New aspects of electronic interactions of keV ions with matter

(1)

ACTA UNIVERSITATIS

UPSALIENSIS UPPSALA

Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1876

New aspects of electronic

interactions of keV ions with

matter

BARBARA BRUCKNER

ISSN 1651-6214 ISBN 978-91-513-0803-6

(2)

Dissertation presented at Uppsala University to be publicly examined in Polhemsalen, Ångströmlaboratoriet, Lägerhyddsvägen 1, Uppsala, Friday, 6 December 2019 at 09:00 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Marika Schleberger (Faculty of Physics and CENIDE, University Duisburg-Essen, Duisburg, Germany.).

Abstract

Bruckner, B. 2019. New aspects of electronic interactions of keV ions with matter. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1876. 61 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-513-0803-6. Low- and medium-energy ion scattering are powerful techniques to perform high-resolution depth profiling with sub-nanometer resolution. Typically, ions with primary energies between a few keV and a few hundred keV are used to probe the sample and backscattered projectiles are detected. To obtain highly accurate composition profiles, knowledge on physical processes governing ion-matter interaction is crucial. Apart from the main (back-)scattering process, which yields a detectable signal, the projectile loses energy in interactions with both electrons and nuclei (stopping) along its path in matter. In all these interactions, also the charge state of the probing particle can be altered. Information on this multitude of interaction mechanisms can be deduced from two different experimental approaches: either in backscattering or transmission geometry. Especially towards lower primary energies, available experimental data are found more scarce. This situation is particularly true for more complex targets, i.e. reactive transition metals and their compounds. This absence of quantitative information on energy loss or charge exchange processes hampers in many cases the quality of characterization despite the high technological relevance of these materials.

To contribute to an improvement of this status quo, this thesis focuses on (i) an analysis of sources of uncertainties in the evaluation of electronic energy loss, (ii) experiments to obtain stopping data for protons and He ions in different reactive samples and (iii) studies of charge exchange between projectile and target.

The first part presents a discussion of two possible sources of systematic errors, i.e. the composition of the investigated sample (thin films of the reactive transition metals often have low Z impurities like H, C, N and O), and deficiencies in the available models for the scattering potential. Concerning impurities in the films, it is shown that a correction according to Bragg's rule yields good agreement with data obtained from clean samples, even for energies down to a few keV, as long as the concentration levels of the impurities are low. In the second part experimentally deduced electronic energy loss data for transition metal nitrides as well as self-supporting Au and W-foils are presented. In the latter study a comparative approach using backscattering and transmission experiments is performed with measurements in both geometries conducted on the same sample, and in the same scattering chamber with only the position of the detector varied. In the final section the influence of surface oxygen on the energy spectra of backscattered ions at primary energies ≤ 5 keV is investigated. Depending on the host material O is found to enhance or suppress sub-surface signals. Additionally, also the change in neutralization efficiency for surface oxides in comparison to clean metal surfaces is studied for single crystalline Al(111) and Ta(111).

Keywords: ToF-MEIS, LEIS, electronic stopping, charge exchange, scattering potential Barbara Bruckner, Department of Physics and Astronomy, Applied Nuclear Physics, Box 516, Uppsala University, SE-751 20 Uppsala, Sweden.

(3)

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I. A note on extracting electronic stopping from energy spectra of back-scattered slow ions applying Bragg’s rule

B. Bruckner, D. Roth, D. Goebl, P. Bauer, D. Primetzhofer. Nucl. In-strum. Methods Phys. Res., Sect. B, 423, 82-86, 2018.

II. On the influence of uncertainties in scattering potentials on quanti-tative analysis using keV ions

B. Bruckner, T. Strapko, M. A. Sortica, P. Bauer and D. Primetzhofer. Manuscript submitted for publication.

III. Electronic interaction of light, keV ions in transition metal nitrides B. Bruckner, M. Hans, T. Nyberg, G. Greczynski, P. Bauer and D. Primetzhofer. Manuscript submitted for publication.

IV. Impact of the experimental approach on the observed electronic en-ergy loss for keV ions in self-supporting, thin films

B. Bruckner, P. M. Wolf, P. Bauer, D. Primetzhofer. In manuscript form.

V. The impact of surface oxidation on energy spectra of keV ions scat-tered from transition metals

B. Bruckner, P. Bauer, D. Primetzhofer. Appl. Surf. Sci., 479, 1287-1292, 2019.

VI. Neutralization of slow helium ions scattered from single crystalline aluminum and tantalum surfaces and their oxides

B. Bruckner, P. Bauer, D. Primetzhofer. Surf. Sci., 691, 121491, 2020.

(4)

My contributions to the included papers

Paper I

I helped with some of the measurements for electronic energy loss in Ni. I performed the simulations and comparisons between the different methods in the paper and wrote the initial manuscript as well as contributed in the revision process.

Paper II

I took part in planning of the study, obtained the experimental data, was in-volved in the analysis as well as data interpretation and wrote the manuscript. Paper III

I participated in the planning, performed all ion scattering experiments, anal-ysed the data and wrote the the initial manuscript.

Paper IV

I took part in the planning process, obtained and evaluated most of the experi-mental data and I wrote the initial manuscript.

Paper V

I participated in the planning of the study, performed the measurements, anal-ysed the data and wrote the initial manuscript as well as contributed to the revision process.

Paper VI

I took part in the planning, measured and analysed the data and wrote the manuscript including the revisions.

(5)

Papers related to the topics of this thesis which are not included:

VII. Electronic interaction of slow hydrogen and helium ions with nickel-silicon systems

T. T. Tran, L. Jablonka, B. Bruckner, S. Rund, D. Roth, M. A. Sortica, P. Bauer, Z. Zhang, D. Primetzhofer. Phys. Rev. A, 100, 032705, 2019. VIII. On the Z 1-dependence of electronic stopping in TiN

M. A. Sortica, V. Paneta, B. Bruckner, S. Lohmann, T. Nyberg, P. Bauer, D. Primetzhofer. Sci. Rep., 9, 176, 2019.

IX. Stopping cross section of vanadium for H+and He+ions in a large energy interval deduced from backscattering spectra

M. V. Moro, B. Bruckner, P. L. Grande, M. H. Tabacniks, P. Bauer, D. Primetzhofer. Nucl. Instrum. Methods Phys. Res., Sect. B, 424, 43-51, 2018.

X. Systematic analysis of different experimental approaches to mea-sure electronic stopping of very slow hydrogen ions

D. Roth, C. E. Celedon, D. Goebl, E. A. Sanchez, B. Bruckner, R. Steinberger, J. Guimpel, N. R. Arista, P. Bauer. Nucl. Instrum. Meth-ods Phys. Res., Sect. B, 437, 1-7, 2018.

XI. Electronic energy-loss mechanisms for H, He, and Ne in TiN

M. A. Sortica, V. Paneta, B. Bruckner, S. Lohmann, M. Hans, T. Nyberg, P. Bauer, D. Primetzhofer. Phys. Rev. A, 96, 032703, 2017.

XII. Electronic stopping of slow protons in transition and rare earth met-als: breakdown of the free electron gas concept

D. Roth, B. Bruckner, M. V. Moro, S. Gruber, D. Goebl, J. I. Juaristi, M. Alducin, R. Steinberger, J. Duchoslav, D. Primetzhofer, P. Bauer. Phys. Rev. Lett., 118, 103401, 2017.

XIII. Electronic stopping of slow protons in oxides: scaling properties D. Roth, B. Bruckner, G. Undeutsch, V. Paneta, A. I. Mardare, C. L. McGahan, M. Dosmailov, J. I. Juaristi, M. Alducin, J. D. Pedarnig, R. F. Haglund, Jr., D. Primetzhofer, P. Bauer. Phys. Rev. Lett., 118, 103401, 2017.

XIV. Charge exchange processes in He+/Cu scattering at low energy K. Khalal-Kouache, B. Bruckner, D. Roth, D. Goebl, P. Bauer. Nucl. Instrum. Methods Phys. Res., Sect. B, 382, 11-14, 2015.

