats, och på så sätt studera vilka fragment som bildas
och hur fort fragmenten fjärmar sig från varandra.
18
Vi använder även en teknik som kallas för röntgenabsorption, med vilken vi
studerar de tomma elektroniska orbitalerna hos atomer, molekyler och kluster.
1.3 Beräkningar
För att bättre förstå de saker vi ser när vi gör våra experiment använder vi
oss av datorberäkningar, med hjälp av vilka vi kan förutsäga både hur den
elektroniska och geometriska strukturen hos en molekyl eller i ett kluster ser
ut. Vi har använt oss av diverse beräkningsmodeller, dels av så kallade abinitio-beräkningar, som förutsäger elektronisk och geometrisk struktur genomkvantmekaniska beräkningar utan att några parametrar anpassas till experi-
ment. Ett kluster är ofta för stort för att kunna behandlas med kvantmekaniska
metoder. Vi använder därför även så kallade molekyldynamiksimuleringar.
I dessa beräkningar använder man en kombination av klassisk- och kvant-
mekanik för att förutsäga, till exempel, strukturen hos ett kluster.
1.4 Resultat
För de första tre arbetena som är inkluderade i denna avhandling har våra
experiment visat följande. Syremolekyler i kluster går fortfarande sönder i
samma omfattning som fria molekyler om de bestrålas med röntgenstrålning
av rätt energi. Metylbromidmolekyler i kluster, å andra sidan, verkar gå sönder
i mindre utsträckning när de bestrålas. Vi kan förklara detta genom att det i
metylbromidklustren verkar finnas vad som brukar kallas för “bandbildning”,
som är ett välkänt fenomen som förekommer i fasta material. För natrium-
molekylen har vi studerat hur avvikelsen från den atomära, sfäriska, geometrin
påverkar de energinivåer som uppkommer. Vi har också karaktäriserat den re-
pulsiva potential som uppstår då ett Augersönderfall sker i en sådan molekyl
som en Coulombpotential.
För de tre följande artiklarna har vi utfört experiment med syftet att stud-
era geometrin hos kluster av diverse atomer och molekyler. För kluster av
argon, dopade med syremolekyler, finner vi att syremolekylerna ligger kvar
på ytan fram tills dess att dopningsgraden (d.v.s. antalet kollisioner mellan
argonklustret och syremolekyler) är tillräckligt hög; då blir argonklustret fly-
tande och syremolekyler kan åka in i det (jfr. figur 1.1). En liknande situ-
ation uppkommer när vi studerar vad som händer då vi dopar argonkluster
med krypton eller xenon. I dessa två senare fall visar jämförelser med vad
som händer när man skapar blandade system genom så kallad “co-expansion”
(när man blandar de två gaserna och sen gör kluster av gasblandningen) att
dopningsproceduren tillåter skapandet av geometrier som är långt ifrån den
termodynamiska jämviktspunkten. Även detta är en indikation på att klustren,
vid låga dopningsgrader, inte har smält. Strukturen hos metylbromidkluster
19
studerades också. Vi kom fram till att strukturen hos dessa molekylära kluster
stämmer väl överens med den man hittar i den fasta fasen.
I de sista två artiklarna undersökte vi vad som sker när man joniserar atomer
som i initialtillståndet är exciterade. Vi använde ett lasersystem för att skapa
dessa exciterade tillstånd. En schematisk bild över hur detta går till kan ses i
figur 1.3. I den första av artiklarna kunde vi i natriumatomer konstatera att ex-
citeringen har som konsekvens att en magnetisk likriktning introduceras i sys-
temet. Vi kunde även observera en effekt som kallas för “shake-down”, ned-
skakning. I en sådan nedskakningsprocess inducerar jonisationen av atomen
en “skakning” i elektronstrukturen, som leder till att den i förväg exciterade
elektronen ger sin energi till den elektron som är på väg ut ur systemet. I och
med detta så ramlar den exciterade elektronen ner till det ställe den kom ifrån
från när den exciterades. Således får man i en nedskakningsprocess från en i
förväg exciterad atom samma elektroniska sluttillstånd som man får vid jon-
isation av en atom i sitt grundtillstånd, men med en elektron som kommer ut
från systemet med högre energi. Den extra energin motsvarar den energi som
tillförts atomen av laserljuset.
I den andra artikeln studerades kaliumatomer som också var exciterade med
hjälp av laserljus. Även här observerades nedskakning. Fenomenet utnyttjades
i detta fall till att karaktärisera de toppar som sågs i våra experimentella spek-
tra. Detta är möjligt, då nedskakningen, trots att sluttillståndet är detsamma för
den som för jonisation av en atom i sitt grundtillstånd, följer en annan sönder-
fallsväg. Symmetrin mellan de olika elektroniska tillstånden påverkar sanno-
likheten för vilka tillstånd som kommer populeras av elektroner. Eftersom den
exciterade elektronen kommer från en elektronorbital med annan symmetri än
den elektron vi senare mäter energin hos, kan vi direkt i våra experimentella
spektra se vilka toppar som är av en sådan karaktär att de blir tillåtna i och
med att vägen till sluttillståndet förändras.
20
2. Introduction
This thesis is based on investigations of the electronic and geometric struc-
ture of gas-phase clusters and metal vapors. Our method of choice is, primar-
ily, photoelectron spectroscopy. Used together with supplementary techniques
such as mass spectroscopy, and with the support of first principles calcula-
tions, this family of experimental methods allow for this and more. Common
to most of the techniques used in this work is that they rely on photons to
create electrons, which can then be measured.
Over the years, new light sources, like synchrotrons, lasers and free electron
lasers (FEL), together with extensive development in instrumentation for elec-
tron detection have made a large array of experiments possible, like studies of
magnetism, ultra-fast processes and charge- and nuclear dynamics. In this the-
sis, synchrotron radiation has been used to study ultrafast phenomena, com-
position of artificially created cluster structures, and fundamental properties
of certain exotic molecules. Lasers were used together with synchrotron radi-
ation in some of the experiments to, for example, prepare and probe aligned
samples and to probe processes occuring in the decay of excited states.
There are many motivations for performing this kind of research, both
within the area of fundamental research and within more practically applied
fields; like miniaturization and improvement of hard-drives, solar cells,
computer chips, exhaust catalysis, hydrogen fuel cells and so on. Gas
phase clusters, which has been the main focus of the work in this thesis
are important model systems, which are said to “bridge the gap” between
the single atom or molecule and the infinite solid [1]. They do so by
presenting an opportunity to study the size-regime where miniaturization of
transistors, magnetic storage etc. becomes influenced by quantum effects.
Figure 2.1 shows a schematic figure of a generic physical property as a
function of cluster size. As depicted in the figure, for small clusters, there are
discontinuous jumps of the property depicted when varying the size. For
larger clusters, the property as a function of size converges towards the bulk
value, and follows a more continuous trend. Exactly where the transition
between the “quantum” and the “bulk” regimes is varies for different
properties, and for different compounds. This research topic is highly active,
and attracts a lot of interest.
Likewise, metal vapors, which have been the second subject of study in
the thesis, present a way of studying the individual properties of the kind of
atoms that form many of the most common solids that are used in everyday
21
life. This leads to a better understanding of how to model solids from existing
knowledge about atomic and molecular properties.
Quantum regime
Bulk regime
Bulk value
Pro
per
tyP
(N)
Size N∞1
Figure 2.1: Schematic overview of a generic property versus cluster size. In the bulkregion, the value P(N) follows a continuous trend towards an asymptote, while in the
quantum region there are significant variations between the various sizes.
Another very appealing property of clusters is that their surface area is very
large, i.e. that almost all atoms or molecules are situated on their surfaces. Ina solid, an infinitesimal percentage of the atoms are situated on the surface,
while in a cluster of 13 atoms 92% are found in the surface layer. Even in
a cluster of a size of several thousands of atoms, � 25% of the constituent
particles are found in the surface layer. This means that, in contrast to the case
of the solid, surface effects considerably affect physical properties such as the
melting point, conductivity, ionization potential and absorption frequencies in
clusters [2, 3].
The cluster part of this thesis deals with rare-gas clusters, molecular clusters
of O2 and CH3Br, and mixed rare-gas/molecular clusters. All of these systems
form clusters mainly by van der Waals interaction. The metal vapor part fo-
cuses on alkali-metals, namely sodium and rubidium, and dimers of sodium.
The information provided by studies of this kind of systems can provide in-
sight into matters not accessible to solid-state experiments.
The approach of this thesis is to begin with a brief overview of some fun-
damental concepts, such as electronic and geometric structure of molecular
and cluster objects, to describe the experimental techniques used to create and
probe such systems, and then to give a more detailed description and discus-
sion of the experiments performed in the papers included in the thesis.
22
3. Fundamental concepts
To facilitate comprehension of the works that are presented in this thesis, the
following overview of various fundamental aspects of related subjects such
as the electronic structure of atoms, molecules and clusters, the geometrical
structure of molecules and clusters and the photoelectric principle will be dis-
cussed briefly in the following chapter.
3.1 The electronic structure of matter
What we commonly think of as ’matter’ consists of large collections of atoms
or molecules. An atom is composed of a nucleus and of electrons. The nucleusconsists of protons and neutrons. In neutral ground state cases, the number of
electrons in an atom is equal to the number of protons in the nucleus. The
number of neutrons can vary, and gives rise to various isotopes of the same
atom. Molecules consist of several atoms bound together by intra-molecular
forces. How the electrons are arranged in the atoms or molecules determine
how these particles interact with each other within the lump of matter, like if
they form a solid iron ingot, or if they form a puddle of water. The arrange-
ment of electrons also determines how the matter interacts with other nearby
pieces of matter through, for example, magnetism. To understand such macro-
scopic properties of large collections of matter, a good understanding of the
electronic structure on the atomic and molecular level is needed.
3.1.1 Atoms
An atom is the most basic system that has an electronic structure. It is from the
understanding of atomic electron structure that molecular electron structure is
inferred. The electrons are arranged in a somewhat peculiar manner, following
what is known as the Aufbau principle, as a consequence of the quantization ofenergy [4]. The most basic structure in the electron arrangement is the shell.This is usually denoted either by a capital letter;K,L,M, ... or by a correspond-ing principal quantum number n = 1,2, ... In this thesis, the number notationwill be used throughout. Electrons in a shell with a lower quantum number
are, typically, more tightly bound to the nucleus, and generally contribute less
to the chemical properties of a compound than electrons in a shell with higher
quantum number.
23
A shell can also be divided into subshells. In this case, what differentiatesthe electronic states from each other are their orbial angular momentum. Con-
ventionally, the subshells are denoted by lower case letters, s, p,d, f , ..., corre-sponding to angular momentum quantum numbers l = 0,1,2, . . . ,(n− 1). Asin the case of the shells, electrons with a lower l quantum number are gener-ally more tightly bound to the nucleus than those with higher ones. The shell
and subshell can be combined into an orbital, which is designated by nl, forexample 2s. In this example, the orbital describes the electrons in shell n = 2
and subshell l = 0.
In many cases, this is enough information to specify the complete elec-
tronic structure of an atom, due to the fact that various constraints such as
the Pauli exclusion principle [5] implicitly determine the remaining quantum
numbers. Such a description is done by utilizing an atomic configuration. Aconfiguration is composed of one or more orbitals, and might look like (1s);corresponding to the H atom in its ground state, where there is only one elec-
tron, which is located in the first shell, and in the first subshell of that shell,
(1s)2; corresponding to the He atom, also in its ground state, which has twoelectrons in the first shell, and in the first subshell. A sodium atom, which has
11 electrons, has the configuration (1s)2(2s)2(2p)6(3s) in the ground state.
s-orbital
px-orbital py-orbital pz-orbital
x
xxx
y
yyy
z
zzz
Figure 3.1: Schematic picture of contour surfaces for s and p orbitals.
One designates the projection of the electron orbital angular momentum�lonto the z-axis of the reference system as ml , the magnetic quantum number.
The definition is such that lz = ml · h. This number is restricted to the values
24
ml = 0,±1,±2, ...,±l. As a consequence, there is only one energy level forelectrons in s-subshells (ml = 0), while there are three for p-subshells (ml =0,ml = ±1). In the case of the p-subshell, ml = 0 corresponds to the orbital
along the z-axis, pz, and ml = ±1 corresponds to the orbitals along the x- andy-axes, px and py. When visualizing orbitals, contour surfaces are normally
used, where the surface is taken to be at a distance from the nucleus within
which there is a 90% chance of finding the electron at any given moment.
An example of s and p contour surfaces is shown in figure 3.1. Each electronalso carries what is known as spin. The spin quantum number for an electron iss = 1/2. Analogue to the case of the angular momentum vector of the electron,the projection of the spin vector �s along the z-axis is known as ms, and can
take values ms = ±1/2. Typically, one also defines a total angular momentum�j =�l +�s, and its projection along the z-axis is designated m j = ml +ms.
