PHOTODISSOCIATION DYNAMICS OF MOLECULES ON SURFACES by Chris C. Johnson B.Sc., University of N orthern B ritish Colum bia, 1998 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF M ASTER OF SCIENCE MCPS-PHYSICS 0 Chris C. Johnson, 2001 THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA O ctober 2001 All rights reserved. This work m ay not be reproduced in whole or in p a rt, by photocopy or other means, w ithout perm ission of th e author. NmUonmi Ubwy 0* Cmnmd# BMiothèque national# du Canada Acquisition# mnd BlMoymphic Servie## Acquisitions el 3#S W#*ngion 8###i services btHographiqu## aas.wwsamglon C#n#d# Csnads 0#m#eON K1A0N# OamwON K1A0N4 The muAor has granted a non­ exclusive licence allowing the N # Ü o o s I L ib E « y o f C m a d 8 ( o reproduce, loan, dWbok or sell copies of ihisAeswmmicm&nn, or electrooic Amnats. T h e m A o r ie A in s o w m e id i ip c f th e cop|ynghtmthisdiesû.Nei*her(he Ae^DorsubetmdmlexIractsûomA may b ep nn led oro6erw se rqmxxluced wiAout Ae author's permismon. L'aokmr a accordé une Hcaice non mmluaive permettant à la BAËoAèqoenadanak du Canada de Mpmdmre, prêter, distribuer ou vendre des oppieg de cette thëse sous lafbnnedenûcrdldieÆ hn,de reprodiKtW am papier eu sur (bnnat ëkctmoiqxm. L'auteur oooaerve la prtpriété & droit d'auteur qui protège cette thèse. hB la Aèaem des extraits substantiels deooDe-cinedQivBmtètreimprnnès ou autrement reproduits sans son autorisatioiL 0-612-80697-9 CanadS APPROVAL Name: Chris C. Johnson Degree: Master o f Science Thesis Title: PHOTODISSOCIATION DYNAMICS OF MOLECULES ON SURFACES \ Examining Committee: ( X L / Chair: Dr. Alex Michalos Professor Emeritus, Political Science Program UNBC Supervisor: D r E ^ ^ a n s l n Associate Professor, Physics Program UNBC Committee Member: Dr. Mark Shegelski Associate Professor, Physics Program UNBC Committee Member: Dr. Samuel W alters Associate Professor, Mathematics & Computer Science Program UNBC Committee Member: Dr. Margot M anny Assistant Professor, Chemistry Program )C E xterW Examiner: Dr. John Hepburn Professor, Chemistry & Physics - Head, Department o f Chemistry University o f British Columbia Date Approved: _______O o H ) h c ^ A bstract M ethyl iodide and m ethyl brom ide on C u(llO ) surfaces have been studied by retarding potential spectroscopy, tem perature program m ed desorption, and time-of-flight mass spec­ trom etry. The A = 337 nm photodissociation of m ethyl iodide on C u(llO ) is found to occur by both charge transfer and direct photodissociation processes. M ethyl brom ide dissoci­ ated exclusively due to charge transfer processes on C u(llO ). Charge transfer results from hot substrate photoelectrons dissociatively attaching to the adsorbate. T he workfunctions of the dosed C u(llO ) surface are at all tim es greater th an th e energy of a single photon, therefore desorption processes had a high probability of being neutral. R etarding potential spectroscopy m easurem ents found th a t the workfunction of th e Cu(110)-I surface increased by 1.2 eV as compared to the C u(llO ) surface = 4.48 eV [26]). M ethyl brom ide on the Cu(110)-I surface did not show any evidence of photodynam ics due to th e A = 337 nm light, which is thought to be due to the increased workfunction of th e sem i-conductor C ul substrate preventing th e form ation of hot photoelectrons of th e right energy for attach m en t. M ethyl iodide dissociation on th e Cu(110)-I surface is assigned to direct neutral photodis­ sociation only, since there is no evidence for charge transfer processes. M ethyl iodide on Cu(110)-I is found to be tilted at 20° off norm al in the [110] direction and is found to be sig­ nificantly more ordered th an on the C u(llO ) surface. On b o th surfaces m ethyl iodide yields were found to m axim ize on completion of the second layer of adsorbate. Yields decreased rapidly for higher coverages on both surfaces. T O P spectra on both surfaces is assigned to desorption a n d /o r dissociation from the exposed top layer. T here is some evidence th a t over­ layers do not com pletely cover underlayers. M ethyl iodide on both C u(llO ) and Cu(110)-I show evidence of altered A = 337 nm photodissociation dynam ics as com pared to sim ilar wavelengths in the gas-phase. On both surfaces m ethyl iodide cross-sections are enhanced by approxim ately two orders of m agnitude. Further evidence of altered neutral photody­ namics is found in th e /* /! branching ratio for dissociation from the ^Qo and ^Qi excited states of adsorbed m ethyl iodide. Dissociation of adsorbed m ethyl iodide is found to favour the higher energy I* channel rath er th an the I channel. T he altered direct photodissociation dynamics in m ethyl iodide is thought to be due to changes in th e excited state potential energy surfaces due to adsorption and proxim ity to th e copper surface. C ontents P h o to d y n a m ic s a t S u rfa c e s 1.1 Introduction . ................................................................................. 1.2 Substrate and Surface . ............................................................................................ 1.2.1 Bulk Crystal S tructure and Cleavage P l a n e s .................................. 1.2.2 B and S tructure of Substrates ................................................ 1.2.3 Im age and Surface Potentials . ........................... 1.3 A d so rp tio n .................................................................. .. ................. 1.3.1 Introduction 1.3.2 Physisorption ................................................................. 1.3.3 C h em isorption............................... 1.3.4 A dsorbate Overlayer Structures on Surfaces .................................... 1.4 Energy Transfer Mechanisms between th e A dsorbate and S u b s t r a t e .................. 1.4.1 In tro d u c tio n ........................... 1.4.2 Charge Transfer from the S u b strate to A d s o r b a t e ..................... 1.4.3 D irect Excitation of the A dsorbate orA dsorbate-Substrate bond . . . 1.4.4 T herm al P r o c e s s e s ........................... 1.4.5 Q u e n c h in g ................................................................................................... 1.5 Surface Dynamics .................................. 1.5.1 Introduction ............................................................. 1.5.2 T he Franck-Condon P r i n c i p l e ............... .............................................. 1.5.3 Potential Energy Surfaces for D esorption and D is s o c ia tio n .................. 1.5.4 Concluding R e m a r k s ......................................................................................... 1 1 2 2 4 5 6 6 6 7 8 8 8 10 12 12 12 13 13 14 15 17 E x p e r im e n ta l M e th o d s 2.1 In tr o d u c tio n ....................................................................................................................... 2.2 A pparati .......................................................... 2.2.1 U ltra-H igh Vacuum C h a m b e r ........................ 2.2.2 Sample and M a n ip u la to r.................................................................................. 2.2.3 H eating the S a m p le ............................ 2.2.4 Q uadrupole Mass Spectrom eter . .............................................................. 2.2.5 L a s e r ....................................................................................................................... 2.3 Standardized Surface Techniques ..................................... 2.3.1 Sample P reparation ......................................................................................... 2.3.2 Low Energy Electron D iffraction(LEED ) ..................... 2.3.3 Auger Electron Spectroscopy ( A E S ) ............................... 19 19 20 21 22 23 24 24 25 25 26 28 in 2.3.4 R etarding Potential Spectroscopy .............................................................. 2.3.5 T em perature Program m ed D e so rp tio n (T P D )............................................. 2.4 Surface Photolysis Experim ents: Time-Of-Flights ...................................... 2.4.1 In tro d u c tio n ........................................ 2.4.2 Yields as a function of to tal photons ........................................................ 2.4.3 Yields as a function of dose ...................................................... 2.4.4 A ngular D istributions ............................... 30 34 37 37 38 40 41 M eth yl H alides on Copper: Standardized Surface E xp erim en ts 3.1 In tro d u c tio n .................................................................................................................. . 3.2 Auger Spectroscopy of C u(llO ) ............................................................. 3.3 LEED pattern s of C u (1 1 0 ),C u (1 1 0 )-I........................................................................ 3.4 Tem perature Program m ed D e so rp tio n (T P D )........................................................... 3.4.1 CH 3l/C u ( 110 ) ....................................................................................................... 3.4.2 CH 3 / / C u ( 110 ) - I ................................................................................................... 3.4.3 C ^ 3B r /C u ( llO ) ................................................................................................... 3.5 R etarding Potential Spectroscopy (R PS) .................. 3.5.1 CH 3I / C u ( l l O ) ................................................................................................... 3.5.2 CH 3I /C u ( 110 ) - I ................................................................................................ 42 42 43 45 49 49 50 52 53 53 56 S u rfa c e P h o to I y s is :T im e - O f-F lig h t E x p e r im e n ts 58 4.1 Introduction .................................................................... 58 4.2 Kinematics and P otential energy s u rfa c e s ........................... 59 4.2.1 K inem atics for Direct Gas-phase Dissociation ......................................... 61 4.2.2 Potential Energy Surfaces for Direct P h o to a b s o rp tio n ........................... 62 4.2.3 K inem atics Equations for Dissociative Electron A ttachm ent Dissoci­ ation in th e G a s - p h a s e 65 4.2.4 Potential Energy Surfaces ............................................ 65 4.3 Surface Photolysis of CJTsBr on Cu(110)and C u ( 1 1 0 ) - I ...................................... 67 4.4 Surface Photolysis of C'i 73/ / C u ( 110 )-I ................. 73 4.4.1 E xam ination of the D etector Resolution ............................ 73 4.4.2 C haracterizing the energy transfer mechanisms and dissociation . . . 75 ................................................................................................ 81 4.4.3 A ngular Yields 4.4.4 A ltered photodissociation dynam ics .................................................... 83 4.5 Further Discussion ................................................................ 90 4.6 Surface Photolysis of C i7 al/C u (l 10 ) ........................ 92 4.6.1 Coverage experim ents ................................... 92 4.6.2 A ngular D ependency E x p e rim e n ts............................... .......................... ... . 98 4.6.3 I* Branching R atio and C T -P D IS ............................... 98 4.6.4 Cross-section M easu rem en t..................... 101 4.6.5 Further D is c u s s io n ........................ 103 4.7 C o n c lu s io n s ................................................... 107 IV List o f Tables 4.1 Significant times on the TOP spectrum for figure 4.10. The energies of CH-i (15 amu) fragments and CH^Br (95 amu) molecules corresponding to those times are also calculated from E = 4.2 Significant times on the TOP spectrum for CH^I/Cvl[ 1 1 Q)-\ in figure 4.15. The energies of CH^ fragments and C H 3 I molecules corresponding to those times are also calculated from E = ~mv^......................................................................................... 4.3 Description of the param eters forequation 4.11 4.4 Significant times on the TOP spectrum for C H 3 I/Cu{110) in figure 4.25. The energies of C H 3 (15 amu) fragments and C H 3 I (127amu) molecules corresponding to those times are calculated with E = V 70 76 83 96 List of Figures 1.1 Simple schematic of light directed onto an adsorbate-substrate system in equilib­ rium with its surroundings............................. 1.2 Electrons leak out of the bulk into the space just above the surface causing a smoothing of the surface and contractive relaxation (arrows) of the layer of atoms nearest the surface.(Diagram from Zangwill [43, pg 29]; Finnis and Heine [14].) . . 1.3 Face-Centered-Cubic(FCC) Unit Cell and the surface of an FCC crystal cut in a (110) orientation................................................. 1.4 Image, surface, and bulk potentials as a function of a line z normal to the surface. The image potential smoothly joins onto the surface potential, which in turn is joined to the bulk potential. Diagram from Inglesfield [18, pg 120]. . .............. 1.5 Hydrogen atom and it image near a metal surface. Van der Waals’ forces arise because of interactions between the charges of the hydrogen atom and their images. Diagram from Zangwill [43, pgl86]. 1.6 Physisorption of Molecules on Surfaces. Inelastic scattering traps a molecule at a surface due van der Waals potential. (Diagram from Zangwill [43, pg361]; Tully [35]) .................................................................................................................................... 1.7 Chemisorption of Molecules on Surfaces. Adsorption potential with a precur­ sor physisorbed well and a deeper chemisorbed well closer to the surface. One­ dimensional model of chemisorption by Lennard-Jones(1932) Diagram from ZangwiU [43, pg366]..................................................................................................................... 1.8 Charge Transfer at a surface. Hot electrons are produced at a surface which then dissociatively attach to the adsorbate. The LUMO is the lowest unoccupied molec­ ular orbital and the HOMO is the high occupied molecular orbital. Diagram from Wolf [40]................................................................................................................................ 1.9 Mean Free Path of Electrons in various Solids. The meanfree path of electrons in solids is of the same order but less than the optical penetration depth of w 100Â. Therefore most excited electrons will undergo scattering processes. The dashed curve is that expected from theory. (Zangwill [43, pg21]; Rhodin and Gadzuk [29]; Somorjai [31]. Theoretical curve from Penn [ 2 8 ] .) ...................................................... 1.10 Frank-Condon transitions on potential energy surfaces(PES) are vertical transi­ tions. The internuclear separation is unchanged during the transition. The FranckCondon transitions have an envelope determined by the localization of molecules in the well (gaussian figure).............................. . 1.11 Potential Energy Surface showing the MGR desorption mechanism. Diagram from Zhou [44]............................................................................................................................... VI 2 3 4 5 7 8 9 10 11 14 16 1.12 Potential energy surface showing a surface dissociation mechanism similar to the MGR model for desorption. One or both of the dissociative products can chemisorb to the surface. Diagram from Zhou [44]. . ............................................................. 1.13 Potential energy surface showing the Antoniewicz desorption mechanism. Diagram from Zhou [44]............................. 2.1 2.2 17 18 Schematic of experimental s y s t e m ................ 20 Cutaway drawing of the UHV Chamber. Apparat! are arranged on the UHV chamber in two tiers facing the central vertical axis. A sample manipulator can raise,lower, rotate the sample in convenient orientations for each experiment. . . 21 2.3 Circuit for controlling the temperature of the sample, including heating and cool­ ing. There are two thermocouple wires attached to the sample which are used for measuring the sample temperature........................................................................ 23 2.4 Ewald sphere for electrons at normal incidence to a crystal surface. (Diagram from Zangwill, [43, pg34];Kahn [20].) 26 2.5 Diagram of Rear-view LEED Optics. Electron gun is at the center of the phos­ phorous screen. Electrons are backscattered off the sample towards the LEED optics............................ 27 2.6 (a)The three-electron Auger process. An inner shell electron is knocked out of its orbital by a high energy bombarding electron (3k eV). An electron from an upper orbital fills the gap and at the same time passes energy to an adjacent electron, which is kicked out of its orbital as a result, (b) Backscattered N(E) distribution. Inset is the N'{E) distribution. (Diagrams from Zangwill [43, pg22,23]; (b) Park and den Boer [27]. ) ........................................ . 28 2.7 The Auger circuit for measuring N'{E). E b = 3 keV is the energy of the bom­ barding electrons. A E w lOV is the peak-to-peak voltage of a small sinusoidal voltage. Vo ramps from « 90 volts to 1000 volts for copper. The phase w = 2irf where f = 1.4 kHz is the frequency of the sinusoidal voltage....................................... 29 2.8 When dipolar molecule attach to a metal surface they may orientate in a particular direction. In this case dipoles are shown orientated in an ‘up’ direction. The negative end of the dipole is closer to the surface.................... 30 2.9 The ideal representation for measuring workfunctions. Since the chemical potential does not differ a great deal from the Fermi potential in metals even at the temper­ atures T ~ 2800 K that the tungsten filament is heated to, I will follow the usual practice of referring to the maximum potential energy as the Fermi energy. Vq is variable in order to maintain a constant space potential difference between the two metals. changes with the addition of molecules to the metal surface. The metal sample receives electrons and is called the anode as a result. The tungsten filament of the electron gun produces electrons and is called the cathode...................... 31 2.10 The electrons produced from the electron gun have an approximately gaussian dis­ tribution. The peak in the distribution is labelled E b - Only electrons with energy greater than E q = will contribute to the current on the sample. Electrons with less energy than E q will be deflected away from the sample............................... 32 vu 2.11 The current on the sample varies with the retarding potential V q - When Vo is increased such that it begins to deflect the electron beam away from the sample, the current begins to drop off............................................................................................ 2.12 This data is proportional to the derivative of N(E), the distribution of electron energies. As the current on the sample decreases due to an increasing V q there is a zero-crossing point on the N'{E) graph. It is marked with E b ........................... 2.13 Circuit design used to measure the workfunction changes of the sample as it is being dosed with molecules. The Lock-In amplifier measures the amplitude of the sin(2wf) harmonic. The amplitude is directly proportional to the derivative of the electron energy distribution N(e)...................................... 2.14 Rotation of the sample through 0 about z-axis is the [001] azimuth. Rotation of the sample through about the x-axis is the [110] azimuth........................................ 3.1 AES spectrum for Cu(llO). The scan is done by measuring the amplitude of the second harmonic {sin2ujt). The signal is proportion to N'(E), the Auger electron signal. Sample cleanliness is confirmed by the appearance of Auger peaks consistent with copper only. .............................................................................. 3.2 AES spectrum for Cu(110)-I. The scan is also done by measuring the amplitude of the second harmonic {sin2LJt). The iodine peaks show up near 510 eV..................... 3.3 Picture of the LEED screen with a clean Cu(llO) surface. The electron beam energy is 105 eV. The reciprocal net is rectangular and is characteristic of wellordered crystalline copper in a (110) cut......................................................................... 3.4 Picture of the Screen in a LEED experiment with Cu(110)-I at an electron beam energy of 101 eV. The reciprocal surface net is similar to a c(2x2) pattern but additional satellite spots are visible in the [110] direction. The additional spots likely indicate longer range ordering............................................... 3.5 This schematic illustrates the LEED patterns is dots, squares and X’s. The dots are the position of the (1x1) copper spots. The iodine also contributes to the dots. The squares are iodine spots. The X’s are the additional satellite spots. Diagram from Johnson et al. [6].............................................. 3.6 A c(2x2) real space overlayer structure is a rectangle with an atom in the center. The length of the sides are 2 times the length of the real space rectangular surface net of the copper atoms on the (110) surface..................... 3.7 (a) shows a c(2x2) overlayer structure, not iodine, in four-fold sites on a Cu(llO) surface. In this case the hypothetical atoms used in the overlayer structure are smaller in size than the iodine atoms, but are the largest th at could be fit into the c(2x2) overlayer, (b) shows the relative sizes of the iodine, copper, and the hypothetical overlayer atoms. ........................................................................... 3.8 TPD spectrum for CH3l/Cu(110). The first layer is complete at a dose of 9.0 L and desorbs from the surface at 140 K............................. 3.9 TPD spectra of Ci73l/Cu(110)-I. The first layer is complete at a dose of 9.5 L and desorbs at 149 K ................................................................................................................ 3.10 TPD spectrum for C iÎ3 B r/C u (1 1 0 )............................................................. 3.11 Workfunction of Cu(llO) surface as a function of CH 3 I d o s e ......................... via 33 33 35 38 44 44 45 47 47 48 48 49 51 52 53 3.12 Workfunction measurements of cesium on single crystal Tungsten surfaces. (Dia­ gram from Zangwill [43, pg293]; Kiejna and Wojciechowski [ 2 2 ] .) ........................... 3.13 Workfunction of Cu(110)-I surface as a function of CH 3 I d o s e .............................. . 3.14 The crystal structure of solid CH 3 I. In the solid C % I,50 % of the dipoles are orientated down and 50 % are orientated up. The dipoles are tilted at 20° from normal. As layers of CH 3 I on Cu(llO) or on Cu(110)-I get thick, they will at some point have this ideal structure. This occurs when the surface effects are no longer significant. Diagram modified from Kawaguchi et al. [21]. ..................................... 4.1 Gas-phase adsorption cross-section of the A-band for CH 3I as a function of wave­ length. Diagram modified from Waschewsky et al. [39]................................................ 4.2 The total gas-phase cross-section for photodissociation of CHgBr as a function of wavelength in the A-band. Diagram modified from Van Veen et al. [38]................... 4.3 Gas-phase potential energy surfaces of CH3 I in the A-band. The ground state and the three excited states are shown. The and the °Qo states curve-cross. (Diagram from Eppink et al. [12]; Gedanken and Rowe [15]............................. 4.4 Gas-phase potential energy surfaces of CHgBr in the A-band. The ground state and the first three excited states are shown. (Diagram from Van Veen et al. [38].) 4.5 Gas-phase potential energy states of CH3X in the A-band. Energy states shown with and without spin-orbit interactions. There are three optically allowed transi­ tions from the ground state to the states with spin-orbit interactions. Diagram from Yabushita et al. [42]............................................................. 4.6 (a)The gas-phase branching ratio for excitation to the °Qo state. 4>q* = N q* /( No -+• No* where absorption of photons leads to a production of initial populations No* in the I* (°Qo) channel and No in the I channel, (b) The probability of curve crossing as a function of wavelength. For both (a) and (b) the dashed and solid lines are alternate interpolations of the data points. In (b) the dashed line is similar to that expected from the Landau-Zener probability curve. Diagrams from Eppink et al. [12]................................ 4.7 CH 3 X potential energy surfaces for the electron attachment mechanism. Diagram modified from Ukraintsev et al. [37]..................................................... 4.8 TOE spectra at various doses of CifaBr on Cu(llO) ...................................... 4.9 Slow and Fast Peak Counts as a function of CH 3 BT dose on Cu(llO) . . . . . . . 4.10 Time-of-flight spectrum of CH 3Br/Cu (110) at a dose of 10 L ................................... 4.11 Time-of-flight spectra of CH 3B r/C u( 110 )-I at doses of 10 L and 20 L. In this case the CH3 collected in the TOF apparatus is background. Apparently CHsBr is not dissociating. The Cul on the surface is likely acting as a barrier to the production of hot electrons. There is insufficient energy in the light photons to produce photoelectrons with enough energy overcome the Cul barrier...................... 4.12 Time-of-flight spectra of 20 L CH 3l/C u ( 110 )-I. In one TOF spectrum the flight path d is 16 cm and in the other d is 8.3 cm. The width of the signal is largely due to the energy distribution of particles desorbing from the sample, however, the resolution of the detector also affects this width................ 4.13 TOF spectra of C H 3 l / C u { 1 1 0 ) - I . ............................................................. IX 55 56 57 60 60 62 63 63 64 66 68 69 71 72 74 76 4.14 The total counts in I and I* peaks were summed and plotted as a function of dose. The C H j signal begins when the coverage is increased past 1 ML, reaches a maximum on completion of the 2 ML, and then rapidly decreases for higher coverages. .............................................. 4.15 Time-of-flight spectrum of CH3l/C u ( 110 )-l at a dose of 20 L ................................. 4.16 Angular experiments in the [110] and [001] a z im u th .................................................. 4.17 Example fit to a TOF distribution CH3l/C u ( 110 )-I. Graph is from Johnson and Jensen [6]. ............................................................... 4.18 4>* branching ratios from the fits to the distributions in the TOFs on CH 3l/C u ( 110 )I as a function of dose. The dashed line is the (f>* = 0.1 branching ratio for gas-phase CH3I at 333nm. Graph is from Johnson and Jensen [6]................................................ 4.19 Depletion spectrum for a 20 L dose on CH 3l/C u ( 110 ) - l ........................................... 4.20 Depletion spectrum with a larger number of photons for a 20 L dose on CH 3l/C u ( 110)- 1 ............................................................................................................................... 77 78 82 84 85 87 88 4.21 Depletion spectrum for a 40 L dose on CH 3l/C u ( 110)-l. The cross-section or slope of the graph changes as the total number of photons increase on the sample. . . . 89 4.22 Various TOF spectra of C H 3 I/Cu{110). From TPD measurements, a dose of 9.0 L .......................... 93 is a monolayer of coverage on the Cu(llO) surface. 4.23 The slow and fast peak counts in the TOF spectra of figure 4.22 are summed and plotted as a function of dose.............................................................................................. 95 4.24 High Resolution Scans on CHsI/Cu{110) at 7 L and 20 L. The counts in these scans were summed in Ifis MCS-TOF bins. The scans demonstrate that for less than 1 ML the CH 3-I- signal from the fast distribution is not sharply peaked. At 2 ML however the CHg-t- signal from the fast distribution is bimodal, asymmetrical and sharply peaked..................... 96 4.25 Time-of-flight spectrum for 20 L of CH 3I on C u(llO )................................................... 97 4.26 Fast Peak Counts as a function of Angle for 20L on C i 73l/C u ( 110 ) ........................ 99 4.27 Slow Peaks Counts as a function of Angle for 20L on C iÏ 3l/C u ( 110 ) 100 4.28 A example fit to the spectrum of Ci 73l/C u ( 110) system using the fitting function, equation 4.11. Graph from Johnson and Jensen [6 ]................................................ 102 4.29 CT-PDIS proportion as a function of coverage of the fast peak on the TOFs C i 73l/C u ( 110 ) system. The estimates were done using the fitting function, equa­ tion 4.11. Graph from Johnson and Jensen [6].................................................................. 102 4.30 Yields from the fast distribution as a function of total number of photons for a dose of 9.0 L ............................................................ 104 4.31 Yields from the slow distribution as a function of total number of photons for a dose of 9.0 L . . . .............................. 104 4.32 Yields from the fast distribution as a function of total number of photons for a dose of 20.0 L ...................................................... 105 4.33 Yields from the slow distribution as a function of total number of photons for a dose of 20.0 L ................................................................................................. . 105 4.34 Yields from the fast distribution as a function of total number of photons for a dose of 40 L. The non-linearity of this graph is making it difficult to interpret the slope as the cross-section............................................. 106 4.35 Yields from the slow distribution as a function of total number of photons for a dose of 40 L. The graph is again non-linear........................................................................ 