Organic Photovoltaic Films:The Commissioning of and Preliminary Measurementson an Organic Molecular Beam Epitaxy SystembyAlexandra TullyB.Sc., Yale University, 2014A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinThe Faculty of Graduate and Postdoctoral Studies(Physics)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2019© Alexandra Tully 2019The following individuals certify that they have read, and recommend tothe Faculty of Graduate and Postdoctoral Studies for acceptance, the thesisentitled:Organic Photovoltaic Films: The Commissioning of and Pre-liminary Measurements on an Organic Molecular Beam Epi-taxy Systemsubmitted by Alexandra Tully in partial fulfillment of the requirementsfor the degree of Master of Science in Physics.Examining Committee:Sarah Burke, Physics and AstronomySupervisorDavid Jones, Physics and AstronomySupervisory Committee MemberiiAbstractOrganic Photovoltaics (OPVs) may provide a means of achieving flexibleand transparent solar cells, comprised of inexpensive materials and createdthrough scalable processes. Compared to today’s dominant silicon-basedsolar cells, OPVs suffer from lower power conversion efficiency, and a prin-cipal barrier to efficient power conversion in OPVs lies in the separationof generated charges. In OPVs, photoabsorption results in a coulombicallybound exciton; in order to generate free charges, we must engineer excitondissociation. Thus, an understanding of the dynamics involved in excitondissociation and the underlying electronic states that drive this separationis requisite to increasing the power conversion efficiency and developingcommercially-viable OPV devices. In order to do this, we intend to mapthe energy landscapes of the system on a femto- to picosecond timescale, aswell as an A˚ngstro¨m length scale. To facilitate this mapping, we will usea combination of time- and angle-resolved photoemission spectroscopy (tr-ARPES) and scanning tunneling microscopy (STM) and scanning tunnelingspectroscopy (STS) to analyze our films. Using a femtosecond pump-probescheme, TR-ARPES measures the dynamic spectral properties of a systemby monitoring a material’s electronic states after excitation. STM/S pro-vides local information on the electronic structure, including both occupiedand unoccupied states. Combined, these measurements will facilitate theunderstanding of energy level alignment, the band structure of the system,and the evolution of the excited states. Because of the quality and purityrequirements for the samples, as well as the fragility of organic thin films,we must grow our films in-situ in a UHV environment. Over the past twoyears, we have designed and commissioned an organic molecular beam epi-taxy (OMBE) growth chamber as well a home-built low energy electrondiffraction (LEED) characterization chamber that is attached to an ARPESsystem. This thesis discusses the motivation and background informationfor this project in further detail, presents the experimental techniques re-quired to understand and operate the OMBE and LEED chamber, describesthe commissioning process of the OMBE, and touches on our preliminarygrowth recipes and data acquisition.iiiLay SummaryOrganic photovoltaic (OPV) materials have the capability to fundamen-tally alter our engagement with solar energy: the devices these materialsenable can be thin, flexible, light-weight, semitransparent (transparent tovisible light), low-cost, and take relatively little energy to produce. How-ever, increases in efficiency are required before OPV devices can become aviable alternative to silicon-based solar cells. One of the principal causesof inefficiency is poor charge transfer, and we do not yet understand thephysics that underlie it. In order to probe the physics of charge transferin these materials, we intend to use a combination of scanning tunnelingmicroscopy (STM), scanning tunneling spectroscopy (STS), and time- andangle-resolved photoemission spectroscopy (TR-ARPES). Such analysis re-quires extremely pure OPV sample films, which we can grow in-house inour organic molecular-beam epitaxy growth chamber (OMBE). This thesisdiscusses the commissioning of our OMBE and our preliminary attempts atfilm growth and characterization.ivPrefaceAll commissioning of the custom ScientaOmicron Lab10 OMBE system wasperformed by me, Erik Ma˚rsell, and