In many-electron systems, one uses the vector sum of the individual elec-
tron angularmomentum vectors; �L = ∑i�li. This angular momentum is quan-
tized by an integer quantum number L such that |L|= h ·√L(L+1). A similarapproach is taken when describing the spin of such systems. The quantum
numbers ML = 0,±1,±2, . . . ,±L and MS = 0,±1,±2, . . . ,±S are the projec-tions of �L and �S along the z-axis of the reference systems also in this case.In addition, ML = ∑ml and MS = ∑ms. The orbital angular momentum and
spin are connected also in this case by the total angular momentum vec-
tor �J. The total angular quantum number J is calculated according to J =L+S,L+S−1, . . . , |L−S| and the projection value MJ = 0,±1,±2, . . . ,±J.In the cases where additional information is needed, such as about the spin
of the electrons, the configuration is usually accompanied by a term symbol.The term symbol is constructed by using the quantum numbers L, S and J, andis taken to be 2S+1LJ . Similar to the simple case of the subshells, the quantum
number L = 0,1,2, ... is designated S,P,D, .... The term symbol notation is
very useful when describing, for example, excited or ionic states. A distinction
is usually made between the more tightly bound electrons in an atom, and
those electrons that are easier to remove. The electrons that are tightly bound
to the nucleus are called core electrons. The remaining electrons, which arenot as strongly bound, are known as valence electrons, and are the electronsprimarily responsible for forming bonds with other atoms. More specifics can
be found in a regular text-book, such as [6].
3.1.2 Molecules
A molecule is commonly defined as being a stable, electrically neutral group
of atoms, held together by chemical bonds [7]. In this thesis, molecules con-
sisting of atoms bound by the sharing of electron pairs have been studied.
Such bonding is known as covalent bonding. Examples of such covalently
bound systems are the O2, H2O and Na2 molecules. Many of the terms used
to describe the electron structure of molecules are conceptually similar to
25
those used in the atomic case, and most understanding of molecular electron
structure stems from using various schemes of combining atomic orbitals into
molecular orbitals.
A way to describe the electron structure of a molecule is by using an
approach known as molecular orbital linear combination of atomic orbitals(MO-LCAO). As is implied by the name, this approach makes use of the
sums of atomic orbitals to combine them into molecular orbitals. In, for
example, a diatomic molecule, the spherical symmetry of the electrostatic
field of the individual atom is broken, and replaced by a cylindrically
symmetric field. For electrons in the molecule, the angular quantum
momentum vector �l precesses around the symmetry axis of this field, witha constant value ml = l, l − 1, l − 2, ...,−l. In molecules, the orbital angularquantum number l is designated by λ , and analogous to the atomic casesubshells λ = 0,1,2, ... is designated as σ ,π,δ , .... Molecular orbitals aredesignated by their principal quantum number and their angular momentum
quantum number. Orbitals that are rotationally symmetric around the z-axisare σ -orbitals. Thus the atomic s- and pz-orbitals both become σ -orbitals in adiatomic molecule. The px- and py-orbitals become π orbitals, and so on.If we take the example of O2, we can construct the molecular electronic
configuration of the molecule by using the electronic configuration for indi-
vidual O atoms. Each such atom has 8 electrons, in a (1s)2(2s)2(2p)4 config-uration. A schematic picture of how the combination can be done is shown in
figure 3.2.
The configuration of the O2 molecule becomes
(1σ)2(1σ∗)2(2σ)2(2σ∗)2(3π)4(3σ)2(3π∗)2
where the ∗ denotes antibonding orbitals. Term symbols are used also in the
case of molecular electron structure, to describe spin and geometry. One then
uses quantum numbers corresponding to the sum of the individual electron
spins and orbital angular momenta. The individual electron spin projection
quantum number, which in the atomic case is denoted ms is denoted σ . Thisshould not be confused with the λ = 0 quantum number designation σ . Thetotal angular momentum is denoted by ω and is defined as ω = λ + σ . Thesums are designated by Λ, Σ andΩ. Note that a simple arithmetic sum is suffi-cient in the case of molecules, since both λ and σ are, inherently, projectionson the molecular axis. The term is given by 2Σ+1ΛΩ. Sometimes, the mirror
symmetry of the spatial part of the electron wavefunction is given by adding
either a + or a - to the term symbol, and sometimes the property of inversion
symmetry for the entire wavefunction is also denoted by a g or u (gerade orungerade) added to the term. The oxygen molecule, for example, is a 3Σ−
g in
its ground state. For an in depth discussion, see for example [4].
26
1s1s
2s2s
2p2p
1σ
1σ∗
2σ
2σ∗
3σ
3σ∗
3π
3π∗
E
OO O2
Figure 3.2: Schematic picture of the electron structure of O2 The atomic levels thatcontribute to certain molecular levels have a dashed line connection to these levels.
The ∗ denotes antibonding orbitals.
3.1.3 Clusters and solids
The step towards bulk solids from atoms or molecules is rather large. In be-
tween the two exists a category of objects known as clusters [1]. A clustercan be loosely defined as a collection of atoms or molecules held together
by intra-molecular forces, such as van der Waals forces, polar forces, metal
bonds etc. Some objects in this category can be classified as both molecules
and clusters, and the nomenclature is not always entirely unambiguous. Com-
mon to all clusters are that they consist of 3 to some 107 atoms or molecules.
The electronic structure of clusters evolve very differently depending on the
type of bonding mechanism(s) present in the cluster. This thesis focuses on
van der Waals clusters, and in these, the electronic structure of the consituent
atoms or molecules is, in most cases, largely unperturbed from the gas phase
case. There are some differences, where atomic and molecular orbitals are
compressed by neighbours in the cluster [8], and there might be breaking of
molecular symmetries in the cluster lattice to some extent [9], but in the ex-
27
periments performed within the framework of this thesis, these differences
cannot be made out due to limitations in the resolution.
When a cluster becomes large enough, it becomes a bulk solid. It has then
acquired all solid state properties, like conductivity, magnetism and heat ca-
pacity. The electron structure of solids is characterized by electron bands,
which are, in principle, overlapping orbitals in which electrons can move
freely in the entire solid. As this is beyond the scope of this thesis, a further
description will not be given here. For a detailed discussion, see for exam-
ple [10]
3.2 The geometric structure of matter
In addition to the electronic structure, matter also possesses geometric struc-
ture. For atoms, one can define certain radii, beyond which interaction with
other atoms is neglible, such as the covalent radius and the van der Waals ra-
dius [11]. These determine at which distances the atoms bind to other atoms
with the respective kind of bonding mechanism. These, together with the elec-
tronic structure determine how the atoms form molecules, clusters and solids.
Molecules are, as previously mentioned, composed of several atoms. This
means that they can be classified within certain geometrical point groups [12].Knowing what point group a molecule belongs to gives information about
the geometry of the electronic orbitals, about the vibrational modes of the
molecule and about which transitions between the different orbitals that are
possible. In this work, molecules of three kinds have been studied. These kinds
are homo-nuclear, diatomic molecules belonging to the point group D∞h, such
as O2 and Na2, and the more complex CH3Br molecule, belonging to the point
group C3v.
As a part of the description of the different electronic and geometric states
of a molecule, potential surfaces are often used. These surfaces map the energy
of the molecule as a function of the internuclear distances, and are often used
to to predict molecular dissociation limits, equilibrium distances in molecules
and to interpret vibrationally resolved spectra. The use of potential surfaces is
motivated by the Born-Oppenheimer approximation, which states that electronand nuclear movements can be decoupled [13]. This is not entirely true, but
in most cases it gives a good agreement to experiments. Figure 3.3 shows
an example of a 1-D potential surface (i.e. a potential curve) for the oxygen
molecule in its ground- and first ionized state.
The horizontal solid lines in the potential wells in figure 3.3 mark the vibra-
tional levels of the oxygen molecules in its different states. Vibrations come
into play for molecules, since they consist of more than one nuclei, which
can move with respect to each other. These intramolecular vibrations can, for
example, be induced by absorption of, or ionization by radiation. A change
in the vibrational state of the molecule is governed by the transition moment
28
Internuclear distance (A)
Pot
ential
ener
gy(a
rb.
unit
s)
X 2Πg
X 3Σ+
g
0 1 2 3 4 5
FC region
Figure 3.3: The potential surfaces of the oxygen molecule in its ground state (bottom)and in its first ionized state (top). The Franck-Condon region is marked with a shaded
rectangle, the horizontal solid lines are some of the vibrational levels.
.
�Rv. The intensity of a given vibrational transition is proportional to the square
of the transition moment, i.e. �Rv2. In a notation where �μ is the dipole opera-
tor, the first vibrational level is denoted by Ψi and the second one by Ψ f , the
transition moment is given by the integral
�Rv =∫
Ψi�μΨ f dR (3.1)
The transition moment depends somewhat on the internuclear distance in
the molecule. If this dependence is disregarded, one arrives at the Franck-Condon (FC) approximation. The approximation states that the transitions be-tween two vibrational levels can be treated as vertical in a potential diagram
like the one in figure 3.3. The square of the overlap integral between the initial
and final vibrational state
F=(∫
ΨiΨ f dR)2
(3.2)
if known as the Franck-Condon factor. Vibrational transitions can be said to
be proportional to this factor, since the equilibrium distance in the molecule,
29
Re, is constant within the FC approximation. The shaded area in figure 3.3
is the so-called Franck-Condon region. Most of the vibrational transitions
upon excitation or ionization takes place within this region. Knowing the FC-
region, and by measuring the relative intensities of the different vibrational
levels, it becomes possible to experimentally determine potential surfaces of
molecules. This is highly desirable, as they can provide information about
equilibrium distances, vibrational frequencies, dissociation limits and a vari-
ety of other molecular properties.
Clusters can be formed in a large variety of geometries, depending on the
thermodynamical conditions during the formation process, on the method of
production and on the type of bonding mechanisms present in the clusters.
Typically, the geometrical structure of clusters resemble the geometrical struc-
ture of the infinite solid more and more as the cluster grows in size. In this
thesis, the radial segregation of oxygen molecules in large Ar clusters have
been investigated.
The infinite solid is characterized by its crystal structure or lack thereof. As
in the case of molecules, one can classify various kind of geometries within
mathematical groups, which determine various properties. This is a subject
discussed in for example [10].
3.3 The photoelectric effect
When light shines onto matter, be it a single atom or a sheet of metal,
electrons can be ejected. This is provided that the light is of sufficient energy
to overcome the so-called work function, Φ of the material, or ionization
threshold in an atom or molecule. This effect was discovered by H. R. Herz in
1887 [14], and explained by A. Einstein in the famous paper from 1905 [15].
A schematic picture of the effect is shown in figure 3.4.
What Hertz found experimentally, and Einstein explained using the quan-
tization of energy, is that electrons are emitted from a material with a kinetic
energy related to the energy of the incident radiation. In principle
Ekin = hν −Φ−|EB| (3.3)
describes the relation. Ekin is the kinetic energy of the emitted photoelec-
tron and EB is the binding energy of the electron. The work function Φ is
well known for many materials. All energies and calibration of spectra in this
work are given with respect to the vacuum level, as is commonly done when
working with atoms, molecules and clusters [16].
The photoelectric principle is the corner-stone of photoelectronspectroscopy (PES), the family of experimental techniques employed in thiswork, since it allows for the determination of the binding energy of electrons
30
Solid Single atom
Figure 3.4: A schematic picture of photoelectrons being emitted from amaterial underirradiation. The leftmost image shows a solid piece of material filled with an electron
gas, and the rightmost shows an atom with well defined energy levels for the electrons.
by measuring their kinetic energy. These various techniques are further
described in a following chapter.
31
4. Experimental equipment
The experimental work in this thesis was performed mostly at the MAX lab-
oratory in Lund, Sweden, and at the department of physical sciences at Oulu
university, Oulu, Finland. The following chapter gives an introduction to the
experimental equipment used in the experiments. The concepts of synchrotron
radiation and laser light are introduced, along with a description of the exper-
imental systems and the way they are prepared and handled. The MAX labo-
ratory and the beam-line I411 will be presented more thoroughly, since this is
were most of the research was done.
4.1 Vacuum
The different kinds of experiments, such as photoelectron spectroscopy, per-
formed in this thesis have high demands for vacuum around the sample. This
is due to the fact that the spectroscopic information in such an experiment is
mediated by electrons, which interact strongly with air. In a vacuum, the elec-
trons are free to fly a long distance without interacting with any other particles
and thus to preserve the spectroscopic information about the probed sample.
Similarly, a synchrotron storage ring requires even higher vacuum, since the
electron beam that is stored in the ring would be quickly depleted and scat-
tered at atmospheric pressure, and since the soft x-rays emitted from it would
be absorbed quickly in an atmosphere.
Typically, the pressure at the experimental station should be at most in the
10−5 mbar range for electron spectroscopy to be feasible. In the synchrotronring, pressures in the 10−10 mbar range are common. To maintain this kind ofvacuum is quite hard, and requires continuous pumping of the experimental
chamber and surrounding equipment [17]. In our experiment, we use standard
oil roughing pumps, and turbomolecular pumps of different sizes to maintain
vacuum even under very demanding experimental conditions (i.e. very high
gas loads). An additional benefit of performing experiments in vacuum is that
the sample contamination is inherently low, since there are fewer particles in
the surroundings that can interact with the sample before it is probed. This
also has the consequence that there will be less residual contributions to the
experimental spectra acquired under vacuum conditions.