106 XI A c k n o w le d g e m e n ts The first person who comes to m ind who deserves recognition for assistance in th e creation of this m aster’s thesis is m y supervisor D r Erik Jensen, who provided m e w ith a great deal of inspiration and direction for this research. I have come to appreciate E rik ’s insight into physics and his cautious and intuitive approach to forming conclusions based on experim en­ tal data. Thanks for everything Erik. My com m ittee also had m any helpful suggestions th a t im proved th e quality of m y thesis. Thanks to Dr. M argot M andy, Dr. M ark Shegelski, Dr. Sam W alters and th e external reviewer Dr John H epburn of UBC. Special thanks to the UNBC physics departm ent for th e various support th a t has been provided to me over the years. I think finding th e funds and resources to have a g rad u ate program has been challenging for the physics departm ent. I am grateful th a t I could persue graduate work in physics here at UNBC. Thanks to Dr Erik Jensen and Dr M ark Shegelski for the graduate courses I took w ith them . I believe graduate courses are taught on top of an already full workload and so th e ex tra work they do is expecially w orthy of appreciation. A very special thanks to Debbie Price, secretary for the physics d ep artm en t, for her help with office supplies and printing th a t are necessary to produce a thesis. Thanks to the U niversity of N orthern B ritish Colum bia for providing m e w ith a graduate teaching assistantship for these past several years and also thanks to C hristine Dom ning, the senior physics lab instructor, who has been my im m ediate supervisor for m any years. Thanks also to my partner M elanie K arkjala,w ho has actually lived an d listened to m e all these years while I did my graduate work. Finally, it should be noted th a t financial support for this research was provided by th e N a­ tional Science and Engineering Council (NSERC) on a grant to Dr. E rik Jensen. Chris Johnson, O ctober 2001 XU C hapter 1 P h otod yn am ics at Surfaces 1.1 Introduction This thesis is an investigation of th e photodynam ics of adsorbate molecules attach ed to a m etal surface. T he photodynam ics of such systems cannot be accurately predicted because the physics of adsorbates on m etal surfaces are not well understood. As a result experim en­ tally verified solutions to the Schrddinger equation for adsorbate-substrate system s are not available in the literature. In adsorbate-substrate physics th e outer orbitals of th e adsorbate interact closely w ith the orbitals of th e substrate surface. The energy states of th e outer adsorbate electrons are perturbed to a greater or lesser extent depending on th e adsorbate and th e substrate. Adsorbates often have been well studied and well-understood in their gas-phase form. Insight into surface physics is often achieved by com paring an ad so rb ate’s experim ental photodynam ics to its gas-phase dynam ics. Differences and sim ilarities be­ tween surface and gas-phase physics will often indicate when th e surface is affecting the adsorbate’s energy states and when it has little or no effect. D ata is obtained in this thesis in time-of-flight (T O F) studies by adsorbing molecules on a substrate surface, forming an adsorbate-substrate system which is in equilibrium w ith its surroundings, and then directing coherent pulsed laser light is onto th e system. See figure 1.1. Since photons can penetrate the surface, there are in general three distinct possibilities in the system for photon absorption. Photons m ay be absorbed in th e adsorbate, in the substrate, or in th e adsorbate-substrate bond. T he last case, absorption of photons in the adsorbate-substrate bond, can only occur when the adsorbate and th e substrate form a strong chemical bond such as a covalent or an ionic bond. The absorbed photons cause a non-equilibrium situation in th e system by in itiatin g m olecular excited states. Typically the excited states are unstable and energy is transferred inter-m olecularly a n d /o r intra-m olecularly in the su b strate and in the adsorbate as the excited system relaxes. Inter-m olecular transfers are exchanges of energy between molecules whereas intra-m olecular transfers are exchanges of energy w ithin a molecules’ energy states. An adsorbate molecule may transfer its energy by (1) radiative decay such as fluorescence and phosphorescence, (2) non-radiative decay such as charge transfer to the substrate, and (3) chemical transform ations [44]. Fluorescence is radiative decay where the transition is between electronic states of the same spin or m ultiplicity; phosphorescence is between elec­ tronic states of different m ultiplicities such as a singlet and a trip let state. Phosphorescence 1 Light y y X y A d so rb a te -O O O O O O O O Substrate Figure 1.1: Simple schematic of light directed onto an adsorbate-substrate system in equilibrium with its surroundings. takes place only in excited states th a t are long lived, since such transitions are forbidden [1]. T he studies in this thesis are concerned w ith process (3). Chem ical transform ations result in the adsorbate undergoing dissociative processes which sever th e inter-m olecular bond to the surface or the internal bonds of the molecule. In either case if th e dissociation is to occur, it m ust proceed on a tim escale th a t is com petitive w ith processes (1) or (2). In th e first adsorbate layer on m etals, the excited state lifetim e is on th e order of fem toseconds (1 femtosecond = 10“ ^^s). Therefore if dissociation is to be a significant process in th e first adsorbate layer, it m ust occur on th e order of femtoseconds. P article ejection occurs as a direct result of dissociative processes. Intra-adsorbate bonds or surface bonds are broken to cause molecular fragm ents or molecules to be ejected. Particle ejection processes are generally referred to as d e so rp tio n in d u c ed b y e le c tr o n ic tr a n sitio n s (D IE T ) . Experim entation is often done on ordered surfaces because th e resulting d a ta tends to be characteristic and reproducible. O rdered surfaces are formed from th e polished surfaces of bulk crystalline solids. C rystals do occur naturally, however, to guarantee th a t th e crystals used in experim ents are very pure and hence well-ordered, experim enters use artificially grown crystals. Crystalline solids are categorized by their electrical properties as m etals, semi-conductors, and insulators. T he electrical properties of th e su b strate crystals are known to have significant effects on th e adsorbate photodynam ics. A dsorbate molecules tend to arrange in characteristic p attern s on the crystal surface. M any species of molecules on crystalline surfaces have preferred bonding sites and typically align in preferred directions. Features of surfaces relevant to this experim ental work are discussed in th e next sections. 1.2 Substrate and Surface 1.2.1 B u lk C ry s ta l S tru c tu re a n d C leavage P la n e s A bulk crystal is m ade up of fundam ental volume units called prim itive cells; th e cells are m ade up of atoms in periodic volume arrangem ents. The prim itive cells th a t make up an ordered periodic crystal occur at regularly spaced intervals. W hen a bulk crystal is cleaved or cut the periodicity of the crystal is broken at th e newly formed surface. R econstruction I-/ Figure 1.2: Electrons leak out of the bulk into the space just above the surface causing a smoothing of the surface and contractive relaxation (arrows) of the layer of atoms nearest the surface. (Diagram from Zangwill [43, pg 29]; Finnis and Heine [14].) of the atom s in the first layers nearest the surface usually follows. In m etals th e surface conduction electrons lower their kinetic energy by rearranging th eir distribution in space [43]. The electrons leak out of the bulk into the space ju st above th e surface as in figure 1.2. The charge density of the electrons decreases in th e solid and results in a sm oothing of surface features [43]. The screening on the outerm ost layer of atom s decreases as a result and the layer is subject to a contractive relaxation in th e dim ension perpendicular to th e surface. The contractive relaxation is opposed by electron-electron repulsive effects ; equilibrium is reached when the two effects are balanced. R econstruction can also affect deeper layers [43]. The crystal structure of surfaces is described w ith surface nets, or two vectors in the plane of the surface. Every lattice point can be reached by the prim itive translation vectors of the surface net T , which has th e form T = m a, + nbg. a@ and bg are the prim itive vectors and m and n are integers. The periodicity in three dimensions of a real periodic crystal is also broken by dislo­ cations, or m isplacem ent of atoms. T he presence of dislocations and th e fact th a t a bulk crystal cannot be perfectly cleaved along any of th e three M iller indices of figure 1.3 results in steps and holes on the surface. Some of these steps and holes are removed by polishing the crystal face to a high degree of smoothness at an angle close to th e (110) cleavage face. The steps and holes th a t are left after th e crystal surface has been polished are electronically smoothed by electrons which leak out of the solid. Because th e electron density is higher in th e area where electronic sm oothing has taken place, surface chem istry can often occur at these sites. Step and holes have more surface area for molecules to in teract w ith because of the local curvature. A dsorbate atom s which bond to steps and holes often have a lowered adsorption energy and a stronger bond. Figure 1.3 shows the placem ent of atom s in an ideal face-center-cubic (FCC) unit cell. The structure of the FC C cell is a cube with atom s on each of the 8 corners and with an atom in the center of each of th e 6 faces. Cleavage of th e FCC unit cell in th e (110) plane shown in the FCC unit cell produces the FCC (110) crystal surface. The surface is corrugated. The [110] and [001] directions on th e surface are shown in th e diagram. FCC (110) Surface FCC Unit Cell 14— b----- t a • [001 ] Cleavage Plane [ 110] Figure 1.3: Face-Centered-Cubic(FCC) Unit Cell and the surface of an FCC crystal cut in a (110) orientation. 1.2.2 B a n d S tru c tu re o f S u b s tra te s The band structures of bulk crystals are well-known. The periodicity of bulk crystal stru c­ tures makes it possible for approxim ate solutions of the Schrddinger equation. The electron wavefunctions th a t participate in th e bonds between atom s in th e bulk are energetically split creating a large num ber of closely spaced discrete energy states. These allowed en­ ergy states form an allowed band of energies. There may be several allowed bands w ith disallowed energy bands between them . The disallowed bands are known as forbidden bulk energy bands. The bands of allowed energy states are full up to a m axim um energy and em pty above this energy. W hen th e crystal is at a tem p eratu re T = 0 Kelvin, th e m ax­ im um occupied energy state is known as the Fermi e n e r g y A t higher tem peratures T > 0 electrons can be therm ally excited above Ep- Cleavage of a bulk crystal leaves th e bulk band structure generally intact at the newly formed surface, however, allowed surface energy states often form in the forbidden bulk energy bands. There are essentially three broad classes of crystals: Insulators, sem i-conductors, and conductors. The location of the m axim um occupied energy determ ines the electrical char­ acteristics of the substrate. At T = 0 the Fermi energy lies at the top of an energy band in insulators. If the band is full, it is referred to as a valence band. The energy gap to the next allowed band of energies is large and is therefore, a large barrier to the therm al excitation of electrons. If electrons cannot be exited into th e conduction band by increasing the tem perature, thereby increasing the conduction of th e m aterial, then the m aterial is an insulator at th a t tem perature. W ith semi-conductors the Fermi energy also lies at th e top of a valence energy band at 0 K, but th e energy gap to th e higher energy conduction band is small enough th a t at higher tem peratures electrons can be therm ally excited across the en- V acuum I3 Bulk V) V Figure 1.4: Image, surface, and bulk potentials as a function of a line z normal to the surface. The image potential smoothly joins onto the surface potential, which in turn is joined to the bulk potential. Diagram from Inglesfield [18, pg 120]. ergy gap. The excitem ent leaves some holes in the valence band and some excited electrons in the conduction band. Light of sufficient energy, depending on th e energy gap between the valence band and conduction band, can induce photoconductivity when directed onto this m aterial. For metals, the conduction band of energies is only partially filled and th e Fermi energy lies far below the top of the band. Due to the close spacing between energy levels w ithin the conduction band, electrons can be excited from discrete energy levels below th e Fermi energy to discrete energy levels above the Fermi level. As a result m etals are good conductors and are very photoconductive. Light of sufficient energy can produce excited photoelectrons w ithin the m etal. Some of these photoelectrons can escape the bulk and be tran sm itted to the surface. 1.2.3 Im ag e a n d S u rface P o te n tia ls A charged particle at a crystalline surface disturbs the electronic equilibrium of th e surface. Charge density of opposite sign to the particle increases on the surface and the particle sees an attractive potential. T he potential far from the surface is th e asym ptotic image potential. Using simplified units; — (X) < z < 0 ( 1 . 1) where Vq is the potential of the vacuum, where e = 0 for a m etal surface and e = 1 for a perfect insulator, and where z defines th e distance of th e particle from th e surface. If |z| is large the potential the electron experiences is th a t of the vacuum . Near the surface where \z\ is small, th e attractiv e image potential is large b u t the repulsive bulk and surface potentials com pete and reduce the attractiv e affect. The result is th a t th e image potential sm oothly m atches onto the surface potential. Near the surface th e real potential is fairly linear and attractive; this potential joins onto the bulk periodic potentials. Figure 1.4 shows the image,surface, and bulk potentials as a function of a line z norm al to the surface. The attractive image potentials m eans th a t work will have to be expended in removing an electron or a charged particle from the surface. The workfunction energy is the m inim um energy an electron needs to escape the surface. It is defined as the energy required to remove an electron from the surface to the vacuum. 1.3 A dsorption 1.3.1 In tro d u c tio n One of the first steps in studying adsorbate-surface system s is in determ ining how molecules adsorb to surfaces. The bonds th a t are form ed result from an interaction betw een the surface and the adsorbate’s valence electrons. If th e interaction is strong, molecules can be broken apart as they approach the surface. W hether an adsorbate is characterized as physisorbed or as chemisorbed depends on th e n ature of th e surface bond. Physisorption is a weaker form of bonding and chem isorption is a stronger form of bonding. A dsorption of a molecule also implies th a t the molecule is not able to p en etrate th e surface. W hen a molecule adsorbs its valence electrons interact w ith su b strate valence electrons. The Fermi exclusion principle keeps the adsorbate electron, and hence th e molecule, out of th e lower energy states of th e substrate. Therefore, adsorption results from attractiv e and repulsive potentials. A dsorbates have vibrational quantum states w ithin these potentials. 1.3.2 P h y siso rp tio n Physisorption results from Van der W aals’ forces. W hen a neutral molecule is near a surface electron-electron interactions are initiated between the surface and th e nearby molecule. Dipole interactions between th e molecule and th e surface arise essentially from image effects. For exam ple figure 1.5 shows a hydrogen atom near a m etal surface. A lthough the hydrogen atom is in neutral state, dipole interactions w ith th e surface result because th e instantaneous distance between electron and th e surface is not th e sam e as th e distance between th e proton and th e surface. T he to tal electrostatic energy for this system ,U , can be defined as the sum of four term s, each term results from th e image potential [43]. For a m etal surface [43, pgl85], Simplifying this w ith a r j z power expansion [43, pgl85], T 3 ( 1 -3 ) 16 Since the r distance is less th an the z distance the leading term of this expansion indicates th a t th e interaction has a ^ dependency. T he m agnitude term s of the ^ variable are not considered to give the correct m agnitude of the dispersion force due to th e simplified treatm en t. Therefore the physisorption potential V(z) is given as, V{z) = ^ (1.4) / s / \ ' \ \ + \ r / e*/ / / F II î Figure 1.5: Hydrogen atom and it image near a metal surface. Van der Waals’ forces arise because of interactions between the charges of the hydrogen atom and their images. Diagram from Zangwill [43, pgl86]. where the C is a constant. On approach to th e surface the Pauli exclusion principle will repel the molecule due to overlapping wavefunctions of th e molecule and th e surface. The net potential on the molecule will be a result of th e repulsive Pauli potential and the attractiv e van der Waals potential and be of the form [43, pg 188]: V{z ) = K n { z ) — C (1.5) W here n(z) is the ground state charge density of the surface resulting from the surface electrons spilling out of the bulk into th e vacuum. K is a constant and z — Zy defines a reference plane from which the van der Waals potential should be m easured from. V(z) invariably has a shallow m inim um a few Angstrom s from th e surface [43]. Figure 1.6 shows the process of a molecule physisorbing to a surface. If th e molecule has sufficient energy it will scatter off th e surface and escape. It can also inelastically scatter and be trapped in the well created by th e van der Waals potential and th e exclusion principle. It will continue to inelastically scatter off th e surface until its translational energy is absorbed by surface. If a large num ber of molecules are physisorbed on a surface, collisions between adsorbed molecules will cause vibrational energy transfers. The transferred energy will allow some molecules to escape the potential well. T he tim e a molecule stays in the well depends on the adsorption potential of the molecule and th e tem p eratu re of the surface. Cooling removes therm al energy from the adsorbate which helps the molecules stick to the surface. 1.3.3 C h e m iso rp tio n Chem isorption is a stronger form of bonding th an physisorption. C hem isorption results in the outer orbitals of the adsorbate molecule and th e Fermi or outer energy states of the surface fusing into combined quantum energy states. Bonding and anti-bonding orbitals are form ed w ith electrons filling the lower energy bonding orbitals first. Fig 1.7 illustrates a schem atic potential for a molecule to chemisorb to a surface. The molecule initially induces B g Ie B £ Distance from surface z Figure 1.6: Physisorption of Molecules on Surfaces. Inelastic scattering traps a molecule at a surface due van der Waals potential. (Diagram from Zangwill [43, pg361]; Tully [35]) van der Waals effects th a t produce a physisorbed potential well. T he molecule m ay be initially trapped in this well and then moved into th e deeper chem isorption well which is localized closer to the surface [43]. Chem isorption is a form of conventional chemical binding with a heat of adsorption on a scale of 1-10 eV [43]. 1.3.4 A d s o rb a te O v erlayer S tru c tu re s on S u rfaces W hen gas-atoms condense on a su b strate they form overlayer structures. There are essen­ tially two types of adsorbate overlayer structures. A com m ensurate overlayer stru ctu re is a two-dimensional space group th a t overlaps the su b strate space group symmetrically. An incom m ensurate overlayer structure has no sym m etry related to th e su b strate space group. The difference between overlayer groups is related to the binding site preferences of ad­ sorbed molecules or lack of preference. An incom m ensurate overlayer generally does not have binding site preferences on th e surface it is attached to. Molecules may absorb on top of substrate atom s, betw een substrate atom s, and in hollow sites. Molecules may stand up, lie down, or tilt at an angle with respect to the surface [43]. 1.4 Energy Transfer M echanism s b etw een the A dsor­ bate and Substrate 1.4.1 In tro d u c tio n There are several means whereby photoenergy can be transferred to th e adsorbate. The sim plest means is for a photon to be directly adsorbed by the adsorbate or by th e adsorbatesubstrate bond. This is referred to as direct photoabsorption. A second means of transferring energy results from absorption of photons in the substrate. An excited electron and hole V(z) Physisorbed Well Chemisorbed Well Figure 1.7: Chemisorption of Molecules on Surfaces. Adsorption potential with a precursor physisorbed well and a deeper chemisorbed well closer to the surface. One-dimensional model of chemisorption by Lennard-Jones(1932) Diagram from Zangwill [43, pg366]. are created, which relax independently. It is possible for either the electron or the hole to be transferred to the adsorbate or adsorbate-substrate bond. A th ird means for transferring energy to th e adsorbate results from therm al excitation of the substrate-adsorbate system. T herm al excitations occur when energy dissipates into vibrational or rotational modes. Excited electrons or holes which are produced in th e sub­ strate and do not a ttach to the adsorbate m ust be therm alized in th e su b strate [44]. This therm al energy can be transferred to the adsorbate and result in ejection of particles when m olecular vibrational or rotational energy is converted to translational energy. Generally we will want to avoid conditions in our photoabsorption experim ents which will in itiate significant therm al energy transfer, since this would com plicate the study. Avoiding signif­ icant therm al effects is relatively simple to do, since th e photoabsorption experim ents in this thesis are all done at light intensities below th e threshold per pulse) [44] considered necessary to initiate therm al effects from a laser. Therefore therm al effects are not considered significant in the photodynam ic experim ents done for this thesis. A fter energy is transferred to the adsorbate, there are relaxation channels available to the excited adsorbate molecules. If chem istry is to occur in th e adsorbate, the relaxation channel for the chem istry m ust be tem porally com petitive w ith the quenching process. Quenching transfers th e electronic excitation from the ad so rb ate’s states to th e energy states in the substrate. On m etals the electronic excitations are quenched on a tim escale of femtoseconds. Quenching is slower or non-existent for semi-conductors or insulators. In this thesis we collect particles from desorption processes as a result of th e surface photodynam ics, therefore these relaxation processes m ust be com petitive on the particular adsorbate-surface system s we are studying. D esorption results in molecules or molecule hot electrons Adsorbate Resonance (LUMO) I (HOMO) Substrate Adsorbate Figure 1.8: Charge Transfer at a surface. Hot electrons are produced at a surface which then dissociatively attach to the adsorbate. The LUMO is the lowest unoccupied molecular orbital and the HOMO is the high occupied molecular orbital. Diagram from Wolf [40]. fragments being ejected from the system . The ejected particles contain inform ation on the final state distributions of the resulting degrees of freedom [45]. Since these degrees of freedom include internal, vibrational, rotational or translational excited states, they contain a great deal of useful inform ation for studying th e surface chemistry. Inform ation can be extracted from the ejected particles by collecting them and analyzing them w ith regard to such variables as species,intensity, energy, and trajecto ry angle. We know th a t two general processes result in desorption from th e surface when energy is transferred to the adsorbate. In one process the energy transferred to the adsorbate causes dissociation of adsorbate molecules which produces fragm ents w ith translational energies. In another process energy transferred to the adsorbate causes in tact molecules to desorb from the surface when internal energy is converted into translation energy. In b o th dissociation and molecular desorption translational energy is a necessary condition for a particle to escape the surface, however it is not a sufficient condition. Factors which affect escape will be discussed later in this chapter. Dissociation and desorption in itiated by photoabsorption in the system are respectively referred to as photodissociation and photodesorption. Their kinetics will be described in detail in th e succeeding sections of this chapter. 1.4.2 C h arg e T ra n sfe r fro m th e S u b s tra te to A d s o rb a te Charge transfer occurs when charge is transferred between th e adsorbate and the substrate. The process for charge transfer, see figure 1.8, results from th e photoelectric effect. Photons p enetrate to various depths in the substrate depending on the energy of th e photons and the dielectric properties of the substrate. For photons w ith energy hu < lOeV th e optical penetration depth is % 100 in m etals [40]. Electrons in th e m etal su b strate absorb and are energetically excited by incident photons creating w hat is known in band structure as electron-hole pairs. R elaxation of th e electron and the hole are independent following 10 100 - A"# 30 "6 1 ! A .. » •^®W Iz 10,- 4:, I I Ill'll 10 ■ I I I 11III 30 100 *#» O,# C» •#*# Mo #A: I 1 I I I 111i _L 300 2000 000 EhclfoocmngyCeV) Figure 1.9: Mean Free Path of Electrons in various Solids. The meanfree path of electrons in solids is of the same order but less than the optical penetration depth of % 100^4. Therefore most excited electrons will undergo scattering processes. The dashed curve is th at expected from theory. (Zangwill [43, pg21]; Rhodin and Gadzuk [29]; Somorjai [31]. Theoretical curve from Penn [28].) creation. The excited electron is often referred to as a photoelectron. R elaxation of the photoelectron and the hole is the result of scattering. Fig 1.9 shows the m easured and theoretical meanfree paths for electrons in various solids. The m eanfree p ath of electrons in solids is of the sam e order bu t less th a n th e optical penetration d ep th of ~ 100/4. Therefore m ost excited electrons will undergo scattering processes. Scattering w ith cold electrons produces secondary photoelectrons and holes [40, 44]. B oth photoelectrons and holes can be transported to th e surface by scattering events [44]. Due to surface screening effects, photoelectrons and holes are not correlated, th a t is they do not form an electron-hole pair known as an exciton [44]. Photoelectrons on th e surface w ith energy less th an th e vacuum energy are further designated as hot photoelectrons. If photoabsorption is characterized by a high probability of single-photon processes and a low probability of m ulti-photon processes, there will be a distribution of photoelectrons at the surface w ith a m axim um kinetic energy equal to the energy of the incident photons. M ultiphoton processes are insignificant when light intensities are low. M ultiphoton processes can be initiated for laser power densities above 10^ MW / cm^ [44] and therefore they can be avoided for intensities much lower th an th a t. Photoelectrons and holes scattered at an adsorbate covered surface are subject to the electronic interface potential between the substrate and the adsorbate. Due to the quantum mechanical nature of electrons, scattering at a surface m ay result in electrons and holes tunnelling through the interface potential causing valence ionization of the adsorbate. An electron attaching to the adsorbate can be represented by 11 A —B G —^ A — B where A-B represents an absorbed molecule and e~ represents an excited electron which attaches to the adsorbed molecule. A hole attaching to the adsorbate can be represented by 1.4.3 D ire ct E x c ita tio n o f th e A d s o rb a te o r A d s o rb a te -S n b s tra te bond Direct photoabsorption by the adsorbate or th e adsorbate-substrate bond can prom ote molecules to repulsive excited states. In the case of direct excitation of a physisorbed adsorbate, valence electrons are excited to higher energy states. The excitations can cause repulsion of the adsorbates’ nuclei which can lead to dissociation or desorption if these pro­ cesses occur on a tim escale com petitive w ith quenching. T he surface bonds of chem isorbed molecules are typically formed by shared orbitals of th e su b strate and th e adsorbate. Pho­ toexcitation in chemisorbed molecules prom otes electrons from bonding orbitals to a n ti­ bonding orbitals which can lead directly to desorption or dissociation [3]. D irect excitation of a molecule A-B by light of energy hv can be symbolically represented by A —B hu —y A — B* Dissociation and desorption are described w ith potential energy surfaces in th e Surface Dynamics section in this chapter. See figure 1.10 for a potential energy surface resulting from a direct excitation from the ground state A-B to an excited A — B* state. 1.4.4 T h e rm a l P ro c e sse s Most photoelectrons excited by photon absorption undergo scattering processes, since the optical penetration depth of the light is larger th an the m ean-free-path of th e scattering photoelectrons [44]. Photoelectrons w ith sufficient energy m ay be photoem itted from the substrate through the adsorbate-substrate interface. However photoelectrons w ith insuf­ ficient energy to be photoem itted will be absorbed in the su b strate through scattering processes [44]. Collisions between holes, electrons, and phonons will cause excitation energy to be transform ed to lattice phonons [44]. Such therm al excitation can transfer vibrational energy to the adsorbate. If sufficient vibrational energy is transferred the adsorbate can dissociate or desorb. For pulsed laser light the rate of dissociation due to therm al excitation depends exponen­ tially on the intensity of the light [44]. Lasers w ith low light intensities and less pulses per second will result in fewer therm al excitations. In this research therm al effects are avoided by using light intensities below the threshold for significant therm al effects to be present. 1.4.5 Q u en ch in g M olecular quenching is a process th a t transfers electronic energy from the states of the adsorbate to the states of the substrate. The rate of quenching depends sensitively on the natu re of the substrate and on w hether resonant tunnelling between the adsorbate states 12 and the substrate states is available. Resonant tunnelling occurs when the energy of the electron state in the molecule is the same as the em pty states th a t are available on the surface. Electrons can tunnel between the molecule states and th e em pty surface states w ithout gaining or losing energy. Resonant tunnelling is not available when th e energy of the electron state in the molecule coincides with an energy in a forbidden bulk band of the substrate or filled surface states. W hen resonant tunnelling is not available quenching m ust proceed by the Auger process, which is considerably slower. In this Auger process, a higher energy electron from the substrate enters a lower energy orbital of an adsorbate ion, displacing an electron already in the state, with some energy is transfer to an electron in an adjacent orbital of the ion. This ex tra energy is sufficient for th e electron to enter an allowed and em pty state of the substrate, leaving the adsorbate molecule in a neutral state. W hether the substrate surface is an insulator, sem iconductor, or m etal can have a large effect on the tim e factor for quenching. Efficient quenching m echanism s do not exist on ideal insulator surfaces because of the large energy gap between th e fully occupied band of energies and the higher energy unoccupied band [45]. M etal and sem iconductor surfaces, in contrast, are often strongly quenched and typically quenching rates are of the order of femtoseconds [40, 44, 45]. 