Vacuum Science and Technology in Accelerators

Transcription

1 Vacuum Science and Technology in Accelerators Lectures are the members of ASTeC Vacuum Science Group: Oleg Malyshev (Lectures 1,6) Keith Middleman (Lectures 2,3) Joe Herbert (Lecture 4) Reza Valizadeh (Lecture 5) January-February 2017 Lecture 5: slide 1

2 Surface and Ultrahigh vacuum Contents: Surface energy and reconstruction Molecular adsorption on surface Langmuir Isotherm describing surface adsorption Analytical Method for surface Characterisation Lecture 5: slide 2

3 Surface What is surface? A special interface between solid/liquid and vacuum Interface: a small number of atomic layers that separate two solids in intimate contact with one another Why surface? Many properties determined by the surface/interface Many processes happens mainly at surface (chemical reaction including catalysis, crystal growth, thermionic emission) Surface/interface may differ significantly from the bulk (Phase Density, Composition, Mechanical, Electronic, Magnetic properties ) Strong interrelation of surface /interface with many other research fields Surface as ultra-thin films presents 2-D model for theory (even 1D eg. Nanostructures - 3D thick film) Many technical applications (semiconductor device) Lecture 5: slide 3

4 Surface Energy and reconstruction There is a direct correspondence between the concepts of surface stability and surface free energy i.e. surfaces with low surface free energy will be more stable and vice versa The total free energy may minimized in several way: 1. By predominantly exposing surface planes which have a low surface free energy 2. By reducing the amount of surface area 3. By altering the local surface atomic geometry in a way which reduces the surface free energy O.B. Malyshev Lecture 5: slide 4

5 Surface energy and reconstruction The total free energy may minimize in several way: 1. By predominantly exposing surface planes which have a low surface free energy Most stable solid surfaces are those with high surface atoms or surface atoms of high coordination number (Note the two are obviously not independent, but inevitably strongly correlated). Consequently, for example, if we consider the individual surface planes for an FCC metal, then we would expect the stability to decrease in the order: FCC CN s (111) = 9 FCC CN s (100) = 8 FCC CN s (110) = 7 FCC CN b = 12 FCC (111) > FCC (100) > FCC (110) Number bond available = 3 Number bond available = 4 Number bond available = 5 Lecture 5: slide 5

6 Surface energy and reconstruction Relaxation is a small and subtle rearrangement of the surface layers which may nevertheless be significant energetically, and seems to be commonplace for metal surfaces. It involves adjustments in the layer spacing perpendicular to the surface, there is no change either in the periodicity parallel to the surface or to the symmetry of the surface. Un-relaxed Surface the first layer of atoms is typically drawn in slightly towards the second layer (i.e. d 1-2 < d bulk ) Relaxed Surface (d 1-2 < d bulk ) balanced, symmetrical set of forces. imbalance forces actin on un-relaxed surfaces. Lecture 5: slide 6

7 Surface energy and reconstruction 3) Reconstruction Unlike relaxation, the phenomenon of reconstruction involves a change in the periodicity of the surface structure - the diagram below shows a surface, viewed from the side, which corresponds to an unreconstructed termination of the bulk structure a reconstructed surface - this particular example is similar to the "missing row model" proposed for the structure of a number of reconstructed (110) fcc metal surfaces. Since reconstruction involves a change in the periodicity of the surface and in some cases also a change in surface symmetry, it is readily detected using surface diffraction techniques (e.g. STM, LEED & RHEED) Lecture 5: slide 7

8 Molecular Adsorption The adsorption of molecules on to a surface is a necessary prerequisite to any surface mediated chemical process. For example, in the case of a surface catalyzed reaction it is possible to break down the whole continuously-cycling process into the following five basic steps : Diffusion of reactants to the active surface Adsorption of one or more reactants onto the surface Surface reaction Desorption of products from the surface Diffusion of products away from the surface Lecture 5: slide 8

9 Molecular Adsorption 1. Physical Adsorption / Physisorption: Not very specific No electron transfer, although polarization of may occur Rapid, non activated and reversible No dissociation of adsorbed species Monolayer or multilayer Only significant at relatively low temperatures Enthalpies are in the region of -5 to -10 KJ/ mol As the temperature increases, process of Physisorption decreases. Lecture 5: slide 9