33
4.2 Synchrotron radiation
In 1947, Elder, Gurewitsch, Langmuir and Pollock, at the General Electric
synchrotron accelerator, observed light emitted from their machine [18]. First
thought to be Cherenkov radiation, it soon became apparent that what was
seen was something else. This radiation turned out to be produced by the ac-
celerated electrons that were travelling at relativistic speeds in the synchrotron
accelerator, and thus it was termed synchrotron radiation. The explanation was
given by Schwinger in 1949 [19]. Some years later, in 1956, synchrotron ra-
diation was also observed in astronomical processes by Burbridge [20]. In the
case of astronomical processes, high-energy electrons are accelerated in the
magnetic fields of cosmic objects, producing jets of synchrotron radiation.
The development of synchrotron light sources have gone on from the hum-
ble beginnings in the General Electric lab, to the large 3rd generation syn-
chrotron facilities in operation at a multitude of places today. Development of
4th generation facilities and the so-called x-ray free-electron lasers (FEL) is
ongoing, with the first such facilities just recently becoming operational.
In a 3rd generation synchrotron facility, the most important part is the elec-
tron storage ring, where electrons, as the name implies, are stored in a cir-
cular orbit at relativistic speeds. The storage ring consists of long, evacuated
tubes, magnetic structures and conditioning devices to control the electron
speed, beam-path and divergence. The velocity of the electrons determines
the amount of beaming, i.e. the brightness of the light, in such a way that the
closer to light speed the electrons travel, the more light is emitted into the for-
ward tangential direction of the electron velocity [21]. This is schematically
shown in figure 4.1.
�v�v
�a�a
a) b)
Figure 4.1: The emitted power at two different electron velocities. In a), β = 0 and
in b), β = 0.9. The emission is beamed in the tangential direction of the electron pathwhen at relativistic speeds.�a and�v denotes acceleration and velocity, respectively.
34
At certain points in the magnetic structure, synchrotron radiation can be ex-
tracted in a number of ways [22]. Either, by using one of the ring bending
magnets, which produces a rather broadband and comparatively low-intensity
radiation, or by introducing a so-called insertion device into the ring. An inser-
tion device can, for example, be an undulator or a wiggler. In these, periodic
magnetic structures are used to extract synchrotron radiation. The characteris-
tics of wiggler- and undulator radiation differ from bending magnet radiation,
and from each other, in the intensity and the wave-length distribution of the
generated radiation. Figure 4.2 shows a schematic picture of the three afore-
mentioned magnetic structures.
e−
e−
hν
hν
Bending magnet Undulator/Wiggler
Figure 4.2: Schematic picture of a bending magnet, and an insertion device of theundulator/wiggler type.
Common to all of these kinds of artificially generated synchrotron radiation
is that they all have some very peculiar characteristics, namely:
• High brigthness, intensity and brilliance.• Tunability over a large energy range, from eV to MeV.• Well defined polarization.• Time structure.
35
For the end-users, this opens up possibilities to perform a wide range of exper-
iments that have hitherto been impossible, with regards to different techniques
and also for very dilute samples. For example, core level photoelectron spec-
troscopy, x-ray absorption spectroscopy and x-ray crystallography are tech-
niques that can be routinely applied to systems such as gas phase samples,
surface films, and protein crystals by utilizing synchrotron radiation.
4.3 Laser Light
Building on the fundaments laid down by Einstein [23] and other important
theoretical works by many prominent researchers, C. H. Townes demonstrated
the first working maser (Microwave Amplification by Stimulated Emission)
in 1953 [24], This device had a non-continuous output of microwaves. The
term “Laser” was coined by R. G. Gould in 1959, in a conference paper titled
The LASER, Light Amplification by Stimulated Emission of Radiation [25]. Asthe name suggests, a laser is a device that emits amplified light, rather than
microwaves.
Typically, a laser consists of two main parts; a gain medium and a oscilla-
tor cavity. The gain medium can be, for example, a crystal, a liquid, or a gas
mixture which is selected to provide optimal characteristics for some experi-
mental parameter, such as light wavelength or intensity. The oscillator cavity
is, in most cases, limited by two mirrors, one of which is semi-transparent,
thus letting a part of the amplified light through. Other designs might include
only one, or no mirrors, but they are limited by large beam divergence. Fig-
ure 4.3 shows a schematic picture of the difference in divergence between
zero, one and two mirror lasers.
a)
b)
c)
Figure 4.3: Three different laser designs. a) has no mirrors, and a large divergence, b)has one mirror and medium divergence, c) has two mirrors and small divergence.
36
Like synchrotron radiation, laser light has some very special properties that
makes it useful to perform a wide range of experiments [26, 27]. Among these
are the facts that lasers are
• Highly collimated.• Spatially coherent.• Quasimonochromatic, i. e. has a very narrow wavelength range.• Very bright, or have a very large radiant power.• Time structure.This makes them highly usable in various kinds of applications, in research
(spectroscopies, quantum optics etc.), in industry (drilling, cutting and so
forth) and in every day life (cd-players, bar code readers and many other com-
mon apparatus).
4.4 The MAX laboratory and beam-line I411
The MAX laboratory in Lund, Sweden, is one of two national research lab-
oratories, the other being the radio telescope facility in Onsala. At the MAX
laboratory, research in a wide range of topics is performed, utilizing the three
synchrotron storage rings and the linear accelerator at the facility. The three
storage rings are the MAX I, a second generation storage ring, the MAX II, a
third generation ring and the MAX III, also a third generation ring. Most of the
work in this thesis was performed at the I411 soft x-ray beamline at MAX-II,
where photon energies of roughly 60-600 eV are accessible. Figure 4.5 shows
an overview of the MAX laboratory.
Figure 4.4: The different elements of the beam-line I411 at the MAX laboratory inLund. The leftmost part is connected to the undulator insertion device, experiments
are performed at one of the experimental stations at the right end of the beamline.
This beamline and its experimental station is specially designed to accomo-
date high pressure experiments, like gas phase samples. This is achieved by
using many vacuum pumps and so-called differential pumping stages, which
provide a steep gradient in the local pressure in the beamline. In the experi-
mental station, pressures can be in the 10−5 mbar range without affecting thevacuum in the synchrotron storage ring measurably. The permanent experi-
37
Figure 4.5: The layout of the MAX laboratory facility in Lund, Sweden.
mental station of the beamline is equipped with a Scienta R4000 photoelec-
tron spectrometer. Just before the permanent experimental station is a part of
the beamline which can be interchanged with user experiments. It is in this
part of the beamline all of the laser experiments have been performed, using
a set-up built in Oulu, Finland [28]. The spectrometer in this case was a Sci-
enta SES-100 hemispherical analyzer. A drawing of the beam-line is shown in
figure 4.4
Beside the beamline is a sealed hutch, where a state-of-the-art laser system
is installed. This system consists of a 10W CW (continuous wave) Coherent
Verdi laser, operating at 532 nm wavelength, most often used to pump one of
the other lasers that are available. These other lasers are a Coherent 899-21
ring laser, which can use either a Ti:Sa crystal or a dye jet as gain medium,
providing an effective output power of up to 2 W CW in a wavelength range
of 550-1000 nm. There is also a pulsed Ti:Sa fs laser, giving optimal output at
around 600 nm wavelength. In addition to this, the equipment includes com-
puterized control of the cavity optics for the Coherent 899-21 laser, which
allows for wavelength scans, and a doubler ring allowing users to reach UV
wavelengths. A schematic picture of the CW laser system which has been used
Figure 4.6: A schematic overview of the different components in the CW laser system
employed in this work.
4.5 Sample production
In this thesis, beams of clusters of atoms and molecules as well as free metal
atoms were used as samples. There is no way to commercially obtain a bottle
of clusters or vaporized metal, so the samples have to be produced in situ. Thesections below will detail how this is done with our different set-ups.
4.5.1 Cluster production
Clusters can be produced in many ways, each with advantages and drawbacks
for certain types of experiments. In our experiments we are limited by the
photon flux at the undulator beam-line we use, which means that we have to
have a relatively high target density to be able to use electron spectroscopy as
an efficient experimental probe. Our method of choice is to produce clusters
from gases through adiabatic expansion, which produces fairly large clusters
in a large abundance. We have built our source in-house, and the set-up is
similar to that of [29].
The source works by feeding gas under high pressure (usually 1-5 bar) into
a stagnation volume, from which the gas is fed through a small, conical noz-
zle to an expansion volume. The pressure in the expansion volume may vary
between 10−5 and 10−3 mbar. The small conical nozzle has an opening of 150μm, and an opening angle of 20◦. Gas coming from the stagnation volume tothe expansion volume through the nozzle is adiabatically expanded, and forms
a supersonic jet [30] in the expansion volume. The clusters are mostly formed
on the axis of the jet, which means that it is possible to suppress the atomic or
molecular residue by introducing a skimmer in front of the nozzle. This skim-
mer also works as a separating stage in the differential pumping of the set-up.
The clusters and the residual atomic or molecular gas that pass through the
skimmer then enters the interaction region, where the synchrotron beam in-
tersects the cluster beam. Clusters formed in the jet-expansion are distributed
around a mean size 〈N〉, which is governed by the expansion parameters; noz-zle temperature, nozzle geometry and stagnation pressure [29, 31, 32]. Higher
pressure and lower temperatures generally give larger clusters.
39
Figure 4.7: Schematic picture of the cluster source (to the left) attached to the experi-mental station (to the right).
Our set-up is designed to allow for high stagnation pressures, liquid nitro-
gen cooling of the nozzle and relatively large nozzle diameters. Two large
turbo-pumps are used to maintain the pressure in the expansion volume. All
these considerations allow for a source that produces a dense enough beam to
perform even very demanding types of electron spectroscopies [33]. We use
calibrated pressure regulators and gauges to control the pressure in the experi-
ment, and a LN2 cooling system together with an electrical heater to maintain
stable nozzle temperatures down to around -170◦C. To give some idea of thecharacteristics of the source, it allows us to perform electron spectroscopy
experiments on Ar clusters in the size range from 100 to 50000 atoms. A
schematic picture of the cluster source is shown in figure 4.7.
4.5.2 Doped cluster production
To create structures of mixed clusters, we have used the post-expansion dop-
ing technique, pioneered by Scoles et al. [34, 35]. In our case, the doping stageconsisted of four needles of 150 μm inner diameter, mounted perpendicularlywith respect to the cluster beam and at a variable distance from the expansion
nozzle. The doping stage in this set-up is mounted before the cluster beam
is skimmed. This design minimizes the deflection of the cluster beam, which
passes through the centre point of the needle “cross”. A schematic picture of
the set-up is shown in figure 4.8. The doping stage is capable of delivering
most kinds of gaseous substances to the cluster beam, thereby allowing cre-
ation of, for example, a chemically reactive sample system.
40
Figure 4.8: Schematic picture of the doping stage. The four needles are mounted per-pendicularly with respect to each other, and with respect to the sample beam that
passes between them.
4.5.3 Metal vapor production
Metal vapors can, just like clusters, be produced in many ways. These differ in
the vapor density they provide and are suited to different applications. In our
experiment, we have chosen to use resistively heated ovens, which give a large
amount of metal vapour. Our experimental set-up with an oven was situated in
a user-endstation built in Oulu [28], which included a Scienta SES-100 elec-
tron energy analyzer. Another similar oven was situated inside the laser hutch
beside the beamline, to provide a sample for the reference chamber. Such a
reference chamber allows for finding laser flourescence lines and thus tuning
the laser system independently of what is going on in the main experimen-
tal chamber. Figure 4.9 shows the flourescence from 3s → 3p excited sodiumatoms, produced with this set-up.
41
Figure 4.9: Photograph of flourescense from laser excited sodium atoms.
42
5. Probing techniques
There are several different techniques within the broad definition of electron
spectroscopy. Those that have been used in this work will be given a thorough
introduction in the following chapter, together with a brief explanation of how
such spectra can be interpreted with the help of, for example, calculations and
curve fitting.
5.1 Ultra-violet Photoelectron Spectroscopy (UPS)
UPS, also known as valence photoelectron spectroscopy [36], is used to probe
the shallower part of the electron cloud, the valence electrons. The way the
photoelectrons are produced is by valence photoionization, a straightforward
photon in–valence electron out process. Figure 5.1a shows a schematic picture
of UPS. For the studies in this thesis, UPS has mainly been used as a tool for
roughly determining the sizes of clusters, but also, in the case of (Ar)m(O2)nclusters, to determine the structure of mixed cluster systems. Due to several
factors contributing to an efficient production of photoelectrons in this energy
region (such as the maximum photon flux of the beam-line I411 and the high
ionization cross-sections of most valence orbitals around 60 eV) it is also an
excellent tool for aligning the cluster beam with the synchrotron radiation
beam, and a good way to see contaminations in the spectra. Two examples of
UPS spectra can be seen in figure 5.2 where the leftmost spectrum is the UPS
spectrum of Argon clusters and the rightmost spectrum is the UPS spectrum
of O2 clusters.
One can note that the cluster feature in the Argon spectrum is very broad
compared to the atomic feature in the same spectrum. This is commonly at-
tributed to band formation of the valence states in rare-gas clusters [37]. In
the case of O2, on the other hand, the width of the cluster feature envelope is
not that much different from that of the molecular ditto. This effect is, so far,
not well studied, but might be due to the fact that the oxygen-oxygen symme-
try inside a cluster lattice prevents band formation or that the oxygen valence
orbitals are smaller than those of rare-gas atoms. UPS energies can be found
easily in the literature, for example in [38].