Sarah Burke.The characterization chamber and its support structure was designed, built,and commissioned by Erik Ma˚rsell and me. Aghigh Jalehdoost, a doctoralresearcher at the University of Freiburg, assisted in the initial design of thechamber.The argon line and attachment devices were designed and built by ErikMa˚rsell and me.I performed the majority of machining work, in the QMI machine shop, withthe assistance of Harish Gautam. As part of this thesis I have begun to learnhow to operate the CNC machine.All sample preparation and characterization was performed by Erik Ma˚rselland me.vTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 OPVs in Brief . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3 Project Objective . . . . . . . . . . . . . . . . . . . . . . . . 31.4 Thesis Objective . . . . . . . . . . . . . . . . . . . . . . . . . 42 Organic Photovoltaics . . . . . . . . . . . . . . . . . . . . . . . 62.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62.2 Mathematical Description of an Excitonic Solar Cell . . . . . 72.3 Excitonic Solar Cells vs. Conventional Solar Cells . . . . . . 82.4 Charge Separation in Excitonic Solar Cells . . . . . . . . . . 102.5 Efficiency in OPVs . . . . . . . . . . . . . . . . . . . . . . . 162.6 Types of OPV . . . . . . . . . . . . . . . . . . . . . . . . . . 202.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 Experimental Techniques . . . . . . . . . . . . . . . . . . . . . 263.1 Organic Molecular Beam Epitaxy . . . . . . . . . . . . . . . 263.1.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . 263.1.2 MBE . . . . . . . . . . . . . . . . . . . . . . . . . . . 27viTable of Contents3.1.3 Measurement Techniques . . . . . . . . . . . . . . . . 293.2 Low Energy Electron Diffraction . . . . . . . . . . . . . . . . 303.2.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . 303.2.2 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . 303.2.3 Electron Scattering in LEED . . . . . . . . . . . . . . 323.2.4 LEED Imaging and Apparatus . . . . . . . . . . . . . 373.2.5 The Ewald Sphere . . . . . . . . . . . . . . . . . . . . 403.2.6 Kinematic Theory . . . . . . . . . . . . . . . . . . . . 464 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . 514.1 The OMBE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514.1.1 Scienta Omicron System . . . . . . . . . . . . . . . . 514.1.2 Characterization Chamber . . . . . . . . . . . . . . . 614.2 Design and Construction . . . . . . . . . . . . . . . . . . . . 644.2.1 Support Structure for Characterization Chamber . . 644.2.2 Argon Line . . . . . . . . . . . . . . . . . . . . . . . . 664.2.3 Sample Plate . . . . . . . . . . . . . . . . . . . . . . . 674.3 Achieving and Maintaining UHV . . . . . . . . . . . . . . . . 724.3.1 Definition and Pumping System . . . . . . . . . . . . 724.3.2 Main Chamber Bake . . . . . . . . . . . . . . . . . . 744.3.3 Characterization Chamber Bake . . . . . . . . . . . . 755 Preliminary Data . . . . . . . . . . . . . . . . . . . . . . . . . . 775.1 Choice of Materials . . . . . . . . . . . . . . . . . . . . . . . 775.2 Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . 785.3 Preliminary Data . . . . . . . . . . . . . . . . . . . . . . . . 795.3.1 Au(001) . . . . . . . . . . . . . . . . . . . . . . . . . 795.3.2 Au(001) with C60 . . . . . . . . . . . . . . . . . . . . . 805.3.3 Au(111) . . . . . . . . . . . . . . . . . . . . . . . . . 816 Conclusion and Future Directions . . . . . . . . . . . . . . . 826.1 Conclusions and Open Questions . . . . . . . . . . . . . . . . 826.2 Future Directions . . . . . . . . . . . . . . . . . . . . . . . . 83Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84AppendicesA OMBE System Construction . . . . . . . . . . . . . . . . . . . 93A.1 Characterization Chamber Manipulator Stage Modifications 93viiTable of ContentsA.2 Characterization Chamber Support Structure Customization 94A.3 Argon Line Fastening System . . . . . . . . . . . . . . . . . . 95B Krypton and Xenon Recycler . . . . . . . . . . . . . . . . . . 96B.1 Background and Motivation . . . . . . . . . . . . . . . . . . 96B.2 System Design . . . . . . . . . . . . . . . . . . . . . . . . . . 96B.3 The Catch . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97B.4 The Purification System . . . . . . . . . . . . . . . . . . . . 