10 Molecular Adsorption 2) Chemical reaction / Chemisorption: Highly specific Electron transfer, leading to bond formation between adsorbate and adsorbent Activate, may be slow and irreversible May involve dissociation Monolayer only Possible over wide range of temperatures Enthalpy are in the region of -200KJ / mol With increasing in temperature, Chemisorption first increases and then decreases Lecture 5: slide 10

11 Adsorption Kinetics-The rate of Adsorption The rate of adsorption, R ads of a molecule onto a surface can be in the manner as in any Kinetic process R ads = A exp ( -E a / RT ). P x Where E a is activation energy for adsorption, and A is the pre-exponential (frequency) factor. P is the partial pressure of molecule and x is the kinetic factor. The rate of adsorption is governed by : 1. The rate of arrival of the molecule at the surface 2. The proportion of the incident molecule which undergoes adsorption. Thus: R ads = incident molecular Flux. Sticking probability The flux of incident molecules is given by Hertz-Knudsen equation. F = P / (2πmkT) 1/2 Where P is gas pressure, m is mass of one molecule and T is Temperature Lecture 5: slide 11

12 Adsorption Kinetics-The rate of Adsorption In general, the sticking probability, S. S = f (θ). exp ( -E a / RT ) Where once again the Ea is the activation energy for adsorption and f (θ) is function of existing surface coverage of adsorbed species. Thus combining the equations for S and F yield to the following expression for the rate of adsorption: If it is further assumed that sticking probability is directly proportional to the concentration of vacant surface sites, then f (θ) = (1-θ) where in this instant, θ is fraction of site which are occupied. Lecture 1: slide 12

13 Case I-Phyisorption: Energy of Adsorption A shallow minimum in the PE curve at a relatively large distance from the surface (typically d > 0.3 nm) before the strong repulsive forces arising from electron density overlap, There is no barrier to prevent the atom or molecule which is approaching the surface from entering this physisorption well, i.e. the process is not activated and the kinetics of physisorption are invariably fast. Case II Physisorption + Molecular Chemisorption The depth of the chemisorption well is a measure of the strength of binding to the surface - in fact it is a direct representation of the energy of adsorption, whilst the location of the global minimum on the horizontal axis corresponds to the equilibrium bond distance (r e ) for the adsorbed molecule on this surface. Lecture 5: slide 13

14 Energy of Adsorption Case III Physisorption + dissociative chemisorption In the example (left) there is no direct activation barrier to non dissociative adsorption There is a substantial barrier to chemisorption. Such a barrier has a major influence on the kinetics of adsorption At this point it is useful to return to consider the effect of such a barrier on the relationship between the activation energies for adsorption and desorption, and the energy (or enthalpy) of adsorption. E des a - E ads a = - ΔE ads E des a >>> E ads a Thus E des a ~ -ΔH ads Lecture 5: slide 14

15 Desorption process The rate constant for desorption process may be expressed in an Arrhenius form: E a des is the activation energy for desorption, ν, can also be considered to be the "attempt frequency at overcoming the barrier to desorption and N,surface concentration of adsorbed species. Lecture 5: slide 15

16 Langmuir Isotherm As with all chemical equilibria, the position of equilibrium will depend upon a number of factors : The relative stabilities of the adsorbed and gas phase species involved The temperature of the system The pressure of the gas above the surface The Langmuir isotherm was developed by Irving Langmuir in 1916 to describe the dependence of the surface coverage of an adsorbed gas on the pressure of the gas above the surface at a fixed temperature When considering adsorption isotherms, it is conventional to adopt a definition of surface coverage (θ). The maximum (saturation) surface coverage of a particular adsorbate on a given surface always to be unity, i.e. θ max = 1 Θ= (surface sites occupied) / (surface sites occupied + surface vacant sites) Lecture 5: slide 16

17 Variation of Surface Coverage with Temperature and Pressure Langmuir isotherm This is illustrated in the left graph which shows the characteristic Langmuir variation of coverage with pressure for molecular adsorption θ bp at low pressures θ 1 at high pressures The value of b is increased by 1. a reduction in the system temperature 2. an increase in the strength of adsorption A given equilibrium surface coverage may be attainable at combination of pressure and temperature (left side). As the temperature is lowered, the pressure required to achieve a particular equilibrium surface coverage decreases. Lecture 5: slide 17