(6)

(7)

Kurzfassung

Die Charakterisierung von Filmsystemen mit Dicken im Nanometerbereich kann mithilfe von Ionenstreuung mit Primärenergien zwischen einigen keV und mehreren hundert keV durchgeführt werden. Diese Energien werden in der Literatur mit nieder- und mittelenergetischer Ionenstreuung assoziiert (LEIS und MEIS). Um Ionenstreuung für quantitive Materialcharakterisierung zu verwenden, müssen die zugrunde liegenden physikalischen Prozesse ver-standen sein. In einem Streuexperiment können unterschiedliche Wechsel-wirkungen von Ionen mit dem zu untersuchenden Material auftreten: Groß-winkelstreuung an einem Atomkern, die zur Rückstreuung führt und damit Grundvoraussetzung ist um ein Signal zu detektieren; Wechselwirkungen mit Elektronen oder Atomkernen der Probe entlang der Trajektorie, welche zu einem kontinuierlichen Energieverlust führen. Zusätzlich kann sich entlang der Trajektorie auch der Ladungszustand des Projektils ändern. Für niedrige Energien gibt es wenig experimentelle Daten über die unterschiedlichen As-pekte der Wechselwirkung von Ionen mit dem elektronischen System. Dies betrifft unter anderem Anregungsprozesse von Projektil und Target als auch die Wechselwirkungspotentiale. Information über die auftretenden physikalis-chen Prozesse können mittels Ionenstreuung in verschiedenen Geometrien er-halten werden: Transmission und Rückstreuung.

In dieser Dissertation liegt der Fokus in (i) der Analyse von möglichen Fehlerquellen bei der Auswertung von Energieverlustdaten, (ii) dem Messen von experimentellen Energieverlustdaten in unterschiedlichen Materialien und (iii) der Untersuchung des Ladungsaustausches zwischen Projektil und Ma-terie. Im ersten Teil wird der Einfluss von zwei verschiedenen, systematis-chen Fehlerquellen berücksichtigt: Zusammensetzung der Probe, denn dünne Filme von Übergangsmetallen weisen oft Verunreinigungen wie H, C, N oder O, auf und Unsicherheiten in den verwendeten Streupotentialen. Im zweiten Teil dieser Dissertation werden Energieverlustdaten für Übergangsmetallni-tride und freistehende, dünne Folien präsentiert. An den freistehenden Folien wurden Messungen in Transmission und Rückstreugeometrie durchgeführt, wobei alle experimentellen Einstellungen gleich waren, nur die Position des Detektors wurde zwischen den Messungen variiert. Im letzten Teil wird der Einfluss von Sauerstoff an der Oberfläche verschiedener Metalle auf das Spek-trum von rückgestreuten Heliumionen mit Primärenergien ≤ 5 keV untersucht. Abhängig vom Metall kann Sauerstoff entweder das Signal aus Sub-Oberflä-chenschichten verstärken oder reduzieren. Weiters wurde an Einkristallober-flächen untersucht, wie sich Sauerstoffadsorption auf die Neutralisationswahr-scheinlichkeit von Heliumionen auswirkt.

(8)

(9)

Sammanfattning

Jonspridning vid låga och medelhöga energier utgör ett kraftfullt verktyg för att genomföra djupprofilering av material med upplösning bättre än en nanome-ter. Joner med primärenergi mellan ett fåtal och några hundratals keV an-vänds typiskt för att undersöka ett prov, och bakåtspridda projektiler detek-teras. För att generera exakta profiler av provets kemiska sammansättning krävs kännedom om de fysikaliska processer som styr interaktionen mellan en jon och ett material. Förutom den huvudsakliga (bakåt)spridningsprocessen som sker i en kollision med en atomkärna och ger upphov till den uppmätta signalen, förlorar projektilen energi genom interaktioner med både elektroner och atomkärnor längs sin väg i materialet. I alla dessa processer kan även jonens laddningstal förändras. Kunskap om interaktionsmekanismerna kan inhämtas genom experiment i två olika geometrier: bakåtspridning och trans-mission. Speciellt vid låga primärenergier är dock den tillgängligen datan begränsad. Detta gäller i synnerhet för komplexa prover såsom reaktiva över-gångsmetaller och deras föreningar. Avsaknaden av kvantitativ information om energiförlust och laddningsutbytesprocesser påverkar därför kvaliteten av materialkaraktärisering, trots den stora teknologiska relevansen av dessa ma-terial.

För att bidra till att förbättra den nuvarande situationen är denna avhan-dling fokuserad på (i) en analys av källor till osäkerhet vid utvärderingen av elektronisk energiförlust, (ii) experiment för att ta fram energiförlustsdata för protoner och heliumjoner i olika reaktiva material och (iii) studier av laddning-sutbyte mellan projektil och material.

I första delen diskuteras två möjliga källor till systematiska fel, nämligen provets sammansättning (tunna filmer kan innehålla föroreningar med lågt atomnummer såsom H, C, N och O) och brister i gängse modeller för sprid-ningspotentialen. Det visas att en korrektion av data för förorenande prover i enlighet med Braggs regel ger god överensstämmelse med data som inhämtats från rena prover. Vid undersökningar av den interatomära potentialens roll för ämnen med högt atomnummer, såsom Hf, observerades betydande skillnader i multipelspridningsbakgrunden mellan uppmätta och simulerade energispek-tra. Speciellt för data från HfN är olika korrektioner för skärmningslängd nödvändiga för att beskriva bidrag från både enkel- och multipelspridning i energispektrumet. Bäst överensstämmelse mellan experiment och simulering överlag fanns dock för TFM-potentialen.

I den andra delen presenteras data för elektronisk energiförlust i nitrider av övergångsmetaller samt fristående folier av Au och W. För att minska system-atiska fel vid utvärderingen av energiförlust användes olika tillvägagångssätt

(10)

vid de presenterade mätningarna: (i) utvärdering av antingen bredden eller höjden hos metallsignalen, och (ii) mätningar i olika spridningsgeometrier. I det senare fallet jämförs data inhämtad från experiment i transmissions- och bakåtspridningsgeometri på samma prov och i samma spridningskammare, genom att endast variera detektorns position. Det observerades att energispek-tra för energispek-transmitterade He-projektiler var extremt känsliga för små variationer i foliens struktur, på grund av skillnader i energiförlust mellan kanaliserande och slumpmässig geometri. Med ökande detektionsvinkel i transmissions-geometri upphör denna struktureffekt. För protoner gjordes inte någon lik-nande observation. Denna skillnad stämmer överens med tidigare rapporter-ade avvikelser från förutsägelser om elektronisk energiförlust för He-joner.

I det sista avsnittet undersöks inverkan av syre vid provets yta på ener-gispektra för bakåtspridda joner vid primärenergier ≤ 5 keV. Förändringen i neutraliseringseffektivitet för ytoxider jämfört med rena metallytor kan ob-serveras. Antalet bakåtspridda joner från enkristallint Al(111) och Ta(111) visade för båda fallen en matriseffekt beroende på ytans kemiska struktur. För Ta observerades en ytterligare förändring i energiberoendet hos neutraliser-ingsmekanismen, vilket huvudsakligen tillskrevs den öppna strukturen hos en bcc(111)-yta. Beroende på värdmaterial kan O dessutom förstärka eller un-dertrycka signaler som härrör från områden under provets yta.

(11)

Statutory Declaration

I hereby declare that the thesis submitted is my own unaided work, that I have not used other than the sources indicated, and that all direct and indirect sources are acknowledged as references.

This printed thesis is identical with the electronic version submitted.