98B.5 Flow Rate Calculations . . . . . . . . . . . . . . . . . . . . . 100viiiList of Figures1.1 Examples of current and near-future OPV applications: a.window at EPFL’s SwissTech Convention Center[7] b. an or-ganic solar window at Michigan State University, transparentto visible light[8] c. the Mercedes-Benz G-Code concept car,which utilizes paint based on organic photovoltaic technologyto harvest solar energy.[9] . . . . . . . . . . . . . . . . . . . . 12.1 Description of charge generation and transfer process in aPV. a. Photoabsorption: an incident photon excites an elec-tron from its ground state in the highest occupied molecularorbital (HOMO) to the lowest unoccupied molecular orbital(LUMO). b. Exciton formation: electron and hole are nowcoulombically bound, forming an exciton. c. Free charges: inany PV, the electron and hole must become free charges ableto reach the cathode and anode respectively. d. In a conven-tional solar cell (CSC), the exciton dissociates spontaneously,immediately generating free charges. e. In an excitonic solarcell (XSC), photoabsorption results in an exciton, the disso-ciation of which must be orchestrated. . . . . . . . . . . . . . 92.2 Exciton dissociation via a donor-acceptor heterojunction: a.Photoabsorption results in an exciton, straddling the HOMOand LUMO of the donor material. b. Exciton diffuses toD-A interface. c. Exciton enters the charge-transfer state.d. Exciton dissociates into free charges that provide chargetransport to electrodes. . . . . . . . . . . . . . . . . . . . . . 13ixList of Figures2.3 I − V curve of an OPV under illumination, depicting the fillfactor (FF) and the power curve. VOC and ISC are deter-mined by the makeup of the solar cell (the internal electricfield). Vmax and Imax are chosen to maximize the power out-put (Pmanx = Imax × Vmax) depending on the I − V curve.The I − V curve is determined by the internal resistance ofthe OPV; in a cell with no internal resistance, Vmax = VOC ,Imax = ISC , and the I−V curve would be a right angle. Thepower curve is given by P = I × V . . . . . . . . . . . . . . . . 182.4 Schematic of a basic dye-sensitized solar cell (DSSC) com-prised of a layer of organic dye adsorbed onto an electronacceptor along with a redox electrolyte to replenish electronsin the dye. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.5 Schematic of a bilayer OPV device comprised of layered elec-tron donor and electron acceptor materials, designed to facil-itate exciton dissociation via the heterojunction interface. . . 212.6 Schematic of a bilayer OPV device comprised of layered elec-tron donor and electron acceptor materials, designed to putan interface within the diffusion length of a greater percent-age of excitons. The electron blocking buffering layer preventselectrons from screening the anode; it collects and transportsholes to the anode while blocking electrons from reaching it. . 222.7 Two of the common acceptor molecules used in OPVs. C60 isthe molecule discussed most prominently in this thesis. Fig-ure adapted from K. A. Cochrane’s PhD Thesis.[51] . . . . . 223.1 Basic MBE chamber; no built-in characterization method. . . 283.2 A basic LEED setup: the electrons are emitted from an elec-tron gun and scatter off of the sample, the grid provides a re-tarding field that filters out in-elastically scattered electrons,elastically scattered electrons pass through the retarding fieldand hit the fluorescent detector screen, the image is then cap-tured by the camera. This process will be described in greaterdetail later in this section. . . . . . . . . . . . . . . . . . . . . 313.3 Example of a LEED spot pattern: Au(001) with electronbeam energies of a. 80 eV and b. 150 eV. (MCP set at480 V in both images. See Subsection 3.2.4 for explanationof an MCP.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32xList of Figures3.4 Illustration of how the observed LEED pattern depends onthe configuration of the LEED optics. Figure adapted fromArthur T. Hubbard’s The Handbook of Surface Imaging andVisualization, p. 291.[81] . . . . . . . . . . . . . . . . . . . . . 