18 Surface Analytical Techniques 1. Photoelectron Spectroscopy 2. Auger Electron Spectroscopy 3. Vibrational Spectroscopy (IR, EELS, Raman) 4. Secondary Ion Mass Spectroscopy (SIMS, SNMS,SIMTOFS) 5. Temperature programmed techniques (TPD,TPRS) 6. Surface structure (STM, LEED, AFM) Lecture 5: slide 18

19 X-ray Photoelectron Spectroscopy Introduction Qualitative analysis Quantitative analysis Instrumentation Lecture 5: slide 19

20 Introduction Photoelectric effect Photoelectric effect Einstein, Nobel Prize 1921 Photoemission as an analytical tool Kai Siegbahn, Nobel Prize 1981 Lecture 5: slide 20

21 Analytical Methods --- X-ray Photoelectron Spectroscopy (XPS) Kinetic Energy Photon h Photoelectron KE = hν - (E B +ϕ) 0 0 E v E f XPS spectrum: Intensities of photoelectrons versus E B or KE VB Binding Energy 3s 2p 2s 1s Elemental identification and chemical state of element Relative composition of the constituents in the surface region Valence band structure Lecture 5: slide 21

22 Binding Energy Reference K.E. = h -B.E. - Vacuum level.f sample e - K.E. = h -B.E. - - ( spec- sample) = h -B.E. - Vacuum level.f sample.f spec spec sample Fermi level Fermi level B.E.F core level B.E. = h - K.E. -.F spec Lecture 5: slide 22

23 Instrumentation Electron energy analyzer X-ray source Ar ion gun Neutralizer Vacuum system Electronic controls Computer system Ultrahigh vacuum system < 10-9 Torr (< 10-7 Pa) Detection of electrons Avoid surface reactions/ contaminations Lecture 5: slide 23

24 Background: Photoelectrons with energy loss Peak: Photoelectrons without energy loss Lecture 5: slide 24

25 Relative binding energies and ionization cross-section for U Lecture 5: slide 25

26 For p, d and f peaks, two peaks are observed. The separation between the two peaks are named spin orbital splitting. The values of spin orbital splitting of a core level of an element in different compounds are nearly the same. Au The peak area ratios of a core level of an element in different compounds are also nearly the same. Spin orbital splitting and peak area ratios assist in element identifications. Lecture 5: slide 26

27 Spin-orbital splitting Peak Notations L-S Coupling ( j = l s ) e - s= 1 2 s= j = l + 2 j = l 1 2 Lecture 5: slide 27

28 Qualitative analysis Gold XPS wide scan spectrum Photoelectron Peaks 4s 4p 1/2 4p 3/2 4d 3/2 4d 5/2 5s 4f 5/2 4f 7/2 5f 1/2 5p 3/2 Binding energies Auger Peaks N 67 O 45 O 45 N 5 N 6 N 67 N 4 N 6 N 67 N 5 N 67 V Binding Energies Lecture 5: slide 28

29 X-ray Induced Auger Electrons K.E. is independent of the x-ray photon energy. However, in the B.E. scale, Auger peak positions depend on the x-ray source. Lecture 5: slide 29

30 General methods in assisting peak identification (1) Check peak positions and relative peak intensities of 2 or more peaks (photoemission lines and Auger lines) of an element (2) Check spin orbital splitting and area ratios for p, d, f peaks A marine sediment sample from Victoria Harbor Si 2p Si 2s Al 2s Al 2p The following elements were found: O, C, Cl, Si, F, N, S, Al, Na, Fe, K, Cu, Mn, Ca, Cr, Ni, Sn, Zn, Ti, Pb, V Lecture 5: slide 30

31 XPS Sampling Depth i = inelastic mean free path of an electron in a solid For an electron of intensity I o emitted at a depth d below The surface, the intensity is attenuated according to the Beer-Lambert law. So, the intensity I s of the same electron as it reaches the surface is I s = I o e -d/ With a path length of one 63% of all electrons are scattered Lecture 5: slide 31

32 Sampling Depth Sampling Depth is defined as the depth from which 95% of all photoelectrons are scattered by the time they reach the surface ( 3 ) Most s are in the range of nm for AlK radiation So the sampling depth (3 ) for XPS under these conditions is 3-10 nm Lecture 5: slide 32