(12)

(13)

Acronyms

ACOLISSA Analysis of the Charge Of Light Ions Scattered from Surface Atoms

AES Auger Electron Spectroscopy AN Auger Neutralization

BS BackScattering

DFT Density Functional Theory DOS Density Of States

EBS Elastic Backscattering Spectrometry ERDA Elastic Recoil Detection Analysis (or ERD) ESA ElectroStatic Analyzer

FEG Free Electron Gas

FWHM Full Width Half Maximum

HIERDA Heavy Ion Elastic Recoil Detection Analysis (or ERD) HOPG Highly Oriented Pyrolytic Graphite

IBA Ion Beam Analysis

ISS Ion Scattering Spectroscopy LEED Low-Energy Electron Diffraction LEIS Low-Energy Ion Scattering MC Monte-Carlo (simulation) MCP MicroChannel Plate

MEIS Medium-Energy Ion Scattering

ML MonoLayer

MS Multiple Scattering NRA Nuclear Reaction Analysis PIXE Particle Induced X-ray Emission

RBS Rutherford Backscattering Spectrometry RI Resonant (re)Ionization

RN Resonant Neutralization SCS Stopping Cross Section

SRIM Stopping and Range of Ions in Matter TFM Thomas-Fermi-Molière

TD-DFT Time-Dependent Density Functional Theory ToF Time-of-Flight

TR TRansmission

TRBS Trim for Rutherford BackScattering TRIM TRansport of Ions in Matter

VB Valence Band

XPS X-ray Photoelectron Spectroscopy ZBL Ziegler-Biersack-Littmark

(16)

(17)

1. Introduction

The first ion-solid interaction experiments with alpha particles from radium can be traced back to Marie Curie in 1898-1899 [1]. She already observed, that alpha particles travelling through matter can lose their speed. Roughly ten years later, first backscattering experiments were presented by Ernest Ruther-ford [2].Over a hundred years later, the irradiation of target atoms with MeV ions is used for material analysis, commonly referred to as Ion Beam Analysis (IBA). Depending on the employed technique different reaction productions of the ion-matter interaction are detected. However, in order to perform material analysis, knowledge on the occurring ion-solid interactions are crucial. Ions can be backscattered from a target nucleus in a large angle scattering event. When traversing matter ions lose energy due to interactions with the target nuclei and electrons, respectively. The physical quantity to describe this slow-ing down process is the stoppslow-ing power. Electronic stoppslow-ing can occur due to ionization or excitation of target electrons, which can result in the emis-sion of photons or secondary electrons, as well as charge exchange processes. Primary ions of a few MeV also can induce nuclear reactions resulting in dif-ferent reaction products, i. e. particles or photons of a characteristic energy. Depending on the detected species specific information on the sample can be deduced.

Typical questions in material analysis address thickness, composition or structure of the investigated sample. A commonly employed technique to ob-tain a composition-depth profile is Rutherford Backscattering Spectrometry (RBS) [3]. Analogue to the first scattering experiments ions (including alpha particles) are used to probe the sample and backscattered projectiles are de-tected. This technique shows a high sensitivity towards elements with higher atomic number. Whereas, it is more difficult to determine the content of light species within a matrix of heavy elements. Additionally, light elements such as H cannot be detected at all. Elastic Recoil Detection Analysis (ERDA) is a complementary technique, which allows the detection of light species includ-ing hydrogen [3, 4]. Here, the sample is, typically, irradiated with ions in a more grazing incidence compared to RBS and recoiled atoms and scattered ions are detected in forward scattering geometry. In both RBS and ERDA depth profiling of the sample can be performed via the known energy loss of the projectile while traversing matter. In Particle Induced X-ray Emission (PIXE), characteristic X-rays induced by light ions of commonly a few MeV are detected and low amount of trace elements can be identified [5]. For pri-mary ions of several MeV nuclear reactions can occur for certain

(18)

projectile-1. Introduction

nuclei combinations resulting in characteristic reaction products. Nuclear Re-action Analysis (NRA) is sensitive to isotopes and depending on the width of the resonance high-resolution depth profiling of certain elements can be per-formed [3, 6]. Commonly, this technique is employed to detect light species such as hydrogen or deuterium. In general, for material characterization a combination of the different techniques is performed to obtain a quantitative depth profile for all occurring elements in the sample [7]. The present meth-ods combine high accuracy with commonly, non-destructive analysis. The first advantage is rooted in the simplicity of ion-solid interaction at MeV ions. Ion trajectories can be understood as a series of adiabatic binary collisions with nuclei and tight bound electrons with minimum influence of target chem-istry. The second advantage is due to typically low number of probing particles ∼ 1013_{− 10}15_{interacting with a much larger number of target atoms.}

For high-resolution depth profiling of nm-thin films, which is increasingly demanded due to the ongoing miniaturization of e. g. electronic devices, lower primary energies are needed. A combination of alternative detector concepts has led to the developement of Medium-Energy Ion Scattering (MEIS) with sub-nanometer resolution [8–10]. For even lower primary energies as em-ployed in Low-Energy Ion Scattering (LEIS) surface and sub-surface contri-butions can be investigated [11–15]. Both MEIS and LEIS are also powerful techniques to study the crystalline structure of the investigated sample [16, 17] and with the use of large, position-sensitive detectors even real-space crystal-lography can be conducted [18–21]. The basic principle for MEIS and LEIS is the same as for RBS, however, the larger scattering cross sections require to properly take multiple scattering into account. Additionally, also electron capture and loss processes between the excited projectile and an electron from the target can be decisive for their application.

In order to perform accurate ion beam analysis as well as calculations of the range and deposited energy of ions, precise knowledge on the electronic as well as nuclear energy loss is of high relevance. Experimental reference data are particularly important, as the lower ion energies make a theoreti-cal prediction of electronic interactions much more complex. However, the amount of data available for certain projectile-target combinations below the stopping maximum especially for materials of high technological interest as transition metals and their compounds (nitrides, oxides) [22, 23] is scarce. Thus, in reality often stopping values based on semi-empirical physical mod-els are used resulting in inter- and extrapolation of experimentally existing data (SRIM [24]). However, for certain transition metals where no stopping data are available significant deviations between this predictions and experimental data were reported [Paper IX - 25, 26]. But not only a lack of data for certain systems but also the spread between different data sets for one projectile-target combination [22, 23] is problematic for applications. One possible source of uncertainty for the large differences is the detailed composition of the inves-tigated sample. Especially for the chemically reactive transition metals, the 2

(19)

fabrication of high quality films or even self-supporting foils is a demanding task and a thorough characterization of the impurities in the films is crucial in order to obtain reliable stopping values. This difficulty to produce high quality thin films can also explain the fact, that most of the published stopping data in literature are for noble metals [27]. Another possible source of uncertainty is given by the different experimental approaches, i. e. transmission and back-scattering geometry. The different probed impact parameters in the two scat-tering geometries could in theory yield different energy losses along the ion trajectory [28] and require in practise different approaches in the evaluation of data. In this context, it is worth to mention, that also scattering potentials are affected by decreasing the ion energy and can be sensitive to details of the electronic structure [29].

For electronic energy loss in compounds typically predictions by Bragg’s rule (additivity of the stopping power) are used. However, for energies around and below the stopping maximum, where energy loss is mainly due to ex-citation of valence electrons, these predictions are expected to be inaccurate due to significant differences in the valence bands for a metal and its com-pounds. These deviations from predictions by Bragg’s rule have already been thoroughly reported in literature [30–33].

In this thesis the focus is on improving the understanding of electronic inter-actions of medium and low energy ions. The influence of different sources of uncertainties as sample impurities and uncertainties in the available scattering potentials on the evaluation of electronic stopping is evaluated. Additionally, energy loss data for transition metal nitrides and self-supporting Au and W foils are presented. The latter measurements have been performed in trans-mission and backscattering geometry to observe a possible impact parameter dependence of the energy loss. In the last part the influence of surface oxygen on the intensity and shape of the spectrum of backscattered ions is investigated with LEIS. Before presenting the results of this thesis, I will give a short in-troduction in the ion-solid interactions in chapter 2, followed by an overview over the employed ion scattering techniques as well as the corresponding set-ups (chapter 3).

(20)

(21)

2. Ion - solid interactions

An energetic ion travelling in matter interacts with the target’s electrons as well as its nuclei. Both interactions are described in this chapter: In the first section, the focus will be on the description of a binary collision between two point charges. Here, scattering kinematics as well as the differential scattering cross section are introduced. In Sec. 2.2 the interaction of the projectile with a collective of target electrons is taken into account resulting in electronic energy loss of the projectile as well as possible charge exchange with the target medium.