343.5 As incident electrons strike the sample surface (illustratedcoming in from the left and reflecting off to the right), theelectron that strikes the second layer of atoms travels 2d sin θA˚ further than the electron that strikes the top layer of atoms.When 2d sin θ is an integer multiple of the electron wavelength(nλ), constructive interference leads to Bragg peaks. When2d sin θ is not an integer multiple of the wavelength, destruc-tive interference means we detect nothing more than noise.Additionally, it’s important to note that because λ is a func-tion of the incident electron’s kinetic energy (λ = h√2mE),changing the energy of the electrons affects the diffractionspot positions. . . . . . . . . . . . . . . . . . . . . . . . . . . 353.6 Model of LEED setup for organics. . . . . . . . . . . . . . . . 393.7 An Ewald circle (a 1-D Ewald sphere). The circle has a ra-dius of 1/λ, where λ is the electron wavelength; its origin iscentered on the incident wave vector. The circle intersects aninfinite series of parallel lines with spacing 1a , where a is thelattice constant of the crystalline sample. . . . . . . . . . . . 413.8 Confirmation that the Ewald circle satisfies the Bragg condi-tion. Triangles are from the example Ewald circle in Figure3.7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.9 A 2-D Ewald sphere, intersecting with 2-D array of rods. . . . 423.10 A 3-D Ewald sphere, where the rods have collapsed to a 3-Darray of dots. . . . . . . . . . . . . . . . . . . . . . . . . . . . 433.11 A quasi-3-D Ewald sphere, with rods of varying intensity. . . 443.12 Ewald sphere representation of the effect of changing beamenergy on LEED spots. Changing the beam energy changesthe radius of the sphere (recall that λ = h√2mEand r = 1λ);this results in a different number of intensity spots intersect-ing the sphere, which causes a different scale of LEED pattern. 454.1 Scienta Omicron customized Lab10 OMBE system. Doesnot include our home-built characterization chamber. Fig-ure adapted from Scienta Omicron Lab10 system manual. . . 524.2 Solidworks mock-up of characterization chamber. . . . . . . . 53xiList of Figures4.3 Temperature of sample relative to temperature of baseplate(which is the temperature reported by MISTRAL). Graphprovided by Scienta Omicron (in Lab10 MBE system manual) 544.4 Sketch of QCS evaporator (effusion cell). Figure provided byMBE Komponenten (in QCS operating instructions manual). 554.5 Drawing of allowed shutter positions for our QCS. Figure pro-vided by MBE Komponenten (in QCS operating instructionsmanual). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564.6 Figure showing allowed shutter configurations for evapora-tors. Configurations represent crucibles open and exposed tothe chamber. . . . . . . . . . . . . . . . . . . . . . . . . . . . 574.7 Drawing of the transfer arm head, showing the process ofthe rotation of the internal shaft (controlled by rotation ofthe external magnet) leading to the opening and closing ofthe clamp on the sample plate. Figure provided by ScientaOmicron (in Lab10 MBE system manual). . . . . . . . . . . . 594.8 Our sample acceptor stage, or dock. This is our sample load-ing and storage dock, and it can hold up to 5 samples. Imagesprovided by Scienta Omicron (in Lab10 MBE system manual)showing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 614.9 Solidworks mock-up of characterization chamber with sup-port structure, attached to OMBE. . . . . . . . . . . . . . . . 624.10 The N300 non-evaporable getter – or NEG – is a pump thatfunctions by presenting a large, coated surface area into thechamber and using it to adsorb contaminant molecules. Im-age provided by Gamma Vacuum (in NEG manual). . . . . . 634.11 Solidworks drawing of manipulator in characterization cham-ber, with both a. side-on view and b. top-down view. . . . . 644.12 Solidworks mock-up of support structure for characterizationchamber, with adjustable means of fixture to OMBE frame.Note that the bracket on the side of the support with 4 holesis actually missing from our final frame; the vertical pieces of80-20 it joined were replaced with a single piece on each sideof the structure. . . . . . . . . . . . . . . . . . . . . . . . . . 65xiiList of Figures4.13 Sputter gun schematic. In our sputter gun, electrons areemitted from the cathode filament, which is kept at a -100V potential with respect to the anode cage. Argon is leakedinto the anode cage, where it is ionized by the electrons. Therepeller that surrounds the anode cage reflects electrons thathave passed through the anode cage without ionizing an ar-gon atom back into the anode cage, thereby increasing theionization efficiency of the sputter gun. The hole in the re-peller allows argon ions to escape into the chamber. Thisfigure is adapted from SPECS’ IQE 11/35 manual. . . . . . . 