33 1 monolayer = 0.3 nm Universal Curve for IMFP nm (nanometers) depends on K.E. of the photoelectron the specific material Lecture 5: slide 33

34 Quantitative XPS: I Some XPS quantitative measurements are as accurate as ± 10% I i = N i i i K where: I i = intensity of photoelectron peak p for element i N i = average atomic concentration of element i in the surface under analysis i = photoelectron cross-section (Scofield factor) for element i as expressed by peak p i = inelastic mean free path of a photoelectron from element i as expressed by peak p K = all other factors related to quantitative detection of a signal (assumed to remain constant during exp t) Lecture 5: slide 34

35 How to measure I measured Worst Accuracy better than 15% using ASF s Use of standards measured on same instrument or full expression above accuracy better than 5% Best In both cases, reproducibility (precision) better than 2% Lecture 5: slide 35

36 Transmission Function Transmission function is the detection efficiency of the electron energy analyzer, which is a function of electron energies. Transmission function also depends on the parameters of the electron energy analyzer, such as pass energy. Pure Au after Ar + sputtering Lecture 5: slide 36

37 Quantitative Analysis: II Scofield Cross-section Factors ( i ) have been calculated for each element from scattering theory, specifically for AlK and MgK radiation Inelastic Mean Free Paths ( i ) varies with the kinetic energy of the photoelectron. It can be estimated from a universal curve or calculated (better). For a multi-element surface layer consisting of elements i, j, k. N i = I i N i +N j +N k ( i i ) I i + I j + I k i i j j k k Lecture 5: slide 37

38 Examples of Quantitation I Lecture 5: slide 38

39 Examples of Quantitation II Lecture 5: slide 39

40 Errors in Quantitation I i = sometimes difficult to separate intrinsic photoelectrons for the extrinsic scattered photoelectrons which comprise the background ( ± 5-100%) i = calculated value (unknown magnitude) i = estimated error ± 50% Lecture 5: slide 40

41 Chemical Effects in XPS Chemical shift: change in binding energy of a core electron of an element due to a change in the chemical bonding of that element. Qualitative view: Core binding energies are determined by: electrostatic interaction between it and the nucleus, and reduced by: the electrostatic shielding of the nuclear charge from all other electrons in the atom (including valence electrons) removal or addition of electronic charge as a result of changes in bonding will alter the shielding Withdrawal of valence electron charge (oxidation) Addition of valence electron charge increase in BE decrease in BE Lecture 5: slide 41

42 Chemical Shifts: Oxide Compared to Metal Li-metal 1s 2 1s 2 2s Density 1s 2 Binding Energy is lower due to increased screening of the nucleus by 2s conduction by 2s electrons 2s 1s 2 Li Li 2 O Li 2 O 2s 6 1s 2 2s 2 0 Li-metal Li 1s 2s 1s 2 Li Binding Energy is higher because Li 2s electron density is lost to oxygen PE spectrum Binding Energy E Fermi Lecture 5: slide 42

43 Photoemission Process can be thought of as 3 steps: (a) Photon absorption and ionisation (initial state effects) (b) Response of atom and creation of photoelectron (final state effects) (c) Transport of electron to surface (extrinsic effects) (one additional +ve charge) B + A B B A B Lecture 5: slide 43

44 Lecture 5: slide 44

45 Koopman s Theorem The BE of an electron is simply the difference between the initial state (atom with n electrons) and final state (atom with n-1electrons (ion) and free photoelectron) BE = E final (n-1) E initial (n) If no relaxation* followed photoemission, BE = - orbital energy, which can be calculated from Hartree Fock. *this relaxation refers to electronic rearrangement following photoemission not to be confused with relaxation of surface atoms. Lecture 5: slide 45

46 Lecture 5: slide 46

47 Examples of Chemical Shifts Lecture 5: slide 47

48 Detailed Iron 2p Spectrum of High Purity Iron x Fe 2p/1 Fe 2 O 3 Metallic Fe Binding Energy (ev) Lecture 5 : slide 48

49 x 10 2 Detailed Spectrum of Fe 2p line for Magnetite (partly oxidized) Fe 2p_HSS2_3/33 35 Fe (III) Fe (II) Binding Energy (ev) Lecture 5: slide 49