43
EE a) b)
γ
γ
Figure 5.1: a): Schematic picture of UPS. b): Schematic picture of XPS. Note thedifference in origin of the photoelectrons.
-1-1 00Relative binding energy (eV)
Inte
nsity
(arb
.un
its)
hν=61 eVhν=61 eV
Figure 5.2: Spectra of the 3p level of Ar clusters and of the X-state in O2 clusters.The dashed lines represent the total envelope in the cluster spectrum, the solid lines
represent the molecular envelope and the shaded areas represent, roughly, the cluster
features in the spectra.
5.2 X-ray Photoelectron Spectroscopy (XPS)
XPS, or core electron spectroscopy, is used to probe the electronic structure
deeper in the electron cloud than UPS. Just like UPS, it is a straightforward
photon in–electron out process, but where in UPS a valence electron is emit-
ted, the XPS electron is emitted from one of the core orbitals of the atom or
molecule. As mentioned in section 3, core electrons does not participate in
chemical bonding, and remain atomic-like even in a solid. Although this is the
case, there is a feature in XPS which makes is an extremely powerful tech-
nique for identifying various compounds. This feature is the chemical shift
of core level binding energies. This concept of chemical shift means that the
44
energy of a core level will be affected by its surroundings, which makes it
possible to identify and study even individual atoms of a certain type in, for
example, an alkane chain, since the atoms at different places in the molecule
will have different surroundings. The principle of XPS is shown in figure 5.1b,
and the XPS spectrum of O2 is shown in figure 5.3. In the case of O2, there
are two peaks in the XPS spectrum. These two components are due to the
paramagnetic splitting in the O2 molecule, which comes about because the
oxygen molecule has only two electrons in its highest occupied molecular or-
bital (HOMO), and because this HOMO has space for two more electrons.
Thus, O2 is said to be an open shell molecule, leading to this kind of splitting.XPS energies for a wide variety of species can easily be found in tables such
as for example [39].
542543544545546547548
Inte
nsity
(arb
.un
its)
Binding Energy (eV)
ClusterMolecularfeatures features
Figure 5.3: The O1s XPS spectrum of O2 clusters. Note that for clarity, the spectrumdisplays only the adiabatic vibrational components as shaded areas, while the solid
black line takes into account all of the vibrational components used in the fit. The
ionizing photon energy was 570 eV.
5.3 Auger Electron Spectroscopy (AES)
Auger electron spectroscopy is different from UPS and XPS in that it isn’t
merely a photon in–electron out process. Instead, AES studies the electrons
that are emitted when a core-hole decays. A schematic picture of this is shown
in figure 5.4. To observe Auger electrons, a core hole is first created by, for
instance, photoionization. The core-ionized state is unstable, and after a short
period of time (typically in the fs range) it will decay to a lower, more sta-
45
ble energy state. This may for instance happen by a valence electron “falling
down” into the core-hole, while another valence electron is emitted with a
certain energy. This latter electron is called an Auger electron and its kinetic
energy can be measured by an electron spectrometer. The bandwidth of the
ionizing radiation is of no consequence for the resolution in the Auger spectra,
since the kinetic energy of the Auger electrons is given by the energy differ-
ence between the energy levels involved in the Auger decay and the second
ionization threshold energy. Rather, the width of Auger lines will be deter-
mined by the lifetimes of the intermediate and final states. Therefore, features
in atomic Auger spectra may often be quite sharp. Auger electron energies can
give valuable information of the relative positions of energy levels in an atom
or molecule.
E
Photoelectron Auger electron
γ
Figure 5.4: Schematic picture of AES. In the first step, the sample is photoionized,and in the second step the Auger process takes place.
5.4 Near-edge Absorption Fine Structure (NEXAFS)
The fourth type of spectroscopy used in this thesis is Near-edge Absorption
Fine Structure spectroscopy, or NEXAFS. This kind of spectroscopy measures
the x-ray absorption near an ionization threshold. In our case, this method was
used to map out unoccupied electron orbitals, that is, eigenstates of the sys-
tem wavefunction that are not populated in the electronic ground state. Com-
monly, NEXAFS is used in surface science to determine molecular orientation
and symmetry in adsorbed molecules, since the technique is sensitive to bond
angles. With the equipment at our disposal, we measure what is known as par-
tial electron yield (PEY). This works in a way that the exciting photon energy
is swept over a certain energy region, and for each step in the sweep, most
46
of the electrons that are emitted, regardless of their source channel, will be
collected and summed by the electron spectrometer. When this sum is plotted
against the photon energy, one obtains the a spectrum that is approximately
proportional to the “real” absorption intensity.
5.5 Resonant Auger Spectroscopy (RAS)
Resonant Auger spectroscopy works much like AES, with the notable excep-
tion that there is no core ionization preceding the Auger decay, but rather a
core excitation. In such an event, a core electron is excited by a photon with
just the right energy to move it out of the core, and into one of the unoccu-
pied orbitals that was mentioned earlier in the description of NEXAFS. When
in such an orbital, the atom or molecule will be in a similarly unstable state
as when core-ionized, but the ways to a more stable situation are somewhat
different. There are two main cases, both of which have the common denomi-
nator that a valence electron falls down to fill the core hole. The first of these
is where the excited electron takes part in the core hole decay, leaving the
atom or molecule in a one-hole state. This is called a participator decay. The
second case is one where a valence electron is emitted as the Auger electron,
leaving the atom or molecule in a one-particle, two-hole state. This is called
a spectator decay. A schematic picture of RAS is shown in figure 5.5. In this
thesis, RAS has been used as a probe of molecular dissociation, and it will be
further described in this context in the results section.
E
Spectator decay Participator decay
γ
Decay types
Figure 5.5: Schematic picture of RAS. The shaded box shows the two different decaychannels, participator and spectator decay, that succeeds the core-excitation.
47
5.6 Interpretation of X-ray spectra
The interpretation of photoelectron spectra is not always a straightforward
task. There may be overlapping or unresolved features, making an unambigu-
ous interpretation of the spectra hard, but with modern tools many of these
difficulties may be resolved. Two common approaches when it comes to inter-
pretation is to either use curve fitting, based upon previous empirical knowl-
edge and known formulas for a given system, or to use ab initio calculationsto predict what should be seen in the spectra (binding energies, lineshapes and
linewidths, differential and total cross-section) and then assign features from
this calculation. There are several ways and methods to perform such calcu-
lations, and if such a method was used it will be described in greater detail in
context of the experiment it was applied to. If nothing else is mentioned, the
spectral analysis was done by curve-fitting. For the sake of simplicity, only
UPS and XPS will be shown in the following discussion. Interpretation of
RAS, AES and NEXAFS spectra will be discussed together with the results.
5.6.1 Koopmans theorem and the chemical shift
A theorem that is commonly used when interpreting photoelectron spectra is
Koopmans theorem [40]. According to this theorem, the ionization energy of
a molecule is equal to the energy of the highest occupied molecular orbital
(HOMO);
EB,k �−εk (5.1)
This allows for ab initio calculation of ionization energies of molecules,since orbital energies can be calculated using methods such for example den-
sity functional theory (DFT) or Hartree-Fock (HF) theory. This theorem does
not take electron correlation or relaxation of the orbitals upon ionization into
account when determining ionization energies, which is a problem in many
systems. However, the relaxation of orbitals upon ionization gives a contribu-
tion to the binding energy that, in many cases, is of opposite sign from that
of electron correlation. Relaxation effects come in many forms; but primar-
ily from two sources. These are relaxation of orbitals in the same atom or
molecule, and charge transfer to or from the surrounding atoms or molecules.
Because of the cancelling character of the relaxation and electron correlation
effects, Koopmans theorem often gives sufficiently good results for spectral
interpretation where there are clearly observable features in the spectra.
One of the most prominent features of photoelectron spectroscopy is its
inherent element and geometric sensitivity. This sensitivity is due to the
so-called chemical shift, which was mentioned earlier. The chemical shift ismainly due to what is known as screening, i.e. differences in the Coloumb
interaction of the vacancy in the ion with the rest of the atom or molecule. A
conceptually simple example is the sodium azide molecule, NaN3. The ionic
character of the various parts of this molecule (two N−, one N+ and one Na+
48
part) give rise to three distinct spectral features. One from the N+ ion, one
from both of the N− ions and one from the Na+ ion. The negative charges
screen the charge from the nitrogen nucleus, reducing the binding energy on
the negatively charged ions; this makes it possible to assign features even
from atoms of the same element, which differ only in position in a molecule.
In a covalently bound system, the chemical shift is mostly related to the
electronegativity difference of various parts of a molecule.
5.6.2 Atomic photoelectron spectra
The simplest case to begin with when discussing the interpretation of pho-
toelectron spectra is the atomic photoelectron spectra. As mentioned earlier,
molecular spectra can be said to be derived from the atomic spectra, and clus-
ter spectra in turn can be understood from the spectra of its constituents. An
ordinary atomic UPS spectrum is shown in figure 5.6a. Here, one clearly sees
two sharp peaks. These can be assigned to an atomic orbital which is spin-
orbit split, and the difference in energy between the two peaks is the differ-
ence in energy that the photoemitted electron has depending on its spin state.
In atomic XPS spectra, the situation is much the same. There is, however
one more thing that one needs to consider in XPS, namely the phenomenon
known as Post-Collision Interaction, or PCI [41, 42]. This is a mechanism that
introduces a kinetic energy dependent asymmetry in many XPS spectra. In a
classical picture, this can be viewed as a consequence of the outgoing photo-
electron being overtaken by the Auger electron after some time. The asymme-
try becomes larger if the photoelectron has a low kinetic energy, meaning that
the photoelectron is overtaken at an earlier stage. Usually, one tries to avoid
this by choosing the photoelectron kinetic energy to be large enough to avoid
the photoelectron being overtaken at all.
5.6.3 Molecular photoelectron spectra
Molecules differ from atoms in that they have at least two nuclei. These nuclei
move with respect to each other depending on how the electron cloud is cur-
rently configured, by vibrations and rotations along and around the bonds in
the molecule. This gives rise to vibrational features in the molecular spectra,
as is shown in 5.6b, as was discussed earlier. From such spectra it is possible
to deduce for example the bond-lengths in a molecule. The relative probability
of a vibrational transition is given by the Franck-Condon factor (see eq. 3.2).
From these it is possible to deduce various things about the molecule, such as
the shape of its potential surfaces. In XPS, the main difference between the
atom and the molecule is that the molecule may have many different core-
levels, each situated in an individual atom in the molecule. These may differ
much in energy, such as the levels C1s and O1s levels in CO, or have lev-els that are close to each other like the two N1s levels in N2O. One of the
49
1516
a b
13 12
hν=61 eVhν=61 eV
Binding energy (eV) Binding energy (eV)
Rel
ativ
ein
tens
ity
(arb
.un
its)
Figure 5.6: UPS spectra of the atomic Ar 3p state (panel a) and of the O2 X-state(panel b). The two peaks in the Ar spectrum are due to the spin-orbit splitting of the
atomic 3p level. The corresponding term symbols are 2P1/2 and 2P3/2 from left to
right. The peaks in the O2 spectrum are due to the intramolecular vibrations.
strongest points with XPS is that it can separate the core levels for each in-dividual inequivalent atom in the sample, by the chemical shift induced by
the neighboring atoms. XPS is, for example, sensitive enough to differentiate
between the core level of the far-end N and the central N in N2O. This fact
can be, and is commonly, used to fingerprint molecules, as mentioned earlier
in section 5.2.
5.6.4 Cluster photoelectron spectra
In addition to all the mechanisms mentioned above, clusters have some more
things that complicates the spectral interpretation. In the systems presented in
this thesis, the main difference between the previous cases is that the neigh-
bors in a cluster induce additional polarization screening of the core hole whenphotoionization occurs [43]. Disregarding initial and minor final state effects,
this means that the effective binding energy of the electrons is lower in a clus-
ter of a given atom or molecule than in the free ditto. The screening is, to
a large extent, determined by the nearest neighbor atoms or molecules. This
leads to that vacancies in bulk atoms are screened to a higher degree than
in surface atoms, since bulk atoms have more nearest neighbors. Figure 5.7
shows a schematic picture of the polarization screening mechanism. In core-
level photoelectron spectra of rare gas clusters, this is seen as two distinct
features of different binding energy, the surface feature being closest to the
atomic feature. This is mainly used to study the difference between surface
and bulk properties, but can also be used to determine approximate cluster
Figure 5.7: A schematic picture of the screening of an electron (core) hole in an in-sulating van der Waals cluster. The positive charge is screened by electrons of neigh-
boring particles being polarized towards it.
Clusters are also, though very small, extended systems. That means that
there is an effective electron attenuation length, which is the distance an elec-
tron statistically travels inside the cluster before being absorbed or scattered
by another atom. This attenuation length is strongly dependent on the electron
kinetic energy [45, 46]. This dependence is often used to vary the contrast
between the cluster surface and the cluster bulk contributions to the spectrum.
Last, cluster features are broader than atomic or molecular features. This
is due to various mechanisms [47], but mainly to variations of polarization
screening depending on the position of the ionized atom. This is due to the
fact that the sample beam that we probe consists of clusters of a lot of dif-
ferent sizes, as has been discussed earlier. Furthermore, the clusters do not
always form as perfect crystals, but rather as a mix between locally well or-
dered chunks and other more randomly ordered parts. The final spectral shape
of clusters will be an average with contributions from all of the possible clus-
ter sizes and geometries. There are other, minor effects which also broaden
the cluster spectra; for example vibrations in the clusters. All this considered,
one has a fairly good idea of how to decompose a cluster spectrum. There are
also recent advances in the theoretical lineshape modeling of clusters that can
potentially ease the burden of cluster scientists by predicting the cluster size
theoretically from experimental spectra [48, 47].