2.1 Binary collision

2.1.1 Scattering kinematics

For a two-particle system a schematic drawing of the scattering geometry is shown in Fig. 2.1 in the laboratory frame. Note, within this thesis the labora-tory frame is used. The incoming projectile of mass m1 and primary energy

E0 is scattered from a stationary target with mass m2. In the scattering

pro-cess, the projectile transfers energy to the target atom resulting in a reduced final energy Effor the scattered particle and an energy of the recoiled atom Er.

The corresponding scattering angle ϑ and the recoil angle φ are indicated in Fig. 2.1. Typically, in ion scattering one defines the kinematic factor for the scattered particle as ksc≡ Ef E0 =   m1cos ϑ ± q m2₂− m2 1sin2ϑ m1+ m2   2 (2.1)

Figure 2.1: Schematic drawing of a two-particle scattering event in the laboratory frame. The projectile of mass m1and energy E0scatteres from a stationary target atom

(22)

2. Ion - solid interactions

and for the recoiled particle, respectively, as krecoil≡ Er E₀ = 4m1m2cos2φ (m1+ m2)2 . (2.2)

Both, kscand krecoilcan be derived via energy- and momentum-conservation;

details can be found e. g. in Ref. [34]. For m2

m₁ ≥ 1 the positive sign applies in

Eq. 2.1, whereas for | sin ϑ | ≤ m2

m₁ ≤ 1 two different scattering kinematics

oc-cur.

In first order, an ion scattering event can be described as binary collision between the projectile nucleus and a target nucleus. Note, the concept of scattering introduced above is also valid for scattering with other particles than atomic nuclei, a collision with the significantly lighter electron, how-ever, would not yield a backscattering event. In a backscattering experiment with known mass and energy of the projectile as well as scattering geomen-try, i. e. the angle of incidence and the position of the detector, the energy of the detected particle contains information on the mass of the scatterer or re-coiled atom via this classical model. However, due to backscattering from a target nuclei one cannot obtain information on the electronic characteristics, such as chemical bonding. For this investigations different probing beams are necessary, e. g. photons [34].

2.1.2 Scattering cross section

The number of target atoms per unit area can be determined via the fraction of detected and incoming particles and the differential scattering cross section

dσ

dΩ, which corresponds to the probability of the projectile being scattered into

a finite detector element dΩ. In classical scattering geometry, the differential scattering cross section depends on the scattering potential V (r), i. e. on the relation between the impact parameter b of the incoming ion and the scattering angle ϑ , and is given by

dσ dΩ= − b sin ϑ db dϑ. (2.3)

The simplest assumption for the scattering potential is the interaction of two point charges, e. g. two nuclei of atomic number Z1and Z2with an interaction

distance r. The resulting scattering cross section was first derived by Ruther-ford [2] and is therefore, called RutherRuther-ford cross section. In the center-of-mass frame it reads dσ dΩ R = Z1Z2e 2 4πε0E0 2 1 sin4(ϑ /2), (2.4) with e representing the elementary charge and ε0 the vacuum permittivity. A

detailed derivation of the Rutherford cross section can be seen e. g. in Ref. [34]. 6

(23)

2.1 Binary collision 2 3 4 5 6 10 0 10 1 10 2 10 3 10 -1 10 0 10 1 10 2 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 (b)

J. Dem arche et al. Sigm aCalc [ m b / s r ]

incident ion energy [MeV] 1 6 O( ) 1 6 O, = 170 Rutherf ord (a) 2.9 3.0 3.1 3.2 10 1 10 2 10 3 energy [M eV] [ m b / s r ] [ Å 2 / s r ]

incident ion energy [keV] Rutherf ord TFM ZBL He + Hf, = 155

Figure 2.2: Deviations from Rutherford cross sections for He at (a) high and (b) low primary energies, due to nuclear resonances and electron screening, respectively. In the right panel, the scattering cross section for He and 16O is depicted for experi-mental data [36] and SigmaCalc simulations [37]. The inset shows the resonance at 3.037 MeV, which is used for EBS measurements. In the right panel the cross sec-tion for He scattered from Hf is shown, for unscreened (Rutherford - black line) and screened potentials (TFM and ZBL, red and blue lines).

In 1913 Geiger and Marsden experimentally verified the (Z2e/v)2sin−4(ϑ /2)

dependence of the scattering cross section from experiments using different targets, i. e. Au, Pt, Sn, Ag, Cu, Al and C foils [35]. However, this simple as-sumption of Coulomb interaction is not suitable for all primary energies. If the distance of closest approach between projectile and scatterer is in the order of the nuclear radius, typically for primary energies of several MeV/u, nuclear reaction between the projectile and target atom can occur resulting in strong deviations from Rutherford cross sections. In Fig. 2.2(a) the cross section of He+ scattered from O is plotted for experimental data [36] and SigmaCalc simulations [37]. The significant deviations from Rutherford cross section due to nuclear reactions between He and O are commonly used in Elastic Back-scattering Spectrometry (EBS) experiments enhancing the sensitivity for low atomic number constituents, such as O.

For primary energies below ∼ 100 keV/u the distance of closest approach even in a large-angle scattering event is larger than the typical radii of K-shell electrons. In this energy regime, screening of the charge of the involved nuclei has to be taken into account. In Fig. 2.2(b) the differential scattering cross section is plotted for the Coulomb potential and two commonly employed screened potentials. A detailed description of the screening models applied in this thesis can be found in the following Sec. 2.2.1.

(24)

2.2 Interactions with electrons of the target

2.2.1 Screened scattering potentials

At large interaction distances, which are of particular importance in low- and medium-energy ion scattering electron screening reduces the strength of the scattering potential. Here, typically screened Coulomb potentials Vscrare

em-ployed Vscr= VC(r) · Φ r a (2.5) with the screening function φ _ar and the screening length a. In literature various models for the screening functions are available. Two commonly employed ones are the Thomas-Fermi-Molière (TFM) [38] and the Universal (Ziegler-Biersack-Littmark - ZBL) [39] potential which are both of the form of φ r a =

_∑

i bie− ci·r a . _(2.6)

Depending on the employed potential, different parameter sets for biand ci

are used. The corresponding screening lengths are either according to Firsov [40] a_TFM= _√0.8853a0 Z1+ √ Z2 2/3 (2.7)

or the universal screening [39] aZBL=

0.88534a0

Z₁0.23+ Z0.23₂ (2.8) where a0 denotes the Bohr radius. In Fig. 2.2(b) the differential scattering

cross sections for He and Hf are plotted for the Coulomb, TFM and ZBL potential. Depending on the employed screening model differences of up to ∼ 15 % can be observed in the scattering cross section obtained for either TFM or ZBL potential.

Especially for low energies even weaker interatomic potentials as predicted by the above screening functions are necessary to properly describe experi-mental spectra. Therefore, the screening length can be modified via a typ-cially linear correction factor caas proposed by O’Connor and Biersack [41].

Screening length corrections for various energies and scattering geometries have been determined experimentally in different studies [42–44] with ca

de-pending on the employed energy and scattering geometry. Semi-empirical calculations give suggestions for screening length corrections for different projectile-target combinations [41, 45]. Note, the above description of the scattering potential does not consider the influence of changes in charge state as well as excitation state of the projectile in the solid [29]. For a more detailed 8

(25)

2.2 Interactions with electrons of the target comparison between the potentials and the effect on data evaluation the reader is referred to Sec. 4.1.2 as well as manuscript II in this thesis.

2.2.2 Energy loss and charge exchange

Energy loss

Ions penetrating in matter lose energy due to continual interactions with both electrons and nuclei of the target. The resulting deceleration process is also referred to as electronic Se and nuclear stopping Sn, respectively. A measure

for the mean energy loss per unit path length is given by the total stopping power S:

S= Sn+ Se= −

dE

dx. (2.9)

To avoid the dependency of the energy loss data on the target atomic density n, commonly, the Stopping Cross Section (SCS) ε is used, which is defined as

ε ≡ 1

nS. (2.10)

Under the assumption that the energy transfer to either electron or nucleus is governed by binary collisions, S can be expressed in terms of the impact pa-rameter dependent energy transfer in a single collision with either an electron or a nucleus T (b) and its corresponding scattering cross section σ (T ):

S= n Z T(b)dσ (T ) = n Z b→∞ b=0 2πbT (b)db. (2.11) Note, b corresponds here, to the impact parameter in either a collision with an electron or a nucleus.