50 x 10 2 Detailed Oxygen 1s Spectrum O 1s / Surface Hydration Metal Oxide Binding Energy (ev) Lecture 5: slide 50

51 Before sputtering After 200eV Ar + sputtering Cubic-BN Crystal B 1s oxide BN B 1s Binding Energy (ev) N 1s N 1s Binding Energy (ev) c/s BNO? BN Binding Energy (ev) Binding Energy (ev) Lecture 5: slide 51

52 High Resolution Spectra Arsenopyrite BE ΔBE FWHM %Area AsFeS As AsFeS As As 2 O As 2 O 5 As 3d AsFeS c/s As 2 O 5 60 As 2 O 3 As um diameter x-ray spot Binding Energy (ev) Lecture 5: slide 52

53 Chemical Shift Lecture 5: slide 53

54 Aluminum Oxide Thickness 2000 Al(2p) Oxide thickness = 3.7 nm 1500 aluminum oxide aluminum metal Counts Binding Energy (ev) High resolution Al (2p) spectrum of an aluminum surface. The aluminum metal and oxide peaks shown can be used to determine oxide thickness, in this case 3.7 nanometres Lecture 5: slide 54

55 Estimation of Oxide Thickness Usually, the binding energies of the oxide and the metallic species are separated by a few electron volts. Thus, when the oxide is thin (< 9 nm), it is possible to distinguish the contribution from both oxide and metal photoelectrons. For aluminum, oxide thickness (d) is given as: d (nm) = 2.8 ln ((1.4(Io/Im))+1) where Io and Imare the intensities (peak areas) of the oxide and metal photoelectron peaks respectively. Lecture 5: slide 55

56 Instrumentation Lecture 5: slide 56

57 Electron Energy Analyzer Concentric hemispherical analyzer (CHA) For an electron of energy E o at S E = 0.63 w 1 E o R o Lecture 5: slide 57

58 Pass Energies and Transfer Lens (1)To resolve a 1000 ev electron to ± 0.5 ev would require an analyser with w=1 mm and R=1.2 metres! Therefore, it is convenient to retard the energy of the incoming electrons so that they have a lower (and constant) energy as they pass through the analyser. The lens system which retards the electron energy also focuses the electrons energy from the sample to increase the throughput. Lecture 5: slide 58

59 Factors Pass energy Analyzer radius Slit width Elements in the transfer lens Energy of the photoelectrons Lecture 5: slide 59

60 PET : Polyethylene terephthalate C 1s 10 ev 20 ev C 1s Different Pass Energies ev C 1s 40 ev 80 ev C 1s Lecture 5: slide 60

61 CHA Analysers Operating Modes CAT Retardation Mode: Constant Analyser Transmission Characteristics: - Constant Pass Energy across spectrum, therefore fixed resolution across spectrum - Easier quantitation since transmission is fixed -However, fixed transmission works against high KE photoelectrons since most electrons here are scattered - narrow acceptance angle -Pass Energy Entendue CRR Mode Constant Retarding Ratio, not used for XPS Lecture 5: slide 61

62 Satellite peaks High energy satellite lines from magnesium and aluminium targets Lecture 5: slide 62

63 X-ray monochromator n =2dsin For Al K 8.3Å Advantages of using X-ray monochromator Narrow peak width Reduced background No satellite & Ghost peaks use (1010) planes of quartz crystal d = 4.25Å o = 78.5 Lecture 5: slide 63

64 Photoelectron spectra of SiO 2 excited with Al K α radiation Unfiltered radiation Monochromatized radiation Lecture 5: slide 64

65 Kratos Axis Ultra at SSW Lecture 5: slide 65

66 Photoelectron Line Widths Contributions to width 1. Inherent linewidth of the photoelectron production event lifetime-dependent temperature-dependent Lorenzian-shape 2. Width of Exciting line MgK < AlK Monochromatised AlK is better. Two component shape is modelled as a Gaussian 3. Analyser Resolution determined by pass energy and slit width, modelled as a box function. Lecture 5: slide 66

67 Commonly used Lecture 5: slide 67

68 Analytical Methods Convolution How to obtain high-resolution XPS spectra? Deconvolution Lecture 5: slide 68

69 The Use of Different Photon Energy (a) ZrLα 2040 ev (b) Mg Kα ev ( c) Ti Kα 4510 ev oxide Si Lecture 5: slide 69

70 SESSION 3 Energy losses: extrinsic and intrinsic Electron attenuation: inelastic scattering Interpretive models: QASES Plasmon losses, shake-up and shake-off satellites Multiplet interactions Depth profiling Lecture 5: slide 70