51
6. Summary of papers
In this section, a summary of the results of all the papers will be presented,
together with a more detailed discussion of the different experiments and the
theory employed to interpret the results.
6.1 Dissociation of molecules
In papers I-III, the dissociation behavior of different molecules (O2, CH3Brand Na2) in different media (O2 clusters and Ar clusters) and in vacuo re-spectively is investigated. In section 3.2, potential curves of molecules were
discussed. These all showed bound states, i.e. states in which, without vibra-
tional excitation, the intra-molecular bonds are not broken. Keeping this dis-
cussion in mind, there are also electronic states in molecules which are purely
repulsive. Figure 6.1(a) shows a schematic comparison between a bound state
potential and a repulsive potential. The steepness of the dissociative potential
in the FC-region will, in principle, determine the time duration of the disso-
ciation. By experimentally measuring the kinetic energy release (KER) of the
dissociation, such potential curves can be characterized [49, 50].
Internuclear distance (A)
Pot
ential
ener
gy(a
rb.
unit
s)
1s−1
σ∗
X 3Σ+
g
0 1 2 3 4 5
FC region
(a) A schematic picture of the ground
state potential and the 1s−1σ∗ excitedstate potential
Inte
nsity
/arb
.un
its
Photon Energy/eV
3σ∗
Molecular TIY
Cluster PEY
530 535 540 545
3π∗
1 2 3
<N>=10000
(b) The x-ray absorption spectrum of
free oxygen molecules (bottom) and clus-
tered oxygen molecules (top).
Figure 6.1: A dissociative potential curve and the oxygen cluster NEXAFS spectrum
A way of preparing such a dissociative state is to perform a core-to-valence
excitation. This means that one excites an electron from a core-orbital to an
unoccupied valence level, by absorption of an x-ray photon as in for exam-
53
ple RAS spectroscopy (see section 5.5). A molecule in such an excited state
will, in some cases, dissociate on a fs time-scale. This means that the disso-
ciation takes place on the same time-scale as an Auger decay. Breaking of
the molecular bond on this time-scale is often termed ultra-fast dissociation
(UFD) [51, 52, 53, 54]. The consequence of this is that one can observe spec-
tral features from Auger decays both in the intact molecule and from decays
taking place in fragments of the molecule. This type of spectroscopy has been
performed in papers I and II. Both of these papers address the question ofwhat effect the surrounding medium has on the dissociative character of the
core-excited state.
6.1.1 Oxygen clusters
In paper I, clusters of oxygen have been studied in an attempt to quantify theeffect of clustering on dissociation after core excitation. Oxygen was chosen
as an experimental system, since it is fairly easy to handle and since here are an
abundance of data already published for the molecule in vacuum and in solid
phase [54, 55]. NEXAFS spectroscopy was employed to locate the 1s−1σ∗
resonance in the clustered molecules, the spectrum is shown in figure 6.1(b).
ΔEv = 0.8 eVΔEv
Inte
nsity
/arb
.un
its
Binding Energy/eV
13 11
546 544 542
ΔEc
ΔEc
ΔEc = 0.7 eV
(a) UPS and XPS spectra of oxygen clus-
ters. The polarization screening shift was
found to be 0.7-0.8 eV. In the UPS spec-
trum (top) the molecular progression is
the spectral feature with the highest bind-
ing energy, and the broad feature at lower
binding energy is the cluster feature. The
XPS spectrum is fit by a molecular com-
ponent and a cluster component.
470 475 480 485 490 495Kinetic Energy/eV
Inte
nsity
/arb
.un
its
ΔE
ΔE
ΔE= 1.0 eV
2P
2D
1
1
2
3
(b) RAS spectra of oxygen clusters. Ex-
citation energies are marked by the num-
bers 1 through 3, and match the ver-
tical bars in figure 6.1(b). The bottom
spectrum is the RAS spectrum of the
free molecule. The other spectra are from
the clustered molecules; for each energy
point there two spectra are displayed. The
topmost of these two is the total spec-
trum while the lower one shows the clus-
ter spectrum only (where molecule has
been subtracted).
Figure 6.2: UPS, XPS and RAS spectra of oxygen clusters.
54
In the NEXAFS spectrum it can be seen that the shape and position of the
1s−1π∗ resonance is virtually unaffected by clustering. This is not the case forthe 1s−1σ∗ resonance, which is somewhat blue-shifted and broadened in theclusters. Figure 6.2(a) shows UPS and XPS spectra of oxygen clusters, which
were used to extract the shift due polarization screening for a singly charged
molecule in the cluster matrix. The shift turned out to be around 0.7-0.8 eV in
these spectra.
A detuning study was performed to characterize to 1s−1σ∗ resonance in theclusters. In the resulting RAS spectra, which are shown in figure 6.2(b), one
can see that there are spectral features that are shifted towards a higher kinetic
energy (i.e. a lower binding energy) relative to those corresponding to decay in
an atomic oxygen fragment in the case of the free molecule. The shift towards
higher kinetic energy is approximately 1 eV. This is slightly larger than the
polarization screening shift for a singly charged molecule seen in XPS and
UPS. An explanation for this difference is that the excited molecule breaks
into fragments which have time to move inside the cluster matrix; a typical
path-length of 1 Å before the Auger decay takes place can be calculated for
the excited fragment. Since this, on average, moves the fragment closer to
its neighbors, the screening increases. These facts, when taken together, are
very strong indications of the UFD process taking place inside the clusters. In
this case it was not possible to distinguish contributions from the surface/bulk
components, due to the very demanding experimental conditions. What one
can observe, though, is that there seems to be no radical differences between
the molecular and the cluster RAS spectra, save that of the shifted features that
can be attributed to polarization screened fragments. These findings indicate
that the valence electronic structure of oxygen molecules is not affected by
clustering to any significant amount.
6.1.2 Bromomethane clusters
Paper II presents a similar study as that of paper I, with the principle dif-ference that bromomethane (CH3Br) clusters have been used in the sample
beam. The extent of the CH3Br valence orbitals is larger than that of O2, and
the interaction between the molecules is governed not only by van der Waals
forces, but also by dipole-dipole interactions. This gives the CH3Br clusters
a local ordering of Pnma packing [56], in contrast to the rather amorphous
packing of oxygen molecules under the experimental conditions employed in
paper I [57]. To characterize the band formation in bromomethane clusters,we have performed ab initio model calculations on the bromomethane dimer.We have utilized the geometries found in [58], and performed anMP2 calcula-
tion, using the LANL2 effective core potentials [59] together with a cc-PVDZ
basis set. All of the calculations were performed usingGaussian 03c [60]. Thesplitting of the outermost molecular 2e valence level due to dimerization canbe viewed as a lower limit to the band-width in the larger cluster system. It
55
is found that the splitting in the dimer becomes � 0.2 eV. The calculations
show also that the valence state in the dimer is formed by orbitals that are
delocalized over neighboring molecules. This makes it rather probable that
even more overlap will occur in a cluster matrix, where there are more nearest
neighbor to each molecule. A picture of a resulting electronic orbital in the
bromomethane dimer is shown in figure 6.3.
Figure 6.3: A resulting molecular orbital (HOMO-3) in the dimer.
The onset of band formation means that the valence electronic structure
is very much affected by clustering. Thus, there might be significant differ-
ences between the cases of bromomethane and oxygen. In the case of bro-
momethane, the core-to-valence excitations of interest are 3d5/2 → 4a1 or3d3/2→ 4a1. Figure 6.4(a) shows the NEXAFS spectra of molecular and clus-tered bromomethane. In spite of the apparent similarity of the 3d → 4a1 ab-sorption spectrum in the molecular and cluster cases, we observe significant
differences in the character between the two, when analyzing the RAS spectra
at different points on the resonance.
In the molecule, both of the spin-orbit states are dissociative. There have
been previous studies on the bromomethane molecule to map this shape reso-
nance [52, 53]. Figure 6.4(b) shows molecular RAS spectra, at different exci-
tation energies. The spectral shape has been assigned in the previous studies
as being due to a mixture of molecular Auger features and atomic Auger fea-
tures. The shift between the two spectral features at the higher and the lower
excitation energies can be attributed to the difference in energy between the
two excited states. The spin-orbit splitting between the 3d5/2 and 3d3/2 levelshas been found to be approximately 1 eV, which is consistent with what has
56
69.5 70.0 70.5 71.0 71.5 72.0 72.5Photon energy (eV)
Inte
nsity
(arb
.un
its)
3d5/2 → σ∗ 3d3/2 → σ
∗
ΔE
(a) NEXAFS of bromomethane
molecules and clusters. Within the
measured statistics, no change in the
peak shape can be observed.
50 52 54
1D
1D
3P
3P
3d3/2 → σ∗
3d5/2 → σ∗
hν=70.2 eV
hν=70.6 eV
hν=71.0 eV
hν=71.4 eV
hν=71.8 eV
Kinetic energy (eV)
Inte
nsity
(arb
.un
its)
(b) CH3Br molecular RAS spectra.
Atomic features are shown as solid lines,
as is the total fit lineshape. The shaded ar-
eas represent molecular Auger features.
Figure 6.4: The NEXAFS spectrum of molecules and cluster, and the RAS spectra ofthe molecule, detuned over a range of energies as indicated in the figure.
been observed in earlier works [52, 61]. This splitting corresponds very well
to the Auger feature shift when exciting the two spin-orbits components.
Figure 6.5(a) shows the XPS spectrum of bromomethane clusters of two
different sizes. These have been used to extract the shift due to polarization
screening for a singly ionized state.
In clusters, if the electronic final state is localized to an individual molecule,
the vacancy will be affected by polarization screening. If this is the case, and if
UFD takes place in the bromomethane clusters, one should be able to observe
spectral features from atomic fragments that are shifted up in kinetic energy.
Figure 6.5(b) shows cluster RAS spectra at the same excitation energies as for
the molecule. These have been treated within the framework proposed above,
i.e. by modeling the cluster lineshape as a shifted and broadened copy of the
features stemming from the free molecule. As can be seen from the figure,
this kind of model works rather well to describe the experimental spectrum
for both of the spin-orbit components.
However, upon analysis of the ratios between decays in the molecule and
the cluster, and between the resonant Auger decays and photoionization, one
comes to several conclusions. There is a clear effect on the dissociative char-
acter of the 3d → 4a1 excitation in clusters when compared to the same exci-tation in the free molecule. The number or decays in screened fragments, i.e.
atomic fragments that are within the cluster matrix is suppressed compared to
the number of decays in fragments from free molecules. Also, the total num-
ber of decays observed in the RAS spectra is suppressed as compared to the
57
hν=110 eV
hν=110 eV
hν=110 eV
ΔEL=0.81 eV
ΔEL=0.72 eV
Inte
nsity
(arb
.un
its)
Binding energy (eV)78 77 76 75
Large clusters
Small clusters
Molecule
(a) XPS spectra of the CH3Br molecule,
small clusters and large clusters. The po-
larization screening shift was found to be
0.72 eV for the small clusters and 0.81
eV for the larger size.
50 52 54
1D
1D
3P
3P 3d3/2 → 4a1
3d5/2 → 4a1
hν=70.2 eV
hν=70.6 eV
hν=71.0 eV
hν=71.4 eV
hν=71.8 eV
Kinetic energy (eV)
Inte
nsity
(arb
.un
its)
(b) The RAS spectra of large clusters,
at the same excitation energies as those
used in 6.4(b).
Figure 6.5: The XPS spectra from the molecule and two sizes of clusters and the RASspectra from large size clusters.
RAS spectra of the free molecule. The latter fact is somewhat surprising, since
the NEXAFS spectra are very similar.
Several possible explanations exist for these observations. However one can
rule out some explanations, like electron recapture, fragment cage exit, neigh-
bor closing and secondary ionization due to various reasons. Remaining expla-
nations include band formation in the initial and final states (charge transfer)
or physical differences between the two spin-orbit components induced by ge-
ometry, for example variations in the molecular field splitting components. As
can be seen, all of the valence states in the dimer are delocalized to some ex-
tent, so it is not unreasonable to believe that this might be the case also for the
RAS intermediate states. Such delocalization would explain the effects seen
above, since it would imply that many of the excited electrons are delocal-
ized, thereby forming an, in principle, core ionized final state which would
then undergo normal Auger decay. Normal Auger features are not visible in
our RAS spectra because of the narrow energy region we examine. However,
they will be counted when performing NEXAFS spectroscopy, which makes
this a plausible explanation for what is seen in the spectra.
58
6.1.3 Sodium dimers
Another way of inducing dissociation of a molecule is by performing core
photoionization. A core-ionized, bound state can relax by an Auger process.
The final state can then be dissociative. In a spectrum where Auger electrons
are measured, a dissociative state will give rise to a broad, featureless bump.
The width of this bump can be related to the slope of the potential surface
of the final state. Figure 6.6(a) shows a schematic picture of this relationship.