Nuclear stopping can be described as a sequence of elastic binary collisions between the projectile and a nucleus, and its contribution to the total energy loss only becomes relevant towards lower primary energies and higher pro-jectile masses. In this thesis, nuclear stopping contributions are modelled by either employing the ZBL or TFM potential, as introduced in the previous section.

The energy transfer from the projectile to electrons of the target can occur via different mechanisms depending on the ion velocity: excitation of the tar-get electrons and charge exchange processes between excited projectile states and electrons in the conduction band of the solid.

Information on electronic energy loss can be obtained in two significantly different experimental approaches: in transmission (tr) or backscattering (bs) geometry. In transmission the measurements are performed on self-supporting foils with typical scattering angles of ϑtr < 2◦. For backscattering

(26)

2. Ion - solid interactions 10 -3 10 -2 10 -1 10 0 10 1 10 2 10 3 10 4 10 5 10 6 10 -1 10 0 10 1 10 2

Bohr & Bethe E -1 e n [ 1 0 -1 5 e V c m 2 / a t o m ] energy [keV] He + Al Lindhard & Scharff E 1/2

Figure 2.3: Stopping cross section of He+ions in amorphous Al with the values from SRIM-13 [24].

ϑbs> 150◦. In both measurements the projectile loses energy due to

elec-tronic and nuclear stopping along its trajectory. However, for backscattering at least one large-angle scattering event, i. e. small impact parameter, is neces-sary in order for the projectile to be detected. Therefore, the impact parameters probed along the ion trajectories differ between transmission and backscatter-ing geometry.

Also the crystallinity of the investigated sample has to be taken into account in ion scattering experiments and therefore, in energy loss investigations. First studies of the influence of a single crystalline structure on the trajectory of en-ergetic charged particles can be traced back to Lindhard in 1965 [46]. When the ion beam is closely aligned with a symmetry direction of a single crys-tal in the bulk of the material, ions undergo only small-angle scattering events (large impact parameters) resulting, typically, in a steering of the beam through the crystal channel [34, 46, 47]. This steering of the ion beam is commonly reffered to as channeling and depends strongly on the projectile-target com-bination as well as their energy. If the detector is aligned with a symmetry direction of the crystal certain trajectories are suppressed. This phenomenon is commonly referred to as blocking. Electronic energy loss in channeling / blocking geometry may significantly differ from energy loss in a random ge-ometry depending on projectile mass and energy [48–51].

Figure 2.3 shows the electronic and nuclear SCS for He+ in amorphous Al as black and red solid line, respectively. The non-trivial scaling of the energy loss with the ion velocity vp, requires different models, which are indicated

in Fig. 2.3. Additionally, it can be observed, that nuclear energy losses are only relevant at low primary energies with εn> εe only for E ≤ 1 keV. The

depicted data are obtained from the Stopping and Range of Ions in Matter (SRIM) database [24], which is a fit function based on physical models. 10

(27)

2.2 Interactions with electrons of the target For swift, light ions with vp v0 (and the Bohr velocity v0 = ₁₃₇c ) the

projectile is typically stripped of its electrons and only weakly perturbs the electronic system of the solid. A quantum mechanical approach obtained with first order perturbation theory is first given by Bethe in 1930 [52]. In this model the stopping power of a bare nucleus with charge z = Z1· e is given by

−dE dx = 4πz2ne mev2_p e2 4πε0 2 · ln 2mev 2 p I ! (2.12)

with the electron mass me, the target electron density ne, the vacuum

permit-tivity ε0 and the mean ionization potential I. The stopping power decreases

with increasing primary energy due to the decreasing time interval where the projectile is in the vicinity of the target atom. The original model from 1930 was improved by a correction for the mean ionization potential [53], and by in-cluding higher order perturbation theory [53–56], resulting in good agreement between theory and experiments for energies of several MeV/u and amor-phous targets [22, 57]. Additionally, also a relativistic correction for the Bethe formula (Eq. 2.12) is available [58].

For single crystals, energy loss along channels formed by atom strings or planes is lower than for random geometries, due to the restricted impact pa-rameters probed along a channeling direction resulting in lower contributions from excitation and ionization of core electrons [47, 48, 59, 60].

This observed difference of the energy loss depending on the impact pa-rameters probed along the ion trajectory for amorphous and single crystalline targets gives rise to the question if the electronic energy loss evaluated from ei-ther transmission or backscattering experiment yields a systematic difference also due to the selective impact parameters probed. However, at elevated ener-gies no systematic differences were observed between different data sets [22], as well as a combined study for energy loss of protons in different materials [61, 62].

Electronic energy loss in a compound, Scompound, can be modelled with

Bragg’s rule [63] corresponding to the weighted sum of energy loss of its constituents Siwith concentrations ci:

Scompound=

∑ici· Si

∑ici

. (2.13)

At high primary energies good agreement between predictions by Bragg’s rule and evaluated energy loss from experiments is achieved with differences typi-cally ≤ 2 % [32].

For low primary energies, i. e. vp< vF(with the Fermi speed of the electrons

(28)

the target is described by a homogeneous Free Electron Gas (FEG) the elec-tronic energy loss is found to be proportional to the ion velocity [64, 65]

S= Q(Z1, ne) · v. (2.14)

The friction coefficient Q depends on the atomic number of the projectile as well as the effective density of the FEG ne, which is typically characterized

by the density parameter rs= (3πne/4)1/3 corresponding to the Wigner-Seitz

radius of a sphere occupied by one electron. A list of rs parameters is given

e. g. in Ref. [66]. The friction coefficient can be modelled with linear response theory [67, 68] or non-linear theories, typically including Density Functional Theory (DFT) to calculate the ion-electron potentials for a FEG [69–73]. Note, both approaches neglect the crystalline structure of the target.

When an effective FEG density is evaluated from experimental plasmon frequencies }ωp,expt, good agreement between DFT based calculations and

experimental data has been observed for protons in several different metals and semiconductors [74–77]. However, already for He DFT is underestimating the electronic energy loss for many targets indicating additional contributions to electron hole pair excitation as charge exchange processes [77–81]. Further details on charge exchange processes for He+projectiles are given on page 14. Also proton stopping in noble metals was observed to deviate from the FEG model at an ion velocity of v ∼ 0.2 a.u. This deviation was attributed to the fi-nite excitation threshold of the d-band electrons [82–84]. The velocity regime < 0.2 a.u. can be described with the FEG model when only s-electrons are considered. Above ∼ 0.6 a.u. also good agreement can be found when an ef-fective density parameter rs,eff as deduced from the plasmon frequencies is

used for the FEG considering contributions of both the 6s1and 5d10-electrons [66, 85, 86].

Electronic energy loss of protons in early transition and rare earth metals of the 6th periode, however, cannot be explained with a homogeneous FEG model [Paper XII]. Electronic stopping in these materials is so efficient that absurdly high electron densities would be required to describe the experimen-tal data. Qualitatively, the high efficiency of excitation of valence electrons was attributed to the combination of both a large number of occupied and un-occupied states at the Fermi level.

The above mentioneld difficulties illustrate the importance of experimen-tal reference data. However, in general data are scarce for the more reactive transition and rare earth metals [27] and their compounds [22], especially for energies below the stopping maximum. One major difficulty is the preparation of high quality thin films without low Z impurities. Therefore, in many appli-cations typically tabulated values from SRIM are used, which are based on inter- and extrapolation of available experimental stopping data, although for certain projectile-target systems significant deviations from these predictions 12

(29)

2.2 Interactions with electrons of the target can be observed hampering e. g. sample characterization with IBA or calcula-tions of implantation depth profiles.

If no experimental stopping data are available for compounds, commonly, predictions by Bragg’s rule are used even for lower primary energies. How-ever, at these energies significant deviations are expected, because energy loss is mainly due to excitation of valence electrons. In general, the structure of the valence band (VB) of a compound compared to the VB of its constituents differs significantly resulting in typical deviations of ∼ 20 % between experi-ments and predictions by Bragg’s rule [32, 33, 87, Paper VII, XIII]. To illus-trate the complexity of ion-solid interactions at lower energies, in particular for compounds, in the following paragraphs a few examples of recent experi-mental investigations are given.