664.14 Photograph of the argon line and valve setup, with gas bottle. 674.15 Standard sample plate/holder mechanisms for a. ScientaOmicron systems and b. the custom Scienta ARPES sys-tem. In a, the sample is either clamped or epoxied to thesurface of the plate. In b, the sample is typically epoxied tothe top of the top hat shaped mount. . . . . . . . . . . . . . . 684.16 Simplified process of transferring the top hat sample postfrom the ARPES bullet to the OMBE double-decker sampleplate, to illustrate their compatibility. a. Insert the top hat(oriented with the flat sides of the brim parallel to the slot)into the slot on the double-decker sample plate. b. Slidethe top hat to the back of the slot. c. Twist top hat untilit wedges under the spring-style metal strips on the top ofthe double-decker sample plate. The flat sides of the brimshould be close to perpendicular to the slot. d. View of tophat aligned to be inserted into double-decker sample plate. e.View of top hat wedged under the spring-style metal stripson top of the double-decker sample plate. . . . . . . . . . . . 694.17 Molybdenum clamp for sample security. Made of .125 mmthick molybdenum foil, this clamp fits over the top hat crys-tal, with the brim of the top hat resting under the ring. The3.5 mm inner diameter is sufficient to allow the main crystalsample surface (the top of the top hat) to extend up throughthe inner circle and protrude above it. The post has a plat-form on top where the crystal sits. We take the spokes of thegear that extend radially outward in the above image, andbend them down around that portion of the crystal and post,clamping them below the underside of the post’s platform. . . 70xiiiList of Figures4.18 Figure showing the pump system for the main chamber (“MBEChamber”), the FEL chamber (“Fast Entry Lock”), and thecharacterization chamber (“LEED Chamber”). Schematicadapted from Scienta Omicron’s Lab10 MBE system manual. 734.19 Image of OMBE bakeout in progress. . . . . . . . . . . . . . . 744.20 The water manifold system for the OMBE. Image providedby Scienta Omicron (in Lab10 MBE system manual). . . . . . 755.1 LEED on Au(001). Example Bragg peaks are circled, withspot (1,0) marked in both a and b for reference. a. Beamenergy of 80 eV, MCP at 480 V. b. Beam energy of 150 eV,MCP at 480 V. . . . . . . . . . . . . . . . . . . . . . . . . . . 795.2 LEED on Au(001) pre- and post- C60 deposition. a. Beamenergy of 30 eV, MCP at 480 V, pre-deposition. b. Beamenergy of 30 eV, MCP at 480 V, post-deposition. Note thepresence of 2 additional spots due to the C60, located betweenthe 4 corner double spots of the Au(001). Two examples ofC60 spots are circled, with 4 others visible in the image. TheseC60 spots are first-order Bragg peaks (in the first Brillouinzone). c. Emphasizes the likely 2 domains of C60 representedin image b. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805.3 LEED on steppy Au(111) crystal at 132 eV, MCP at 480 V.The Bragg peaks are labeled. . . . . . . . . . . . . . . . . . . 81A.1 Solidworks drawing of manipulator rod with .35 mm cutout. . 93A.2 Custom clamp and brackets to hold and stabilize our char-acterization chamber. All construction of these pieces wasprimarily performed by me, with some assistance from Har-ish Gautam. a. Clamp clamp cut from a single block of316 stainless steel using a waterjet cutter, with tapped holesand polished using a combination of a grinding stone and aDremel. b. Aluminum brackets cut using a waterjet cutterand polished using a grinding stone. . . . . . . . . . . . . . . 94A.3 Fasteners for attaching the argon line to our system’s supportframe. The circular hole at the top of the fastener is whatattaches to the frame itself, and the elongated hole allows usto securely hang the argon line below the frame at a chosendistance from the frame. . . . . . . . . . . . . . . . . . . . . . 95A.4 Aluminum plates used to a. support the argon bottle and b.securely mount our argon line valves. . . . . . . . . . . . . . . 