Such an analysis was performed in paper III. A beam of sodium dimers (or,really, Na2 molecules; the dimers are covalently bound with the outermost
3s electrons forming a binding molecular σ orbital) was chosen as a proto-
type system, since its electron structure is very simple, with only two valence
electrons, and since it is a case where, in the atom, Auger decay from the
core-ionized 2p−1 state is not allowed while it becomes possible in the dimer.
1 2 3 4 5
FC-region(ground state)
FC-region(excited state)
Augerwidth
Final state
Ground state
Excited state
Ene
rgy
(arb
.un
its)
Interatomic distance (A)
EffectiveFC-region
0
1
(a) A schematic picture of how the
width of an Auger feature can be esti-
mated from knowledge about the poten-
tial curves of a molecule. In this case, an
“effective” FC-region was defined, to ac-
commodate the fact that the transition to
the dissociative state takes place from an
excited state.
Dimer feature
16 18 20 22 24
Inte
nsity
(arb
.un
its)
Kinetic energy (eV)
Auger width
1,3S 3P
0
1
(b) The Auger spectrum from the 2p-ionized Na2 molecule. The width of the
Auger feature is well estimated using the
model with an effective FC-region.
Figure 6.6: Potential curves and Auger spectrum for the 2p-ionized Na2 molecule.
In this work, we have modeled the Auger final state by a pure Coulomb
potential on the form
Vcol =q1 ·q2
r(6.1)
where the charges q1 and q2 were each set to one. This is based on the as-sumption that, in the final state, the two positive charges in in the valence
orbitals repel each other and end up on different Na atoms in the molecule.
Using the effective-FC-region approximation, we come to the conclusion that
59
a pure Coulomb potential represents the Auger final state very well in this
case. Another point in favor of this conclusion is the fact that the Na+ ions in
the final state have smaller radii than the the neutral atoms that constitute the
molecule; thus the overlap between the outermost Na orbitals will be much
smaller in the final state than in the initial state (2.32 Å [12] as compared to
3.08 Å [62]).
The main point of this paper, however, is to assign the molecular field split-
ting and life-time of the Na2 molecule Na 2p core hole. This was done usingXPS, in combination with ab initio calculations. Density functional theory(DFT) [63] was employed to calculate the inter-nuclear distance in ground
state, and in the core-ionized state (using the Z+1 approximation [64]). The
experimental spectrum was subsequently fitted using the calculated vibra-
tional progressions. Figure 6.7(a) shows 2p binding energy region from boththe atom and the molecule, while figure 6.7(b) shows only the molecular spec-
trum with the fit components. The molecular field splitting was determined to
be 42±10 meV and the core-hole lifetime was found to be 15±8 fs from theLorentzian width of the fitted peaks. The spin-orbit splitting corresponds well
to previous studies of solid sodium [65], with a value of roughly 160 meV.
However, the lifetime of the core-hole is much shorter than what has previ-
ously been reported [66]. Even though many competing processes take place
in the solid state that can affect the lifetime, this might indicate that the life-
time in the solid state might have been overestimated.
39 38.5 38 37.5 37 36.5 36
x 15
1P1
3P0
3P1
3P2
2p1/2
2p3/2
hν = 61 eV
Inte
nsity
(arb
.un
its)
Binding Energy (eV)
1
0
(a) An overview spectrum of the 2p bind-ing energy region for the Na atom and for
the Na2 molecule.
36.3 36.2 36.1 36 35.9 35.836.4
hν = 61 eV
2p1/2
2p3/2
Inte
nsity
(arb
.un
its)
Binding Energy (eV)
Em
0
1
(b) The XPS spectrum of the molecule.
The solid line represents the fitted line-
shape, the dotted lines represent the in-
dividual fit components and the vertical
bars represent the vibrational bar spec-
trum.
Figure 6.7: XPS spectra from the sodium atom and the sodium dimer molecule.
6.1.4 Conclusions
Synchrotron radiation based techniques were used to probe dissociation
dynamics in various systems. It was shown that dissociative states of
60
molecules in the free phase is not necessarily dissociative when the same
kind of molecule is in a cluster. The reason for this is, as of yet, not entirely
clear, but with more theoretical effort and better experimental set-ups, the
future of this kind of experiments is very promising. It was also shown that it
is possible to determine the core-hole lifetime and molecular field splitting
in alkali-metal dimers with a good accuracy. This is highly interesting, since
there are no solid-state limitations and effects to take into account when
performing the analysis of effects such as molecular field splitting and
lifetime.
6.2 The structure of doped and molecular clusters
Paper IV-VI deals with the geometric structure of doped and pure free molec-ular clusters. Not much is known about these structures a priori, since it isrelatively recently that experimental and computational tools have become
available that allow for reasonably accurate structure determination of clus-
ters [47, 57, 67, 68, 69]. Performing calculations on large clusters (several
thousands of atoms or molecules) is still prohibitively computationally ex-
pensive, but the feasible size limit has been pushed upwards to at least a few
thousands of atoms or molecules. In such calculations, a classical or semi-
classical approach is most often taken, using molecular dynamics (MD) or
Monte-Carlo methods to predict the geometric structures of clusters. Limita-
tions to these models are mainly that electronic effects, such as magnetism, are
not properly taken into account. This is not always a problem, but, for exam-
ple, in the case of oxygen clusters, the paramagnetic nature of the constituent
molecules dramatically influences the phase diagram [57].
In paper IV, the diffusion behavior of oxygen molecules adsorbed on large(N � 〈8000〉) Ar host clusters has been investigated. There are many compre-hensive theoretical studies of the diffusion behavior of atoms and molecules
doped on inert host clusters [68, 69], and also a range of measurements per-
formed by fluorescence spectroscopy [34, 35]. Recently, XPS spectra of polar
molecules adsorbed on Kr clusters were also published [70]. Figure 6.8(a)
shows the X-state UPS spectrum of oxygen molecules which are doped on
Ar. An important point here is that intra-molecular vibrations can still be re-
solved even in the cluster feature. This indicates that there is no significant
band formation between the different oxygen molecules in this case. That it is
the case does not come as a surprise, since the oxygen valence levels are fairly
confined to the molecule, and since the oxygen molecules probably are fairly
sparse in the Ar matrix.
This allows us to use UPS as our structural probe, instead of using XPS,
which is the most commonly used technique for this purpose [47, 48, 71, 72,
73, 74, 75, 75, 76]. Using UPS has several advantages, especially in the case
of oxygen. First, the photon flux at the beamline where the experiment was
61
13 12.5 12 11.5 11 10.5
Binding energy (eV)
Inte
nsity
(arb
.un
its)
hν=60 eV
A
B
Pd=18 mbar
*
(a) The O2 X-state in O2/Ar clusters. The
vertical solid bars mark the molecular FC
profile. The dashed vertical bars mark the
same FC profile shifted to a lower bind-
ing energy to try to match the cluster
feature. The * marks the region of the
spectrum which cannot be described us-
ing just one vibrational progression.
Vertex
Edge
Face
Interface
Bulk
13 12.5 12 11.5 11Binding energy (eV)
Inte
nsity
(arb
.un
its)
Pd=9 mbar
Pd=14 mbar
Pd=18 mbar
Pd=29 mbar
hν=60 eV
(b) Spectra of the O2 X-state in the
O2/Ar clusters at different doping con-
ditions. The shaded areas represent ion-
ization at different sites in the cluster. As
the doping pressure increases, the black
(bulk) feature grows dramatically.
Figure 6.8: UPS spectra of the O2 X-state in O2/Ar clusters.
performed is optimal at energies that access the valence region. Second, the
resolution is not limited by the lifetime of the vacancy, since the typical decay
pathway of a valence ionized state is radiative decay. In the case of oxygen,
it is very hard to probe the core-levels with an acceptable resolution and still
have a sufficient photon flux at the beamline utilized in this experiment. A
complicating fact is that the valence states are quite rich in vibrational struc-
ture; however, this is not such a big problem as one might first think, since the
valence spectra of molecules are often well known and extensively studied.
From figure 6.8(a) it can be seen that the simple shift-and-broadening ap-
proach that is usually applied in the case of XPS does not work very well
to describe the oxygen cluster spectral feature. Instead, we have employed
a modified version of this approach. To be able to quantify where the oxy-
gen molecules are sited in the Ar matrix, we have used electrostatic calcu-
lations (Tinker package, using the AMOEBA polarizable forcefield [77]) to
get the polarization screening shift values for different sites in and on the
Ar cluster. With the help of depth-profiling, (pure) surface/bulk splitting has
been ruled out as the source of the discrepancy between the simplest shift-
and-broadening model and the experimental spectrum at low pressures. Fig-
ure 6.8(b) shows oxygen cluster UPS spectra as a function of the doping pres-
sure. As the doping pressure increases, it can be seen from the fits that the
62
feature at the bulk energy (black) increases drastically. This is expected, since
at some point, the Ar host cluster will become liquid, and oxygen molecules
will, in principle, be able to migrate freely in the Ar host. At lower doping
pressures, the oxygen molecules does not penetrate into the bulk at all, but
stay in the surface and interface layers.
In general, the results of this paper agrees well with how one would expect
that oxygen behaves when doped on Ar by considering its van der Waals in-
teraction potential. That no band formation is observed is a fact that can allow
UPS to be used as a structural probe in many other mixed cluster experiments
with similar properties. The results of this study indicate that a high degree of
control of the radial structure and diffusion behavior of molecules adsorbed
on inert host clusters can be achieved.
Paper V presents a doping study, using XPS of Ar clusters doped with Kr
and Xe. This technique has the great advantage that it provides a (in principle)
linear relation between the observed spectral intensity and the actual com-
position of the sample. This is a modified truth, because of issues regarding
the electron effective attenuation length of electrons in the cluster [45, 46].
Figure 6.9 shows XPS spectra of the two systems, at different doping condi-
tions. One can clearly see the evolution of the dopant features in these spec-
tra. For Kr, there is a clear radial structuring, where most of the Kr ends up
on the surface of the cluster, in contrast to the co-expanded case, where Kr
atoms are mostly found in the bulk, and Ar atoms are mostly found on the
surface [73, 74].
surface
surface
bulk
bulk
0.0 0.0-1.0 -1.0
pure
pure
doped
doped
co-e
xp
co-e
xp
Ar 2p−1
3/2 XPS, hω = 270 eV Kr 3d−1
5/2 XPS, hω = 115 eV
Inte
nsi
ty[a
rb.u
nit
s]
Inte
nsi
ty[a
rb.u
nit
s]
Relative binding energy [eV ] Relative binding energy [eV ]
(a) XPS spectra of Ar clusters doped and
co-expanded with Kr.
-1.00.00.0 -0.5-0.5 -1.0
pure
pure
Doped:
40
mbar
Doped:
40
mbar
Doped:
2m
bar
Doped:
2m
bar
co-e
xp
co-e
xp
Ar 2p−1
3/2 XPS, hω = 270 eV Xe 4d−1
5/2 XPS, hω = 112 eV
Inte
nsi
ty[a
rb.u
nit
s]
Relative binding energy [eV ]Relative binding energy [eV ]
(b) XPS spectra of Ar clusters doped and
co-expanded with Xe.
Figure 6.9: XPS spectra of Ar/Kr and Ar/Xe clusters.
For Xe, the situation is more complex. At low doping pressures, the dopant
remains in the surface layer, with a polarization screening that can be traced
to the Ar matrix. For higher doping pressures, the Xe XPS feature looks very
much like the pure Xe clusters, which are also shown in the figure. This has
been interpreted as island formation on the Ar surface; a schematic picture of
this is shown in figure 6.10. This is also very different from what is seen from
63
the co-expanded case, where there is a clear interface feature in the spectra,
indicating a very high radial segregation [72].
Low doping pressures High doping pressures
Figure 6.10: A schematic picture of how Xe forms islands as it aggregates on the Arclusters.
The main conclusion is that it is, indeed, possible to create free clusters
by post-expansion doping that do not have a thermodynamically favorable
structure, such as has been predicted in, for example, [68]. The other method
of creating mixed rare-gas clusters, co-expansion, gives a cluster distribu-
tion in which clusters have a much more thermodynamically favorable struc-
ture [72, 73, 74]. This is not surprising, considering that the cluster flight dis-
tance from the expansion nozzle to the probing region is in the order of several
cm, allowing the clusters time to cool down and minimize their energy by at-
taining a more stable configuration. In the doped case, the situation is such
that the host clusters have time to cool down even before the doping takes
place. Thus the doping process might or might not heat the host cluster up to
the melting temperature, allowing for control of the radial distribution of the
dopant atoms in the Ar host.
Another subject about which very little is known is the structure of molec-
ular clusters. These clusters can be formed by a combination of a wide range
of inter-molecular forces, such as van der Waals forces, dipole interactions
or hydrogen bonds. In paper VI, the structure of CH3Br clusters was studiedby XPS and UPS spectroscopy. The polarization screening shift value trend
in both XPS and UPS follow a similar pattern; the orbitals that are localized
on the methyl group have a larger polarization screening shift, while the ones
that are localized on the bromine atom have a smaller polarization screening
shift. States involving orbitals that are delocalized over the molecule have a
polarization screening shift in between the two. Figure 6.11 shows UPS and
XPS spectra from large bromomethane clusters. The shift values are marked
in the figure.