71 Intrinsic and Extrinsic Losses --- Variation of Al2p energy loss structure Origins of the XPS background Extrinsic losses (electron-phonon event) inelastic scattering Intrinsic losses (electron-electron event) A part of photoemission event Alternative final states Why study intrinsic backgrounds? Background B.E. (ev) Information about the depth and lateral distributions of elements using the QUASES method developed by Sven Tougaard Lecture 1: slide 71

72 Inelastic Scattering Background Tougaard developed a fitting procedure for the inelastic scattering tail, which may give some information about the structure of the surface layer, such as, complete coverage by a metal layer or formation of metal clusters. Lecture 5: slide 72

73 Analysis of XPS Spectra Using QUASES Traditional XPS quantification assumes Outer surface of sample is homogeneous Outer surface concentration is directly proportional to the peak intensity More accurate quantification should include peak intensity, peak shape and background energy In photoelectron spectroscopy electrons detected result from two processes the intrinsic electrons from photoelectron process the extrinsic electrons from scattering of photoelectrons passing through surrounding atoms Depending on the depth of the emitting atom within the surface, as well as its lateral distribution, the extrinsic portion will change dramatically The figure shows a theoretical calculation of the extrinsic portion of a copper 2p spectrum as a function of the position and distribution of the emitting copper atoms within a matrix of another element 0.1 nm a 3 nm 5 nm 2.5 nm c d The above example courtesy of b Lecture 5: slide 73

74 Plasmon Loss Peaks For some materials, there is an enhanced probability for loss of a specific amount of energy due to interaction between the photoelectron and other electrons. Al 2s spectrum of sapphire (Al 2 O 3 ) S: surface plasmon B: bulk plasmon Some photoelectrons lose more than once Lecture 5: slide 74

75 Depth Profiling Ar + Sputtered materials Peak Area Sputtering Time Lecture 5: slide 75

76 Peak Area Concentration Sputtering Time Depth Calibration of depth scale 1. Sputtering rate determined from the time required to sputter through a layer of the same material of known thickness. 2. After the sputtering analysis, the crater depth is measured using depth profilometer. A constant sputtering rate is assumes. Lecture 5: slide 76

77 Atomic Concentration (%) Depth profile of Architectural Glass Coating O 1s O 1s O 1s Ti 2p 40 Si 2p Nb 3d Ti 2p N 1s Si 2p 20 N 1s Al 2p 0 0 Surface 200 Sputter Depth (nm) Lecture 5: slide 77

78 Atomic Concentration (%) Depth profile of Architectural Glass Coating O 1s O 1s O 1s Ti 2p 40 Si 2p Nb 3d Ti 2p N 1s Si 2p 20 N 1s Al 2p 0 0 Surface 200 Sputter Depth (nm) Lecture 5: slide 78

79 Factor Affecting Depth Profiling Instrumental factors Sample characteristics Adsorption from residual gas atmosphere Redeposition of sputtered species Impurities in ion beam Non-uniform ion beam intensity Time-dependent ion beam intensity Depth information (IMFP) Original surface roughness Crystalline structure and defects (Channelling ) Alloys, compounds, second phases (Preferential sputtering and induced roughness ) Radiation-induced effects Primary ion implantation Atomic mixing Sputtering-induced roughness Preferential sputtering and decomposition of compounds Enhanced diffusion and segregation Electron-induced desorption Charging of insulators (analysis distortion; electromigration) Lecture 5: slide 79

80 X-ray damage Some samples can be damaged by the x-ray For sensitive samples, repeat the measurement twice to check for x-ray damage. Lecture 5: slide 80

81 Shake-up Peaks K.E. =K.E.-ΔE B.E. =B.E.+ ΔE e - For some materials, there is a finite probability that the photoelectronic process leads to the formation of an ion in its excited state with a few ev above the ground state. Unfilled levels Valence levels E 1s Polystrene Lecture 5: slide 81

82 Cu (II) Shake-up Peaks A feature for the identification of Cu (II) Lecture 5: slide 82