95xivList of FiguresB.1 Overall system design for Kr and Xe recycler. “The Catch”is the means of capturing the contaminated gas, the “LiquidNitrogen Trap” is the purification system, and the “KryptonGas Reservoirs” are the means of storing gas that facilitatesreuse. We commonly refer to a gas reservoir as the “insert”when it is not actively storing purified Kr or Xe, due to theremovable container’s role in the purification system; for prac-tical purposes, “insert” and “reservoir” can be treated as syn-onymous. Note that the “O2 Exhaust” line will likely requiresome sort of pump, the details of which have not yet beendecided. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97B.2 Schematic of ”The Catch” integrated into the vacuum cham-ber system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98B.3 Critical temperatures of the gasses relevant to our system. . . 98B.4 Figure of insert immersed in LN2 bath, with Kr freezing tothe inner walls, and liquid O2 pooling at the bottom. TheO2 would likely form a slush-like consistency that might fall,almost snow-like, through the gas mixture (there have beenseveral requests for the design of this system to incorporatetransparent containers so as to watch this process; sadly,transparent dewers and pressurized containers are very ex-pensive). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99B.5 Molar flow rate of Kr gas in our system for both a 150 µmand 100 µm diameter nozzle. . . . . . . . . . . . . . . . . . . 100xvGlossaryA˚ AngstromARPES Angle-resolved photoemission spectroscopyCT Charge transferCuPc Copper phthalocyanineeV Electron voltsFCC Face centered cubich¯ Plank constantHOMO Highest occupied molecular orbitalLEED Low energy electron diffractionLN2 Liquid nitrogenLUMO Lowest unoccupied molecular orbitalMBE Molecular beam epitaxyMO Molecular orbitalOMBE Organic molecular beam epitaxyOPV Organic photovoltaicsPTCDA (3,4,9,10)-perylenetetracarboxylic dianhydrideQCM Quartz crystal monitorRHEED Reflection high energy electron diffractionSTM Scanning tunneling microscopy/eSTS Scanning tunneling spectroscopyTSP Titanium sublimation pumpUHV Ultrahigh vacuumZnPc Zinc phthalocyaninexviAcknowledgementsSarah Burke and David Jones – I am immensely grateful for the opportunityyou have offered me. Thank you for teaching me, supporting me, and foralways finding time when there is none.Erik Ma˚rsell – Thank you for your endless patience with a never endingstream of questions. I honestly cannot imagine a better colleague to havehad over these past two years. Wherever you go next, I hope you continueto teach – your future students will be better for it.LAIR – You are the best mix of kind and competent, of helpful and driven.Each of you has made my time here happier, and I feel incredibly fortunateto be your colleague and friend.The Ultrafast Spectroscopy Lab – Thank you for teaching me about yoursubfield, for patiently explaining countless new concepts to me, and for beingwonderful people all the while.Deborah and Alan – You have done more for me than I could ever recount,repay, or even know. All of the adventures are thanks to you.Tim – You are the most unexpected and extraordinary turn of my life sofar. I love you, and I am grateful for your love every day. Thank you.xviiChapter 1Introduction1.1 MotivationA recent review in Nature described photovoltaics as “the most elegantdemonstration of renewable energy generation,” calling organic photovoltaics“arguably the most radical approach.”[1] While silicon-based solar cellscurrently dominate the photovoltaic industry,[2] organic photovoltaics, orOPVs, are nearing commercial-viability with companies like Heliatek[3] andRaynergy Teck[4] working on this technology alongside countless academicinstitutions and research groups throughout the globe.[5, 6]πd.e. f.a.b.c.Figure 1.1: Examples of current and near-future OPV applications: a. win-dow at EPFL’s SwissTech Convention Center[7] b. an organic solar windowat Michigan State University, transparent to visible light[8] c. the Mercedes-Benz G-Code concept car, which utilizes paint based on organic photovoltaictechnology to harvest solar energy.[9]11.2. OPVs in BriefAlthough silicon-based solar cells are efficient, stable, and increasinglycheap, they have several drawbacks: they require high-temperature process-ing with environmentally harmful materials, are heavy (at approximately 10kg/m2), and inflexible.