These shift values tell us that the bromine end of molecules are mostly
screened by methyl groups, which have a small polarizability compared to
bromine atoms. It also tells us that the methyl carbon is mostly screened by
bromine atoms, which are highly polarizable. All things considered, this is
strong evidence for a packing in the cluster matrix similar to that of the bulk
material [78], and to that which has been proposed for the bromomethane
dimer [58]. Recent calculations on the local structure of bromomethane clus-
ters also support these findings [79].
64
Inte
nsity (
a. u
.)
78 77 76 75Binding Energy (eV)
3d3/2
≈ 0.94 eV
(a) XPS Br 3d
74
3d5/2
≈ 0.94 eV
Molecule
Cluster
293 292 291 290Binding Energy (eV)
(b) XPS C 1s
Inte
nsity (
a. u
.)
≈ 1.25eV
Molecule
Cluster
(a) XPS spectra of the Br 3d and C 1sedges in bromomethane clusters.
2e
11.0 10.0 9.0
Binding Energy (eV)
10.5 9.5 8.5
Inte
nsity (
a. u.)
≈ 0.9 eV
11.0 10.010.5
(a) UPS
Molecule
Cluster
18 16 14Binding Energy (eV)
17 15 12
Inte
nsity (
a. u.)
17 1416
(b) UPS
13
15 13
1e
3a1
Molecule
Cluster
≈1.1 eV ≈1.0 eV
(b) UPS spectra of the 1e, 3a1 and 2e va-lence states of bromomethane clusters.
Figure 6.11: XPS and UPS spectra from pure bromomethane clusters.
6.2.1 Conclusions
The structure of neutral free clusters is hard to determine using experiments,
but with the help of ab initio and MD calculations, XPS and UPS spectra cangive accurate information. The fact that the techniques are inherently surface-
sensitive give a distinct advantage when investigating the structure of nano-
scale objects such as clusters. In this work, the behavior of rare-gas atoms
and molecules adsorbed on rare-gas cluster hosts has been studied using these
techniques together, giving a more clear picture of what influences the final
structure of the cluster system. XPS has also been used to characterize the
structure of a cluster consisting of polar molecules; the same reasoning can be
applied to clusters of other polar molecules to investigate local geometry and
packing.
65
6.3 Laser excited metal vapors
Papers VII and VIII deal with x-ray photoelectron spectroscopy on vapors oflaser-excited sodium and potassium, respectively. In photoemission, there is
a type of processes known as electron shake. These shake processes can beclassified as either shake-off, shake-up or shake-down. Figure 6.12 shows a
schematic picture of the final states in each of these processes. In a shake-off
type process, the outgoing photoelectron loses enough energy to the valence
electronic configuration to cause valence ionization. Thus the final state con-
sists of a photo-electron, a shake-off electron and a doubly charged ion. A
shake-up process is similar to the shake-off in that the outgoing photoelectron
loses energy to the valence electrons. It does not, however, cause ionization,
but rather a valence excitation; i.e. a valence electron is excited into an unoccu-
pied orbital. Both of these processes reduce the kinetic energy of the outgoing
photoelectron [80]. The third possibility, shake-down, is only possible when
photo-ionizing an excited state. In this case, the photoelectron does not lose
energy to the valence electrons, but rather gains energy while a de-excitationof the initial state takes place.
E
Ground state Shake-up Shake-off Excited state Shake-down
hνhν
Ground state shake Excited state shake
Figure 6.12: Three different shake processes. On the left hand side, shake-up and
shake-off from the ionized ground state, on the right shake-down from the ionized
excited state.
The first experimental observations of the shake-down process are pre-
sented in paper VII. The relative simplicity of the sodium atom (only s andp orbitals, and only one electron in the valence shell) and the fact that the3s− 3p gap is in the optical range (588.995 nm, 2.1 eV) makes it an idealcandidate for studies of the shake-down process itself. It turned out to be pos-
sible to directly compare the direct photoemission lines from the ionization
2p63s + hν → 2p53s+ e− to the shake-down photoemission lines, 2p63p +hν → 2p53s+ e−. The difference in kinetic, and thus binding energy of the
66
two kinds of photoelectrons was shown to be 2.1 eV, corresponding to the
excitation energy of the laser light.
39 38.5 38 37.5 37 36.5 36 35.5binding energy (eV)
Figure 6.13: Overview and detailed decomposition of the shake-down features of
laser-excited Na atoms.
Figure 6.13(a) shows the measured 2p XPS spectrum. At higher bindingenergy, the ground state multiplet features can be seen, at higher energies are
the shake-features; these are, as expected, a shifted copy of the ground state
multiplet lines. As can be seen from figure 6.13(b), there is another feature
overlapping the shake-down lines, this is attributed to the ground state 2pphotoemission from the Na2 molecule, as was discussed in paper III. To con-serve parity between the initial and final states, it turns out to be necessary
that the outgoing 2p photoelectron wave is p-symmetric, and that the excited3p electron undergoes a dipole transition to the 3s orbital. It was also possibleto probe the alignment of the excited atoms; it is described well by the frame-
work of linear dichroism [81]. The most prominent part in the description of
the dichroism in this case is the term which is independent on the relative po-
larizations between the laser light and the synchrotron light. This shows that
the photoelectrons are primarily emitted in the direction of atomic alignment.
Figure 6.14(a) shows the shake feature measured at different polarization an-
gles, and figure 6.14(b) shows the fitted dichroism function.
Shake-down was studied also in paper VIII, but with the purpose of char-acterizing final states that are only weakly excited in direct photoemission.
In this case, 3p photoemission from potassium atoms was examined. For thelaser part of the experiment, the exciting radiation was tuned to 769.9 nm,
1.61 eV. This pumped the K 4s 2S1/2 → 4p 2P1/2 transition. In the direct 3pphotoemission from potassium, there is a significant configuration interac-
tion between the 3p54s and 3p53d states. There are also a number of extrafinal states from the 3p53d configuration, namely the 3p53d 3F states. The
ground state of potassium is S-symmetric, and thus one would not expect tosee F-symmetric final states in the direct p-photoelectron emission. Even so,
Figure 6.14: Angle dependence of the shake-down spectral features.
such states are visible. Figure 6.15(a) shows the ground state 3p photoelec-tron spectrum. The observed F-symmetric features stem from configuration
interaction, shake-up photoemission or higher-order photoemission.
25.5 25 24.5 24binding energy (eV)
inte
nsity
(ar
b. u
nits
)
K 3p PES× 200
3p54s
3P
2
3p53d
3p54s
1P
1
3p5(3d 4s)
1 3P
3F
2
3F
3
3F
4
(a) Ground state K 3p direct photoe-
mission spectrum. The dashed line is
the same spectrum as the solid line, but
scaled by a factor of 200 to show the very
weak F-symmetric final state features.
25.5 25 24.5
inte
nsity
(ar
b. u
nits
)
24 23.5 23binding energy (eV)
inte
nsity
(ar
b. u
nits
)
ground state K
laser excited K
3P
1P
3F
(b) Comparison between the shake-down
features (lower panel) and the direct 3pphotoemission. The relative intensity of
the F-symmetric final states to the P-
symmetric final states varies drastically
between the two cases.
Figure 6.15: XPS spectra of the K3p level.
We have also observed the 3p53d 1,3D lines in the direct photoemission
spectrum. In the shake-down spectra, a difference in excitation scheme from
the direct photoemission case (in direct photoemission, the 3d excited finalstates can only be reached via shake-up or by configuration interaction with
the 4s states, whereas in the shakedown process, both the 4s and the 3d fi-nal states are excited via the shakedown of the 4p electron) changes the ratiobetween the 3d and 4s excited states drastically. Figure 6.15(b) shows the 3pspectrum from the laser excited initial state. The 3d final states are much more
68
likely to be populated in the excited case. By studying the variations in inten-
sity ratio between features from the same final states in direct photoemission
and from shakedown, it was thus possible to quantify which final states that
are likely to have an electron in the 3d orbital, and which final states havean electron in the 4s orbital. The shake-down structure of potassium is con-
siderably much more complex than that of sodium, which is not surprising
considering the larger number of electrons.
6.3.1 Conclusions
The combination of synchrotron and laser radiation has been utilized to
perform spectroscopy on excited metal atoms. This excitation and probing
scheme allowed for observation of the shake-down process, which is
observable only from excited states. The difference in decay pathway
between the single photoionization and the shake-down process allows
characterization of the final states in a complex system such as potassium,
even though the final states in the two cases are the same. For sodium, it was
seen that an aligned ensemble of excited atoms were created during the laser
excitation, through performing a study of the linear dichroism. These results
provide a good basis from which it is possible to begin to interpret other
laser-induced effects in similar spectra.
69
7. Future outlook
The synchrotron has facilitiated an almost exponential growth of knowledge in
the surface science community. New facilities and lightsources will continue
to open up new possibilities for new types of experiments and for experiments
on systems that have, hithterto, been impossible to investigate for various rea-
sons. The type of experiments that have been performed in this thesis will
benefit greatly from these developments, since today, they are really touching
on the limits of what is possible.
The combination of synchrotron radiation and lasers is especially promis-
ing, because of the unique ability to study the core electron structure of sys-
tems that are involved in a lot of processes in nature. A few such very in-
teresting prospects are the studies of atmospheric photochemistry, chemical
catalysis and various biological systems. The knowledge about such systems
today is very limited, and there are almost endless possibilities to imagine new
ways of studying them in the future.
The continued development of theory goes hand-in-hand with the experi-
ments, and is progressing at a quicker pace than ever before; on one hand,
computers are getting faster and faster, allowing for studies of larger systems
with “brute force” methods, on the other hand, very extensive work is being
performed in the areas of paralellization and algorithmic science, speeding
up computations by orders of magnitude in the best cases. Quantum molec-
ular dynamics is also becoming more usable because of these facts, which is
especially useful for systems such as clusters.
The interdisciplinary aspect of work like this is also highly fruitful, bring-
ing together the communities of biology, chemistry and physics to tackle ques-
tions that span these subject matters. As for the close future, the cluster studies
at the Uppsala group serve not only as interesting systems in themselves, but
also as finite model systems for the development of an understanding of pro-
cesses that take place in liquid systems; a key point if biologically important
systems are ever to be studied by techniques like the ones employed in this
thesis. It is the belief of many, including myself, that atomic, molecular and
cluster science has a very bright future, considering the need for minaturiza-
tion and development occuring in the nano-science and technology areas.
71
8. Acknowledgments
The number of people involved in bringing this thesis work to its conclusion is
large, and I thank each and everyone who has, in some way, contributed to it.
There are, of course, people who deserve special mention. First and foremost, I
would like to thank my supervisors, Olle, Gunnar andMaxim for always being
available to answer questions about all things practical and for all their encour-
agement over the years. I would also like to thank Svante, for his unflagging
confidence in his graduate students and for all our discussions. Joachim, thank
you for the laser work, it was a pleasure even if it was sometimes frustrating.
All of my fellow Ph. D. students, who provided fun and joy and a positive
atmosphere everyday at work, thank you. Especially, Andreas, with whom I
spent most of my time, writing, teaching, traveling and discussing. I would
like to thank the Finnish group from Oulu, and the Bergen group, for allowing
me to join in their beamtimes, and for allowing me to visit their universities
for experiments and course work. All of the people at Fysikum, for providing
an open and productive working environment, thank you! Also, thanks to the
MAX-lab staff, who provided the most odd bits and pieces for the experiments
to run smoothly.
I would also like to thank all of my friends, none mentioned, none forgotten,
for making living life a joy each day and for always being there when there
was a rough patch. Wherever and whenever you are around, there is always
fun to be had. All of the people working on a non-profit basis, producing the
fantastic stage shows known as spex, in which I have taken part over the years,
thank you. All members of HKF, who are just as deranged as one should be,
and who keep coming up with the most brilliant get-rich-quick-schemes all
the time. How come we’re not rich yet?
Finally, I’d like to thank my family, Mamma, Pappa, Mia and Sanna, for
being who you are, and for always encouraging me to do what I wanted to do.
73
Bibliography
[1] H. Haberland, editor. Clusters of Atoms and Molecules I & II, volume 52 ofSpringer Series in Chemical Physics. Springer, Berlin, 1994-1995.
[2] M. Schmidt, R. Kusche, B. von Issendorf, and H. Haberland. Irregular variations
in the melting point of size-selected atomic clusters. Nature, 393:238, 1998.
[3] G. Ganteför, G. Bröker, E. Holub-Krappe, and A. Ding. Photoionization thresh-
olds of rare gas clusters. J. Chem. Phys., 91:7972, 1989.
[4] P. W. Atkins and R. S. Friedman. Molecular Quantum Mechanics. OxfordUniversity Press, 3rd edition, 2001.
[5] M. Massimi. Pauli’s Exclusion Principle: The Origin and Validation of aScientific Principle. Cambridge University Press, 1st edition, 2005.
[6] B. H. Bransden and C. J. Joachain. Physics of Atoms and Molecules. PrenticeHall, 2nd edition, 2003.
[7] A. D. McNaught and A. Wilkinson. Compendium of Chemical Terminology.Blackwell Science, 2nd edition, 1997.
[8] D. Kedziera and A. Avramopoulos and M. G. Papadopoulos and A. J. Sadlej.
Electronic spectrum of the confined auride ion. Phys. Chem. Chem. Phys.,5:1096, 2003.