1.5 Surface D ynam ics 1.5.1 In tro d u c tio n Surface dynamics in this thesis are studied prim arily by collecting desorbing particles, th ere­ fore understanding the desorption processes is im portant. There are only three processes which can transfer photoenergy to th e adsorbate: therm al transfer from the su b strate to th e adsorbate, charge transfer from the substrate to the adsorbate, and direct photoabsorption by th e adsorbate. The photodynam ic experim ents are done at laser light intensities below th e threshold needed to in itiate therm al effects so we need only consider charge transfer and direct effects in th a t particular experim ent. However therm al energy transfers will be significant in the tem perature desorption experim ents. The particles ejected from the sur­ face as a result of direct and charge transfers in the photodynam ic experim ents can either be neutral or charged. N eutral particles require less energy to escape a m etal surface th an charged particles due to the attractiv e image effects on ions. For an ion to escape the surface the energy transfer creating th e ion m ust be greater th an the workfunction of the surface. The workfunction is th e m inim um energy an electron and any singly charged particle requires to escape the surface. At th e laser light intensities th a t we work at, energy transfers occur as a result of single-photon processes and each photon has less energy th an the workfunction. Therefore in the direct photoabsorption process less energy than the workfunction is transferred. In the charge transfer process th e m axim um energy of any electron transferred to the adsorbate is the energy of a single photon. If charged particles are to escape the surface additional energy m ust be provided. Chemical reactions and adsorbate electron affinities can provide th e additional energy. If a particle does not have sufficient energy to escape as an ion, it m ay neutralize and escape by neutral desorption mechanism s. B oth direct photoabsorption and charge transfer to the adsorbate can give rise to neutral particle ejection. T he processes are described with 13 Photon Capture 0 c til Quenching a + B Interatomic Distance (A-B) Figure 1.10: Frank-Condon transitions on potential energy surfaces(PES) are vertical transitions. The internuclear separation is unchanged during the transition. The Franck-Condon transitions have an envelope determined by the localization of molecules in the well (gaussian figure). potential energy surfaces(PES), although knowledge of adsorbate p o ten tial energy surfaces is very lim ited. Knowledge of excited FES is m ore lim ited th a n th e ground sta te FES [44]. Potential energy surfaces derived from detailed quan tu m m echanical calculations are not available, therefore as a substitute, simplified models are proposed to account for the observed behaviors [44]. Gaining inform ation about a surface FES is an objective of surface studies. It should be clarified, th a t we do not have th e m eans to create accurate simplified surface FES for the system s in this study. We can only gather q u alitative inform ation th a t will give us b e tte r insight. Gas-phase FES are often used in th e analysis for insight into an adsorbates' chemical behavior, w ith th e understanding th a t adsorption changes th e FES m ore or less. 1.5.2 T h e F ran c k -C o n d o n P rin c ip le W hen a gas-phase diatom ic molecule absorbs a photon th e excitation causes th e molecules nuclei to reposition. Franck and Condon realized th a t th e energy transfer to th e diatom ic molecule occurs on a m uch faster tim escale th a n th e repositioning of th e nuclei [1]. Therefore th e energy transfer can be approxim ated as a vertical transfer of energy, w ith th e nuclei of th e diatom ic atom s in th e ground-state position. In figure 1.10 th e potential energy of a molecule attach ed to a surface is represented by a ground state potential energy surface. A transfer of energy to th e molecule results in a sudden Franck-Condon transition,represented as a vertical line, to an excited potential energy surface. The intranuclear separation does no t change during the transition. An envelope for the Franck-Condon transitions is determ ined by th e w avefunction th a t describes 14 the vibrational energy state in the ground state wavefunction. W hen a molecule undergoes a Franck-Condon transition the shape of the excited dissociative curve and the FranckCondon envelope result in a broadening of the energy distributions of dissociated particles. Steep potential energy curves in the Franck-Condon region and wider envelopes resu lt in broader energy distributions. 1.5.3 P o te n tia l E n e rg y S urfaces for D e so rp tio n a n d D isso c ia tio n A widely accepted model of desorption induced by electronic tran sitio n (D IET) is th e m odel of figure 1.11 proposed by M enzel,Corner and Redhead (M GR) in 1964 [44]. In figure 1.11 a Franck-Condon transition takes an adsorbate from th e v=0 vibrational state of ground state wavefunction to a repulsive excited state. On the excited state th e adsorbate experiences nuclear motion and is repulsed from th e surface. Escape from th e surface on th e excited state curve is possible, however due to efficient decay channels the adsorbate m ay undergo a de­ excitation back to the ground state potential energy curve. T he de-excitation is a quenching process and the energy is converted into substrate excitation. As a result of excitation to the dissociative state and the subsequent quenching, th e adsorbate A gains potential and translation energy. The adsorbate A returns to the ground state w ith potential energy plus vibrational energy equal to E'^.. If in figure \ .11 E'^. > D' sX th e de-excitation point, then th e adsorbate A may desorb in the ground state and can be detected by time-of-flight mass spectrom etry(TO F-M S)^The to tal energy gained after retu rn to th e ground state m ust be greater than D q to overcome the a ttra ctiv e potential of th e ground state. Do is th e energy the adsorbate requires to desorb from th e v = 0 vibrational state in th e ground state. If th e energy is insufficient, E'^ < D ' , th e adsorbate vibrates in th e ground state well and can lose th e energy it gained from the excitation by coupling with phonons of th e surface. The coupling can return the adsorbate to th e v = 0 vibrational state in th e ground state. On m etal surfaces efficient fast quenching mechanisms result in th e probability of direct desorption on the excited state PES being low [44]. Nuclear m otion occurs on a m uch slower tim escale th a n electronic quenching. T he probability of desorption is determ ined from th e initial excitation probability tim es th e probability of survival on th e excited state to obtain an energy > D' [44]. The M enzel,Com er and Redhead model is particularly useful when describing th e direct excitation process of photoabsorption. In th e case of direct photoabsorption, an electron in an adsorbate molecule absorbs a photon and is excited to a dissociative state where quenching processes can occur. Desorption by th e M enzel,Com er and R edhead m odel can occur as a result. For some chemisorbed adsorbates, valence ionization can lead directly to repulsive states [3]. Valence ionization occurs when th e adsorbate gains or loses an electron creating a negative or positive ion. The M C R model can be extended to describe intra-adsorbate bond dissociation [44]. In figure 1.12 an adsorbed molecule is excited to a repulsive excited state. T he molecule m ay dissociate in the excited state or quenching can bring th e m olecule back to th e ground sta te potential energy surface. If th e energy gained on th e excited state is sufficient, the molecule A-B may dissociate in the ground state or one or both elem ents A, B can enter ^TOF-MS is described in the experim ental section and is one of principle m eans o f obtaining data in this thesis. D ata obtained by TOF-M S is described in Chapter 4 o f this thesis. 15 O) Distance from surface Figure 1.11; Potential Energy Surface showing the MGR desorption mechanism. Diagram from Zhou [44]. the chemisorption well. If the energy gained is insufficient for dissociation, th e molecule A-B may also rem ain in the physisorption well. T he chem isorption well in this case results from one or both of the fragm ents forming a chemical bond (chemisorbing) to the surface. One or both of the fragm ents A, B m ay escape the surface after entering the chem isorption well if th e molecule acquires sufficient energy after transferring to th e ground state potential energy surface from the excited state, T he Antoniewicz desorption m echanism in figure 1.13 describes desorption from bound excited states. This m echanism occurs as a result of a Franck-Condon transition from a bound ground state wavefunction to an excited bound state. T he excited state is attractiv e with the result th a t th e molecule is moved towards th e surface. W hen the distance between the surface and the molecule is small enough, the molecule is repulsed by the Fermi exclusion principle. Franck-Condon de-excitation transitions back to the m olecular ground state result from quenching. If sufficient energy has been gained as a result of th e excitation and quenching processes, th e molecule m ay desorb in the ground state. This m echanism is particularly useful for describing desorption processes due to charge transfer between a m etal substrate and an adsorbate. In th a t case th e excitation to the excited state is the result of an electron attaching to th e adsorbate or an electron from the adsorbate escaping to th e substrate in a process referred to as valence ionization. In one case a negative ion is formed and in the other a positive ion is formed. A bound state is formed as a result of ionization because the ion is attra cted to its im age charge in th e substrate. Ionization of an adsorbate can result in dissociative states of th e molecule or in stable ionic states; desorption can generally occur for both dissociative and stable ionic states. W hether an unstable ion will dissociate or desorb can depend on th e com petition between th e two processes and on th e Franck-Condon transition energies. E xcitation to the bound excited 16 n I Reaction Coordinate Figure 1.12; Potential energy surface showing a surface dissociation mechanism similar to the MGR model for desorption. One or both of the dissociative products can chemisorb to the surface. Diagram from Zhou [44]. state followed by quenching to the ground state can also lead to containm ent of m olecule in the ground state well if there is insufficient energy gained. Coupling to su b strate phonons can retu rn the molecule to the v = 0 vibrational energy state. 1.5.4 C o n clu d in g R e m a rk s This concludes the section on potential energy surfaces for desorption and dissociation and also th e chapter on surface photodynam ics of adsorbates on surfaces. The p o ten tial energy surfaces described above can generally describe th e surface dissociation and desorption pro­ cesses. D etailed potential energy surfaces,however, are not available. D etailed PE S would provide a predictive m eans for understanding th e photodynam ics on surfaces. Q u an tu m m e­ chanical calculations of detailed FES require b e tte r understanding of th e surface processes on individual surface systems. T h at is w hat this thesis represents. It is an a tte m p t at a b etter understanding of several specific surface system s. These system s are introduced in C hapter 3. Some of the factors which are significant to surface photodynam ics have been reviewed in this chapter although the review is by no m eans exhaustive. There are m any detailed factors which result in th e photodynam ics of adsorbate-surface system s. 1 have ju s t described some of th e im portant structural and electronic factors. T he next chapter will describe the experim ental apparatus and experim ental techniques whereby d a ta is obtained for th e study of surface system s for this thesis. 17 >« e 0) c LU Reaction Coordinate Figure 1.13: Potential energy surface showing the Antoniewicz desorption mechanism. Diagram from Zhou [44]. 18 C hapter 2 E xperim ental M eth od s 2.1 Introduction The experim ental m ethods chapter discusses th e experim ental ap p aratu s and th e experi­ m ental techniques for acquiring data. B ut before I begin a description of the experim ental details I m ust answer an im portan t question, “w hat is th e aim of th e experim ents?” As a surface experim entalist I m ust be able to form carefully controlled ad so rb ate-su b strate sys­ tem s and m ust have precise tools for m anipulating those system s. In a very general sense, the aim of the experim ents is to modify the ad sorbate-substrate system in some significant and m easurable way. W ith th e experim ental d ata, I can say som ething significant about the surface physics. Saying som ething significant about th e surface physics m eans I m ust know w hat molecules are on th e surface. T he sample used in th e experim ents m ust be of a high p u rity and it m ust also be clean when I dose molecules onto it. T he molecules I dose m ust also be of high purity and I m ust be able to m aintain a clean ad sorbate-substrate system at least while the experim ent is in progress. Cleanliness is of course a relative te rm and is lim ited by the experim ental apparatus. There will always be low concentrations of undetectable species in the experim ental environm ent. U nderstanding th e surface processes in an experim ent is to a large degree predicated on assum ptions about which molecules are significantly present in the experim ent. Good assum ptions make good understanding m ore probable. In this thesis surface experim ents are done in an ultra-high-vacuum ( UH V ) environm ent w ith a base pressure of 5 x 10~^° torr. High tem p eratu re heating and A r+ ion bom bardm ent is available in th e UHV to clean th e sample. Auger spectroscopy and Low Energy E lectron Diffraction (LEED) are available to check th e sam ples’ cleanliness and good orderliness of th e surface structure. T he molecules dosed onto th e sam ple are bought from a m anufacturer in a high purity concentration. Once an adsorbate-substrate system is form ed in th e UHV environm ent th e system is modified by th e addition of th erm al energy or light photons. T herm al energy transfers to th e sam ple are m onitored by m easuring th e sample tem p eratu re and light photons are m onitored by m easuring th e num ber of photons p er crn^ on th e sample. As energy is added to the system , detectors m easure species desorbing from th e surface of the system . 19 Diverging .Lens Muror Nitrogen Laser Converging I^ns Light Path UV Grade Silica Window LEED/RFA Computer Ion Pump i Channeltron Ionizer UTI lOOC Mass Spectrometer Q j —8.3cm TuiboMCS Mass Spectrometer Heating Filament Sample Ion Pump Lock-In Amp Figure 2.1: Schematic of experimental system 2.2 A pparatus Various d ata acquisition electronics and com puter software are used in this thesis work. All com puter functions were controlled using a M acintosh Pow erPC clone( PC120) running com m ercially available software, although m any of th e program s are custom w ritten. A commercial visual constructs program (LA BV IEW ) was used to w rite com puter program s to control various lab equipm ent such as Auger, Q uadrupole Mass Spectrom eter, th e work­ function experim ent, and the laser pulses. These control program s were w ritten by Dr Jensen. A commercial graphing program (Igor) was used to plot th e d a ta after it had been acquired in an experim ent. T he experim ental system (see fig 2.1 ) is composed of a custom -built U ltra High Vac­ uum (UHV) cham ber, a crystalline sam ple, and is equipped w ith a rear-view LEED /R FA Auger analyzer (O CI Vacuum engineering), a quadrupole m ass-spectrom eter (U T I lOOC), a low energy electron gun, and an ion gun. The system has th e ability for doing stan­ dard surface science techniques including Low Energy Electron D iffraction (LEED ), Auger E lectron Spectroscopy (AES), Q uadrupole Mass Spectroscopy (Q M S), T em perature Pro­ gram m ed Desorption (T P D ), and Work Function (A $ ) m easurem ents. T he Ar~^ ion gun is used for sputtering the sample. A heating system and a cooling system is available for 20 Cooling Hose attached J] Micrometer Raise/Lowers Bellows -hollow tube X-axis(Coming out of the page) Ion Pump Mass Spec Sample Ionizer Channeltron LEED/RFA Figure 2.2: Cutaway drawing of the UHV Chamber. Apparatus is arranged on the UHV chamber in two tiers facing the central vertical axis. A sample manipulator can raise,lower, rotate the sample in convenient orientations for each experiment. m aintaining the tem p eratu re of th e sam ple from ~ 95FT to tem p eratu res in excess of 1200 K. The upper lim it is not known since th e samples used in th e cham ber m elt before this tem perature is reached. T here is a laser, not p a rt of th e UHV system , which produces light of wavelength A = 337 nm . The various p arts of th e experim ental system are described below. 2.2.1 U ltra -H ig h V acu u m C h a m b e r The u ltra high vacuum (UHV) cham ber (See figure 2.2 for a cutaw ay drawing) has a base pressure in the low 10“ ^° to rr range and a working pressure in th e m id 10“ ^° to rr range. It is pum ped using a m echanical pum p (Edwards RV5) and a turbom olecular pum p (Varian V250). The turbom olecular pum p has a num ber of tu rb in e blades which ro tate at a high rpm . This pum p can only be operated when th e cham ber pressure is low (approx. 10“ ^ torr). W hen molecules in the cham ber come in contact w ith th e tu rb in e blades they are knocked into th e m echanical pum p, which removes th em from th e UHV system . Because th e m echanical pum p is used to initially reduce the pressure to % 1 0 ~^torr it is frequently referred to as the roughing pum p. An interlock valve is used betw een th e cham ber and the roughing pum p to prevent m echanical pum p oil from leaking into th e cham ber. T here are two ion pum ps (Perkin Elm er) attached to th e quadrupole m ass spectrom eter (QMS ). The pum ps ionize residual gas and cause it to collide w ith reactive m e tal plates. This ‘rem oves’ 21 the gas from the UHV environm ent. There is an external heating system for baking th e cham ber. T he cham ber is baked at 120 degrees C for at least 48 hours in order to remove w ater {H 2 O) molecules th a t condense on its inner walls. W ater molecules are always present in air and en ter the cham ber when it is opened. T he w ater vapour pressure prevents th e cham ber from pum ping down to UHV pressures. W ater molecules adsorbed on th e inner cham ber walls outgas very slowly . Heating the cham ber hastens their removal. W hen th e w ater is gone low cham ber pressures are attainable. Gas handling was done using a glass gas rack and leak valves on the cham ber. The glass gas rack allows storage of chemical liquids in small glass flasks (100ml capacity). The gas rack is pum ped by a m echanical vacuum pum p(E dw ards RV5). A variable leak valve ( Duniway) is used to control the admission of gases into th e cham ber from th e glass rack. A high volume all-m etal (MDC) valve was used for flooding th e cham ber w ith gaseous nitrogen when bringing it up to air for various m aintenance and m odification work. To bring gaseous nitrogen into th e cham ber a line was connected to coiled copper tubing im m ersed in water. One end of the copper tubing was connected to a line im m ersed in liquid nitrogen in a container. W hen th e high volume valve on th e cham ber was opened liquid nitrogen was sucked into the copper tubing, acting as a heat exchanger, where it was converted to gaseous nitrogen before entering the cham ber. An ion gauge (Varian) and an ion gauge controller (G ranville-Phillips 350) was used to m easure th e UHV pressure. Several UV com patible windows were m ounted on th e cham ber. One was used to allow UV laser light into th e cham ber. The other was used to visually align the laser light onto the sam ple using a HeNe laser( Scientific SLC). T he visible red He-Ne laser beam (0.95mW o u tput) was aligned w ith th e invisible UV laser beam , and th e n the combined beam s were visually aligned onto th e sam ple using th e visible He-Ne beam . 2.2.2 S am ple a n d M a n ip u la to r The C u(llO ) sample is m ounted on a ta n ta lu m plate by tun g sten wires th a t pass through the edges of the sam ple and also through th e ta n ta lu m plate. The ta n ta lu m p late is attach ed to a m anipulating rod w ith sapphire spacers betw een th e m ounts of th e ta n ta lu m p la te and th e m anipulating rod. T he sapphire spacers allow th erm al contact b u t electrically insulate the sample from the m anipulating rod. The m anipulating rod is hollow and liquid nitrogen at its boiling tem perature is passed through th e rod. T he base te m p e ratu re of th e sam ple is ~ 96 K. Two K -type therm ocouples are spot welded to th e tu n g sten m ounting wires next to the sample. An additional wire used for grounding th e sam ple as necessary is spot welded to another tungsten m ounting wire. A pparatus is arranged on the UHV cham ber in two tiers facing the central vertical axis. A ppropriate placem ent of apparatus on the cham ber ports is necessary if two or m ore apparatus are required in an experim ent. T he m anipulating rod th a t th e sam ple is attach ed to is p a rt of a translation stage and ro tary feedthrough allowing it to be moved in X-Y-Z directions and rotated about an axis. Translation is lim ited to 15 cm in a vertical Z direction and 3 cm in a horizontal X and Y plane; ro tatio n about th e Z-axis is 360 degrees. Therm ocouple wires and an additional wire for grounding are attach ed to th e sam ple and then externally exited from the vacuum cham ber using com m ercially available connectors. 22 UEÊD/Aug®r Thernwcouoe Wire CN30Ù0 Temperaturs Controller DC Vodmo# Supply □ Sample 'Nceratnic DC Offmet Couplings Figure 2.3: Circuit for controlling the temperature of the sample, including heating and cooling. There are two thermocouple wires attached to the sample which are used for measuring the sample temperature. The additional wire allows the sample to be grounded to th e cham ber potential, or to have a potential placed on it.A tungsten filam ent is m ounted in a filam ent holder behind the sample to allow heating of the sample. 2.2.3 H e a tin g th e Sam ple The tem perature control circuit is shown in figure 2.3. H eating of th e sam ple is controlled by a program m able tem perature controller(Omega CN3000) using a PID (proportional integral derivative) algorithm . Sample tem p eratu re is m easured by the controller using a Chrom alAlumal (type K) therm ocouple, which is attached to the sample. T he controller heats the sample by controlling a variable DC current to the coiled tungsten wire m ounted behind sample. T he filament is m ade by coiling a tungsten wire and is connected to external con­ nectors on the vacuum cham ber. T he DC current (0-8 am peres) is pu t through th e filament using a DC Power Supply(H P Harrison 6286A). Electrons and photons produced from the tungsten wire radiate in all directions. Some are therm ally absorbed by th e grounded sam ­ ple. W hen electrons from the filament strike the sample, their kinetic energy is transferred resulting in heating of th e sample. The filament can be biased from ground to -800 volts by a DC offset (Sorensen DCRB Power Supply). The negative bias is used when flashing or annealing th e sam ple to high tem peratures. T he bias repels electrons produced at th e filam ent, increasing their kinetic energy. The tem p eratu re controller can linearly ram p th e sam ple tem p eratu re as a function of tim e to a specified value. However, it has not been possible to linearly ram p the tem p eratu re using the DC bias offset in the current set-up. W ithout the bias, heating of 23 the sample is lim ited to about 400 C. W ith the bias sample tem peratures can be reached in excess of 1200 C. 2.2.4 Q u a d ru p o le M ass S p e c tro m e te r The quadrupole mass spectrom eter (QMS) (UTI lOOC) is m ounted in a stainless steel tu b e on the UHV chamber. The tube is part of the UHV system and is also pum ped by two magnetic ion pumps (Perkin-Elm er) w ith a combined capacity of 50 L /s. T he QMS is separated from the m ain cham ber by a grounded m etal shield w ith a 4-m m -diam ap ertu re surrounding the QMS ionizer. There is m etal grid over th e ap ertu re th a t is electrically insulated from the m etal shield. W hen th e ionizer is turned on, th e grid is charged up by electrons produced in the QMS ionizer. The potential on th e grid repels electrons back into the ionizer and prevents them from entering the cham ber where they could cause reactions in an experim ent. As a result the current incident on th e sam ple from the ionizer is not m easurable (less than 0.01 pA ) [6]. Ions in th e cham ber are also prevented from entering the ionizer region of th e QMS. As a result, only neutral particles produced in th e time-of-flight experim ents can enter the ionizer. In m ost experim ents th e distance from th e center of the chamber, where the sample is situated in an experim ent,to the QMS ionizer is 8.3 cm. The U TI lOOC QMS can m easure masses in th e range from 0 to 300 atom ic mass units (amu). N eutral molecules and neutral fragm ents produced from the sam ple enter the ionizer by passing the electrically floating m etal grid and going through the 4m m aperture in the m etal shield. The small aperture in th e m etal shield selectively increases th e probability th a t the molecules entering the ionizer subtend a small solid angle. Particles which come from large solid angles are scattered away from th e ionizer by th e shield. In the ionizer molecules and fragm ents can be positively ionized and th en m ass-selected by th e quadrupole massfilter to enter the channneltron. There is a delay between tim e when th e molecule or atom is ionized and the tim e when it enters th e channeltron. For th e U TI lOOC th e delay tim e for m /e = 115 am u ions has been determ ined to be ~ 20p s [32]. In the channeltron electron m ultiplier, the positive particles cause electron cascades. T he electron cascades in tu rn initiate a m easurable electric signal th a t indicates th a t a positive particle entered the channeltron. The channeltron can be operated in pulse counting m ode or in an analogue mode. In pulse counting mode each selected fragm ent th a t enters the ionizer is registered as a pulse , whereas in analogue mode the intensity of fragm ents entering the channeltron are registered as an analogue current signal. 2.2.5 L aser The laser used in the experim ents is a pulsed nitrogen laser (M olectron UV-12,A = 337.1 nm) for surface photochem ical experim ents. It delivers an energy of 0.87m J/cm ^ per pulse and is operated at lOHz w ith a pulse tim e w idth of 10ns. The laser is not p art of the UHV system . A commercially available light m eter(M olectron PowerM ax 500D) w ith a light m eter probe (M olectron PowerMax PM 3Q) was used to m easure th e light power of the laser. 24 2.3 Standardized Surface Techniques 2.3.1 Sam ple P re p a ra tio n The C u(llO ) sample surfaces were initially prepared by cycles of Argon ion (A r+) bom bard­ m ent and annealing to high tem peratures in the UHV cham ber. The argon ion bom bardm ent process began by lowering the sample in th e cham ber to the level of the Argon ion gun, which consists of a tungsten filament and focussing lenses. The power supplies to th e fil­ am ent and the lenses were turned on. The cham ber was backfilled w ith Argon gas to a pressure of 6 x 10“ ^ torr. To make sure th a t the Argon ion beam was actually striking the sample th e current on the sample was m easured. This was done by p u ttin g a current m eter between the sample and ground. Typically a current of 1^ A was detected when th e beam was on the sample surface. The sam ple was th en biased to -250 eV in order to increase the kinetic energy of the bom barding ions to 750 eV. A fter 20 m inutes the bias and the Argon gas supply was turned off. T he sam ple was annealed to a high tem perature(920° C) and allowed to cool. Auger Electron Spectroscopy was done to check if the sam ple was clean. If the Auger scan showed only the presence of substrate atom s, the surface was then checked by Low Energy Electron Diffraction(LEED). The appearance of a sharp LEED p attern , in conjunction w ith a clean Auger spectrum , verifies th a t the sam ple is crystalline and th a t there are few, if any, foreign atom s on th e surface. It is “clean” in other words. Large and fuzzy LEED spots are a good indication th a t foreign atoms are disrupting th e periodicity of th e surface. This can occur even when the Auger scan shows no sign of foreign peaks due to th e lim ited sensitivity of the Auger to some elements. The cycles of bom bardm ent for 20 m inute periods and annealing of the sample were continued until the sam ple was determ ined to be clean. The cleaning process could take several days or a week depending on how dirty the sam ple was when cleaning began. This is because the cycles of cleaning and heating increase th e UHV background pressure which can take a day to recover to norm al levels. D uring an experim ent, the sample is cooled using liquid nitrogen to ~ 95 K. A t th a t tim e the CiLaBrfSigma Aldrich 99.5%) or CHsl{Sigm& Aldrich 99.5%) gas introduced into the cham ber. M ethyl bromide is a gas in a pressurized lecture b o ttle attached by a teflon line to the gas rack. M ethyl iodide is a liquid which is held in glass container attached to the gas rack. The vapour pressure of the liquid at room tem p eratu re is used to introduce the CHsl into the glass handling rack. Degassing of th e glass rack system is done by pum ping the m ethyl iodide or m ethyl bromide gas away several tim es ju st before an experim ent. The unit used to m easure the am ount of gas let into the cham ber is th e Langm uir (1 L = 10~® torr • seconds) and is referred to as th e “dose” . Molecules from the gas phase condense onto the sam ple in the UHV and in th e ideal case form solid layers. Each layer of condensed molecule on the sample is referred to as a M onolayer(ML). Because th e num ber of molecules th a t actually stick to the sample is small com pared to th e num ber th a t is let into the cham ber, the num ber of monolayers formed is independent of the surface area of the sam ple. The num ber of layers formed as a function of dose is governed by the sticking coefficient of the particular molecule injected into UHV and its ionization efficiency in th e ion gauge. Ion gauges m easuring pressure are typically standardized for nitrogen m easurem ent. The gas doses in this thesis are all based on the uncorrected ion gauge readings. 