In LiF, a large band gap isolator, a threshold velocity was observed for elec-tronic energy loss of protons and He ions in both experimental [88–90] and theoretical studies [91, 92]. Below this velocity no electron-hole pair excita-tion is possible and the energy loss is exclusively due to nuclear losses. Ex-periment and Time-Dependent-DFT (TD-DFT) calculations show good agree-ment. However, the absolute value of the threshold is slightly overestimated in the theoretical studies. The electronic energy loss of protons and also for heav-ier projectiles in grazing incidence could be described surprisingly well with a FEG with an effective density parameter and DFT calculations, although the valence band in LiF is no free electron gas [93]. The authors explained this observation with the strong perturbation caused by the low energy ion in the crystal resulting in a metallic character of the target.

For oxides with band gaps between 0 eV to 9 eV proton stopping below 10 keV was attributed mainly to the excitation of the O 2p electrons [26, Pa-per XIII] independent of the electronic configuration in the target (metallic or insulating character). Investigations of different nitrides, which feature lower ionic and more covalent character in the bonds, are presented in Sec. 4.2.1 as well as Paper III.

In the FEG model only excitation of target electrons is considered without taking into account the structure of the target material. Additionally, at low energies dynamic processes, such as charge exchange between the electrons in the solid and an excited projectile state [94] as well as formation of molec-ular orbitals [79], contribute to electronic energy losses. Therefore, TD-DFT calculations are necessary to properly describe energy losses at low energies for both crystalline and amorphous targets. In recent years electronic stopping powers for different projectile-target combinations - mainly protons, He ions and self-irradiation of metals as well as semiconductors - have been modelled with TD-DFT simulations [95–100]. Data for electronic energy loss of both protons and He ions in Ni show good agreement between experiment and TD-DFT models concerning the velocity scaling, the absolute values, however, are underestimated in the calculations by ∼ 15 % [99, Paper VII].

(30)

2. Ion - solid interactions Charge exchange

As already indicated in the previous section, for slow He ions additional en-ergy loss processes to electron-hole pair excitation contribute to electronic stopping. When a He ion approaches the sample surface the projectile state interacts with the conduction band as well as bound electron states resulting in a shift of the He 1s level as a function of the distance between ion and surface atom. Depending on the He-target nucleus distance different electron capture or loss processes can occur. In this thesis, only Auger neutralization (AN) and resonant processes in close collision, i. e. neutralization (RN) and re-ionization (RI) are adressed. For more details on possible charge exchange mechanism the reader is referred to Ref. [94, 101].

The physical quantity describing the efficiency of the charge exchange pro-cesses is the ion fraction P+, which corresponds to the fraction of the yield of backscattered ions Y+vs. the yield of all backscattered particles:

P+= Y

+

Y+_+Y0, (2.15)

with the yield of neutrals Y0. Note, here any contributions from negative or double charged ions are neglected.

Auger neutralization is a non-local effect and therefore, can occur at all en-ergies, whenever there is an unoccupied projectile level below the Fermi level of the target. In this case, an electron from the conduction band can tunnel through the potential barrier and fill the empty core level of the projectile re-sulting in a release of energy, which can lead to excitation of a second electron in the conduction band. A schematic drawing of the process is depicted in the left image of Fig. 2.4.

If AN is the only active charge exchange mechanism the ion fraction corre-sponds to a probability for the projectile surviving neutralization P+ and can be calculated via the rate equation [102]

dP+ dt = −P

+

ΓA(~r) (2.16)

with the position dependent Auger-neutralization rate ΓA(~r). For straight

tra-jectories both on the incoming and outgoing trajectory as well as the assump-tion of no planar dependence of the neutralizaassump-tion, the ion fracassump-tion can be expressed in terms of the characteristic velocity vc ≡RΓA(z)dz and is given

by P_i+= exp − Z ΓA(z) dt dzdz ≈ exp − 1 v⊥,i Z ΓA(z)dz = e− vc v⊥,i (2.17) 14

(31)

2.2 Interactions with electrons of the target

Figure 2.4: Schematic drawing of the AN, RN and RI process.

Here, the subscript i indicates the ion fraction for either incoming or outgoing trajectory and 1/v⊥=_dzdt represents the inverse perpendicular velocity.

Depending on the distance between projectile and target nucleus, the level of the projectile is shifted due to interactions with electrons of the target [103]. If the He level is in resonance with the conduction band of the solid, an elec-tron from the conduction band can tunnel through the potential barrier to the core level of the projectile. This process is referred to as resonant neutral-ization (RN), as indicated in the second image in Fig. 2.4. If the He level is shifted above the Fermi level, resonant ionization (RI) of the projectile is possible, indicated in the last image in Fig. 2.4.

The minimum distance rminrequired to enable resonant processes depends

on the projectile-target system. For a fixed scattering geometry ϑ , rmin

corre-sponds to a threshold energy Ethand resonant neutralization and re-ionization

can only occur for E ≥ Eth. Ideally, Eth is determined by experiments with

a primary beam of He0atoms and the detection of exclusively backscattered ions [104, 105]. However, also more indirect evaluations of the threshold en-ergy are reported, as deviations from the velocity scaling according Eq. 2.17 [106] or the onset of the re-ionization background in the ion spectra [107]. Typical values for Eth reported in literature range from 0.3 keV to 2.1 keV for

Al [105, 108] and Cu [106, 107], respectively.

Matrix effects in the neutralization efficiency depending on surface struc-ture [109, 110], surface crystallinity [111] as well as chemical strucstruc-ture in the surface, e. g. for graphitic and carbidic C [112], are reported in literature complicating straight-forward surface analysis with low energy ion scattering.

(32)

(33)

3. Experimental methods, set-ups

& data evaluation

In the first sections of this chapter the applied experimental techniques and the used set-ups are presented. For sample characterization Rutherford and Elastic Backscattering Spectrometry (RBS and EBS, see Sec. 3.1) as well as Time-of-Flight Elastic Recoil Detection Analysis (ToF-ERDA, see Sec. 3.2) were used. This combination of different ion beam analysis techniques allows for high resolution composition and depth profiling of light and heavy constituents of the samples [3, 7, 113].

The energy-loss measurements were performed with Time-of-Flight Medium-and Low-Energy Ion Scattering (ToF-MEIS Medium-and ToF-LEIS) set-ups as de-scribed in Sec. 3.3 and 3.4, respectively. Both methods follow the same basic principles as RBS, however, the lower primary energies decrease the informa-tion depth. From an analytical point of view this allows sample characteriza-tion with sub-nm depth resolucharacteriza-tion of nm thin films, if the electronic energy loss of the projectile-target system is known. The investigations of the charge state of slow ions were mainly performed with an ElectroStatic Analyser (ESA) LEIS set-up.

In Sec. 3.5 a short description of the sample preparation is given includ-ing other surface characterization methods employed such as Auger Electron Spectroscopy (AES) as well as Low Energy Electron Diffraction (LEED).

The final Sec. 3.6 in this chapter addresses data evaluation for all employed techniques. Additionally, simulation programs used in this thesis are pre-sented.

3.1 Rutherford & Elastic Backscattering Spectrometry

RBS is a commonly employed ion beam analysis technique to determine thick-ness and stoichiometry of thin films. Typically, light ions, i. e. H+or He+, with primary energies of a few MeV are used as projectiles and ions backscattered from the target nuclei are detected. The backscattering angle is typically cho-sen between 150◦to 180◦. Due to energy loss of the projectile while traversing matter, a depth profile of the sample can be obtained from the energy spectrum. This technique is particularly sensitive to heavier elements, whereas the detec-tion of light elements in a matrix of elements with high atomic number is more difficult. To enhance the sensitivity for light ions nuclear resonances in the dif-ferential scattering cross section are used, e. g. the16O(α, α0)16O resonance

(34)

3. Experimental methods, set-ups & data evaluation

at 3.037 MeV [36, 37] as plotted in Fig. 2.2. Due to the significant deviations from the Rutherford cross section, these measurements are typically referred to as EBS measurements.