[9] A. A. Pavlychev and E. Rühl. Dynamic localization of inner-shell excitations in
free atomic and molecular clusters. J. Electron. Spectrosc. Relat. Phenom.,106:207, 2000.
[10] N. W. Ashcroft and N. D. Mermin. Solid State Physics. Brooks Cole, 1976.
[11] L. Pauling. The nature of the Chemical Bond. Cornell University Press, 3rd
edition, 1960.
[12] P. W. Atkins and J. de Paula. Physical Chemistry. Oxford University Press,7th edition, 2002.
[13] M. Born and R. Oppenheimer. Zur Quantentheorie der Molekeln. Ann.Physik, 84:457, 1927.
[14] H. Hertz. Über einen Einfluss des ultravioletten Lichtes auf die elektrische
Entladung. Ann. Physik, 31:983, 1887.
75
[15] A. Einstein. Über einen die Erzeugung und Verwandlung des Lichtes be-
treffenden heuristischen Gesichtspunkt. Ann. Physik, 17:132, 1905.
[16] Stefan Hüfner. Photoelectron Spectroscopy. Springer, New York, 3rd edition,2003.
[17] J. F. O’Hanlon. A Users Guide to Vacuum Technology. John Wiley & Sons,3rd edition, 2003.
[18] F. R. Elder, A. M. Gurewitsch, R. V. Langmuir, and H. C. Pollock. Radiation
from Electrons in a Synchrotron. Phys. Rev., 71:829, 1947.
[19] J. Schwinger. On the classical radiation of accelerated electrons. Phys. Rev.,75:1912, 1949.
[20] G. R. Burbridge. On Synchrotron Radiation from Messier 87. Astrophys.J., 124:416, 1956.
[21] J. D. Jackson. Classical Electrodynamics. John Wiley & Sons, 3rd edition,
1998.
[22] D. Attwood. Soft X-rays and extreme ultraviolet radiation. CambridgeUniversity Press, 1999.
[23] A. Einstein. Strahlungs-Emission und -Absorption nach der Quantentheo-
rie. Deutsche Physikalische Gesellschaft, 18:318, 1916.
[24] J. P. Gordon, H. J. Zeiger, and C. H. Townes. Molecular microwave oscillator
and new hyperfine structure in the microwave spectrum of NH3. Phys. Rev.,95:282, 1954.
[25] R. G. Gould. The LASER, Light Amplification by Stimulated Emission of
Radiation. In The Ann Arbor Conference on Optical Pumping, June 1959.
[26] E. Hecht. Optics. Addison Wesley, San Francisco, 4th edition, 2002.
[27] W. T. Silfvast. Laser Fundamentals. Cambridge University Press, 2nd edition,
2004.
[28] M. Huttula, M. Harkoma, E. Nõmmiste, and S. Aksela. A multipurpose electron-
ion spectrometer for measurements with synchrotron radiation. Nucl. Instrum.Methodes Phys. Res. A, 467-468:1514, 2001.
[29] R. Karnbach, M. Joppien, J. Stapelfeldt, J. Wörmer, and T. Möller. CLULU: an
experimental setup for luminescence measurements on van der Waals clusters
with synchrotron radiation. Rev. Sci. Instrum., 64:2838, 1993.
[30] H. Haberland, U. Buck, and M. Tolle. Velocity distribution of supersonic nozzle
beams. Rev. Sci. Instrum., 56(9):1712, 1985.
76
[31] O. F. Hagena and W. Obert. Cluster formation in expanding supersonic jets:
effect of pressure, temperature, nozzle size, and test gas. J. Chem. Phys.,56(5):1793, 1972.
[32] C. Bobbert, S. Schütte, C. Steinbach, and U. Buck. Fragmentation and reliable
size distributions of large ammonia and water clusters. Eur. Phys. J. D, 19:183,2002.
[33] M. Tchaplyguine, R. Feifel, R. R. T. Marinho, M. Gisselbrecht, S. L. Sorensen,
A. Naves de Brito, N. Mårtensson, A. Svensson, and O. Björneholm and. Selec-
tive probing of the electronic structure of free clusters using resonant core-level
spectroscopy. Chem. Phys., 289:3, 2003.
[34] T. E. Gough, D. G. Knight, and G. Scoles. Matrix spectroscopy in the gas phase:
IR spectroscopy of argon cluster containing SF6 or CH3F. Chem. Phys. Lett.,97:155, 1983.
[35] S. Goyal, G. N. Robinson, D. L. Schutt, and G. Scoles. Infrared spectroscopy
of SF6 in and on argon clusters in an extemded range of cluster sizes: finite-size
particles attaining bulklike properties. J. Chem. Phys., 95:4186, 1991.
[36] J. W. Rabalais. Principles of Ultraviolet Photoelectron Spectroscopy. JohnWiley & Sons, 1977.
[37] R. Feifel, M. Tchaplyguine, G. Öhrwall, M. Salonen, M. Lundwall, R. R. T.
Marinho, M. Gisselbrecht, S. L. Sorensen, A. Naves de Brito, L. Karlsson,
N. Mårtensson, S. Svensson, and O. Björneholm. From localised to delo-
calised electronic states in free Ar, Kr and Xe clusters. Eur. Phys. J. D,30:343, 2003.
[38] K. Kimura et al. Handbook of He I Photoelectron Spectra of FundamentalOrganic Molecules. Halsted, New York, 1981.
[39] A. Thompson and D. Vaughan, editors. X-Ray Data Booklet. LBNL, Berkeley,USA, 2001.
[40] T. Koopmans. Über die Zuordnung vonWellenfunktionen und Eigenwerten
zu den Einzelnen Elektronen Eines Atoms. Physica, 1:104, 1933.
[41] G. B. Armen, J. Tulkki, T. Åberg, and B. Crasemann. Quantum theory of post-
collision interaction in inner-shell photoionization: Final-state interaction
between two continuum electrons. Phys. Rev. A, 36:5606, 1987.
[42] P. van der Straaten, R. Morgenstern, and A. Niehaus. Angular dependent post-
collision interaction in auger processes. Z. Phys. D: At. Mol. Clusters, 8:35,1988.
77
[43] O. Björneholm, F. Federmann, F. Fössing, and T. Möller. Core level binding en-
ergy shifts and polarization screening: A combined experimental and theoretical
study of argon clusters. J. Chem. Phys., 104:1846, 1996.
[44] M. Tchaplyguine, R. R. Marinho, M. Gisselbrecht, N. Mårtensson, S. L.
Sorensen, A. Naves de Brito, R. Feifel, G. Öhrwall, M. Lundwall, S. Svens-
son, and O. Björneholm. The size of neutral free clusters as manifested in the
[45] A. Jablonski and C. J. Powell. Relationships between electron inelastic mean
free paths, effective attenuation length and mean escape depths. J. Elect. Spec-trosc. Relat. Phenom., 100:137, 1999.
[46] A. Jablonski and C. J. Powell. The electron attenuation length revisited. Surf.Sci. Rep., 47:33, 2002.
[47] H. Bergersen, M. Abu-samha, J. Harnes, O. Björneholm, S. Svensson, L. J.
Sæthre, and K. J. Børve. Size of neutral argon clusters from core photoelec-
tron spectroscopy. Phys. Chem. Chem. Phys, 16:1891, 2006.
[48] M. Abu-samha, K. J. Børve, G. Öhrwall L. J. Sæthre, H. Bergersen, T. Rander,
O. Björneholm, and M. Tchaplyguine. Lineshapes in carbon 1s photoelectron
spectra of methanol clusters. Phys. Chem. Chem. Phys., 21:2473, 2006.
[49] O. Björneholm, A. Nilsson, A. Sandell, B. Herrnäs, and N. Mårtensson. Deter-
mination of time scales for charge-transfer screening in physisorbed molecules.
Phys. Rev. Lett., 68:1892, 1992.
[50] A. Naves de Brito, A. Naves de Brito, O. Björneholm, J. Soares Neto,
A. Machado, S. Svensson, A. Ausmees, S. J. Osborne, L. J. Sæthre, O.-P. Saira-
nen, A. Kivimäki, E. Nõmmiste, and S. Aksela. Fast dissociation of resonantly
core excited H2S studied by vibrational and temporal analysis of the auger spec-
tra. J. Mol. Struct., 394:135, 1997.
[51] P. Morin and I. Nenner. Atomic Autoionization Following Very Fast Disso-
ciation of Core-Excited HBr. Phys. Rev. Lett., 56:1913, 1986.
[52] P. Morin and I. Nenner. Photoionization Decay of Core Excited Molecules.
Phys. Scr., T17:171, 1987.
[53] I. Nenner, P. Morin, P. Lablanquie, M. Simon, N. Levasseur, and P. Millie. Pho-
todissociation of Core Excited Molecules. J. Electron Spectrosc. Relat.Phenom., 52:623, 1990.
[54] I. Hjelte, O. Björneholm, V. Carravetta, C. Angeli, R. Cimiraglia, K. Wiesner,
S. Svensson, and M. N. Piancastelli. Constant-atomic-final-state filtering of
dissociative states in the O1s−1σ∗ core excitation in O2. J. Chem. Phys.,123:064314, 2005.
78
[55] P. Kuiper and B. I. Dunlap. The σ∗ absorption peak at the oxygen 1s edge of O2:Exchange splitting, ultrafast dissociation, and atomiclike Auger spectra.
J. Chem. Phys., 100:4087, 1994.
[56] A. Rosso, T. Rander, H. Bergersen, A. Lindblad, M. Lundwall, S. Svensson,
M. Tchaplyguine, G. Öhrwall, L.J. Sæthre, and O. Björneholm. The role
of molecular polarity in cluster local structure studied by photoelectron spec-
troscopy. Chem. Phys. Lett., 435:79, 2007.
[57] F. Calvo and G. Torchet. Onset of crystalline order in oxygen clusters. J. Crys.Grow., 299:374, 2007.
[58] Y. Futami, S. Kudoh, F. Ito, T. Nakanaga, and M. Nakata. Structures of methyl
halide dimers in supersonic jets by matrix-isolation infrared spectroscopy and
quantum chemical calculations. J. Mol. Struct., 690:9, 2004.
[59] P. J. Hay and W. R. Wadt. Ab initio effective core potentials for molecular
calculations. Potentials for K to Au including the outermost core orbitals. J.Chem. Phys., 82:299, 1985.
[60] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R.
Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M.
Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scal-
mani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota,
R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai,
M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo,
J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi,
C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Sal-
vador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Fores-
man, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov,
G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith,
M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill,
B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, and J. A. Pople. Gaussian 03,
[72] M. Tchaplyguine, M. Lundwall, M. Gisselbrecht, G. Öhrwall, R. Feifel,
S. Sorensen, S. Svensson, N. Mårtensson, and O. Björneholm. Variable sur-
face composition and radial interface formation in self-assembled free,
mixed, Ar/Xe clusters. Phys. Rev. A, 69:031201, 2004.
[73] M. Lundwall, M. Tchaplyguine, G. Öhrwall, R. Feifel, A. Lindblad, A. Lind-
gren, S. Sörensen, S. Svensson, and O. Björneholm. Radial surface segregation
in free heterogeneous argon/krypton clusters. Chem. Phys. Lett, 392:433, 2004.
[74] M. Lundwall, H. Bergersen, A. Lindblad, G. Öhrwall, M. Tchaplyguine,
S. Svensson, and O. Björneholm. Preferential site occupancy observed in co-
expanded argon-krypton clusters. Phys. Rev. A, 74:043206, 2006.
[75] M. Lundwall, A. Lindblad, H. Bergersen, T. Rander, G. Öhrwall, M. Tchap-
lyguine, S. Svensson, and O. Björneholm. Preferntial site occupancy of krypton
atoms on free argon cluster surfaces. J. Chem. Phys., 125:014305, 2006.
80
[76] M. Lundwall, W. Pokapanich, H. Bergersen, A. Lindblad, T. Rander, G. Öhrwall,
M. Tchaplyguine, S. Barth, U. Hergenhahn, S. Svensson, and O. Björneholm.
Self-assembled heterogeneous argon/neon core-shell clusters studied by photo-
electron spectroscopy. J. Chem. Phys., 126:214706, 2007.
[77] J. W. Ponder. TINKER – software tools for molecular design, version 4.2,
2004. Washington University School of Medicine, St. Louis, U.S.A.
[78] P. N. Gerlach, B. H. Torrie, and B. M. Powell. The crystal structures and phase
transition of methyl bromide. Molec. Phys., 57:919, 1986.
[79] J. Harnes, M. Abu-samha, M. Winkler, L. J. Sæthre, and K. J. Børve. Single-
component CH3Cl and CH3Br clusters: Intermolecular interactions and struc-
ture studied by photoelectron spectroscopy and modeling. In manuscript.
[80] K. Siegbahn, C. Nordling, G. Johansson, J. Hedman, P. F. Héden, K. Hamrin,
U. Gelius, T. Bergmark, L. O. Werme, R. Manne, and Y. Baer. ESCA appliedto free molecules. North Holland Publishing Company, 1971.
[81] J. Schulz, Ph. Wernet, K. Godehusen, R. Müller, P. Zimmermann, M. Martins,
and B. Sonntag. Linear dichroism in the 4d photoionization of atomic europium.J. Phys. B, 35:907, 2002.
81
Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 375
Editor: The Dean of the Faculty of Science and Technology
A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally through theseries Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)