25 (04) (03) (02) (01) (00) (01) (02) (03) (04) Redprocal lattice rods —4 g ,—3 g ,—2g, —g, 0 ga 2g, 3g, 4g, Figure 2.4: Ewald sphere for electrons at normal incidence to a crystal surface. (Diagram from Zangwill, [43, pg34];Kahn [20].) 2.3.2 Low Energy Electron Diffraction(LEED) Low Energy Electron D ifïraction(LEED) is a diffraction technique which can be used to study crystal surfaces. Electron diffraction of surfaces is sim ilar to X-ray diffraction of three-dim ensional crystals. Surfaces however are periodic in only th e two-dimensions of the planar surface and the 3-D diffraction conditions relax to allow diffraction in two dimensions. A principle reason for th e relaxation of th e diffraction condition is th e short m ean free p ath of low energy electrons in solids. A diffraction experim ent requires th a t the probe m ust have a wavelength A th a t is sm aller th an interatom ic spacing. T he interatom ic spacing between molecules on a surface are typically ~ 1A° and so electrons used as th e probe m ust have a m inim um energy E = (h/A )^/2m ) ~ 150 eV in order to be sensitive to the crystal structure [43]. As it tu rn s out the m inim um on th e universal curve for the m ean free path of electrons in solids is near this energy (see figure 1.9). Electrons at th e m inim um on this curve have a m ean-free-path on the order of several lattice spacings. Electrons of this energy th a t penetrate deeper th an several lattice spacing will likely be inelastically scattered and therefore diffraction in 2-dimensions becomes possible. Elastically backscattered electrons from crystal surfaces in the range 20-500 eV form Fraunhofer diffraction patterns [43]. The LEED p attern is a reciprocal image of the surface. The distance betw een bright spots in a LEED image is inversely proportional th e lattice spacing in th e real crystalline surface. In reciprocal space, electrons are typically described by th e wave vector k. T he de Broglie relation A = ^ (2.1) can be used to convert k, where k is the m agnitude of the wavevector k, to the wavelength A. 26 retarding grids Observer o Sample Filim ent Window Ultra High Vacuum Phosphorous Screen Figure 2.5: Diagram of Rear-view LEED Optics. Electron gun is at the center of the phosphorous screen. Electrons are backscattered off the sample towards the LEED optics. The surface net which defines the crystal surface lattice was previously defined as T = m a g -f nbg, m and n are integers. Ug and bg are the prim itive vectors of th e surface net. The two-dimensional Lau equations are satisfied when (k ; — k f ) ■&S = 27rm and ( k ; — k f ) • bg = 27rn (2.2) where kj and k f are the incident and outgoing wave vectors for th e electron, and m and n are integers. T he surface net in reciprocal space is described by gg = hAg + fcBg, where h and k are integers. Ag and Bg are the prim itive vectors of th e reciprocal space net. The Lau equations are graphically illustrated by the Ewald sphere of figure 2.4. T he radius of the sphere is given by the m agnitude of the incident wavevector k ; and a reciprocal rod passes through every point defined by the reciprocal net gg. T he two-dim ensional diffraction condition is m et where the sphere intersects the reciprocal rod. The LEED optics are shown in figure 2.5. The electron gun is at the center of the phosphorous screen,where th e (00) spot would appear. Low energy electrons of th e order of 100 eV are directed towards a single crystal sam ple surface in a narrow beam . The surface atom s scatter electrons inelastically and elastically. Some are backscattered tow ards th e phosphorous screen. T he retarding grids in front of th e screen have a potential which is sufficient to repel the lower energy inelastically scattered electrons. The higher energy elastically scattered electrons are unaffected by th e grids and they pass through to th e phosphorous screen. Bright spots appear where th e diffraction condition is m et. T he rest of screen is dark. M ultiple scattering processes affect the intensity of the various spots but it does not affect the num ber or the placem ent of spots. 27 peak 200 Electron energy(eV) (a) lO O O Ep (b) Figure 2.6: (a)The three-electron Auger process. An inner shell electron is knocked out of its orbital by a high energy bombarding electron (3k eV). An electron from an upper orbital fills the gap and at the same time passes energy to an adjacent electron, which is kicked out of its orbital as a result, (b) Backscattered N(E) distribution. Inset is the N'{E) distribution. (Diagrams from Zangwill [43, pg22,23]; (b) Park and den Boer [27]. ) 2.3.3 Auger Electron Spectroscopy (AES) Auger Electron Spectroscopy(AES) is a surface sensitive technique where core-level electrons in atom s are ionized and th e subsequent electron emission is studied. In this technique a surface is bom barded with relatively high energy electrons of 3 keV. The bom barding electrons cause elastic and inelastic scattering of electrons at the surface and in th e bulk. Electrons th a t are backscattered from th e surface are collected in a detector. The Auger process is only one of a num ber of processes which is occurring due to electron scattering. Fig 2.6(a) shows th e steps in an Auger process. This particular one is referred to as a KLL Auger process because a K orbital and two L orbitals are involved. The Auger process is a three electron process. T he first electron is scattered out of an inner atom ic orbital, creating a “core hole” . A second electron decays from a higher energy orbital to fill th e core hole in the low energy orbital by transferring energy to a th ird electron in th e same or an adjacent orbital. The transferred energy is equal to the energy difference between the two orbitals and it scatters th e third electron out of its orbital. Auger electrons produced in th e first few layers of a surface will have a much sm aller probability for scattering th an those produced deeper in the bulk. Auger electrons th a t inelastically scatter contribute to th e background. Therefore th e Auger process is highly surface sensitive. The m easured Auger signal is derived from those atom s in th e first few layers of the surface. Typically the energies of th e Auger electrons are on the order of hundreds of eVs whereas th e workfunction of a surface is on th e order of a few eV. Therefore Auger electrons have sufficient energy to escape the surface. Figure 2.6(b) shows a schem atic spectrum of the electron energy distribution N(E)^ of backscattered electrons collected from a T itan iu m ^N(E) = where is the # o f electrons with energy E, e is the electron charge and t is time. 28 V)+ AV sini cot) retarding grids Auger Controller S a m p le Fihm ent E b = 3KeV Lock-in Amplifier Figure 2.7: The Auger circuit for measuring N'{E). E g = 3 keV is the energy of the bombarding electrons. A V % lOV is the peak-to-peak voltage of a small sinusoidal voltage. Vb ramps from » 90 volts to 1000 volts for copper. The phase u = 27t/ where f = 1.4 kHz is the frequency of the sinusoidal voltage. surface as a function of th e energy. The energy Ep is th e energy of the bom barding electrons. The Auger electrons show up as small peaks on a large background of inelastically scattered electrons. The energies E where th e peaks appear are characteristic of the elem ent, in this case T itanium , th a t is being bom barded and so each atom ic species has a characteristic Auger spectrum . The inset in figure 2.6(b) shows a spectrum of the derivative d N (E )/d E of the electron energy distribution N (E). The Auger peaks are enhanced in this spectrum and are easy to to recognize. For this reason Auger spectrum s are typically done w ith the derivative dN (E )/dE . Auger spectra are done w ith a circuit th a t uses the LEED optics as a retarding field analyzer (REA) (see figure 2.7). The electron gun in the center of the LEED optics bom bards the surface w ith E b = 3 keV electrons. The backscattered electrons are collected as a current I on the phosphorous screen. The current at th e collector is affected by the retarding potential on the retarding grids. If the potential on th e grids is V o, th an all electrons w ith less energy than E q = eVb will be repelled away from collector. In th a t case the current on the detector can be expressed as an integral [41]. rEp /= / AT(E)dE (2.3) J Eo Electrons arriving at any tim e t on th e detector have energy from Eq to the bom barding energy Ep. If the retarding potential Vb on th e grids is m odulated w ith a sinusoidal potential A R sm (cjf), then at any tim e t the m inim um energy of electrons arriving at the detector is E = e(Vb + AVsin{ujt)). T he current I arriving at th e detector can be expressed as a sum 29 î î î î ÎÎT Sample Figure 2.8: When dipolar molecule attach to a metal surface they may orientate in a particular direction. In this case dipoles are shown orientated in an ‘up’ direction. The negative end of the dipole is closer to the surface. of harmonics using a Taylor series expansion, th a t is rEp I = j N { E ) d E = Ao + Aisin(u)t) + A 2 sin { 2 u t) + ......... (2.4) where A q is the d.c. current {Ao = N { E ) d E ) and A i, A g ,... are the am plitudes of the harmonics [41]. The am plitude of the first harm onic A i can be shown to be Ai = AEÆ(Eo) + + ....... O and th e am plitude of the second harm onic Ag can be shown Ag = ^ T V '( E o ) + 4 4o T .......... (2.5) (2.6) In practice we do not m easure N(E) bu t rath er its derivative N '{ E ). The lock-in am plifier is a phase sensitive detector and is able to lock-in on th e 2w phase. In figure 2.7 th e LockIn amplifier is used to measure the am plitude Ag of the second harm onic sin{2ut) when th e retarding voltage on the detector is m odulated by a sinusoidal potential of frequency f. T h at is the retarding voltage is V = Vo -|- A V s i n { u t ) , where A V ~ 1 0 V and w = 2 tt x f (f = 1.4kHz). The Auger controller is used to ram p Vq from 90 volts to 1000 volts, which is sufficient to get all th e Auger peaks in an Auger spectrum of crystalline copper. Since A E = e A V is a small constant num ber significantly sm aller than unity, the lower order term s of Ag dom inate and N ' [ E q) o c Ag 2.3.4 (2.7) Retarding Potential Spectroscopy R etarding potential spectroscopy is a surface technique for m easuring the changes in the w orkfunction of a surface as molecules are dosed onto it. The w orkfunction changes are significant because they are an indication of surface processes. D ipolar molecules on a m etal surface, for exam ple, m ay have a predom inant orientation which can increase or decrease the workfunction. The change in the workfunction will likely indicate th e predom inant dipole 30 I Ô0 Fermi Metal Sample Anode — E F erm i T Tungsten Filament Cathode Figure 2.9: The ideal representation for measuring workfunctions. Since the chemical potential does not differ a great deal from the Fermi potential in metals even at the temperatures T ~ 2800 K that the tungsten filament is heated to, I will follow the usual practice of referring to the maximum potential energy as the Fermi energy. Vo is variable in order to maintain a constant <5$ space potential difference between the two metals. changes with the addition of molecules to the metal surface. The metal sample receives electrons and is called the anode as a result. The tungsten filament of the electron gun produces electrons and is called the cathode. orientation on the surface. Figure 2.8 shows a layer of dipolar molecules orientated on the surface in an ‘u p ’ direction. The negative end of the molecule is close to th e surface and the dipole is pointed away from the surface. This orientation will decrease the workfunction because the electric dipoles are assisting the electrons away from the surface. Dipoles predom inantly oriented in th e ‘down’ direction, not shown, would increase th e workfunction. The basic principle for m easuring changes in the workfunction of the sam ple as it is dosed w ith molecules is as follows. Figure 2.9 shows th e ideal representation for m easuring the changes in the workfunctions at T = 0 Kelvin. Ideal in this case ju st m eans th a t the m axim um potential in the m etal is called th e Fermi potential. In actual fact this is only strictly true at T = 0 Kelvin. For T > 0 Kelvin th e m axim um p o ten tial is somewhat greater than the Fermi energy b u t not significantly so. In figure 2.9 the space potentials of th e m etal sample and the tungsten filam ent of th e electron gun are m aintained at a constant difference Since space is insulating, the space potential of a m etal in a vacuum is at $ above the m axim um potential energy of an electron on the surface. $ is th e workfunction potential. The relative position of the space potential of a m etal in a vacuum can be m anipulated by applying a DC potential to the m etal. Increasing th e potential on the m etal raises th e space potential. The Vq potential in figure 2.9 is variable and it is used to m aintain the constant potential difference between the space potential of th e tu n g sten cathode and th e space potential of the m etal surface anode. Vq is variable because th e m etal sam ples’ workfunction changes as molecules are dosed on to it. In order to m aintain th e constant difference 6$, Vo is varied by the exact am ount th a t th e workfunction changes by. T he changes in Vq are the retarding potential difference th a t we m easure in th e w orkfunction experim ents. 31 Energy(eV) Figure 2.10: The electrons produced from the electron gun have an approximately gaussian distribution. The peak in the distribution is labelled E b - Only electrons with energy greater than E q = e<5$ will contribute to the current on the sample. Electrons with less energy than E q will be deflected away from the sample. In the real system , a low energy beam {E b ~ SeV) of electrons is directed towards the m etal sample. T he electron gun th a t produces th e electrons is com prised of a hot tungsten filament and focussing elements. Electrons in the tungsten are therm ally excited above the vacuum potential. Electrons of energy E are produced in an approxim ately gaussian distribution about the energy E b , the nom inal energy of th e beam as in figure 2.10. N(E) is the electron energy distribution. In order for any electron to reach th e m etal surface it m ust have more kinetic energy th a n E q = e6$. All electrons w ith less energy th a n E q will be repulsed away from th e m etal. The current on th e sam ple can be described by an integral rEmax / = / ( 2 .8 ) J Eq where Emax is th e largest electron energy in th e beam . Figure 2.11 shows the current on the sam ple as a function of the retarding potential V q. W hen the retarding potential is sufficiently strong, the lower energy electrons in th e beam are deflected away from the sample and the current begins to decrease. T he definition of the current I on th e sample in equation 2.8 is rem iniscent of Auger theory where the current on th e LEEED /RFA collector due to backscattered electrons from the sam ple has a sim ilar integral. The retarding potential m easurem ents can also be done with th e use of harm onics by m odulating Vq w ith a sm all sinusoidal potential à . V s i n { u ï ) . This m odulates th e space potential and results in a altern atin g current on th e sample. In this case the current I can be expressed as a sum of harm onics I f J' Ee ,o + e A V sin(Ljt) N { E ) d E = Ao + A ism (w f) + A2sin{2iot). (2.9) where Aq is the DC current and Ai, A 2, ... are the am plitudes of th e harmonics. Using th e 32 < 0 3 u EB Retarding Voltage Vq (V) Figure 2.11: The current on the sample varies with the retarding potential Vq. When Vq is increased such that it begins to deflect the electron beam away from the sample, the current begins to drop off. 80 5040- s75 20- £ i *20 — 3 4 6 5 7 8 Retarding PDlenttal Vq (V) Figure 2.12: This data is proportional to the derivative of N(E), the distribution of electron energies. As the current on the sample decreases due to an increasing Vq there is a zero-crossing point on the N'{E) graph. It is marked with E b 33 Lock-in amplifier with an appropriate circuit (Figure 2.13)we can m easure the am plitudes A i , A 2..... Figure 2.13 shows the circuit used to measure th e retarding potential of the sam ple. T he sinusoidal voltage couples to the % variable voltage through a transformer. The 10 Kfl resister in the circuit does not affect the DC voltage on the sample, it is there to separate the inputs of the lock-in amplifier. The DC voltage on the sample is the retarding voltage Vq. We used the circuit to lock-in on the sin{2u>t) signal, which is proportional to N' {E ), the derivative of the N(E) electron distribution of figure 2.10. As it turns out it is easier to follow the peak in the N(E) distribution by using th e derivative N '{ E ) . In figure 2.12 N' {E ) is shown as a function of V q - The significance of this graph is th a t there is a zero crossing at Eb- We know th a t AT'(Eo) cx Ag (2.10) Therefore we vary Eq such th a t we m easure a zero am plitude on th e sin{2uit) harm onic, which is the zero crossing point. All we require at th a t point is to m ain tain the DC retarding potential Vq on the sample at this zero crossing point as the surface is dosed and th e workfunction changes. By tracking the zero-crossing feature we can m easure th e changes in the workfunction. If Vq is the initial potential on th e sample to m ain tain th e zero crossing, then the retarding potential differences A V = V q — V q . V q are th e m easurem ents for the zero-crossing on th e dosed surface. 2.3.5 T e m p e ra tu re P ro g ra m m e d D e s o rp tio n (T P D ) T em perature program m ed desorption(T PD ) is used to exam ine th e tem p eratu re dependent desorption of atom s and molecules from a surface. T he T P D experim ent uses the sample heating circuit and the mass spectrom eter. A com puter program w ritten to do T PD spectra controls the heating circuit (figure 2.3) for warming th e sam ple and th e QMS for sam pling particles of mass m. The analogue mode of the mass spectrom eter produces a voltage signal where the size of the signal depends on intensity of m ass-selected particles. T PD m easurem ents start with a cooled (100 K) and dosed sam ple. The sam ple is warmed at a controlled ra te with respect to tim e using th e tem p eratu re control circuit. Molecules and atom s desorbing from th e sample enter the ionizer of th e mass spectrom eter. A spectrum of mass-selected intensity as a function of tem p eratu re is produced. Peaks arise in the T PD spectra due to th e strength of th e binding in various layers. For exam ple, the bond of the first layer of molecules to the surface is usually stronger th a n the bond of succeeding layers to under layer s. Since the bond strengths are different, different energies will be required to break th e bonds. If the first layer is m ore strongly bound than overlayers, as is usually the case, then the first layer will require a higher tem p eratu re for removal than the overlayers. If the tem peratures are sufficiently different th an the massselected intensity spectra will show distributions as a function of tem p eratu re. One of the uses of T P D spectra is for determ ining the dosage th a t corresponds to a monolayer, a single layer, on the surface. As gas is let into th e UHV cham ber the vacuum pum ps are pum ping it out and only a small fraction of gas let into th e cham ber actually strike and stick to the sample. The sticking coefficient relates th e fraction of molecules th a t stick to the sample after striking it. The sticking coefficient depends sensitively on such 34 Lock-In Amplifier Inputs T ransform er DC Ramp Voltage UHV C ham ber Electron Gun Sam ple Figure 2.13: Circuit design used to measure the workfunction changes of the sample as it is being dosed with molecules. The Lock-In amplifier measures the amplitude of the sin(2wt) harmonic. The amplitude is directly proportional to the derivative of the electron energy distribution N (e). 35 factors as tem perature, surface, and coverage which will vary depending on th e experim ent. Most of the gas let into the cham ber is pum ped away. Because only a small fraction of the molecules let into the cham ber actually strike and stick to the sam ple, the num ber of Langmuirs th a t corresponds to a monolayer is independent of the size of the sample. 36 2.4 Surface P h otolysis Experim ents: T im e-O f-F lights 2.4.1 In tro d u c tio n The Time-of-Flight (T O F) experim ental apparatus is composed of th e UHV cham ber, a crystalline sample, a quadrupole mass spectrom eter, a m ultichannel scaler, a pulsed n itro ­ gen laser, and a PowerM ac com puter. (See Fig 2.1.) In addition th ere is a provision for interfacing the com puter to various lab equipm ent. During th e T O F experim ent, th e dosed sample surface inside the UHV cham ber is illum inated w ith pulses of photons from th e laser. The laser light enters the cham ber through a UV grade fused silica window along a pathw ay th a t is at an angle of 45 degrees to th e m ass spectrom eter detector. T he sam ple can be rotated as described in the experim ental section. Therefore, T O F d a ta can be obtained w ith the sample surface norm al at various angles to th e QMS. Light th a t is incident on th e dosed sample surface can cause photodissociation and desorption processes on th e dosed surface. The laser is a pulsed nitrogen laser (M olectron UV-12) th a t delivers energy of 0.8 7 m J/ cm? per pulse. Each pulse has a 10ns (nanosecond) duration and th e pulses occur at a lOHz frequency. In a typical T O F experim ent hundreds or thousands of laser pulses are used. The laser beam enters the cham ber through a UV grade fused silica window at an angle of 45° to the QMS. T he QMS is operated in pulse counting m ode in th e T O F experim ents. Molecules and dissociation products entering th e QMS are ionized. Ions are m ass selected to go through the channeltron where they cause electron cascades th a t produce a single pulse signal for each ion. The pulse signals from ions in th e QMS are amplified by a fast pream plifier(EG & G VT120) and are processed by a m ultichannel scaler(M CS, EG & G Turbo MCS). T he MGS is a high speed counting device which records th e arrival tim es of th e pulse signals from the channeltron in sequential tim e bins. T he bins are typically a tim e interval of 2/xs; the first bin coincides w ith th e laser pulse tim e. For higher resolution T O F spectra, 1 jjis bins were used. The QMS detector is a ‘density’ detector since th e T O F signal is a distribution in space [45]. Therefore th e signal in a T O F spectra does not directly represent an energy distri­ bution of particles. The m easured counts in the T O F spectra differ from th e actu al T O F distribution because th e probability of a detecting any p article decreases as its speed in­ creases. Slower particles stay in th e ionizer longer th a n faster particles and therefore have a greater probability of being detected. The relationship betw een T O F spectra and T O F distributions can be m athem atically defined in the following way. T he T O F signal S results from a T O F distribution of particles entering the mass spectrom eter per tim e t at a p ar­ ticular radial distance r. A t any point r, th e T O F distribution of particles from a sample surface is given by the flux I(t). The flux I(t) is th e num ber of particles per area p er tim e; I(t) = vp where v is th e velocity distribution and p is th e num ber of particles p er volume. Since the probability of detecting particles is inversely proportional to their speed, or the tim e it takes to arrive at th e detector. S{t) oc p(f), and therefore I{t) oc This last equation tells us th a t m ultiplying th e T O F spectra by 1 / t converts it to a distribu tio n in tim e. The counts of th e fast particles are been enhanced in com parison to th e counts of the slower particles. The 1 / t factor m ust be taken into account when com paring th e intensities of fast and slow particles in a T O F spectrum T O F experim ents in this thesis are done as a function of sam ple angle to th e mass 37 1001' D irection D e te c to r [1 10] Direction Figure 2.14: Rotation of the sample through 0 about z-axis is the [001] azimuth. Rotation of the sample through $ about the x-axis is the [110 ] azimuth. spectrom eter detector (angular-dependency), as a function of dose on the sam ple (dosedependency), and as a function of to tal laser shots on th e sample (depletions). T he angulardependency experim ents were done w ith respect to th e two azim uths of th e sam ple as in figure 2.14. The coordinate system is fixed and th e detector is also fixed on the y-axis. In the diagram the sample can be ro tated through 0 or $ . The [110] direction of th e sample is on the x-axis and the [001] direction of th e sam ple is the z-axis. R otation of th e sample through $ about th e x-axis is the [110] azim uth. R otation of the sam ple through 0 about z-axis is the [001 ] azim uth. 2.4.2 Y ields as a fu n c tio n o f to ta l p h o to n s The yield as a function of total photons on th e surface system were done w ith th e idea of finding a cross-section for depletion in th e surface system . A cross-section for depletion m easurem ent is a m easure of the probability of m olecular dissociation as photons are directed onto the system. T he num ber of molecules N on the sample available for dissociation and the num ber of molecules th a t dissociate dN are related by d N = —Nd{aO) (2.11) where 0 is the num ber of photons per unit area on th e system and a represents th e probability for dissociation. We assume th a t cr th e cross-section is positive and independent of 0. a has units of area in order th a t the fraction d{a 0 ) is unitless. There are surface effects th a t make it it difficult to m easure cr. Surface reordering effects resulting from dissociation m ay make more molecules available for dissociation. In th a t case yield N m ay effectively increase w ith respect to th e initial yield No and the relationship in equation 2.11 may not be applicable. In addition in the real surface system cr is not generally independent of large 0. The accum ulation of dissociative by-products m ay affect local potentials with the end result th a t th e overall cross-section changes [44]. Dissociative by-products are generally more im p o rtan t for large 0 . By separating the variables in equation 2.11 and integrating we get the exponential decay relationship: 38 AT = ( 2 . 12 ) where N q is the Initial num ber of molecules on the surface and N is the num ber after photodepletion. By combining equations 2.11 and 2.12 we get dfV = (2.13) and then the natural logarithm of this equation is taken ln{dN) = ln{—Nocrd6 ) — cr0 (2.14) If reordering effects do not affect the yield significantly and if a is constant and positive, In(dN) should be linear with respect to 6 and represent a depletion of Wo. T h at is, a plot of In(dN) vs 0 should produce a line with slope a. In{—Nocrd0) in th a t case, is also a constant. Experim entally we do not directly m easure dN. The signal S we m easure in th e T O F spectra is a small fraction of dN, th a t is S= q dN. As long as q does not have a 0 dependency the linear relationship rem ains valid, ie ln{S) = constant — a0 (2.15) In the yield as a function of to tal photons experim ent, th e angle of the sample is fixed and a num ber of T O F spectra are done. The angle where th e sam ple is fixed at is usually th e angle where th e yields were greatest. Each of th e T O F spectra is done w ith th e same num ber of laser pulses and each is num bered in a consecutive fashion. A fter the d a ta has been taken the counts S in the T O F spectra are sum m ed and p lo tted on a sem i-logarithm ic scale as a function of total photons on th e sample. T h at is loge{S) vs 0 is plotted. Below is an exam ple calculation of how a laser pulse is converted to num bers of photon. Before beginning an experim ent, th e laser power was m easured to be 5.4 m J /s w ith the pulse frequency at lOHz. The surface area of the detector is 0.6cm^. The to tal energy per pulse IS E p u l s e ' Epulse — 0.54m J 0 .9 m J ; 2 ~ 7 2 pulse • O.Gcm^ pulse • cm^ . (2.16) The wavelength A = 337.1 nm and th e energy Ephoton of each photon is he Ephoton = — = 5.9 X J I photon = 3.67 eVjphoton (2.17) A where h is Planck’s constant and c is the speed of light. Therefore th e to tal photons num ber of photons in each pulse is given by Epulse _ 9.9m J I {pulse ■c m ? ) Ephoton 5.9 X J / photon 1.5 x \9^^photon pulse ■ Therefore if each spectrum is done w ith 500 laser pulses, for th e first spectrum 0 = 500 x 1.5 X 10^^ photons, for th e second spectrum 0 = 1000 x 1.5 x 10^^ photons, and so on. If the slope of the line in the /ope(S) vs 0 is reasonably linear and negative, then we m ay with confidence interpret the slope of the line to be the cross-section a fordepletion. 39 The power intensity of the laser pulse is found by dividing th e energy in a single pulse by th e tim e for a single pulse to occur. The power intensity is 0.9 X 1 0 -V ...................... 1 I— pulse ■cm? 0.1 pulse ■cm? ( 2 . X9 1 This value is significantly less than th e 10® MW jc m ^ [44] required to initial m ulti-photon effects. The energy transfer processes are single-photon processes. 