Sample characterisation was performed with a 5 MV 15SDH-2 Pelletron accelerator from National Electrostatic Corporation (NEC) located at the Tan-dem Laboratory at Uppsala University. Typically, for ‘standard’ RBS 2 MeV He+ ions are used, due to a combination of various factors: (i) Rutherford cross sections are still applicable for all elements, (ii) large energy separa-tion and (iii) well known electronic stopping as well as the applicability of Bragg’s rule. The used scattering chamber features two passivated implanted planar silicon (PIPS) detectors positioned at different scattering angles to de-tect backscattered particles.

3.2 Time-of-Flight Elastic Recoil Detection Analysis

Due to the limitations of RBS to detect light constituents in a matrix of heavy elements, additional techniques such as ERDA are used to properly determine the composition of a sample. In this method the target is irradiated with ions in a more glancing angle and elastically recoiled particles as well as scattered projectiles can be detected [4, 114, 115]. Note, in ERDA measurements the detector has to be in the forward scattering direction - a schematic drawing is depicted in Fig. 3.1.

In this thesis, only heavy ions are used allowing simultaneous multi-element profiling [3]. This technique is also called Heavy Ion ERDA (HIERDA). The sample characterization included in this thesis were performed with a 36 MeV

127_I8+_{ion beam irradiating the sample typically under α = 67.5}◦_{with respect}

to the surface normal and recoiled and scattered particles with an exit angle of β = 67.5◦are detected.

The ERDA measurements were also performed with the 5 MV Pelletron accelerator at the Tandem laboratory. Two different scattering chambers are equipped with a telescope tube to perform time-of-flight/energy coincidence measurements enabling mass-resolved depth profiling of films up to ∼ 1 µm thickness. In both ToF-ERDA set-ups the time-of-flight of the recoiled parti-cles is obtained from the detection of secondary electrons from a set of two carbon foils. The difference is in the final detection of the particles by either

sample detector projectilesscattered atoms recoiled primary ion beam

Figure 3.1: Schematic drawing of the forward scattering in ERDA.

(35)

3.3 Medium Energy Ion Scattering a silicon surface barrier detector [116] or a segmented gas ionization chamber [117]. The latter is preferred for the detection of heavy elements, because the surface barrier detector suffers from radiation damage due to both heavy re-coiled and scattered primary particles. More details on the set-ups are given in Ref. [116–118].

3.3 Medium Energy Ion Scattering

For studies of energy loss and scattering potential performed in the the medium energy regime the ion beam is provided by a 350 kV Danfysik High Current Implanter Model 1090 [119] located at the Tandem Laboratory at Uppsala University. Three different ion source modes (gas, oven and sputter) provide positive ion beams for a series of chemical elements, including also molecular beams, with high particle currents up to 40 mA [120]. The extraction voltage of the ion source is typically set to 20 kV and therefore, limiting the minimum kinetic energy of the ion beam per charge. The measurements included in this thesis are exclusively performed with the gas ion source. The extracted beam is mass analyzed by a 90◦magnet and subsequently accelerated to the desired final energy - up to ∼ 350 keV for single charged ions.

Via a switching magnet, the ion beam can be deflected into three differ-ent beam lines to perform ion implantations, ToF-MEIS and low energy IBA measurements. In this thesis, only the ToF-MEIS beam line has been used.

A key feature in the MEIS beam line is the electrostatic chopper, which gen-erates pulsed ion beams with typical 1 ns to 2 ns width. Before and after the chopper two sets of slits are installed to define the size of the beam (typically ≤ 1 × 1 mm2_{) as well as the pulse width. To additionally decrease the pulse}

width, a drift tube buncher is installed so that pulses of 0.3 ns can be achieved. For the contributions in this thesis, however, the buncher has not been em-ployed. After focussing of the ion beam via an electrostatic quadrupole triplet, it is steered electrostatically through a 7◦tilted beam line to avoid neutral pro-jectiles from entering the scattering chamber.

The scattering chamber with a typical base pressure of ≤ 5 × 10−8mbar fea-tures a 6-axis goniometer, which allows three translational and three rotational movements of the target. The employed RoentDek detector consists of a stack of MicroChannel Plates (MCP) in chevron configuration with a detector solid angle of 0.13 sr and a diameter of 0.12 m. The typical target-detector distance is r ∼ 0.29 m with small variations ±0.005 m depending on the used sample holder for either thin films on a substrate or freestanding foils. The angularly rotatable and position sensitive MCPs allow the detection of backscattered and transmitted projectiles - neutrals as well as ions - within scattering angles of 0◦ to 160◦. A schematic of the scattering geometry is depicted in Fig. 3.2. Additional to particle detection, MCPs can also detect electrons and X-rays. The time of flight of the projectile is obtained between the particle hitting the

(36)

3. Experimental methods, set-ups & data evaluation

primary ion beam

sample

detector

Figure 3.2: Schematic drawing of the MEIS chamber with the rotatable, position-sensitive MCPs to cover scattering angles from 0◦to 160◦.

MCPs and the time signal from the chopper. To determine the time ‘zero’ necessary to convert the ToF-spectrum in an energy spectrum, the peak of promptly emitted photons is used [121]. For further details on the MEIS beam line as well as the scattering chamber the reader is referred to Ref. [20, 122].

3.4 Low Energy Ion Scattering

In low energy ion scattering typically noble gas or alkali ions are used as pro-jectiles with primary energies ≤ 10 keV [101, 123, 124]. Depending on the employed set-up - time-of-flight or electrostatic analyzer - either all backscat-tered particles or only ions are detected. In this thesis, a ToF-LEIS set-up was employed to investigate electronic energy loss of slow, light ions (H and He) in different target materials. The ESA-LEIS set-up was used for investigations of charge exchange between He+ions and different metallic and oxidized sur-faces.

3.4.1 Time-of-flight set-up

The ToF-LEIS setup ‘ACOLISSA’ (Analysis of the Charge Of Light Ions Scat-tered from Surface AStoms) features a Colutron ion source of model G-2, which can provide beams of atomic and molecular ions within a primary en-ergy range of ∼ 0.5 keV to 10 keV. For this thesis H, D and He have been used as operating gas. The mass separation of the projectiles is performed with a Wien-Filter, allowing a mass resolution of _∆mm ∼ 400. The ion beam from the ion source is focused by a set of 6 beam steering plates. Additionally, the ion source is slightly tilted with regards to the beam line, to prevent neutral parti-cles from reaching the target. A key feature in the primary beam line is again an electrostatic chopper to generate pulsed ion beams. Additionally, the beam line features focusing elements and apertures to obtain a well defined beam spot of ∼ 1 × 1 mm2on the sample.

The scattering chamber with a typical base pressure of 1 × 10−10mbar, which can be additionally improved with the use of LN2traps, features a

man-ual 5-axis target manipulator (3 translations and 2 rotations) which has space 20

(37)

3.5 Sample preparation for two targets. The manipulator is equipped with a heating filament and a thermo-couple to monitor the target temperature.

The detector is located at a scattering angle of 129◦with a circular aperture defining the detector solid angle of Ω = 2 × 10−4sr. Additionally, a post ac-celeration system is installed to separate scattered ions from neutral particles allowing also charge state investigations in the ToF set-up. The detector con-sists of a set of MCPs in chevron configuration and it should be considered that the detection efficiency for ions and neutral particle is different, η+ and η0, respectively. For further details on the ToF-LEIS set-up, the reader is referred to Ref. [125, 126].

3.4.2 ESA-LEIS

The ESA-LEIS set-up ‘MINIMOBIS’ is a compact ion scattering system on the basis of the NODUS set-up [127]. The Leybold-Heraeus ion gun can pro-vide ions with a primary energy range of ∼ 0.3 keV to 5 keV. In this work, only He and Ar have been used as operating gas, where the former was applied for measurements, and the latter for sputter-cleaning of the surface. Again the ion gun is slightly tilted to prevent contributions from primary neutral projec-tiles.