2.4.3 Y ields as a fu n c tio n o f d ose The yield as a function of dose experim ent m easures the yields in T O F spectra as the dose is varied on a prepared surface. If reordering effects as th e experim ent proceeds do not make more molecules available for dissociation and if th e cross-section is a is constant and positive then the yields are governed by equation 2.15 ie ln{S) = constant — crd ( 2 .20 ) T O P spectra done as a function of dose are each done w ith th e same num ber of pulses, therefore 6 is the same and constant for all th e coverages. Therefore the signal S = qdN in the T O F spectra has three dependencies: the fraction q th a t is collected to create the signal S, the num ber of molecules No th a t are in dissociative sites, and the cross-section for dissociation. Interpreting the counts as a function of dose should be done w ith respect to those three variables. In the yields vs dose experim ent th e angle of th e sample is fixed, usually at an angle where the density of desorbing particles is greatest. There are two m ethods of doing yields vs dose spectra. T he first m ethod is to do each T O F spectrum w ith a new dose on the sam ple and w ith a large num ber of laser pulses(~ 2000 ). T he first T O F spectrum is done on th e smallest dose because sequential dosing and flashing to high tem peratures increases the residual pressure in the UHV cham ber. The increase in pressure results in an increase in the background counts in the T O F spectra. Flashing cleans th e dosed molecules off the sample and recooling to < 100A' readies it for th e next dose. By keeping the sequential dose larger th an the previous dose, th e increased background counts have less significance as the signal in the T O F spectra gets larger. Flashing to high tem peratures has an additional negative effect however. The C u(llO ) sam ple is of high pu rity b u t is contam inated by low concentrations of foreign atom s. Sulfur is one of th e more common contam inants. H eating the sample causes contam inants in the bulk to transfer to th e surface. The second m ethod avoids heating of the sam ple during th e experim ent by dosing on top of previous doses. In this m ethod T O F spectra are taken w ith a relatively sm all num ber of laser pulses( % 500) in order th a t th e dose be m inim ally depleted. The sm allest dose to be studied is put on the sample and a T O F spectrum is done. Then the sam ple is dosed w ith additional molecules. In this case th e to tal dose is considered to be th e first dose plus the second dose. A nother T O F spectrum is taken w ith the same num ber of laser pulses. The process is repeated with the to tal dose being the sum of all doses on the sample. 40 O ther advantages of this m ethod over th e first m ethod is th a t it can be done faster and the background counts in the T O F spectra are kept relatively low. A disadvantage is th a t the statistical error in the counts as a function of dose is larger th an in th e first m ethod due to th e lower num ber of counts in each T O F spectrum . 2.4.4 A n g u lar D istrib u tio n s Angular distributions were done by taking T O F spectra of th e dosed sample at various angles. The angles were done w ith respect to th e two azim uths [1Ï0] and [001] of a 3-D coordinate system. All angles were m easured as with respect to th e surface norm al. There are two m ethods for investigating an angle dependence. T he first m ethod is easiest to describe. In this m ethod the sam ple is dosed, an angle is chosen and th e T O F spectrum is taken. The num ber of counts is m easured in th e resulting peaks. T hen th e sam ple is cleaned by flashing to a high tem perature, redosed and a new m easurem ent is taken at a different angle. This process is repeated until enough d ata is accum ulated. As stated in the Yields vs Dose section, heating can bring contam inants to the surface from th e bulk which can affect the photodynam ics. The second m ethod avoids flashing between angular m easurem ents. T he second m ethod uses a single dose for all the angles m easured. In this m ethod each T O F spectrum was collected w ith 1000 laser pulses bu t every th ird T O F was done at a norm alizing angle , for exam ple 0°. The T O F spectra done at th e norm alizing angle were used to norm alize the depletions of the signal. A lthough depletion norm ally follows an exponential curve, for a small num ber of pulses this curve is fairly linear as com pared to th e curve for a large num ber of pulses. Therefore counts in the T O F spectra done at angles between the norm alizing m easurem ents could be increased by some m ultiplication factor based on extrapolation of a linear depletion between m easurem ents. We found th a t this was a m ore accurate m ethod for doing angular experim ents. 41 C hapter 3 M ethyl H alides on Copper: Standardized Surface E xperim ents 3.1 Introduction The object of this research is to work on model systems to advance th e understanding of the surface photochem istry of adsorbed molecules. There is a general them e in recent ( less then 15 years) surface research papers th a t small molecules have been extensively studied in the gas-phase both w ith theoretical means and w ith experim ental m eans [25, 13, 30] but the research on the photochem istry of small molecules on surfaces is not as well developed [19, 30]. This is a m otivation for this thesis work. My research is on th e m ethyl halide group of molecules on crystalline m etal surfaces. As the nam e implies m ethyl halide molecules are a com bination of a m ethyl and a halide. The form ula is C H 3 X where the X can be a F ,C L ,B r , I. My research d ata is drawn from research done by my supervisor. Dr Erik Jensen, and m yself in th e sum m er of 1999. This research d a ta has also been used to publish two papers by D r Jensen, w ith myself as a coauthor [6 , 5]. T he research d a ta used in this thesis is prim arily of m ethyl iodide on crystalline copper cut in th e (110) orientation, although I have included some d a ta for m ethyl bromide on the same surface. W hy are m etal surfaces used for this research? M etal surfaces are interesting because of the strong quenching and strong photoelectric effects th a t occur on m etal surfaces. W hen light of sufficient energy is directed onto a m etal surface, excited electrons are produced by the photoelectric effect. By using light w ith energy less th an th e w orkfunction of th e surface at an intensity below th a t needed to initiate m ulti-photon adsorption, excited electrons can be produced at the surface w ith m axim um energies less th an th e vacuum energy of the surface. In single-photon adsorption, th e m axim um energy of th e hot photoelectrons produced on the surface will always be the energy of a single photon. This is a difficult task if the surface is bom barded with an external source of electrons. E xternal electrons on approaching a surface are influenced by the attractiv e im age potential of the surface and gain energy as they approach, by th e tim e these electrons reach the surface they will have gained energy on the order of a few eV. There is a great deal of interest in electrons with energy with very low energies because in the gas-phase m any molecules have large dissociative resonances for these electrons. 42 Of all molecules th a t are available for study, why study m ethyl halides on surfaces? One answer is th a t if the excited electrons are to cause dissociation in absorbed molecules, th e absorbed molecules m ust dissociate faster th an the quenching rate of th e m etal. Close to the m etal surface, quenching is on the order of femtoseconds. M ethyl halides also have a dissociation rate, at least in the gas-phase, th a t is on the order of femtoseconds. Therefore dissociation of m ethyl halides on m etal surfaces is often able to com pete w ith quenching. Research focusses on the first several layers of adsorbate molecules on clean surfaces. This is because the photochem istry of the adsorbate is usually only perturbed close to the surface. M ethyl halides are described as rod-like shaped w ith a dipole m om ent [8 ]. Because of this rod-like characteristic and th e dipole, they typically have a predom inant orientation in the first layer on a m etal surface. T he molecules dipole typically orientates w ith the halogen close to the m etal surface and at an angle w ith respect to th e surface. This orientation can be affected by the coverage. C H 3 I, CffgBr, and C H 3 CI on P t ( l l l ) all lie down on the surface for low coverages b u t change their orientation towards th e surface norm al as th e first layer is filled [7]. 3.2 Auger Spectroscopy of C u (llO ) Auger spectroscopy was prim arily used to determ ine sam ple cleanliness. This was done by taking an Auger spectrum of the sam ple and com paring th e scan to an accepted A uger scan of clean copper. Figure 3.1 is an Auger scan of clean C u(llO ). The position of the peaks in an Auger scan are characteristic of the atom s in th e first several layers on the surface. An Auger scan is sensitive only to the first several layers of atom s. If th e only peaks showing in the Auger scan are those which are characteristic of copper, th en th e sam ple was determ ined to be clean within the sensitivity of the Auger. If th e sam ple was not clean, th e Auger would show peaks at energies not consistent w ith copper and these could be used to determ ine the necessary sample cleaning process. T he elem ent contam inating th e sample could be identified by comparing the energy of the additional peaks to standardized Auger scans of elem ents. Knowing which elem ent was on th e sample determ ined th e cleaning process. C ertain elem ents like iodine or oxygen could be removed from the sam ple by flashing to high tem peratures but other elem ents, like sulfur, required ion bom bardm ent for removal. Auger scans were also used to confirm th a t iodine was left on th e copper surface after CH 3 Ï was therm ally dissociated on th e surface. An iodided copper surface was form ed by dosing 20 L of C H 3 I onto a clean C u(llO ) surface. T he surface was then warmed to 525 K, below the tem perature needed to remove iodine from copper but above th e tem p eratu re to remove m ethyls and their byproducts [24]. T he iodided surface is referred to as iodided C u(llO ) or as Cu(110)-1. Figure 3.2 is an Auger scan of a Cu(110)-1 surface, done after the surface has cooled somewhat. W ithin th e sensitivity of the Auger spectrom eter, th e surface shows only iodine and copper peaks. 43 20- 2— -4 - 200 6 00 400 1000 8 00 SiîclKS». lr>icfgv|eV) Figure 3.1: AES spectrum for Cu(llO). The scan is done by measuring the amplitude of the second harmonic {sin2u:t). The signal is proportion to 7V'(E), the Auger electron signal. Sample cleanliness is confirmed by the appearance of Auger peaks consistent with copper only. Cu 0- Cu Cu -2 “ -6x10 200 400 600 Elec[ronEnergy(eV) 800 100 0 Figure 3.2: AES spectrum for Cu(110)-I. The scan is also done by measuring the amplitude of the second harmonic [sin2u!t). The iodine peaks show up near 510 eV. 44 Figure 3.3: Picture of the LEED screen with a clean Cu(llO) surface. The electron beam energy is 105 eV. The reciprocal net is rectangular and is characteristic of well-ordered crystalline copper in a ( 110 ) cut. 3.3 LEED patterns o f C u(110),C u(110)-I LEED was prim arily used in conjunction w ith Auger spectroscopy to confirm th a t th e sample was clean and ordered. If the Auger showed only copper peaks, th e possibility still existed th a t the sample surface was disordered due to various contam inants or poor annealing. Fuzzy spots or an irregular pattern in th e LEED indicate some surface disorder. A sharp and characteristic LEED p a tte rn for a surface determ ined to be clean by an Auger spectrum is additional confirm ation th a t the surface is really clean. Figure 3.3 shows a picture taken of the phosphor screen in a LEED experim ent w ith clean C u(llO ). The LEED of clean C u(llO ) shows the surface net characteristic of well-ordered crystalline copper in a (110) cut. T he LEED was also used with Auger to confirm th a t iodine was left on the surface after a 20 L dose of C H 3 I on a cold C u(llO ) surface was warm ed to 525 K. This tem p eratu re is high enough to desorb m ethyl fragm ents which adsorb to th e surface but not high enough to remove the iodine [24]. Figure 3.4 is a picture taken of the LEED w ith a Cu(110)-I sample. The graphics in figure 3.5 illu strate th e spots as dots,squares and X ’s. T he dots are in the position of th e copper spots as on the clean surface, although in this case the intensity of these spots have contributions from iodine and copper atom s. T he squares and X ’s are spots th a t are not apparent on the clean surface. The square spots are bright and the X ’s are dim m er. The dots and squares form a reciprocal surface net th a t could come from a real space c(2x2) p attern. Figure 3.6 illustrates a c(2 x 2 ) overlayer surface net. A c(2x2) structure is a rectangle with an atom at the center. The sides are 2 tim es the length of the real space rectangular surface net of, in this case, th e copper atoms on th e ( 110 ) surface. However in figure 3.4 additional dim m er spots (the X ’s in figure 3.5) are observed in the[lÏ0] direction. These spots indicate th a t there is likely some additional longer range order in this direction. 45 A possible reason for the longer range ordering is th a t there is a lattice m ism atch of the larger iodine atom s on th e smaller copper atom s [6 ]. T he sizes of iodine and of copper atom s were estim ated from chemical bond lengths. An I 2 molecule has bond length of 2.66A°[26]. The diam eter of each iodine atom is therefore ~ 2 .66 A®. T he copper atom diam eter was determ ined from the FCC unit cell size. The crystalline copper FCC unit cell has each side of 3.61A° [2]. Therefore the diam eter of a Cu atom is give by |(3.61).sm (45°)A ° = 1.27A°. Therefore the diam eter of the I atom is ~ 2 x as large as th e diam eter of the Cu atom . Using these atom diam eters an a tte m p t was m ade to produce a c(2 x 2 ) overlayer stru ctu re on a C u(llO ) surface. It was found th a t iodine was too large to fit into such a stru ctu re. A c(2x2) LEED pattern from an overlayer on C u(llO ) can only be produced if each overlayer atom is centered between two copper atom s on th e substrate in a c(2 x 2 ) surface net. One possibility has the overlayer in two-fold sites on th e ridges of th e ( 110 ) rows, and another possibility has the overlayer in four-fold sites in th e troughs of the rows. Figure 3.7 (a) shows a c( 2 x 2 ) overlayer structure, not iodine, in four-fold sites on a C u(llO ) surface. In this case the hypothetical atom s used in th e overlayer stru c tu re are sm aller in size th an the iodine atom s, bu t are the largest th a t could be fit into th e c( 2 x 2 ) overlayer. Figure 3.7 (b) shows the relative sizes of the iodine, copper, and th e hypothetical overlayer atoms. If the real space surface net of th e iodine overlayer on C u(llO ) is generally c( 2 x 2 ), and if the size of iodine is twice as large as copper, then it is likely th a t significant changes m ust take place in th e positioning of th e copper atom s. T h a t is, the surface m ust reorder to accom m odate the iodine atom s. The exact overlayer stru ctu re and the reordering of the surface is at this point is unknown. There is no inform ation in th e literatu re at this tim e on the exact overlayer structu re of iodine on C u(llO ). 46 Figure 3.4; Picture of the Screen in a LEED experiment with Cu(110)-I at an electron beam energy of 101 eV. The reciprocal surface net is similar to a c(2x2) pattern but additional satellite spots are visible in the [1Ï0] direction. The additional spots likely indicate longer range ordering. # X#X OX # X D X# # OX * Xo X #x # Figure 1 Figure 3.5: This schematic illustrates the LEED patterns is dots, squares and X’s. The dots are the position of the (1x1) copper spots. The iodine also contributes to the dots. The squares are iodine spots. The X’s are the additional satellite spots. Diagram from Johnson et al. [6]. 47 -2b- C(2X2) Figure 3.6: A c(2x2) real space overlayer structure is a rectangle with an atom in the center. The length of the sides are 2 times the length of the real space rectangular surface net of the copper atoms on the ( 110 ) surface. f • b- ► A 2b [001] I 2a ^ I ▼ ÜJJÏ (b) Figure 3.7: (a) shows a c(2x2) overlayer structure, not iodine, in four-fold sites on a Cu(llO) surface. In this case the hypothetical atoms used in the overlayer structure are smaller in size than the iodine atoms, but are the largest that could be fit into the c( 2 x 2 ) overlayer, (b) shows the relative sizes of the iodine, copper, and the hypothetical overlayer atoms. 48 1.5 - > 20.0L 14.BL0.5 ■9.9L-9.0L-7.0L- tS.OL; 0.0 120 140 160 ISO 200 220 Sample Temperature(K) Figure 3.8: TPD spectrum for CH3l/C u ( 110 ). The first layer is complete at a dose of 9.0 L and desorbs from the surface at 140 K. 3 .4 Tem perature Program m ed D esorption (T P D ) D uring a TPD experim ent the prepared surface is warmed at a controlled rate. The mass spectrom eter samples for a species of ion produced in the QMS ionizer from particles or molecules desorbing from the surface. A graph is produced o f th e ion intensity as a function of tem perature. We m easured th e principle fragm ent C H ^ produced in the ionizer by fragm entation of the CH^l and C ifsB r molecules. One problem is th a t various hydrocarbon species as well as m ethyl halides produce C H ^ upon contact w ith th e ionizer. Therefore when a C signal is found it m ust be determ ined if molecules are desorbing from the sam ple or if fragments from dissociation on the surface are entering th e ionizer. 3 .4 .1 C H 3 l/C u (1 1 0 ) Tem perature program m ed desorption spectra for doses of 5.0 L to 20.0 L of CH 3/ on C u(llO ) are shown in figure 3.8. In these spectra the principle fragm ent CH 3 is sam pled in th e mass spectrom eter. At low coverage, less th an 3 L, a single peak was evident at 160 K. W hen th e dose is increased past 3 L there are two desorption peaks, one of which is centered at 160 K and the additional second peak is centered at 140 K. The 160 K peak shows up as a small step in th e spectrum s of figure 3.8. The intensity or size of th e second peak at 140 K increases as the dose is increased to 9.0 L , where th e peak size saturates. For doses larger than 9.0 L there are three peaks, two of which are centered at 140 K and 160 K. The additional third peak is centered at 137 K. The intensity of this peak increases w ith the dose on the sample and we saw no evidence of saturation of this peak. All th e d ata is assigned to m olecular desorption of CH 3/ from the surface rath er th a n pro­ 49 duction of CH3 from dissociation of C H 3 / on th e surface based on th e following. We looked for hydrocarbon production, such as m e th an e(C % ) , ethane(C 2ff 6), and ethylene(C2iÏ4) , from the CH3 //C u ( 1 1 0 ) system in separate TPD experiments but none was found for temperatures below 300 K. This is consistent w ith findings from other works [9, 24]. Prod­ ucts from thermal dissociation of m ethyl iodide have only been shown to be dissociated from copper at tem peratures in excess of 300 K [9, 24]. Chiang et al. adsorbed m ethyl iodide on C u(llO ) at 110 K and then studied th e m ethane, ethylene, and propylene(C 3 ife) production from the surface in TPD s. They found these products are desorbed from the surface at tem peratures of 350-500 K, although m ethyl iodide dissociation occurs at much lower tem peratures. They also reported th a t they saw molecular desorption at 135 K for exposures above 3.0 L of CH 3I [9]. Lin et al. did a sim ilar study of m ethyl iodide adsorbed on C u ( lll) . They noted th a t they saw two m olecular desorption peaks at 135 K and at 165 K for exposures above 3 L of CH 3I [24]. They estim ate m ethyl groups are form ed on the surface by 200 K b u t hydrocarbon products(m ethane, ethylene, and propylene) desorb at 300 K. Based on our experim ents and the previous work, th e CH 3 m easured in the mass spec­ trom eter for figure 3.8 m ust come from m olecular desorption. As already outlined above, these TPD spectra show dose dependent m olecular desorption peaks at 160 K,140 K, and 137 K. Saturation of th e peak at 160 K has tw o possibilities. S aturation of this peak may indicate completion of the first layer. A lternatively this peak m ay indicate satu ratio n of stronger bonding sites in the first layer. One such possibility is step and defect sites [6 ]. Molecules attached to step or defect sites have more contact w ith th e surface , therefore, the bonds at defect sites are expected to be stronger and require more energy to break. Previous work [44] on A g ( lll) has shown evidence of adsorption of m ethyl halides at defect sites in the first layer. Zhou et al. found th a t increasing the defects on A g ( lll ) , by sputtering the surface w ithout subsequent annealing before adsorption, enhanced th e TPD desorption of CH 3.Br and CH 3C / in the first monolayer at higher tem p eratu res. Zhou et al. did not report a similar finding for CH 3/ on A g ( lll) bu t they reported th a t CH 3/ adsorbed more strongly to A g ( lll) in the first layer than th e monolayers of CH 3B r and CH 3C / to A g ( lll) . The strong chem isorption of CH 3 / in th e first layer through th e I atom to A g ( lll) m ay m ask adsorption at defect sites. C hem isorption is indicated by th e much higher tem p eratu re re­ quired for desorption of the first layer; CH 3 / on A g ( lll) desorbs and dissociates from the complete first layer near 210 K [44]. Tim e of flight d ata and retarding potential spectroscopy of CH 3/ / C u ( 110 ) will show th a t the peak at 160 K is consistent with attach m en t at stronger bonding sites in th e first layer and saturation of th e 140 K peak at 9.0 L indicates com pletion of the first monolayer. Therefore the peak at 137 K on CH 3l/C u ( 110 ) is assigned to desorption of th e second and any additional layers. It is referred to as th e m ultilayer peak. 3.4.2 C H 3/ / C u ( 1 1 0 )-I Tem perature program m ed desorption spectra from C iÏ 3l/ C u ( 110 )-I are shown in Fig 3.9. For low coverage of 5.0 L on th e sample, th e desorption peaks at 144 K. As th e coverage increases, the center of this peak shifts to slightly higher tem p eratu res. At 10.0 L the center of the peak is at 149 K. S aturation of this peak is taken as com pleted at about 9.5 L since 50 TTDCH3I/Cu(llD)-I 2.0 g 1.5 Im -^Q.DL—= -15.5L-----12.0L---- : - 10.01 0.0 : —SUL—= 0.5 — 8.01-----------— 7.01_____ — 8.01 : — 5.01----- : 120 140 160 ISO 200 M11111 m 220 Sample Tem|3a-ture(K) Figure 3.9; TPD spectra of CH^l/C\x{llQ)-\. The first layer is complete at a dose of 9.5 L and desorbs at 149 K th e second peak can be seen emerging at 10.0 L bu t is not evident at 9.0 L. The second peak is centered at 137 K for 12.0 L but as coverage is increased the peak center shifts to higher tem peratures. This peak size increases w ith coverage and is not observed to be satu rated w ithin the m easured coverages. The signal in figure 3.9 is assigned to m olecular desorption for th e same reasons th a t the T P D spectra of CHsl on clean C u(llO ) in figure 3.8 were determ ined to arise from m olecular desorption. The peaks in both figures 3.9 and 3.8 show up at tem peratures below the observed tem peratures for the production of m ethyl fragm ents from copper in other research [24, 9]. We interpret th e d a ta for C // 3l/C u ( 110 )-I in figure 3.9 as follows: th e high tem p eratu re peak corresponds to the desorption from th e monolayer and th e low tem p era­ tu re peak corresponds to the m ultilayer. Saturation of th e first peak for C i/ 3l/C u ( 110 )-I at 9.5 L is taken as completion of the first layer on th e surface. The second peak results from th e second and m ultilayers on top of the first monolayer. R etarding potential spectroscopy and time-of-flight m easurem ents, which will be discussed later, will support this in terp reta­ tion. T he shift in the center of the peak to higher tem peratures as coverage is increased is indicative th a t the binding energy of molecules to th e surface is increasing as the layer is filled. 51 4 — 3 - •1 8 .3 L --1 4 .3 L — - 1 0 .2 L — 9.2L 110 120 130 140 T em perature(K ) 150 160 Figure 3.10; TPD spectrum for C H 3B r/C u(llG ) 3.4.3 C ^ 3B r / C u ( 1 1 0 ) It will be useful to discuss the T P D spectra of C H ^ B t on C u(llO ). C ifaB r was not as extensively studied as C H 3 I was, however, th e surface dynam ics of CH^Bx on th e C u(llG ) surface are later used for analysis of th e C if 3 l/C u ( 110 ) and C iJ 3 l/C u ( 110 )-I system s. Tem­ perature program m ed desorption spectra from CFf3B r/C u ( 110 ) are shown in Fig 3.10. For low coverage there is a single peak at 140 K. W hen th e coverage increases to 5.0 L a second peak appears at 128 K. This second peak saturates at 9.2 L, where the sym m etry of the peak begins to break w ith the third peak growing in, and th e tem p eratu re for desorption increases to 130 K. The th ird peak grows in relation to th e coverage, and tem p eratu re for desorption increases from 124 K at 10.2 L to 129 K at 18.3 L. This peak is assigned to the second and multilayers. The increasing tem p eratu re for desorption in th e first and second layers indicate th a t interm olecular a ttractio n is increasing as the layers are filled. T he second peak th a t occurs at 128 K to 130 K is assigned to the first layer. The com pletion of th e first layer is approxim ated to occur at 9.0 L. A ttachm ent at step and defect sites are the m ost likely origin of the peak at 140 K. 52 0.0 •0 .1 - I 0.2 - 0.3 DC - 0.4 - 0.5 10 15 Dose(L) Figure 3.11: 3.5 Workfunction of Cu(llO) surface as a function of CH 3 I dose R etarding P otential Spectroscopy (R PS) R etarding Potential spectroscopy (RPS) was used to m easure th e changes in th e workfunc­ tion of a prepared surface as it was dosed. T he w orkfunction is th e energy required to remove an electron from the surface. As molecules are added to th e surface changes can occur in the workfunction. M ethyl halides are dipolar molecules and their addition to a m etal surface in the first layer has often been shown to decrease the workfunction. This is consistent w ith halide end of the CH 3X molecule being closer to th e surface on average. T he halide end, X , of CH 3X is negatively charged. A dsorption of m ethyl halide w ith th e negative end down creates a dipole layer near th e surface. This arrangem ent tends to assist electrons away from th e surface. 3.5.1 C H 3I /C u ( llO ) Fig 3.11 is a graph of the changes in workfunction of th e clean C u(llO ) surface as function of CH 3I dose. The workfunction decreases rapidly to a dose of 5 L. A fter 5 L th e workfunction decreases somewhat to 7 L, then begins a slow increase to 16 L. The workfunction rem ains roughly constant thereafter. The initial decrease in the workfunction up to a dose of 5 L is good evidence th a t the dipolar CH 3I is being attached to th e surface w ith th e I end down. This is consistent w ith the finding of other authors of CH 3I on other surfaces [24, 44]. Lin and Bent found th a t CH 3I adsorbed m olecularly on C u ( lll) at 120 K in th e first layer w ith the I end of th e molecule closer to the surface. Zhou et al. found th a t various m ethyl halides adsorbed to A g ( lll) in subm onolayer coverages caused a significant decrease in the workfunction, indicating th a t the molecules are orientated with the halide closer to the surface. Figure 3.11 provides evidence to help determ ine the dosage th a t corresponds to one monolayer for this system . The TPD s of figure 3.8 showed th a t for doses larger th an 3 L there were peaks at 160 K and at 140 K. For doses less th a n 3 L there was only a single 53 peak at 160 K. The argum ent made in the T PD section was th a t saturation of the 160 K peak at 3 L was either indicative of m ethyl iodide attachm ent at stronger bonding sites such as at step and defect sites or it indicated com pletion of the first monolayer. If we exam ine th e figure 3.11 we see th a t the workfunction decrease does not change noticeably until after 5 L; there is no significant change in th e workfunction decrease at 3 L. This is significant because in many studies of m ethyl halides on clean m etals a significant difference has been observed in the workfunction once the first layer is complete. Therefore, com pletion of the first monolayer at 3 L is unlikely. Saturation of the 140 K T P D peak occurs at 9 L, which we assigned to the completion of th e first monolayer. Figure 3.11 shows th a t at 9 L there is again no significant changes in the workfunction; it is slowly increasing till coverage reaches 15 L, where it remains constant thereafter. The only significant change occurs at 6 L; there are no such significant changes after 9 L. The variations in this workfunction is in fact typical of dipole system s on clean m etal surfaces. Figure 3.12 is the workfunction of Cesium on various surfaces of single-crystal Tungsten , referred to as the C s/W system [43]. W hen Cs is deposited on W an electron from the Cs atom is donated to the m etal. The electron stays on th e W surface in th e vicinity of the Cs ion; a dipole m om ent is created th a t causes a decrease in the workfunction. As a monolayer of Cs approaches the close-packing of th e Cs atom s causes a depolarization of nearby neighbours and causes the workfunction to increase som ew hat. The monolayer is com plete in figure 3.12 near 6 x lO^^cm^ T he workfunction graph for the first layer of C s/W is very similar in shape to the CH 3l/C u ( 110 ) system . Figure 3.11 is interpreted in a sim ilar fashion. At 6 L the CH 3I molecules begin to close-pack and depolarize on C u(llO ). The depolarization reduces th e effective charges on th e I atom s and on the C atom s, and thereby allows the CH 3I molecules to close-pack with th e I end down. Atoms w ith sim ilar charges tend to repel each other. The depolarization of the layer increases the w orkfunction slightly as th e dose is increased past 7 L. 54 5 Cs/W 4 £ 3 100 2 no |kn2 111 0 2 4 W.(10‘Van*) 6 Figure 3.12: Workfunction measurements of cesium on single crystal Tungsten surfaces. (Diagram from Zangwill [43, pg293]; Kiejna and Wojciechowski [22].) 55 0.0 > (D c 0) ï a. -0.2 CD c æ q: - 0 .4 0 5 10 15 20 25 30 D o se(L) Figure 3.13: Workfunction of Cu(110)-I surface as a function of CH 3 I dose 3.5.2 C H 3I / C u ( 1 1 0 )-I The Cu(110)-I surface is formed by w arm ing th e CH 3I dosed C u(llO ) surface to 525 K. A fter recooling the surface the workfunction was m easured to have increased by 1.2 eV as com pared to the clean C u(llO ) surface. The increase in the workfunction is consistent w ith the form ation of dipolar Cu-I orientated w ith th e positive end of th e dipole closest to th e surface. This is a typical result for electronegative atom ic adsorbates on m etal surfaces.[6 ] Figure 3.13 is a graph of the w orkfunction changes of Cu(110)-I surface as a function of CH 3I dose. The graph shows th a t the w orkfunction decreases non-linearly in th e first, second and third layer. T he m ost rapid decrease is in th e first layer, followed by a sm aller and slower decrease in the second. The th ird layer has a sm aller and slower decrease in th e workfunction than the second layer. T he rapid workfunction decrease in the first layer is suggestive th a t m ost of th e CH 3I adsorb w ith the I end down. The non-linear decrease in th e first layers suggests th a t some CH 3I are adsorbing w ith the I end up. This is supported by th e absence of depolarization in th e first layer on Cu(110)-I ; evidence for depolarization was seen in th e first layer on the clean C u(llO ) with the local m inim um at 7 L in th e w orkfunction scan. The continued decrease in workfunction for th e second and th ird layers on C u ( 110 )-I suggests th a t m ore th a n half of the CH 3I m ust adsorb w ith th e I end down in those layers. Further m ore, th e decrease in the w orkfunction changes in going from the first layer to the third layer suggests th a t th e CH 3I absorbs w ith the I end down progressively less from th e first layer to th e third. This is a n atural progression considering th e bulk stru ctu re of CH3I. Figure 3.14 shows the ideal crystal structure of solid CH3I. In the solid CH3I is 50 % down and 50% up. As layers of CH3I on C u(llO ) or on Cu(110)-I get thicker they m ust at some point have this ideal structure. It is likely th a t on the Cu(110)-I sem i-conductor surface, CH3I is able to adopt the alternating position of th e m olecule’s dipoles earlier th an on the clean surface. T h a t is, the up and down arrangem ent begins in the first layers, and as the layers progress th e 50:50 ratio is achieved. This would help to explain the observed tilt of the molecules on 56 20° 164' 124 124 20° H Figure 3.14; The crystal structure of solid CH3I. In the solid CiJsFSO % of the dipoles are orientated down and 50 % are orientated up. The dipoles are tilted at 20° from normal. As layers of CH3I on Cu(llO) or on Cu(110)-I get thick, they will at some point have this ideal structure. This occurs when the surface effects are no longer significant. Diagram modified from Kawaguchi et al. [21]. the Cu(110)-I surface. It is possible th a t alternation of some of th e dipoles on the surface lead to a surface structure where all of th e dipoles are tilted w hether or not they alternate. 57 C hapter 4 Surface P hotolysis:T im e-O f-F light E xperim ents 4.1 Introduction Time-of-flight (T O F) experim ents are done w ith C H 3 / on clean C u(llO ) surfaces and also on Cu(110)-I surfaces. Some TO F surface experim ents were also done w ith C H ^ B r for comparison to the C H 3 / photodynam ics. T he adsorbate photodynam ics observed in these T O F experim ents are the result of photoenergy transferred to th e adsorbate by th e direct and by the charge transfer processes. Only neutral particles can be observed in th e T O F spectra due to th e apparatus set-up. We cannot exclude the possibility, though sm all, th a t some ions are escaping the surface but we are not able to detect th em with th e QMS T O F apparatus. As a result all T O F spectra in this chapter result from neutral pathw ays. A principle reason th a t prototypical system s are formed from C H 3 / and C H sB r adsorbed on m etal surfaces is the fast intra-adsorbate bond dissociation tim es of the molecules. These dissociation tim es are on the order of femtoseconds and are com parable and, consequently com petitive, w ith electronic quenching tim es on m etal surfaces. C H 3 I dissociation occurs in 6 5 /s [16] and C H ^ B r is somewhat faster at ~ 15 fs [44]. Intra-adsorbate bond dissociation tim es are determ ined from th e speed of the fragm ent v and th e bond extension / th e tim e for dissociation is then l / v [44]. Typically I is ~ 2A and v is on th e order of lO ^cm /s, therefore bond dissociation tim es are ~ 200 fs [44]. C H 3 / and C B s B r have similar gas-phase photodynam ics and kinetics. In th e gas-phase both C H 3 / and CH 3^ r undergo photodissociation due to direct adsorption of light and also due to dissociative electron attachm ent(D B A ). Direct dissociation is represented by /»/ -f CFTaX C -k % (4.1) where hu is the energy of th e light th a t is adsorbed by th e C H ^ X molecule. A bsorption of a photon by a C H 3 X molecule results in a non-bonding tt electron centered on th e X atom excited to a dissociative a* state,ie C H ^X * [25]. D uring dissociation th e excitation energy hu dissipates into vibrational, translational, and internal energy modes. Dissociative electron attach m en t (DE A) in m ethyl halide molecules is sym bolically represented by 58 e+ -4- C ^ 3 + (4.2) where the electron e has some kinetic energy and is captured into an anti-bonding unoccupied molecular orbital of the C H 3 X molecule [17]. An anionic state C H ^ X ^ is created by the attachm ent which can dissociate into a m ethyl radical and a halogen ion. DEA as a result of incident light on adsorbed CH 3/ and C H aS r occurs because of charge transfer processes and is referred to as charge transfer photodissociation(C T -PD IS). As discussed in C hapter 1, photons incident on substrates can produce photoelectrons in single-photon processes w ith a distribution of energies w ith a m axim um equal to th e photon energy. Photoelectrons produced at th e su b strate can attach to th e adsorbate and cause it to dissociate. In the gas-phase CH 3/ and C H gB r have a high probability, th a t is a large crosssection, for attachm ent of low energy electrons. The gas-phase DEA of CH 3 / is 10 x larger than the DEA of gas-phase C U s B r [10]. In order to assist in distinguishing in th e surface studies between photolysis resulting from direct photoabsorption and from C T-PD IS, the kinem atics of gas-phase dissociation is reviewed. In the gas-phase, absorption of ultra-violet light is continuous for CH 3/ and CH 3J5 r in their respective A-band energies. T he gas-phase A -band adsorption cross-sections as a function of wavelength A for m ethyl iodide and m ethyl brom ide are shown in figures 4.1 and 4.2. Adsorption cross-sections are a m easure of th e probability th a t photons will be absorbed. A large cross-section indicates a high probability of adsorption whereas a small cross-section indicates a low probability. The CH 3/ A-band (3.5 e V < hu < 5.5 eV) is centered at 4.8 eV (A = 260 nm ) [13] , where the cross-section is ~ 140 x 10 ~^°cm^. [39] At 3.5 eV the cross-section is less th an 2 X 10“ ^^cm^. [39] The C H sB r A-band (4.4 e V < hu < 6.9 eV) is centered at A = 200 nm (6.2 eV) where the cross-section is ~ 60 x 10“ ^°cm^ [23, 38]. At 4.4 eV the cross-section is less th an 10~^^cm^ [23]. In other words th e C H sfir A-band is at m uch higher energies than th e CH 3 / A-band. The wavelength,A = 337 nm , at which the T O F experim ents are done in this thesis is ju st on the edge of the CH 3/ A-band bu t it is far outside the CH 3jBr A-band. We do not see evidence in the A = 337 nm T O F experim ents of direct photoabsorption in C ïla Br adsorbed on the C u(llO ) surfaces. 4.2 K inem atics and P oten tial energy surfaces T he kinem atics for m ethyl halide gas-phase direct photodissociation are well understood. In th e gas phase, dissociation of the CH 3X molecule occurs w ith th e C H s fragm ent and the X fragm ent separating at 180° to each other in a process referred to as axial recoil. Dissociation occurs on a faster tim escale th an the rotational period ~ 3 x 10 “ ^^ s of the molecule [16]. If the m om entum of th e electron or photon th a t causes the dissociation is small, the conservation of m om entum equation is given by mxvx - mcHaVcHs = 0 (4.3) where vch ^ and v x are the respective velocities of th e CH 3 and X fragm ents, m% ^ the ^ mj = 127 amu , m g r - SOamu 59 140 CH 120 o 100 •H €0 40 20 240 250 260 270 260 2*0 300 310 *mv#lmgCh (nm) Figure 4.1: Gas-phase adsorption cross-section of the A-band for CH 3 I as a function of wavelength. Diagram modified from Waschewsky et al. [39]. hv + CH,Br CHqBr 600 c O u s u2 180 200 222 55000 50000 45000 10 n m 400 0 0 Wdwtlatigth (cm'*) Figure 4.2: The total gas-phase cross-section for photodissociation of CHsBr as a function of wavelength in the A-band. Diagram modified from Van Veen et al. [38]. 60 mass of the halide atom , and m c% ^ the mass of th e CH^ fragm ent. Since the to tal energy for translation ,TavaiU h split between th e Chfg fragm ent and the X fragm ent, th e C H z translational energy, is related to the X fragm ent translational energy T x by 1 1 Tavail = T chs + T x = ^^CHsVcHs^ + -m x v x ^ (4.4) Using this relationship and equation 4.3 we can solve for the fraction of to tal energy available for translation of the C H 3 fragm ent, Tchs ■ Tc% = ----- ^ ----- Tavai, mcHs + nix 4.2.1 (4.5) K in e m a tic s fo r D ire c t G as-p h ase D isso ciatio n N eutral direct photodissociation of C H 3 X is described by th e kinem atics equation , Tavail = Ephoton " D o - - E%""" (4.6) where Tava.ii is the total energy available for translation, Do ^ is th e energy required to cause dissociation of the C H 3X molecule from its ground state vibrational energy, and Ephoton is th e energy of the photon causing th e dissociation. and are th e internal energies of the fragm ents. In this equation the C H 3X molecule is assumed to be a t rest and in its vibrational ground state before dissociation. The photon is assumed to have zero m om entum , since it is small compared to th e m olecular m om entum . Using equations 4.5 and 4.6 we can estim ate th e range of CH s velocities. Dissocia­ tion can proceed w ith th e iodine atom in a ground state = 0 ) or in an excited s ta te (E /”*®'’"“^ = 0.943 eV). Each of these possibilities is referred to as a dissociation chan­ nel. They are respectively referred to as th e ground state (^^ 3/ 2) I channel and th e excited state C P 1 / 2 ) I* channel. As the m ethyl halide molecule begins to dissociate, vibrational motions can be induced in the CH3 fragm ent. The internal energies of the CH3 fragm ent, ie include possible vibrational states such as the um brella mode and th e C -H stretch mode. No evidence has been found for a C-H stretch mode being significant [16]. However, a sym m etric vibrational um brella mode has been found to be significant [16]. T he CH 3 fragm ents vibrate w ith q uanta v ~ I b m e V in gas phase dissociation [34]. For th e A = 248 nm and A = 266 nm dissociation of gas phase C H 3I, CH3 fragm ents predom inantly vibrate in the um brella mode at v= 2 in th e ground state I channel and at v = 0 in th e excited state T channel [34, 12]. There is no vibrational d a ta available in th e lite ra tu re for C H 3I dissociation at th e wave­ length of 337.1 nm. Therefore we m ust estim ate th e vibrational energies of th e fragm ents. The um brella mode energies at 337.1 nm will likely be sim ilar to those in th e gas phase at 248 nm and 266 nm. Therefore we assum e C H 3 um brella m ode vibrational energies in the I and I* channels of C H 3I dissociation to be less th a n v = 4 quanta. We use v = 0 to V = 4 quanta for the internal energy of th e CH 3 fragm ent, Using equation 4.5, we find th a t CH 3 fragm ent energies, 7^%, for the I channel have a distribution ranging from ^m cir 3 = 15 amu ^Do(C7f3 - 7) = 2.39 eV 61 I % I 20- Figure 4.3: Gas-phase potential energy surfaces of CH 3I in the A-band. The ground state and the three excited states are shown. The ^Qi and the ^Qq states curve-cross. (Diagram from Eppink et al. [12]; Gedanken and Rowe [15]. 1.15 eV for V = 0 to 0.89 eV for v = 4, and sim ilarly for th e I* channel th e d istribution ranges from 0.31 eV for v = 0 to 0.046 eV for v = 4. 4.2.2 P o te n tia l E n e rg y S urfaces fo r D ire c t P h o to a b s o rp tio n The potential energy surfaces for dissociation due to direct photoabsorption in GH^l and C i/sB r are described in this section. C ffsl and C % B r exhibit very sim ilar direct photo­ dynamics except th a t C % B r dissociation requires higher photon energy. Figures 4.3 and 4.4 are potential energy surfaces for m ethyl iodide and m ethyl brom ide direct dissociation in the gas-phase. In each figure th ere is a ground state configuration labelled ^Ay. This state describes the potential energy of th e CH 3X molecule (X = I or X = B r), as a function of C-X bond length. The molecule has a non-zero vibrational energy in this state. The higher energy dissociative states in th e graphs are reached by prom otion of a non-bonding 7T electron centered on the halogen atom to a C-X anti-bonding state [39]. In th e A-band there are five excited states but only three optically allowed transitions from the ^Aj state to higher energy dissociative states for both C % I and CA^Br. Figure 4.5 shows an energy level diagram of the five excited states. There are two weak perpendicular transitions to the 2E ^Q\ and the 3E ^Q\ states th a t correlate to production of I{^Ps.) and there is a strong parallel transition to the 2 Ai ^Qo state th a t correlates to production of [12 ]. The {^Qi, ^Qi, ^Qo) term s are the Mulliken notation and the (2A i, 2E, 3E) term s are the corresponding Czy notation [12 ]. Dissociation of th e ^Qi and ^Qi states produce C H 3 A I whereas dissociation of th e state produces CHs + /*. Due to a curve-crossing of the potential energy curves of the ^Qi and the ^Qo states, dissociation which initially proceeded in th e ^Qo state m ay transform to the ^Qi state [12, 13]. Curve-crossing involves a non-adiabatic transition in the conical curve-crossing region [12]. Curve-crossing has been found to be significant in m ethyl-iodide 62 70 fiO SO 20- 4 .0 Figure 4.4: Gas-phase potential energy surfaces of CHsBr in the A-band. The ground state and the first three excited states are shown. (Diagram from Van Veen et al. [38].) 3e i 2Qi ) ’e - Conical Intaraection 2Ai(3Qn) 3E— r(Zp,/ 2) . CHa 2 cO) I (2P 3/2) ♦ CH 3 lAi------------ llAl---without spin-orbit with spin-orbit Figure 4.5: Gas-phase potential energy states of CH3X in the A-band. Energy states shown with and without spin-orbit interactions. There are three optically allowed transitions from the ground state to the states with spin-orbit interactions. Diagram from Yabushita et al. [42]. 63 1.0 0.8 0.6 0.4 0.8 ce 0.4 0.0 240 260 280 300 320 340 Wavelength (nm) Figure 4.6: (a)The gas-phase branching ratio for excitation to the ^Qo state. (j>o* = N q* /{ N q + No* where absorption of photons leads to a production of initial populations N q* in the channel and N q in the I channel, (b) The probability of curve crossing as a function of wavelength. For both (a) and (b) the dashed and solid lines are alternate interpolations of the data points. In (b) the dashed line is similar to that expected from the Landau-Zener probability curve. Diagrams from Eppink et al. [12]. dissociation [12, 13, 42]. The yield of fragm ents from the I vs I* channels is strongly dependent on th e curve crossing. A bsorption of photons leads to a production of initial populations No* in the I* channel and No in the I channel. The population change is described by AT" = (1 - and N = Ao + (4.7) where N and N* are the final populations in th e I and I* channels and P^c is th e probability for curve crossing [12]. The total populations are norm alized to unity (N N* = No +No* = 1) [12]. There are two branching ratios of interest. There is the branching ratio after dissociation is complete, le (f>* = N* / { N N * ) and th e branching ratio of th e initial population, ie 4>q* = No* / { No + No*). Figure 4.6 shows th e m ethyl iodide gas phase ^ 0* branching ratio and curve crossing probability Pec as a function of wavelength. For wavelengths less than A ~ 280 nm, th e branching ratio 4>o* of th e initial absorption is close to unity and the curve crossing probability Pec ~ 0.25 [12]. Excitation is predom inantly to the ^Qo state at these wavelengths. Above 280 nm th e 4>o* branching ratio decreases rapidly and the probability P ^ for curve crossing increases to 0.85 [12]. T he reason for the decrease in the 0o* value in the long red tail of the A B and is th a t more excitation is taking place to the ^Qi state [13, 12]. E xcitation to th e state is insignificant above 280 nm [12]. More curve crossing is taking place as excitation to th e dissociative ^Qo state nears the conical curve crossing region [12 ]. M easurem ent of th e dissociative products after dissociation is com plete allows determ i­ nation of the 4>* = A^* /( N + N*) branching ratio. From 222 nm to 304 nm the branching ratio 4 >* for dissociation of the states is significant, being greater than 0.43 [12 ]. A lthough 64 there is no d ata in the literature for the exact wavelength we are working a t (A = 337 nm ) there is data near th e wavelength we are working at. At 333 nm the cff branching ra tio in the /* channel is 4>* = 0.1 [13]. The dissociation of m ethyl brom ide has characteristics similar to m ethyl iodide dissocia­ tion although higher energy photons are required to prom ote th e ground state to th e excited states. (See figure 4.4. ) The ^Qq and th e ^Qi states dissociate to produce CH^ T B ri^P Ç j and the state produces CHj, + Br*{^P\). A lthough th e and th e ^Qo states curvecross, for m ethyl bromide dissociation curve crossing has not been found to be significant [38]. Since we saw no evidence of direct photodissociation of C i/aB r on th e C u(llO ) and Cu(110)-I surfaces in the T O F spectra at A = 337 nm , no further m ention is m ade of the direct dissociation of this molecule. 4.2.3 K in em atics E q u a tio n s fo r D isso ciativ e E le c tro n A tta c h m e n t D isso ciatio n in th e G as-p h ase A kinem atics equation for the dissociative electron attach m en t m echanism to gas-phase CH 3X can also be found. In this case the energy for translation, Tavaih is given by Tavai, = " Do + EAy- - - E%""^ (4.8) where Eeiectron is the energy of th e attaching electron and E A x - is th e electron affinity of the halide atom . D q* is the energy required to cause dissociation from th e lowest vibrational state of the M i ground state. The internal energy of th e halide atom — 0 for both iodine and bromine. In the gas phase the m ost im p o rtan t resonance for DEA for CH 3I is at low energy {Eeiectron = 0.15 cV) [33], although there are other resonances below this energy. The electron affinity E A j ^ for iodine is 3.06 eV [26]. We assum e a range of 0 to 4 q u an ta for the internal energies of the C H z fragm ent and using equation 4.8, we arrive at translational energies, Tavaii , of 0.82 eV to 0.52 eV. If th e electron m om entum is ignored th en we can use equation 4.5 to solve for the C H z translational energies. A range of 0 to 4 q u an ta for the internal energy of the C H z fragm ent gives a translational energy distribution, T c Hs , of 0.73 eV to 0.47 eV. For C H zB t the gas-phase peak in the DEA cross-section occurs at an incident electron energy of 0.4 eV. [37, 17], and th e electron affinity of brom ine E A s r - is 3.36 eV [37, 26]. Using a range of 0 to 4 quanta for the vibrational energy of th e C H z fragm ent and using equation 4.8 we find for gas phase CHzBv, th a t Tavaii ranges from 0.89 eV for v = 0 to 0.59 eV for v=4. We use equation 4.5 to solve for th e C H z translational energies and find th a t T c Hs is 0.75 eV for v= 0 and is 0.5 eV for v = 4. 4 .2.4 P o te n tia l E n e rg y S urfaces Figure 4.7 shows potential energy surfaces for excitation of th e ground state of C H z X by adsorption of a photon and for dissociative electron attachm ent(D E A ) to dissociative states. One of the m ain points in this diagram is to show th a t th e potential energy of the C H z X "Do(C^ 3 - / ) = 2.39 eU, - Br) = 2.87 ef/ 65 OÛ Nuclwf S#p«mUon, (A) Figure 4.7: C H ^ X potential energy surfaces for the electron attachment mechanism. Diagram modified from Ukraintsev et al. [37]. molecule is lowered in the DEA dissociative states com pared to the direct dissociative states for large C-X bond distances. There are two potential energy surfaces for th e DEA energy transfer mechanism. One potential energy surface shows C H 3 and X dissociation w ith both fragments in the gas-phase, and the other shows CH 3 and X dissociating w ith th e X adsorbed. The adsorption of the X fragm ent lowers th e potential energy of th e state. As a result of the lowered energy of th e DEA potential energy surfaces, charge transfer is expected to be more com petitive w ith quenching th an th e direct photoabsorption pro­ cess. A fter electron attachm ent prom otes th e ground sta te of th e C H 3 X molecule to the dissociative state, separation of th e C-X fragm ents quickly results in th e potential energy of the molecule being lower th a n in the ground state [6 ]. In order to retu rn to th e ground state, energy would have to be put into th e system [6 ]. This is opposite to th e situ atio n in direct dissociation, where after prom otion to th e excited state, th e potential energy of the excited state is higher th a n the ground state. W ith direct dissociation there is m ore tim e for quenching of the excited state back to the ground state, therefore it is more likely th a t quenching will occur th an for CT-PDIS. 66 4.3 Surface P hotolysis of CJ^gBr on C u(110)and C u (llO ) I The m ethod for doing TO F experim ents was described in the experim ental section. In brief each TO F experim ent is done w ith a prepared and dosed surface, laser pulses are directed onto the surface system, and CH 3 ions are counted in th e T O F analyzer. A C H j ion signal results from fracture and ionization of particles th a t desorb from th e sample and en ter th e mass spectrom eter ionizer. There is a delay between tim e when th e molecule or ato m is ionized and the tim e when it enters th e channeltron. For the U TI lOOC this delay tim e for m /e = 15 amu ions has been determ ined to be ~ 20p s [32]. True arrival tim es are found by subtracting this 20 ps delay. The surface photolysis of CFfgBr on C u(llO ) is presented first, although this system is not as well-studied as C E 3I on C u(llO ) and Cu(110)-I. One of th e reasons for presenting this system first is th a t the energy transfer dynam ics of th e system is easy to understand. Gas-phase CH^Bv does not have a significant cross-section for direct dissociation anyw here near A = 337 nm, therefore unless adsorption changes the potential energy surfaces signifi­ cantly, C i/sB r adsorbed on a m etal surface will dissociate prim arily via the charge transfer m echanism. C H 3 BT adsorbed on P t ( l l l ) w ith hydrogen was found to photodissociate by CT-PDIS w ith T O F studies at A = 308 nm [36]. Therefore it is not surprising th a t the evidence also points to the charge transfer m echanism in the A = 337 nm T O F sp ectra for dissociation of CH^ B t adsorbed on C u(llO ). In figure 4.8 the A = 337 nm T O F spectra for various doses of C i/sB r on clean C u(llG ) are shown. TO F spectra for all doses of CH 3B r/C u ( 110 ) show CH 3 signal from a fast distribution for times less th an 110/xs and a slow distribution for tim es greater th an llO p s. T P D experim ents indicated th a t the first layer was com plete on this system at a 9.0 L dose. The T O F signal from the fast d istribution is generally featureless and broad for all doses. The lack of sharp features in the fast T O F signal is consistent w ith an energy transfer m echanism with a range of energies, ie the charge transfer m echanism . In figure 4.9 th e counts in the spectra of figure 4.8 have been sum m ed as a function of th e fast d istrib u tio n and the slow distribution and plotted as a function of dose. In figure 4.9 both the fast and th e slow C H ^ signals peak at com pletion of th e first monolayer. From T P D m easurem ents, a single monolayer was found to be 9.0 L of C i/aB r on clean C u(llO ). The peak in the spectra at 1 ML can be understood as resulting from com petition between quenching and charge transfer. Both charge transfer and quenching are expected to be m ost proficient in th e first adsorbate monolayer on a m etal surface. A dditional layers on th e m onolayer are expected to provide a barrier to the m ovem ent of electrons and reduce th e efficiency of both quenching and charge transfer. The peak at 1 ML in figure 4.9 results because dissociation is reduced for coverages greater than 1 ML. Dissociation of molecules is reduced for coverages greater than 1 ML because the proficiency of quenching is reduced less th an the proficiency of charge transfer processes. In figure 4.10 the significant tim es in the T O F spectrum of 10 L of CHgBr on clean C u(llO ) have been labelled with letters A ,B,C, D and E. These tim es are listed in Table 4.1. Each tim e has been converted into a CH3 fragm ent and a CH sBr molecule arriving a t the QMS detector from the sample w ith th a t tim e. K inetic energies, Em, have been 67 hv +CH3BrA:u(110) 2000 50DL 300 L 1500180 L cc c .5) in + I o 1000 10OL 500 20 L 0 0.0 0.2 0.4 0.G Time of Flight 0.8 1.0ms Figure 4.8; TOF spectra at various doses of CffsBr on Cu(llO) 68 hv + C3^Br/Cu(110)-* J3 o 0 m 1 Ü Fast Peak 10 30 20 40 50 Dose (L) Figure 4.9: Slow and Fast Peak Counts as a function of CIIsBr dose on Cu(llO) 69 Significant Point A B C D E T im e(p s) 12 34 110 300 830 (eV) 3.7 0.46 0.04 0.006 0.0008 (eV) 23.6 2.94 0.28 0 .038 0.005 Table 4.1: Significant times on the TOF spectrum for figure 4.10. The energies of C H 3 (15 amu) fragments and CH^Br (95 amu) molecules corresponding to those times are also calculated from E = calculated for CHs fragm ents and C H ^ B r molecules arriving at th e significant tim es, where _ 1 Em = %mv 1 , d \2 1 ,8 .3 x lO -\2 f (4.9) This relationship is used in later sections for calculating energies for CH 3I on C u(llO ) and C u ( 110 )-I as well. T he distance d betw een th e sam ple and th e ionizer is 8.3 cm. W here m is th e mass of the CHz fragm ent(15 am u) or C H z X molecule {niBr = BOamu, m j = 127amu) and t is its arrival tim e(A ,B ,C ....) in th e mass spectrom eter. From the energetics in th e table 4.1 m any of th e CH^ fragm ents from the fast dis­ tribution between tim es corresponding to the significant points A and C could only come from C H z fragm ents from dissociation of th e C-Br bond. The significant point B in figure 4.10 has an energy of 0.46 eV for G H z fragm ents and 2.94 eV for C H z B t molecules. We determ ined earlier th a t DEA resulting in C-Br bond dissociation of th e gas-phase CHzBv molecule produced GHz fragm ents w ith energies of 0.75 eV(v = 0) and 0.5 eV(v = 4). Since the fast distribution is peaked near an energy we expected G Hz fragm ents to have, th e fast distribution is assigned to production of GHz fragm ents from dissociation of th e C-Br bond. T he slower distribution is assigned to m olecular desorption of CHzBv due to DEA. This is the extent of our interest in this system . We tu rn now th e photodynam ics of C H z B v on Cu(110)-I. In figure 4.11 T O F spectra are shown for CHsBr on Cu(110)-I at doses of 10 L and 20 L. The spectra show no evidence of dissociation or desorption of CHgBr molecules. The signal is all background counts. We know th a t on the clean C u(llO ) surface th a t CHsBr was dissociating and desorbing at these coverages. The coverage of CHgBr on Cu(110)-I is probably very sim ilar to th e coverage on clean C u(llO ). The lack of dissociation and desorption on this surface indicates th a t charge transfer from the surface to the adsorbate CHsBr is not taking place. The iodine on th e C u(llO ) is likely form ing a layer of Cul. Cul is a sem i-conductor and on the C u(llO ) surface it likely presents a barrier to th e production of hot electrons. In th e retarding potential experim ents in C hapter 3 we found th a t th e retarding potential of th e Cu(110)-I surface decreases by 1.2 eV relative to the clean C u(llO ) surface. This means th a t th e workfunction of the surface has increased due to adsorption of iodine [6 ]. Photoelectrons formed in the bulk of th e C u(llO ) w ith A = 337 nm photons likely do not have sufficient energy to tunnel through the sem i-conductor C ul layer, which would explain the lack of photodynam ics of CHzBv on Cu(110)-I. At A = 337 nm direct and charge transfer mechanisms are not available for this system . 70 800 S 600 400 200 0 [ui— L 0 200 400 000 800 'Î'îîw-ol-fligfeî (IJS) Figure 4.10: Time-of-fiight spectrum of CH 3B r/C u ( 110 ) at a dose of 10 L 71 1400- 20LCHa-ÆUi;iio)-i 1200 - 1000- 1DLCH,B'/QJ(11Q)-I 800- J V HD- d 400- 200- 1----------- 1-------0.0 0.5 1.0 ~r~ 1.5 2.0 TinitM 'f-H ishtlrns ,1 Figure 4.11: Time-of-flight spectra of CH 3B r/C u( 110 )-I at doses of 10 L and 20 L. In this case the CH 3 collected in the TOF apparatus is background. Apparently CHgBr is not dissociating. The Cul on the surface is likely acting as a barrier to the production of hot electrons. There is insufficient energy in the light photons to produce photoelectrons with enough energy overcome the Cul barrier. 72 4.4 Surface P hotolysis of C773//Cu(110)-I 4.4.1 E x a m in a tio n o f th e D e te c to r R e so lu tio n The signal from C H 3 I on Cu(110)-I is particularly strong and therefore it can be used to study the resolution of the TO F apparatus. The resolution is studied by doing two T O F spectra, one with a short flight path and one w ith a long flight path. In one case th e flight path d is measured to be 8.3cm and in the other it is 16 cm. Resolution error in the T O F spectra is not due to th e tim e w idth of the laser pulse. The tim e w idth of the laser pulse is 10 ns; after subtraction of th e 20/is delay, all flight times are more than 10fis. The ratio of th e laser pulse w idth to the flight tim es is less than = 0.001 or 0.1%. Therefore the laser pulse w idth is insignificant to the flight tim es. There is resolution error due to th e length of th e ionizer. T he m ean or average tim e C H 3 fragments are m easured is 20ps after entering th e ionizer. However there is some variation from this mean tim e, which is the prim ary cause of th e resolution error. The significance of the resolution error is prim arily affected by th e ratio of th e length of the ionizer region (A x % 1cm) in the mass spectrom eter to th e length of the flight p ath , d. Smaller ratios ^ result in b etter resolution of the signal , although they also result in decreased signals. The tim e t a particle is m easured is t ± At = (4.10) V where v is its velocity. Since A d will be less th a n th e length of th e ionizer, increasing d decreases the significance of A d and ,consequently, also th e significance of At. The resolution is analyzed by looking at the fastest and slowest particles in a distribution of particle energies. The real tim e t a particle of velocity v arrives in th e ionization region is t = ^. For the short flight path, a distribution of fragm ents w ith velocities Vi to V2 should arrive at th e ionizer w ith a tim e interval A t short = t\ — t 2 — . Increasing the length of the flight p ath results in the distribution of velocities ui to ug arriving at th e ionizer in a wider tim e interval. The tim e w idth of the velocity distribution for the long flight p ath of d = 16 cm should be Ationg ~ 2A t shortFigure 4.12 shows two T O F spectra of C H 3 I / Cu{110) — I done respectively for flight paths of 8.3cm and 16 cm. M easurem ent of the full-w idths at half th e m axim um (FW HM ) for the fastest of th e two fast peaks give Ationg = (16 ± 2 )fis and Atghort = (12 ± 2 )fis, and for the slower of the two fast peaks Ationg = (26 ± 2)fis and Atghort — (17 ± 2)fis. If the w idth of the FW HM were entirely due to th e w idth of th e T O F distribution, th en the FW HM in each peak should double when th e length of th e flight p a th is doubled. Since the FW HM do not quite double w ithin error, the T O F spectrum for the short flight p ath is somewhat less resolved th an the long flight path. From equation 4.10, th e T O F spectrum w ith the long flight p ath is expected to be more resolved th an the T O F spectrum w ith short flight path. The increased w idth of th e FW HM for the longer flight p ath clearly indicates th a t th e T O F distributions have a tem poral w idth. This confirms th a t th e w idth in the T O F signal is due to an energy distribution w ith some resolution error. This is significant because it asserts th a t we should see a broadening of the T O F signal if th e T O F distribution were to vary significantly due to surface conditions. A lthough the resolution was poorer for the short flight p a th of d = 8.3cm, th e signal 73 4000 h v + CHjî/CuC 1 1 0 H - ^ C H j I g ) 3000 d=16cm Ü3 c3 5 2000 JZ u A t = 1 7 }jS 1000 A t= 1 2 p S 0 100 300 200 400 Tim e o f Flight Figure 4.12: Time-of-flight spectra of 20 L CH3l/C u ( 110 )-I. In one TOF spectrum the flight path d is 16 cm and in the other d is 8.3 cm. The width of the signal is largely due to the energy distribution of particles desorbing from the sample, however, the resolution of the detector also affects this width. was larger. This is the reason th a t th e short flight p a th of 8.3 cm was used in all o th er T O F spectra. 