The scattering chamber has a typical base pressure of < 5 × 10−10mbar which can be improved with the use of an additional LN2 cooling trap. The

samples are located on a carousel and irradiated under normal incidence. The carousel can hold up to 6 targets simultaneously and is equipped with a heating filament to allow also the preparation of single crystals.

To obtain energy spectra of backscattered ions a cylindrical mirror analyzer (CMA) is used accepting only projectiles with a backscattering angle of ϑ = 136◦± 1◦with an azimuthal acceptance angle of 2π. For the detection of the backscattered ions a stack of MCPs in chevron configuration is used.

The large azimuthal acceptance angle in the ESA-LEIS set-up allows mea-surements with significant lower primary ion doses compared to the ToF-system. However, for experiments in channelling or blocking geometry the ToF set-up is more powerful allowing different alignments to block sub-surface contributions.

3.5 Sample preparation

For the energy loss and charge exchange investigations presented in this thesis different samples have been used. The energy loss measurements were per-formed on different transition metal nitrides as well as on self-supporting Au and W foils. For the charge exchange investigation polycrystalline Zn and Ta as well as single crystalline Al(111) and Ta(111) have been employed.

(38)

3. Experimental methods, set-ups & data evaluation 0 100 200 300 400 500 600 -2 -1 0 1 2 3 4 O Ta d N / d E electron energy

Ta with surface impurities Ta (cleanest possible surface)

C

Figure 3.3: AES spectrum for a Ta surface with surface contaminations and the clean-est possible surface after several sputter-annealing cycles depicted as black and red line respectively.

For the transition metal nitrides both thin films with thicknesses ranging between ∼ 10 nm to 30 nm as well as bulk samples have been investigated. All samples were fabricated by reactive DC magnetron sputtering in Ar/N2

atmosphere [128, 129].

The self-supporting Au foils have been custom purchased from Lebow [130], whereas the W foils were fabricated with a MED 010 mini thin film deposi-tion set-up from Balzers: first, a NaI layer was thermally evaporated on a Si substrate. This evaporation process is typically performed at a base pressure of ∼ 1 × 10−4mbar. Subsequently, the W layer is fabricated by magnetron sputter deposition using Ar as sputter gas. The base pressure before the sput-ter deposition was ∼ 1 × 10−5mbar, and the Ar pressure during sputtering ∼ 1 × 10−2mbar. Note, NaI is highly hygroscopic and when the samples are stored in air, the W layer typically cracks within a short time-period.

For measurements in the ToF-LEIS set-up the powerful target preparation chamber, which is connected to the ToF-LEIS set-up, was used. This sepa-rate chamber has a base pressure of 1 × 10−10mbar and features a sputter gun, a heating filament as well as a triple e−-beam evaporator. Additionally, the chamber is equipped with an AES and a LEED system to study the composi-tion and structure of surfaces.

For the AES system typically 3 keV electrons are used as primary beam and the energy spectrum of the Auger electrons is recorded via variation of the en-ergy in a CMA. The low mean free path length of electrons yields elemental information from the near surface regime. In Fig. 3.3 the differentiated elec-tron yield is shown for the cleanest possible Ta surface reached after several sputter-annealing cycles (red line), and for a Ta sample with surface contam-inations (black line). The peaks at ∼ 270 eV and ∼ 500 eV correspond to C 22

(39)

3.6 Data evaluation

Figure 3.4: LEED image of an Al(111) single crystal.

and O, respectively. The high energy lines for Ta corresponding to the MNN transition are not depicted in this figure due to their low sensitivity factors.

In the ESA-LEIS set-up the scattering chamber is equipped with a heating filament and the ion source can be used as sputter gun with normal incidence. The single crystalline samples have been prepared by sputter-annealing cy-cles. An in-situ LEED system allows to check the crystallinity of the samples. Again the low mean free path length of electrons allow for surface sensitivity. In Fig. 3.4 a LEED image of an Al(111) single crystal is depicted, which was used for ion fraction measurements.

The polycrystalline Ta has also been cleaned by sputter-annealing cycles, whereas the Zn sample has only been sputter-cleaned, due to its low melting temperature.

The oxidation studies included in this thesis were performed in the main scattering chamber of the ESA-LEIS set-up. To this aim, initially clean metal surfaces were exposed to molecular oxygen with exposure doses of ∼ 10 L for Ta [131, 132], ∼ 1200 L for Al [133] and ∼ 104L for Zn [134] (with 1 L = 1.33 × 10−6mbar) until a stable ion signal for the metal was reached.

3.6 Data evaluation

3.6.1 Sample composition analysis

Rutherford Backscattering Spectrometry

The spectrum of backscattered ions contains information on the mass of the scatterer via the kinematic factor (see Eq. 2.1), as well as the depth distribution due to electronic energy loss of the projectile along the trajectory in the target. In Fig. 3.5(a) a typical RBS spectrum is shown for He+ ions scattered from a self-supporting Au foil. The measurement was performed with a beam of He ions with 2 MeV primary energy. The width of a signal ∆E, as indicated in Fig. 3.5(a), is proportional to the areal density of the target atoms Ns (in

(40)

3. Experimental methods, set-ups & data evaluation 1400 1600 1800 2000 100 200 300 400 500 500 1000 1500 2000 10 1 10 2 10 3 10 4 2 M eV He + , = 170 H kE 0 (a) 1209 Å Au foil experim ent: = 0 = 10 SIMNRA N ( E ) [ e V -1 ] energy [keV] E Zr C O (surface) N Ar O (interface) N ( E ) [ e V -1 ] energy [keV] (b) 262 Å HfN / C experim ent: = 45 = 55 SIMNRA Hf

Figure 3.5: RBS spectra obtained with He+_{ions scattered from (a) a 1209 Å}

self-supporting Au foil and (b) a 262 Å thin HfN film on top of an HOPG substrate. The red solid line corresponds to SIMNRA [135] simulations. In panel (a) a schematic drawing of a general ion scattering geometry is depicted with the incident angle α and the exit angle β , both with respect to the surface normal.

at/cm2) as well as the electronic stopping cross section factor [ε], which cor-responds to the stopping cross section on both incoming and exiting trajectory [3]:

∆E = Ns· [ε] . (3.1)

For thin films [ε] is given by

[ε] =kε(E0) cos α +

ε (kE0)

cos β . (3.2)

The height of a signal H, also indicated in Fig. 3.5(a) at the high energy edge of the Au peak, is inversely proportional to the stopping cross section factor. In a single scattering model it can be written as

H= N0 cos α ∆Ec [ε] dσ dΩ∆Ω. (3.3)

Here, N0corresponds to the number of incoming projectiles, α to the angle of

the incident beam with respect to the surface normal of the target as indicated in Fig. 3.5(a), ∆Ec to the energy loss per channel in the detector, ∆Ω to the

detector solid angle and_dΩdσ to the differential scattering cross section.

To obtain correct depth profiles from RBS, knowledge on the electronic stopping power, which depends on the composition of the material, is cru-cial. However, in typical RBS measurements, primary energies of a few MeV are used, where the electronic stopping in a compound can be modelled with Bragg’s rule (Eq. 2.13) [63]. Commonly, the spectrum of backscattered ions as obtained in RBS measurements is compared to computer simulations. In this 24

New aspects of electronic interactions of keV ions with matter

New aspects of electronic

interactions of keV ions with

matter

List of Papers

Kurzfassung

Sammanfattning

Statutory Declaration

Contents

Acronyms

1. Introduction

2. Ion - solid interactions

2.1 Binary collision

2.1.1 Scattering kinematics

2.1.2 Scattering cross section

2.2 Interactions with electrons of the target

2.2.1 Screened scattering potentials

∑

2.2.2 Energy loss and charge exchange

3. Experimental methods, set-ups

& data evaluation

3.1 Rutherford & Elastic Backscattering Spectrometry

3.2 Time-of-Flight Elastic Recoil Detection Analysis

3.3 Medium Energy Ion Scattering

3.4 Low Energy Ion Scattering

3.4.1 Time-of-flight set-up

3.4.2 ESA-LEIS

3.5 Sample preparation

3.6 Data evaluation

3.6.1 Sample composition analysis

_∑