74 4.4.2 C h a ra c te riz in g th e en e rg y tra n s fe r m ech an ism s a n d d isso ­ ciatio n TOF (A = 337 nm ) spectra for various coverages of CH^l adsorbed on Cu(110)-I are shown in figure 4.13. The yield of CH3 fragm ents from the surface depends on th e CH3I coverage on th e Cu(110)-I sample. ( Only neutral particles can enter th e QMS; they are ionized after they enter th e ionizer region.) C H 3 fragm ents are not observed above the background counts for doses less th an 9 L, which from T PD experim ents is 1 ML. For coverages greater than 1 ML, there is a C H ^ signal and it is bimodal. The signal maximizes when the second layer is complete at a 20 L dose but decreases rapidly for higher coverages. The distinct shape of the two peaks in th e T O F spectra rem ains consistent from low coverage to high coverage. The T O F signal does not broaden due to high coverage, which indicates th a t C originate from the top layer only. C H ^ originating as a result of desorption from deeper layers would undergo scattering w ith overlayer molecules, which would broaden th e energy distribution and spectra at higher coverages. M ethyl iodide molecules are effectively prevented from dissociating in buried layers. This effect is referred to as caging in th e literature. Caging of CH 3 I has been observed in C H 3 I dissociation in rare gas m atrices [4]. In a T O F study of CH 3 I on A g ( lll) at A = 248 nm all C H 3 fragm ents from surface dissociation were attrib u te d to the upperm ost layer w ith the underlayers caged and prevented from dissociating [19]. We looked for particles such as CHj", CH 4 , , C 2 HQ in th e T O F analyzer as a result of neutral desorption from th e C H 3 I / Cu{ l l O) - l system . We saw no evidence for them at any coverage. Since none of these particles appeared to be present, th e CH 3 signal in the T O F spectra could only come from C H 3 fragm ents from surface dissociation or from m olecularly desorbed C H 3 I. In figure 4.14 the CH 3 yields in th e T O F spectra of figure 4.13 have been sum m ed and plotted as a function of dose. T he to tal counts in I and I* peaks were sum m ed and plotted as a function of dose. T he CH 3 signal clearly begins when th e coverage is increased past 1 ML, reaches a m axim um on com pletion of th e 2 ML, and th en rapidly decreases for higher coverages. The distribution of th e yield as a function of coverage is asym m etrical. The counts increase much faster as the coverage is increased to 20 L th en they fall off after the 20 L dose is reached. In order to assist in the determ ination of w hether the T O F signal results from dissociated C H 3 fragments or from desorbed CH 3 I th e significant tim es for figure 4.15 were tabulated. Table 4.1 lists the significant tim es (w ith electronic delay tim e rem oved) and kinetic energies th a t correspond to the A ,B ,C,... tim es in the T O F spectrum of figure 4.15. The significant tim es are the onset , the peak(s), and th e end of a signal. In figure 4.15 th e peaks at tim es B and D correspond to arrival tim es 21/is and 42 jj.s. From table 4.2, the energies of m ethyl fragm ents arriving at these tim es are 1.2 eV for the fast peak and 0.3 eV for the slow peak. These values are the same, with somewhat more error, as calculated for the gas-phase m ethyl iodide. Recall th a t m ethyl fragm ents were calculated to have energies of 1.15 eV and 0.31 eV for dissociation of m ethyl iodide in the I and I* dissociation pathw ays at 337 nm for v = 0 q uanta of um brella vibrations. From the energetics in table 4.2, particles which arrive from th e first fast peak dis­ tribution between tim es A and C m ust be C H 3 fragm ents from intra-adsorbate surface dissociation. Molecules distributed about th e peak tim e of 21 fxs would have a peak energy 75 60L 400 0 40L : 30 0 0 3ÜL 2000 hv + CH3I /Cu( 1 1 0 ) - 1 —> CH](g) o 1000 20L 1ÜL 0 200 400 600 SOOjUS Time of Flight Figure 4.13: TOF spectra of CH^I/Cu{llQ) - I Significant Point A B C D E T im e (/2 s) 12 21 (eV) 3.7 1.2 0.7 0.3 0.05 28 42 106 E c HsI (eV) 35.2 11.5 6.5 2.9 0.5 Table 4.2: Significant times on the TOF spectrum for C iÏ 37/C u ( 110 )-I in figure 4.15. The energies of CH^ fragments and C H 3 I molecules corresponding to those times are also calculated from E = -mv'^. 76 3000 2500 2000 o f O 1500 1000 500 0 10 30 20 40 50 60 Dose(U Figure 4.14: The total counts in I and I* peaks were summed and plotted as a function of dose. The CHg signal begins when the coverage is increased past 1 ML, reaches a maximum on completion of the 2 ML, and then rapidly decreases for higher coverages. 4000 hv + CHgI/Cu(110)-I 3000 O 5 1000 20. OL 1 50 200 290 300,u s Tim e of R ight Figure 4.15: Time-of-flight spectrum of CH 3l/C u ( 110 )-I at a dose of 20 L 78 of 11.5 eV and a m inim um of 6.5 eV. This is much larger th a n the 3.7 eV expected to be available in the single-photon processes believed to be occurring at the light intensities we are using. The intra-adsorbate dissociation th a t give rise to th e C H 3 fragm ents m u st also occur as a result of direct energy transfers. Fragm ents from th e CT m echanism can arrive as early as fragments from the direct m echanism, however, th e charge transfer m echanism cannot explain the origin of a distribution about 1.2 eV in figure 4.15. A com parison can be made to the T O F spectra of C % B r on clean copper. In figure 4.10 T O F sp ectra of a monolayer of CH 3B r/C u ( 110 ) system resulted from a fast and a slow distribution; th e fast distribution is assigned to hot electron DEA causing intra-adsorbate dissociation. T he peak in th e spectra from the fast distribution is not sharp and distinct like the peaks in figure 4.15. Sharp T O F peaks arise because of a very narrow energy distribution, which are not characteristic of charge transfer processes. Therefore th e first peak is assigned to direct energy transfers from intra-adsorbate dissociation. From th e table 4.2, the second peak in figure 4.15 centered at a tim e of 42 fis seconds can have energy of 0.3 eV for C i /3 fragm ents or 2.9 eV for C H 3 I molecules, both w ithin the 3.7 eV available. However the evidence is against molecules contributing to th e peak. C T is the most likely m echanism to cause m olecular desorption. Electron attach m en t resonances to C H 3 I in the gas-phase are m ost significant at energies below 0.5 eV. It is doubtful th a t th e CT m echanism would cause molecular desorption at 2.9 eV. Molecular desorption also results in molecules preferentially leaving perpendicular to the surface whereas dissociation of th e C-I bond results in th e C H 3 fragm ent having a trajecto ry in th e direction of the dipole m om ent before dissociation. Spectra of CH3 yields as a function of angle, figure 4.16, indicate th a t th e preferred dissociation angle is not perpendicular to th e surface. ( See the angular yields section.) Since m ethyl iodide dissociation occurs by axial recoil, where th e C-I bond dissociation occurs on a much faster tim escale th a n the ro tatio n al m otion of th e CH3I molecule, th e preferred desorption angle a t 20 ° in th e [110 ] azim uth is highly indicative of the molecules tilte d orientation on th e surface and C-I bond dissociation. Therefore th e second peak is assigned to C H 3 fragm ents from dissociation of th e C-I bond. The energetics of the CH3 fragm ents in th e second peak indicate the possibility of a direct photodissociation m echanism a n d /o r a charge transfer mechanism. T he translation energy of 0.3 eV is consistent w ith direct processes in th e gas-phase for 337.1 nm photons. In addition, charge transfer processes for m ethyl iodide in th e gas-phase were calculated to give rise to 0.7 eV CH3 fragm ents, also near th e 0.3 eV peak energy. Strong evidence against a charge transfer m echanism contributing is in figure 4.11, th e T O F spectrum of CH 3Br on Cu(110)-I. There is no CH 3 signal above the background in this system; this is different th a n CH 3Br on clean C u(llO ) in figure 4.10. Photolysis of CHgBr on C u(llO ) indicates a charge transfer m echanism where hot electrons from the su b strate attach to th e CHsBr overlayers. CHaBr is very sensitive to electrons in th e gas-phase and dissociates readily for attaching electrons near 0 eV as does CH 3I. The lack of signal in the CH 3B r/C u ( 110 )-I system indicated th a t Cu-I on the surface acts as a surface barrier, as evidenced by the increased workfunction (A $ = 1.2 eV), to th e photoelectric process th a t produces hot electrons. C ul is typically referred to as a sem i-conductor surface which does not tran sm it electrons as readily as a clean copper surface. Therefore hot substrate electrons are likely unavailable in the CH 3l/C u ( 110 )-I system also. 79 Other evidence against a charge transfer m echanism operating in th e CH 3l/C u ( 110 )-I system is in figure 4.13. T he T O F spectra of CH 3l/C u ( 110 )-I as a function of coverage show th a t the shape and relative sizes of the two CHg signal peaks in each spectrum rem ains consistent as the dose is varied. This is contrary to what is expected to occur when a charge transfer mechanism is operating. As adsorbate layers increase, it is more difficult for electrons to be tran sm itted through the layers and the num bers tra n sm itte d should decrease. If charge transfer were contributing to the intra-adsorbate dissociation, th e decrease in available electrons would affect the yield from dissociation. If two distinct m echanism s were operating in the peaks, it is likely th a t their relative sizes would differ as th e dose is increased. Cross-sections for molecules dissociating due to the charge transfer m echanism would have a dependence on the thickness of the adsorbate [6]. A dditionally th e shape of the second peak is not consistent w ith other observations of m ethyl fragm ents arising because of charge transfer mechanisms. Typically th e CHg fragm ents from charge transfer processes arrives in much broader distributions as for CHgBr on clean C u(llO ). Therefore th e second peak of the CH 3l/C u ( 110 )-I system is also assigned to intra-adsorbate dissociation resulting from direct photodissociation. 80 4.4.3 A n g u la r Y ields T he angular yields T O F experim ent th a t we do here is similar to a technique known as electron stim ulated desorption ion angular distributions. ESDIAD is a technique which has been used to determ ine molecular orientations on surfaces. A high energy electron beam (~ 100 eV) directed onto a surface induces dissociation of adsorbates. Ions resulting from dissociation are detected in angle resolved experim ents. As long as th e dissociative process occurs on a faster tim escale than adsorbate rotation, the repulsion in the dissociation is along th e bond th a t is breaking [8]. If the ions produced by dissociation have a high probability of desorption, then the angular yield of ions will be roughly proportional to the adsorbate bond orientations. If th e dissociation do not have high probability of desorption, th en a significant num ber of ions are undergoing collisions in the adsorbate layers w ith th e likely result of changing surface chemistry. In th a t case the angular yield cannot be a reliable indication of the bond orientations prior to th e experim ent. We can interpret our neutral angular yields experim ents in a sim ilar fashion. We know th a t dissociation is prim arily taking place in th e exposed top layer, since th e CH 3 fragm ent signal shows little evidence of inelastic collisions. The workfunctions for this system also indicated a preference for th e iodine end of the molecule to be closer to th e surface. The signal in the angular yields experim ent in figure 4.16 is significant for angles less th an 40°. Therefore desorption of CH 3 fragm ents are very likely to follow dissociation. D issociation of th e CH3I molecule occurs on a faster tim escale th a n m olecular ro tatio n , therefore the yield indicates the direction of the C-I bond orientation prior to dissociation. CH3I dissociation occurs in 65fs whereas the molecular rotational period is ~ 3 x 10“ ^^s [16]. Figure 4.16 shows the T O F yield of CH3I on Cu(110)-I as a function of angle in two azim uths. O rientational ordering is indicated by th e peaks in the angular yields. In th e [110] azim uth the yield of CH 3 has a significant peak when th e sam ple angle is 20°. In the [001] azim uth the yields are greatest at w ith th e sam ple angle is at 0°. The [001] azim uth depen­ dencies indicates th a t th e molecules are preferentially norm al in this orientation although th e orientational ordering in this azim uth is less pronounced th a n in th e [110] azim uth. The [1 Î 0 ] azim uthal dependencies indicate th a t th ere is a pronounced preference for the CH3I m olecular dipole to be tilte d at 20 ° off surface norm al in th e direction of the [110 ] rows on the surface. Interestingly enough, the bond tilt of 20° on th e surface is close to the bond tilt found in solid CH 3 /.(S ee figure 3.14) In th e solid, th e molecules C-I bonds orient in an alternating up and down structure. R PS experim ents indicated th a t in th e second layer, the C H 3 / molecules orientate preferentially w ith th e I end down. Since there are no yields for less th an 1 ML we do not have direct evidence th a t the molecules are tilted in th e first layer. However it is possible th a t preferential tiltin g in th e first layer is responsible for preferential tilting in overlayers. 81 5 -1 hv + CHjIÆJud 1D)-I -> CHjf^ 4 - # [1 1 0 ] A z 'ïm u îh A [0 0 1 ] A z im u t h i2 c 3 €00 CD C CD \f> X CJ 0-1 0 10 20 30 40 Angle of Sample to Mass Spec Figure 4.16: Angular experiments in the [110] and [001] azimuth 82 Param eter n B A, V F Vq i D escription num ber of peaks Background counts A m plitude velocity of a fragm ent as a function of tim e t effective tem p eratu re param eter stream ing velocity param eter for im proving th e fit Table 4.3: D escription of the param eters for equation 4.11 4.4.4 Altered photodissociation dynamics Evidence for altered photodissociation dynam ics are found in th e 4>* branching ratio. The two fast peaks are an representation of th e distribution of CH 3 fragm ents from dissociation in the I and I* pathways. It is approxim ate because there is resolution error from th e T O F apparatus. By separating th e distributions and counting the am ount of CH3 fragm ents in each peak we can infer th e relative yield from th e I and I* pathw ays. T h e peaks in th e T O F spectra as a function of dose were deconvolved using the following fittin g function [6 ]: F{t) = B + ^ Ai exp t=i —m{v — VQif (4.11) Each of the variables in th e function is defined in table 4.3. T h e m axim um sum m ation index n is the num ber of peaks in th e T O F spectra. This fitting function is based on a variation of the B oltzm ann function [6 , 46]. There is no physical basis for why th e peaks would be fit by a modified B oltzm ann distribution, it ju st happens th a t th e function fits well to th e spectra. This allows a good estim ate of th e yield in each peak. An exam ple of th e fit to a T O F spectrum at 20 L is shown in figure 4.17. A fter the peaks were deconvolved and th e counts in th e I and I* channels were sum m ed, the counts were m ultiplied by | , where t is th e peak tim e. By using th e fitting function to approxim ate the counts in each peak, th e relative yield of th e * = ( N is the m easured counts in each channel) branching ratio is p lo tted as a function of coverage in figure 4.18 [6 ]. It is found th a t (jf is close to 0.66 for all coverages [6 ]. T he * value is not as reliable at high coverage since T O F yields decrease rapidly as th e coverage is increased. However, th e cjf branching ratio is clearly larger at all coverages th a n th e gas-phase value at 333 nm where cjf = 0.1. T he I* excited state is more active in th e surface experim ents and therefore, the dynam ics of CH 3I on Cu(110)-I are p ertu rb ed in all layers as com pared to the gas-phase CH 3I in th e long A tail of th e A band [6 ]. Further evidence for altered photodissociation dynam ics is found in th e depletion yields as a function of to tal photons. Depletion yields as a function of to ta l photons are done by taking successive T O F spectra on th e sam e dose. T he logarithm of th e signal S is p lotted as a function of to tal photons 6. T he relationship between th e signal S in th e T O F spectra and th e to tal photons 6 on the surface system is investigated to see if it fits to equation 2.15 ie 83 W 200 =1—y..r_T- r\ ' ' ' ' ..I -T—r—r—r= h - 2500 %» 500 0 100 50 150 200ps Time of Flight (ps) Figure 4.17: Example fit to a TOF distribution CH3l/C u ( 110 )-I. Graph is from Johnson and J en sen [6]. 84 "I Q ^1 i n 111n 11I I 11n 11111111111 n 111i n Ti 11n I n f " I p. 0.8 13 0.6 0 0.4 e 0.2 lllllllllllllllllllllT 20 40 60 80 100 CH3 IDose (L) Figure 4.18: 4>* branching ratios from the fits to the distributions in the TOFs on CH 3l/C u ( 110 )-I as a function of dose. The dashed line is the * — 0.1 branching ratio for gas-phase CH 3I at 333nm. Graph is from Johnson and Jensen [6]. 85 ln{S) = constant —aO (4.12) The 2 ML coverage seemed to be m ost significant since at this coverage the yields were largest. Figure 4.19 shows a graph of th e photodepletion of 2 ML CH3I on Cu(110)-I on semi-logarithmic scale. Over th e range of photons put onto the surface, th e graph is generally linear and therefore it is likely th a t equation 2.15 can be considered valid. From the slope of this graph and others like it, the cross-section was estim ated to be (1.5 ± 0.5) x 10~^®cm^. The gas phase cross-section of CH3I at the sam e wavelength is 2 x 10 ~^^cm^, therefore the surface cross-section is larger by a factor of ~ 100 . The enhanced cross-section cannot be attrib u te d to adsorption of m ethyl iodide. Don­ aldson et al. reported th a t the A-band adsorption spectrum blue shifts in the gas-phase when C H 3 I forms clusters [6 , 11]. Increasing the concentration of C H 3 I decreased th e prob­ ability of adsorption. We know from the yields as a function of coverage experim ent th a t the first layer in the surface system does not dissociate, therefore only the second layer is contributing. It m ight seem surprising therefore th a t a single layer of solid C H 3 I has at least 100 tim es the probability of adsorbing a photon th an a gas of C H 3 I molecules in a three-dim ensional volume. Gas-phase cross-sections are given in term s of adsorption crosssections rather than dissociation cross-sections th a t we m easure here. A lthough we m easure dissociation cross-sections of C H 3 I, because we work at laser light intensities where only single-photon process are available, we know each dissociation is a result of adsorption of a single photon. T he cross-section determ ined by m easuring the depletion of th e signal is th e to tal crosssection for depletion of th e molecules on th e surface. The to tal cross-section m ay be a com bination of dissociation processes where some C H3 desorbs after dissociation and some attaches to the surface after dissociation. Lam ont et al. observed th a t CH3 attaches to a C u ( lll) surface after the C H ^ B r I C u { \ \ \ ) surface is illum inated w ith UV light [23]. It is most likely th a t iodine rem ains attached to th e surface after th e molecules dissociate. Iodine was not observed in T O F spectra. A depletion spectrum at 20 L w ith larger num bers of photons th a n in figure 4.19 is shown in figure 4.20. As the num ber of photons gets large, at % 12 x 10^®/cm^, th e slope of th e semilogarithm ic graph changes. The slope is m easured to be 5.6 x 10~^°cm^. The change in the slope of graph indicate changes in th e surface chem istry due to dissociation and adsorption of byproducts. The change in the slow peak could simply indicate a reduced C H 3 I crosssection for photon capture and dissociation due to changes in th e surface chemistry. We m ust be cautious about accepting this interpretation. The d a ta is showing some non-linear tendencies as the num ber of photons gets large. Recall th a t th e num ber of fragm ents th a t dissociate d N = —Noe~'^^ad9 and th a t the yield of fragm ents collected in the detector is S—qdN, where q is a fraction. Reordering effects can change the num ber of molecules Nq available for dissociation and also change th e fraction q th a t we collect, by changing the angular distributions, between subsequent T O F spectra. It is quite likely th a t th e yields S at large photon num bers are affected by reordering effects. The slope of the semi-logarithm ic graph at large photon num bers is likely not strictly representative of th e cross-section. A sem i-logarithm ic graph of the yields vs to tal photons for a 40 L dose of C i/ 3l/C u ( 110 )I is shown in figure 4.21. Over the range of to tal photons the graph shows some non-linear tendencies at very low photon num bers. It is quite likely th a t this is not due to counting error 86 g g u | | I I I 1 1 1 1 1 1 111 M j 11111 u n 111 l'i r n " n ['ï 'n T p 11 1111111111 1111 r r | T r n 111 h v 4 rC 3 i;% c ïK iio )-i -0.5 O =15x1 - 1.0 X -1.5 % e X ■2 . 0 H 1 111 1 1 1 1 1 1 1 111 ,i„J,i 1,11 1 H I 1 1 j I, l i . I j h I i l 11 i l u . x . i , l i 11,11 11 1 1 L j 1 1,1, H I 1 1 1 1 1 1 1 1111 8 10 12 14 Photons/cm Figure 4.19: Depletion spectrum for a 20 L dose on CH 3l/C u ( 110)-I 87 16x10 13 1—I—I—I—ΗI—I—I—I—i—I— 1—I—I—I—I—r—I—I—I—I—1—r~|—I—I—I—r sl(>;>e = 4— I— I— I— !— I— I— I 1 I I I L_J I I L_J 1] I 1 I ! <:nnf I 1 20 1 I 25 I I 1 I 30x10 I ia Total AccumulatecI Photons/cm" Figure 4.20: Depletion spectrum with a larger number of photons for a 20 L dose on CH3l/Cu(110)-I 0.0 In , 4 - -| 0.2 2!&x1 () -0.4 crrf t z z—I - 0.8 0 5 15 10 20 25 30x10 Photons/(cm^ Figure 4.21: Depletion spectrum for a 40 L dose on CH 3l/C u ( 110 )-I. The cross-section or slope of the graph changes as the total number of photons increase on the sample. 89 in the yield but rather to a surface phenom ena. The yield as a function of to tal photons for C i/a l on the clean C u(llO ) surface showed similar non-linearity for low num bers of photons. In figure 4.21, the slope changes from 20 x to 4.1 x lO'^cm^, which is a reduction in slope by a factor of 5. Due to th e low photons num bers where th e yields S rapidly decreased it is quite likely th a t the num ber of molecules available to dissociate has rapidly decreased. Since th e accum ulation of byproduct is small at low photon num bers, a possible explanation is th a t at a dose of 40 L the coverage of the fourth layer is incom plete leaving the third and possibly second layers exposed. If th e exposed layers rapidly deplete and reordering of the surface covers them up the slope 0.41 x 10“ ^®cm^ may represent th e cross-section for the fourth layer. This value is about 4 x sm aller th an the cross-section estim ate at 20 L. T he initial slope of 20 x 10“ ^°cm^ is possibly the cross-section value for depletion of these sites, since it is w ithin error of the cross-section for 2 ML coverage. It is also quite possible th a t other surface reordering effects could be responsible for th e rapid initial decrease in th e yields at the low photon numbers. 4.5 Further D iscussion The fact th a t all layers have a similar branching ratio of 4>* = 0.66, is interesting in com par­ ison to the observation th a t the yields are peaked at 2 ML in th e coverage experim ents. It is clear from the dose dependency spectra th a t proxim ity to th e copper surface is somehow responsible for the enhanced yields and large cross-sections. Yields increase when th e cov­ erage increases to a com plete second layer and then decrease as the coverage increases. T he peak in the yields at 2 ML is a ttrib u te d to th e effect of the surface on the potential energy surfaces for intra-adsorbate dissociation. The potential energy surfaces are modified by the presence of th e surface. It is apparent th a t this effect weakens as additional layers are added to the second layer. However the surface effect th a t causes the cj)* = 0.66 branching ratio on the surface apparently is not weakened by additional layers. This is further evidence th a t additional layers do not fully cover the underlayers so th a t the yield comes from several different layers. In th a t case the most significant proportion of th e yield would come from exposed portions of th e second layer, since this layer likely has the largest cross-section. The large cross-section for dissociation, especially in the second layer, indicates th a t the excited states have been m ade more accessible. The difference in energy between the vibrational state of th e molecule in the ^A\ ground state and th e excited state m ust equal the energy of the photon in order for excitation to occur. In order for the excited states to become m ore accessible the relative positions of the excited state and the ground state well m ust have changed. One possibility is th a t th e excited state(s) have moved downwards in potential energy tow ards the ^A\ ground state. A nother possibility is th a t bond lengthening has resulted in the ^Ai ground state well widening or moving to th e right, which would result in lengthening of the C-I bond. Lengthening of the C-I bond would require substantial changes to the ground state wavefunction, which is not likely to occur in the physisorbed second layer. Therefore th e large cross-section is a ttrib u te d to th e repositioning of th e excited states. The repositioning of the excited state(s) is presum ably also responsible for th e cf)* branch­ ing ratio being significantly larger th an the 0.1 value of the gas-phase CH 3 I at 333 nm. In 90 the gas-phase at 333 nm, only a fraction of the photoexcitation is to the ^Qo state, a large proportion of the excitation takes place to ^Qi state [13]. In figure 4.6 from Eppink (et al) gas-phase CH 3 I excitation to the ^Qo state decreases to less th an 50 % and th e probability of curve crossing increases to 90 % as th e wavelength is increased to 340 nm. To account for the (f)* ~ 0.66 value, excitation to the ^Qo state likely has increased and likely curvecrossing is less significant. The repositioning of th e excited states could have resulted in excitation from the ground state to the dissociative state taking place very near th e curve crossing region of the ^Qi and the ^Qo states. The excitation to th e ^Qo state in th a t case either takes place ahove or helow the curve crossing region. If excitation is below the curve crossing region, then the <^* branching ratio represents the ratio of molecules excited to the two different states and the ^Qo, which then directly dissociate respectively via th e I and I* channels. However, if excitation is above th e curve crossing region to the ^Qq state, then we may expect th a t the relevant factors which affect curve crossing to be significant. After excitation to the ^Qo state, th e probability for curve crossing to th e ^Qi sta te to form C H 3 -f I* is approxim ated by the Landau-Zener(LZ) transition probability P [5, 38]: = 1- = 1 - exp e x p I\ / (4.13) where V12 is the singlet-triplet coupling between th e ^Qi and th e states, u is th e velocity of th e fragm ent at the curve crossing point, and A F is th e difference in slopes of th e two potential energy surfaces at the curve crossing point. A first approxim ation has V u proportional to th e spin-orbit coupling in iodine [38]. ^ T he velocity v of the C H 3 fragm ents in our surface study on Cu(110)-I at A = 337 nm is very sim ilar to the velocity of the C H 3 fragm ents in the gas-phase at 333 nm , where th e curve crossing probability ^Qi and th e states to produce C H 3 + 1* is F = 1- c i-y (g ) 40. OL 2000 1500 JS o Ü 20. OL 1000 7.0L 1—1 I 0.0 . I I I I I I I I II I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1 I 1 I I I I I0.2 0.4 0.6 0.8 1.0ms Time of FKglt Figure 4.22: Various TOF spectra of C H 3 I/Cu{110). From TPD measurements, a dose of 9.0 L is a monolayer of coverage on the Cu(llO) surface. 93 at 140 K; the 140 K peak is indicative of the reduced binding energy sites for higher doses in the first layer. It is quite likely th a t th e reduced photodissociation efficiency observed in the 0-2 L dosages is the result of higher binding energy sites observed at these doses and a more rapid quenching m echanism for these molecules [6 ]. These sites are quite likely more efficiently quenched. The orientation of the molecules on the surface at low coverage may be different th an at higher coverages. This could either inhibit dissociation by m aking quenching more efficient, or reduce the probability for detection of th e dissociation. If th e CH 3 I molecule is in a lying down orientation on the surface, dissociation could result in th e CHz fragment attaching to the surface, or escaping at a large enough angle from norm al th a t it is not detected. At completion of the second layer, production of C H ^ fragm ents m axim ized in b o th the fast and the slow distributions. (See figure 4.23) The counts from th e slow d istrib u tio n were slower to fall off than the faster distribution. It is interesting th a t C fragm ents maximized at completion of th e second layer for C H 3 I on both th e clean C u(llO ) and th e Cu(110)-I surfaces. The A = 248nm photolysis of GH 3 I on A g ( lll ) showed th a t C H 3 fragm ent production saturated when the coverage was 3 layers[19]. In th a t study C H 3 fragm ent production began when the coverage was greater th a n 1 ML and increased when the coverage was increased to 3 layers, after which th e C H 3 yield was constant [19]. For lack of b etter understanding, the enhanced yields in th e second layer on C u(llO ) and Cu(110)-I are simply attrib u ted to changes in th e excited state potential energy surfaces. High resolution scans of the A = 337.In m T O F of C H 3 I on clean C u(llO ) are shown in figure 4.24. The scans dem onstrate th a t for a coverage of 7 L, which is an incom plete first layer, the CH 3 signal from the fast d istribution (time-of-fiight less th a n 1 2 0 fis) is not sharply peaked. The signal peak is centered about 64 /rs, which corresponds to a C H 3 fragm ent energy of 0.13 eV. The lack of sharp features and th e energetics are consistent w ith C H 3 fragm ents produced from C T-PD IS from dissociative resonances of gas-phase C H 3 I. In th e gas-phase, C H 3 I has dissociative attach m en t resonances for electrons of energy less th a n 0.15 eV and produces C H 3 fragm ents from CT-PD IS of energy less th a n 0.7 eV. In figure 4.24, the CH 3 signal from th e fast distribution a t 20 L, where the second layer is complete, is bim odal, asym m etric and sharply peaked. Significant points in the C H 3 l C u { 1 1 0 ) system are m arked w ith th e letters A,B,C... in figure 4.25. The tim es th a t correspond to these points are tab u lated in table 4.4 and energies are calculated for CH3 fragm ents and CH3I molecules arriving at these tim es in th e table. T he sharp peaks in the T O F signal th a t correspond to points B and D have tim es 22 /is and at 42 p s and result from particles arriving in a narrow energy distribution. CH3 fragm ents arriving at these tim es would have energies of 1.1 eV and 0.25 eV, which are very sim ilar to the CH3 fragm ent energies of 1.15 eV and 0.31 eV in the I and I* channels for direct neutral dissociation of gas-phase C H 3 I . The asym m etry of th e bim odal signal at 20 L coverage indicates th a t there are also particles arriving in a broad energy distribution. It is likely th a t this broad distribution is CH3 fragm ents from C T-PD IS. In contrast to this study CH3I adsorbed on A g ( lll ) did not show CH3 yield at 248 nm for less th an 1 ML [19]. For low coverages above 1 ML the CH3 yields were a ttrib u te d to a m ixture of CT-PD IS and direct photodissociation. This is evidence th a t C ifsl on C u(llO ) is less strongly quenched th an C H 3 I on A g ( lll ) , at least in the first layer. This is consistent w ith th e observation of th e higher T P D desorption tem peratures for GH 3 I on A g ( lll ) and th e characterization of CLfsI as chem isorbed on 94 hv + CHgTCu(110 ) -* 160x1 [ f N , 140 120 100 V |g o w i% a k | 80- 40 IFas; Peak 20 - Figure 4.23: The slow and fast peak counts in the TOF spectra of figure 4.22 are summed and plotted as a function of dose. th a t surface [44, 19]. The first layer of C H 3 I on A g ( lll) desorbs at ~ 190/T as com pared to 140 K for CH3I on Cu(110)[44]. M ultilayers on both surfaces desorb at sim ilar tem peratures - 136 [44]. Further understanding of the A = 337.1 nm T O F dynam ics of th e C i/ 3l/C u ( 110 ) system can be achieved by comparison to the T O F spectra of C H sB r/C u { llO ) and of C f 73l/C u ( 110 )-I. T he A = 337.In m T O F of CH^Br on clean C u(llO ) as a function of coverage was shown in figure 4.8. The sharp bim odal CH 3 signal, indicative of direct pho­ todissociation processes and evident in the C // 3l/C u ( 110 ) and th e C % I /C u ( 110 )-I systems for coverages greater th an 1 ML, do not appear at any coverage for CH 3B r/C u ( 110 ). The dissociation of the adsorbed CHsBr was a ttrib u te d to charge transfer processes. T he CH 3 signal was the result of intra-adsorbate dissociation of adsorbed CH 3.Br producing a fast distribution for tim es less th a n llO/zs. The CH 3 signal at tim es greater th an 1 1 0 / i s were as­ signed to m olecular desorption of C H sB r resulting from charge transfer processes. Charge transfer processes can result in desorption via th e Antoniewicz m echanism.(See the first 95 2000 1500 - 20LCH3l/Cu(110) 3 scans #3 O O 1000 f ü 500TLCHgl/OXIIO) 6 scans ' ' ' I ' ' ' ' ' ' ' I I 40 g 80 jI 1t i a I s _i. 120pS Time of Flight (ps) Figure 4.24: High Resolution Scans on C % 4/C u(110) at 7 L and 20 L. The counts in these scans were summed in Ifis MCS-TOF bins. The scans demonstrate th a t for less than 1 ML the CH 3+ signal from the fast distribution is not sharply peaked. At 2 ML however the CH 3+ signal from the fast distribution is bimodal, asymmetrical and sharply peaked. Significant Point A B C D E F G Time(/i s) 12 (eV) 3.7 22 1.1 30 42 113 300 800 0.59 0.30 0.04 0.006 0.0008 E chs I (eV) 35.2 10.5 5.6 2.9 0.4 0.056 0.008 Table 4.4: Significant times on the TOF spectrum for C H 3 I/Cu{110) in figure 4.25. The energies of C H 3 (15 amu) fragments and C H 3 I (127amu) molecules corresponding to those times are calculated with E — 96 1 4 0 0 p -i 1200 - I I I [ "I | - ' r " i - | '" r T ' T ~ r - ; - ' T ~ i "i " ' T " [ ' T - T - r - i - r i i i i | i i i i | - ’i t "i h v + C H ,I /C u ( l 1 0 )-» C H j d ) 1000 800 - 600 400 200 - 0 Li I I I I I I I I I I I.J I n J -i I I 200 I I I 400 I I I I 1 . 1 .1 I I I 800 I I I I 800 Figure 4.25; Time-of-flight spectrum for 20 L of CH 3I on Cu(llO ). chapter for a description) The sim ilarities of th e C if 3l/C u ( 110 ) spectra to the C /f 3 B r/C u ( 110 ) spectra indicate th a t the slow distribution for tim es greater th an 120^s in th e A = 337.1 nm T O F sp ectra of the C iÎ 3l/C u ( 110 ) system is likely th e result of molecular desorption from charge transfer processes. The clean C u(llO ) w orkfunction is 4.48 eV [26] and on th e C H 3 I dosed sur­ face the m axim um workfunction decrease is A $ = —0.53 eV, therefore th e C H 3 I /C u { llQ ) workfunction exceeds the photon energy by 0.27 eV [6 ]. The m ost likely desorption and dissociation channels for th e C T-PD IS m echanism will be neutral, since photoelectron at­ tachm ent resonances to gas-phase CH3I is far below the photon energy at 0.15 eV [33]. It is unlikely th a t direct neutral photoabsorption is causing m olecular desorption in the C H 3 IICu{110) system . T he absence of m olecular desorption in th e A = 337.In m TO Fs of C i 73l/C u ( 110 )-I indicates th a t m olecular desorption due to direct photoabsorption is an ineffective process on this system and likely for C iÏ 3l/C u ( 110 ) also. T he m olecular desorp­ tion counts are for all coverages m uch larger th a n th e intra-adsorbate dissociation counts in the T O F spectra, however, due to detection efficiencies this is not a reliable indication of the num bers of fragm ents in each distribution. Normalizing th e spectra by y, where t is tim e, would significantly enhance the faster intra-adsorbate dissociation yields. T here is also an additional factor to consider. T he ionization efficiencies of C H 3 I molecules and C H 3 fragm ents are different and have not been quantified. The yields from C T-PD IS are m ost significant in th e first and second layers. The decrease in the yield of the fast peak after th e second layer is com plete indicates th a t C T-PD IS and 97 direct photodissociation are decreasing. T he percentage of th e yield a ttrib u te d to C T-PD IS is about the same in all layers.(See the section on I* Branching R atio and C T-PD IS. ) The yield of molecular desorption however are nearly as large in the th ird layer as they are in the second. The yields for the A = 337.In m CH 3B r/C u ( 110 ) in figure 4.9 as a function of coverage showed th a t molecular desorption also rem ains large at coverages for where the intra-adsorbate dissociation, due to C T-PD IS, are decreasing. The photodynam ics on this system is entirely a ttrib u tio n to charge transfer processes. The m ost likely desorption process to explain these effects in b o th CH^l and CH^Bv on C u(llO ) is photoejection. Photoejection is a m echanism where an overlayer molecule is ejected from the surface by caged ionic molecules in underlayers. W hen a photoelectron attaches to a C H 3 X { X = l or X = Br) molecule, the C-X bond distance increases and at the same tim e the negatively charged molecule a ttra cts an overlayer molecule. Dissociation is halted in the caged molecule when the electronic excitation is quenched, however if the translation energy of th e overlayer C H 3 X molecule is sufficient it can overcome th e van der Waals forces th a t bind it to the surface. In the CH 3l /C u ( 110 )-I system hot photoelectrons are unavailable so ionic photoejection is not possible. However th ere are also n eu tral pho­ toejection processes where overlayer molecules are ejected as a result of excited underlayer molecules. The fact th a t molecular desorption are not evident in th e CH 3l/C u ( 110 )-I system indicates th a t a neutral photoejection process is not as significant. 4.6.2 A n g u la r D e p e n d e n c y E x p e rim e n ts Figures 4.26 and 4.27 show the results of th e angular experim ents a t a coverage of 20L in th e [110] azim uth and the [001] azim uth. The angular dependency was sim ilar for all coverages. Angular experim ents show th a t the T O F CH^ signal is m axim ized by having th e surface norm al at 0 ° to the detector in both th e [110 ] azim uth and the [001 ] azim uth. B oth CT-PD IS and direct photodissociation are fast processes, faster th an m olecular rotation, so the direction of the dissociating C H z fragm ent is indicative of the instantaneous bond direction before dissociation. The fact th a t th e yields are largest w ith th e surface norm al at 0 ° to the detector in both azim uths indicates th a t th e m olecular C-I bond axis are predom inantly norm al to the surface and therefore intra-adsorbate dissociation and m olecular desorption are b oth preferentially norm al to th e surface. The distribution of bond directions is quite broad in this system however. In th e C i 73l/C u ( 110 )-I system the largest angle where significant yield was obtained in b o th azim uths was 40° off-normal. (See figure 4.16.) In contrast the C H 3 l/C u ( 110 ) system showed significant yields at larger angles than 40° in both the azim uths for both th e fast and th e slow peaks. This larger distribution of bond angles can be explained by greater disorder in this system th a n for the C JÏ 3l/C u ( 110 )-I system , where bonds showed a preference for tiltin g at 20° in th e [110] azim uth. More disorder in this system also explains the sm aller num ber of counts in the T O F spectra. 4.6.3 I* Branching Ratio and CT-PDIS The fitting function,equation 4.11, th a t was used to deconvolve th e peaks in th e Cfif3 l/C u ( 110 )I system , was also used to deconvolve th e three dissociation channels in the fast distribution 98 1 . 2 -, hv + CH]I/Cu(l 10) -> CH](g) 1. 0 - @ [ 1 1 0 ] A zim uth A [001] Azimuth 0 .8 c 3 CD CD C. CD to X U 0.2 - 0.0 -20 0 40 20 GO 80 D e t e c t i o n A n gl e(d eg ree s) Figure 4.26: Fast Peak Counts as a function of Angle for 20L on C iÎ 3l/C u ( 110 ) 99 hv + CH]DCü(l 10) -> CHgdg) @ [ 1 1 0 ] Azim uth A [DOT] A z i m u t h 0. 8 CO •M c 3 € (D 0 .6 - CD C DO 0) u 0.4 0.2 0.0 -20 0 20 40 D etection A ngle(degrees) GO 80 Figure 4.27; Slow Peaks Counts as a function of Angle for 20L on C iÏ 3l/C u ( 110 ) 100 in the C % I/C u (1 1 0 ) system. The counts in the I, th e /*, and the CT-PD IS channels were separated using th e fitting function for tim es less th an llOps. The fitting function did not produce good fits to the slow photodesorption peak. An exam ple of a fit is shown in fig­ ure 4.28. Using th e counts in the I and I* channels, the 4>* ratio of was estim ated. The values for * were found to be in the range 0.4-0.6 , bu t w ith m ore uncertainty th a n for C % I/C u(110)-I [6]. As a result it is not clear th a t th e cf)* ratio for neutral dissociation of CH 3 I on clean C u(llO ) is different th an on Cu(110)-I, although it is clear th a t the (f>* value is larger than th e 0.1 value at 333 nm in th e gas-phase [6 ]. Large values for cf)* have also been found for C H 3 I on the m etal surface A g ( lll ) [19]. The * =0.40 and cf)* was larger for lower coverages [19]. The large cf)* values in layers near th e surface for CH 3 I on C u (llO ), Cu(110)-I and A g ( lll) suggest an addition possibility th a t the I* channel is less efficiently quenched th a n the I channel. Figure 4.29 shows th e estim ated CT-PD IS yield in the fast peak as a function of coverage. The fits indicate th a t the proportional yield of C T-PD IS is significant and varied w ith coverage from 20 to 60 % [6 ]. This graph is interesting because th e CT-PDIS percentage of the yield apparently does not decrease as the dose is increased. This is contrary to w hat we would expect the top adsorbate layer com pletely covers th e underlayers. The decrease of hot electrons through increasing layers of adsorbed molecules has been noted in previous studies [37]. We would expect th a t as additional layers are added, th a t charge transfer would become a less effective process and more of the signal would result from direct photodissociation. In figure 4.29, the efficiency of CT-PD IS, in com parison to direct photodissociation, appears to be as high in 1 ML as it is in ~ 4 layers (40 L). Due to th e large error in th e d ata, this appearance m ay not be correct. If it is correct this is additional evidence th a t th e top layer does not com pletely cover the underlayers. In th a t case th e exposed underlayers could continue to dissociate by the charge transfer m echanism. 4.6.4 C ro ss-sec tio n M e a su re m e n t The total depletion cross-section of C H 3 I jC u {l\Q ) includes C T-PD IS, neutral photodisso­ ciation, and photodesorption of C H 3 I. It also possible th a t th e to ta l cross-section includes processes th a t cannot be m easured, such as m ethyl fragm ents bonding to the surface after dissociation. Figures 4.30 to 4.35 show th e logarithm of th e depletion m easurem ents as a function of total photons for various coverages. The depletion m easurem ents have been done with respect to the fast distribution of C H 3 fragm ents and th e slow distribution of C H 3 I m olecular desorption. The depletion m easurem ents for 9 L and 20 L on a sem i-logarithm ic scale are generally linear as a function of 6 , the to ta l photons on th e sample. A t 9 L the cross-sections as determ ined from the slope give 8.2 x 10“ ^°cm^ and 9.1 x 10~^°cm^ for th e fast and the slow peaks of the T O F spectra. At 20 L th e cross-sections are determ ined to be 2.2 X 10~^^cm^ and 2.4 x 10~^®cm^ for the fast and th e slow peaks. Since the m easurem ents of the m olecular desorption and the intra-adsorbate dissociation are done on the sam e system s, the to tal cross-sections are similar as they are expected to be. The cross-section determ ined at 20 L is significantly larger than th e gas-phase cross-section at th e sam e wavelength and is also larger th a n the cross-section for C H 3 I/C u {11 0) —I. It is clear th a t photodissociation is significantly enhanced although it is not clear th e enhancem ent can be a ttrib u te d to either 101 b) , 1000P @ 800 g ü f O 400 50 100 200ps 150 Time of Flight (ps) Figure 4.28: A example fit to the spectrum of C % I/C u ( 110) system using the fitting function, equation 4.11. Graph from Johnson and Jensen [6]. I' I • I I I I I I I 1 I ] I"I I I I I ' I I ■4' § 0.8 k E o o 0.6 - j I . . 0. 5 0.2 k 10 20 30 40 50 CH3I Dose (L) 60 Figure 4.29: CT-PDIS proportion as a function of coverage of the fast peak on the TOFs C F 3l/C u ( 110 ) system. The estimates were done using the fitting function, equation 4 . 11 . Graph from Johnson and Jensen [6]. 102 charge transfer or to direct neutral photodissociation. We do not have enough inform ation to separate the to tal cross-section into individual cross-sections. At 40 L the yields as a function 0 on a sem i-logarithm ic scale show rapid decrease in th e yields and non-linearity at low photon num bers. This was a typical result for doses over 20 L. A possible explanation of the rapid decrease is that the coverage of the top layer is insufficient to completely cover the underlayers. At 40 L it is possible th a t th e th ird and possibly second layers have been left somewhat exposed. However th e slope is m easured to be 4.80 X and 4.85 x respectively from the fast and the slow distributions. This is 2 x as large as th e cross-section determ ined at 20 L on th e sam e system . It also possible th a t reordering on the surface is responsible for th e rapid depletion. In any event the non-linearity of the graph make estim ating the cross-sections from th e slope a doubtful process. The average of the slope may have little to do w ith th e cross-section and m ore to do w ith site depletion a n d /o r reordering effects. 4.6.5 F u rth e r D iscussion The photodesorption of CH 3 I molecules from the C /f 3l/C u ( 110 ) system is a ttrib u te d to th e charge transfer m echanism, since photodesorption of CH3I did not occur in the C % I/C u (1 1 0 )I system. Photodesorption due to charge transfer was observed on th e C % B r/C u (1 1 0 ) system also. The lack of direct PDIS in the first layer of C i/ 3l/C u ( 110 ) is likely due to fast quenching of the electronic excitation. CT-PD IS is a faster process th a n direct neutral photodissoci­ ation and is therefore b e tte r able to com pete w ith th e fast surface quenching. T he reason CT-PDIS of C H 3I is a faster process th an direct neutral photodissociation was dem onstrated by the potential energy surfaces in figure 4.3. Direct neutral photodissociation of ground state CH3I results from a Franck-Condon transition to a higher energy dissociative poten­ tial energy surface where th e C- I bond lengthens. Quenching can result in a tran sitio n from the excited state back to the ground state potential well. C T-PD IS also results in a Franck-Condon transition to a negative ion potential energy dissociative state bu t C-I bond lengthening quickly results in the dissociative state being at a lower potential energy th an the neutral ground state [6 ]. Quenching processes are ineffective a t this point. In order for the molecule to retu rn to the ground state, th e m olecule’s kinetic energy would have to be turned into potential energy say as th e result of a collision. T he overall tim e for direct dissociation of the CH3I molecule is 50 fs, and therefore th e n eu tral/io n ic curve crossing is reached in a fraction of the dissociation tim e. For adsorbed CH3I the dielectric prop­ erties of the surface result in image charge stabilization th a t lower th e potential energy of th e CT-PD IS dissociative curve and therefore make th e dissociation process more rapid [6 ]. Therefore CT-PDIS is m ore likely th an direct photodissociation to be able to com pete w ith quenching and cause dissociation of CH3I in direct contact w ith th e m etal su b strate when th e total coverage is less th an a single layer. The lack of direct PD IS signal from the first ML support this interpretation. W hen coverage is greater th a n a single layer th e efficiency of quenching is reduced largely because the propagation of electrons through the adsorbed layers to the surface is less efficient. As a result direct dissociative processes are b e tte r able to com pete with quenching. Therefore in the second layer of C iJ 3 l/C u ( 110 ) we see the two characteristic sharp peaks of the direct I and I* channel dissociation processes. 103 0.0 - 5 9LCH31/Cu(110) 0.2 -0.4 0 =8.2 xlQ-^cm' 3 - 0.6 - 0.8 - 1.0 16x10 Total Accumulated Photons/cm^ Figure 4.30: Yields from the fast distribution as a function of total number of photons for a dose of 9.0 L 0.0 - 0.2 -0.4 - 0.6 - 0.8 - 1.0 3 0=9.1 xlO cm" 16x10 Total Accumulated Photons/cm^ F ig u re 4.31: Yields from the slow distribution as a function of to tal number of photons for a dose of 9.0 L 104 0.0 20L CH3I/Cu(110) 0.2 - -0.4 0=2.15 xlO ^cm z - 0.6 - 0.8 4x10 Total Accumulated Photons/cm^ Figure 4.32; Yields from the fast distribution as a function of total number of photons for a dose of 20.0 L 0.0 20L CH 3I/C u(110) - 0.2 -0.4 0 =2.4 x 10' cm I 3 - 0.6 - 0.8 4x10 Total Accumulated Photons/cm^ F ig u re 4.33: Yields from the slow distribution as a function of to tal number of photons for a dose of 20.0 L 105 0.0 40L CH3I/Cu(110) -0.5 - 1.0 slope = 4.80 X 10' cm -1.5 - 2.0 -2.5 10x10 Total Accumulated Laser Pulses/cm^ Figure 4.34: Yields from the fast distribution as a function of total number of photons for a dose of 40 L. The non-linearity of this graph is making it difficult to interpret the slope as the cross-section. 0.0 -0.5 - 40L CH3I/Cu(110) 1.0 s lo p e = 4 .8 5 X 10 ' cm 5 -1.5 - 2.0 - 10x10 Total Accumulated Laser Pulses/cm^ F igure 4.35: Yields from the slow distribution as a function of to tal number of photons for a dose of 40 L. The graph is again non-linear. 106 4.7 Conclusions The study of the C H 3 I on C u(llO ) and Cu(110)-I surfaces has shown some unusual proper­ ties. M easurem ents of the cross-sections show th a t photodissociation is greatly enhanced in the second layer at A = 337 nm com pared to gas-phase studies. On th e Cu(110)-I surface the enhancem ent is solely due to direct n eu tral photodissociation. O n clean C u (llO ), the enhancem ent was a com bination of direct photodissociation and of C T-PD IS, where hot electrons tunnel into dissociative states of th e adsorbed CH3I molecules. T he enhancem ent on this system cannot be attrib u te d to charge transfer or to direct n eu tral photodisso­ ciation since we do not have enough inform ation to separate th e to ta l cross-section into individual cross-sections. T he presence of C T-PD IS for subm onolayer coverage of GH3I on clean C u(llO ) and the lack of direct photodissociation for subm onolayer coverage o f CH3I on Cu(110)-I indicates th a t C T-PD IS is a faster dissociation process and b e tte r able to com pete with quenching. CH3 yields on b o th surfaces were found to peak on com pletion of the second layer and to decrease significantly as th e coverages were increased p a st two layers. The cf)* = N * / { N -f N*) branching ratio was found to be altered at all coverages in both systems from the A = 333 nm gas phase value of 0* = 0.1. This branching ratio indicates more fast CH3 fragm ents from dissociation of adsorbed CH3I were produced from the I* dissociation channel. A dsorption of CH3I is stru ctu rally different depending on the substrate;Cu(110) or C u(llG )-I. CÆ3I adsorbed on Cu(110)-I is found to have a preference for tilting ~ 20° off-normal in th e [1Î0] azim uth. CH3I adsorbed on C u (llO ) showed a preference for orientating th e C-I bond norm al to th e surface, although th e counts fell off significantly more slowly from this system as th e angle of th e surface was increased th a n for the C .% I/C u(110)-I system . The difference is a ttrib u te d to th e greater o rientational or­ dering th a t occurs on th e iodided C u(llO ) surface. There was some evidence th a t overlayer adsorbate layers do not com pletely cover underlayers. T he depletion yield m easurem ents as a function of total photons often showed significant rapid non-linearity at very low photon num bers when the coverage was greater th a n two layers. It is possible th a t exposed sites were rapidly depleting. On the iodided surface, th e charge transfer p o rtio n of th e CH3I dissociation signal was constant as a function of coverage although th e error is large. This was additional evidence th a t th e underlayers were left som ew hat exposed. T he A = 337 nm photolysis of C iïsB r on C u(llO ) and Cu(110)-I surfaces was used to further understand th e C % I/C u (1 1 0 ) and C % I/C u (1 1 0 )-I system s. C i/a B r on C u (llO ) was found to dissociate and desorb by th e charge transfer m echanism b u t C H 3 BT on Cu(110)-I did not dissociate or desorb. T he m ost likely explanation of this effect is th a t hot electrons are unable to tunnel through a C ul barrier when th e photon wavelength is A = 337 nm. T he workfunction of Cu(110)-I surface was found to have increased by 1.2 eV as com pared to th e clean C u(llO ) surface. The direct n eu tral dissociative states of C % B r on th e surface are not accessible at this wavelength. 107 B ibliography [1] Robert A. A lberty and R obert J. Silbey. Physical Chemistry. Jo h n W iley & Sons, Inc., New York, 1992. [2] Neil W. Ashcroft and N.David M ermin. Solid State Physics. CBS Publishing Asia LTD, Hong Kong, 1976. [3] Phaedon Avouris and R obert E. W alkup. Fundam ental M echanism s of D esorption and Fragm entation Induced by Electronic Transitions at Surfaces. Annu. Rev. Phys Chem, 40:173, 1989. [4] L.F. Brus and V .F. Bondybey. Cage Effects and Steric hindrance in van der Waals solids, with application... Journal of Chemical Physics, 65(1):73, 1976. [5] C.C. Johnson and E .T . Jensen. Photodissociation of CH 3I on Cu(110)T: Surface Effects in Dissociation Pathw ays. Surface Science, 451:261, 2000. [6 ] C.C.Johnson and E .T .Jensen. Photolysis of CH 3I on C u(llO ) at 337nm: D irect and charge-transfer photodissociation. Journal o f Chemical Physics, 112(13):5919, 2000. [7] C.French and I. Harrison. O rientation and decom position kinetics of m ethyl iodide on P t ( l l l ) . Surface Science, 342:85, 1995. [8 ] C.French and I. H arrison. Site preferences, energetics and orientation of C H ^ B r on P t ( l l l ) : investigation by RAIRS, T P D and photofragm ent angular distributions. S ur­ face Science, 387:11, 1997. [9] Chao-MIng Chiang, T im H.Wentziaff, and B rian F. B ent. lodom ethane D ecom position on C u(llO ): Surface Reactions of Ci Fragm ents. Journal o f Chemical Physics, 96:1836, 1992. [10] St.John Dixon-W arren. Photochem istry of adsorbed molecules. XII. Photoinduced ionmolecule reactions at a m etal surface for C i / 3X /R C L /A g (lll)(X = B r,I). Journal of Chemical Physics, 98(7):5954, 1992. [11] D .J. Donaldson, V.Vaida, and R. N am aan. U ltraviolet absorption spectroscopy of dissociating molecules :Effect s of cluster form ation on th e photodissociation of C H sI. Journal of Chemical Physics, 87:2522, 1987. [12] Andre T. Eppink and David H. Parker. M ethyl iodide A-band decom position study by photofragm ent velocity imaging, jcp, 109(12):4758, 1998. 108 [13] D.Howard Fairbrother, K. A. Briggman, Eric W eitz, and P.C. Stair. U ltra-V iolet pho­ todissociation dynamics of m ethyl iodide at 333nm. Journal o f Chemical Physics, 101(5):3787, 1994. [14] M.W. Finnis and V. Heine. . J. Phys. F, 4:L37, 1974. [15] A. Gedanken and M.D. Rowe. . cpl, 34:39, 1975. [16] Heinz H erm ann and Stephen R. Leone. P hotofragm entation dynam ics of C H 3 I at 248nm and 266nm: V ibrational distributions in th e C H ^{v 2 ) ’’um brella mode. Journal of Chemical Physics, 76(10):4766, 1982. [17] N.Camillone III, K.A. K han, P.J. Lasky, L. Wu, J.E . Moryl, and Jr. R.M.Osgood. The wavelength dependence of photoinduced hot electron dissociative attachm ent to m ethyl bromide adsorbed on gallium arsenide(llO ). Journal o f Chemical Physics, 109(18):8045, 1998. [18] J.E.Inglesheld. Electronic structure of m etal surfaces. In Physics o f Solids and Liquids, volume 5 of Interaction o f Atoms and Molecules with Solid Surfaces, pages 117-153. Plenum Press, New York, 1990. [19] E.T. Jensen and J.C . Polanyi. Dynamics of Photodissociation of CH 3l / A g ( l l l ) at 248nm. Journal o f Physical Chemistry, 97:2257, 1993. [20] A. K ahn. . 3:193, 1983. [21] Takemi Kawaguchi, M asaya Hijikigawa, Yoshinori Hayafuji, Mas as hi Ikeda, Reizo Fukushim a, and Yujiro Tomiie. The C rystal S tructure of M ethyl Bromide and M ethyl Iodide. Bulletin o f the Chemical Society o f Japan, 46(53-56):53, 1972. [22] A. K iejna and K .F. Wojciechowski. . Prog. Surf. Sci., 11:293, 1981. [23] C.L.A. Lam ont, H. Conrad, and A.M. Bradshaw. T he w avelength dépendance of the photodissociation cross-section of C H ^ B r chem isorbed on C u l l l . Surface Science, 280:79-90, 1993. [24] J. Lin and B.E. Bent. lodom ethane dissociation on C u ( lll) . TecAnoIog;/. A, 10(4):2202, 1992. Journal o f Vacuum [25] L.Poth, Q. Zhong, J.V . Ford, and A.W . C astlem an Jr. Fem tosecond dissociation dy­ namics of m ethyl iodide clusters. Journal o f Chemical Physics, 109(12):4791, 1998. [26] CRC H andbook of C hem istry and Physics 6 U^ edition. Chemical Rubber Publishing Company. CRC Press, Florida, 1980. [27] R.L. P ark and M.L. den Boer. Chemistry and Physics o f Solid Surfaces. Cleveland, CRC, 1977. [28] D.R. Penn. . Phy. Rev B, 13:5248, 1976. 109 [29] T.N. Rhodin and J.W . Gadzuk. The Nature o f the Surface Chemical Bonds. A m ster­ dam, N orth-Holland, 1979. [30] B. Roop, Y.Zhou, Z.M. Liu, M.A. Henderson, K.G. Lloyd, A. Cam pion, and J.M . W hite. Surface Photochem istry. IX. Photochem istry of small molecules on m etal sur­ faces. Journal o f Vac. Sci. Technol. A, 7(3):2121, 1989. [31] G.A. Somorjai. Chemistry in Two Dimensions: Surfaces. Ithaca, Cornell, 1981. [32] S.R.Coon, K.B. Myli, and V.H. Grassian. P hotoproduct C haracterization and D y­ namics in the 248nm Photolysis of C H 3 I T hin Film s on A g ( lll) . Journal o f Physical Chemistry, 99:16416, 1995. [33] J.A . Stockdale, F .J. Davis, R.N. C om pton, and C.E. Klots. Physics, 60:4279, 1974. . Journal o f Chemical [34] Toshinori Suzuki, H ideto Kanam ori, and Eizi H irota. Infrared diode laser stu d y of the 248nm photodissociation of C H 3 I . Journal o f Chemical Physics, 94(10):6607, 1991. [35] J.C. TuUy. . 111:461, 1981. [36] V.A. U kraintsev and I. Harrison. Photochem ical hydrogenation of m ethyl radicals on P t ( l l l ) . Journal o f Chemical Physics, 98:5971, 1993. [37] V.A. Ukraintsev, T .J. Long, T. Gowl, and I. H arrison. Photoinduced dissociative electron attachm ent of C H ^ B r on P t ( l l l ) : The role of the local work function. Journal o f Chemical Physics, 96:12, 1992. [38] G.N.A. Van Veen, T. Bailer, and A.E. De Vries. Photofragm entation of C H ^ B r in th e A Band. Journal o f Chemical Physics, 92:59, 1985. [39] G abriela C. G. Waschewsky, R obert Horanksky, and Veronica Vaida. Effect of Dimers on the Tem perature-dependent A dsorption Cross Section of M ethyl Iodide. Journal of Physical Chemistry, 100:11559, 1996. [40] M artin Wolf. Femtosecond dynam ics of electronic excitations at m etal. Surface Science, 377-379:343, 1997. [41] D.P. Woodruff and T.A. Delchar. Modern Techniques o f Surface Science, 2nd Ed. C am ­ bridge Univ. Press, Cam bridge, 1994. [42] Satoshi Y abushita and Keiji M orokum a. P otential Energy Surfaces for R otational E xcitation of C H 3 product in Photodissociation of C H 3 I. cpl, 153(6):517, 1988. [43] A. Zangwill. Physics at Surfaces. Cam bridge U niversity Press, Cam bridge, 1988. [44] X.-L. Zhou, X.-Y. Zhu, and J.M . W hite. Photochem istry at ad so rb ate/m etal interfaces. Surface Science Reports, 13:73, 1991. [45] Frank M. Zim m erm ann and W. Ho. S tate Resolved studies of photochem ical dynam ics a t surfaces. Surface Science Reports, 22(4-6): 127-248, 1995. 110 [46] Frank M. Zim m erm ann and W.Ho. Velocity distributions of Photodesorbed Molecules. Journal of Chemical Physics, 100:7700, 1994. Il l