[1] OPV devices can be created at room temperaturewhere plastic substrates are coated with a carbon-based thin films (on theorder of 100 nm thick) in a roll-to-roll process; these devices are thereforethin, lightweight (at approximately 500 g/m2), flexible, and inexpensive.Additionally, OPVs can be semi-transparent and of tuneable colour,[10] andperform better under lower light intensities and at higher temperatures.[11]The combination of these qualities enables a degree of adaptation and in-tegration of solar harvesting capabilities into common infrastructure, thatappeals to architects, commercial construction firms, the automobile indus-try, environmental policy groups, and your average homeowner alike.1.2 OPVs in BriefLike conventional silicon-based solar cells (CSCs), organic solar cells (abbre-viated as OSCs, but otherwise known as organic photovoltaics – OPVs) aredesigned to use photoabsorption (along with an applied voltage) to gener-ate electrical power. The process occurs when the absorbed photons exciteelectron-hole pairs called excitons; these electrons and holes are initiallycoulombically bound, but when separated, the excitons become free elec-trons and holes that form a current due to the internal electric field. Themanner in which this exciton dissociation occurs differs between types of so-lar cell, depending on the energetic structure of the materials. While excitondissociation occurs spontaneously in CSCs, in OPVs it must be engineered.This is because excitons in OPVs – called Frenkel excitons1 – have a bindingenergy of a few 100 meV to 1 eV,[12] which is too high to be overcome byan internal electric field or thermal energy. Because of this, OPVs are con-sidered excitonic solar cells (XSCs), and engineering them presents severaldifferent challenges.Over the past thirty years, much progress has been made in managingthese difficulties. We currently have OPVs based on both polymers andsmall molecules as donor material, with fullerene- and non-fullerene-basedacceptor materials. There are now dye-sensitized solar cells[13] and tandem1Conventional solar cell excitons are called Wannier excitons and have a binding energyof ∼10 meV, which is less than thermal energy at room temperature (kBT ≈ 25meV) andcan thus be considered negligible.21.3. Project Objectivesolar cells,[5] OPVs created using solution processing2 and using vacuumprocessing,3 and those created using both.[14] Despite this progress, how-ever, OPVs have not yet reached efficiencies competitive with conventionalsolar cells, and a better fundamental understanding of loss processes in OPVsis required to determine future possibilities of the field.1.3 Project ObjectiveEfficiency in an OPV comes down to absorbing photons, having those pho-tons excite electrons, separating bound electron-hole pairs into free chargesefficiently,4 and preventing those charges from recombining (either throughradiative recombination or non-radiative recombination at a charge trap ordefect). Mitigating the loss processes in this system requires that we un-derstand the microscopic processes that occur in these cells.[1] In order tobetter understand the charge transfer process (specifically, how the excitondisassociates), we must map the energy landscapes of the system on a femto-to picosecond timescale, as well as a length scale on the order of nanometersor A˚ngstro¨ms. This sort of mapping requires spectroscopic techniques thatprovide information on these time and length scales, either simultaneouslyor separately. Using a combination of time- and angle-resolved photoemis-sion spectroscopy (TR-ARPES), scanning tunnelling microscopy (STM) andscanning tunnelling spectroscopy (STS), we are able to map both the tem-poral and the spacial dynamics of electrons (both excitonic and unbound) inour thin films. TR-ARPES measures the energy and momentum of electronsas a function of time since excitation. Here, we use two ∼120 fs5 light pulsesin a pump-probe scheme, sending either a near-infrared or a visible6 pumppulse into our sample (exciting the electrons) followed by a variably-delayedUV pulse that photoionizes the excited state electrons. By analyzing theenergy and momentum of the emitted electrons, the dispersion and timedynamics of the occupied (excited) states can be measured. In other words,TR-ARPES enables us to excite an exciton, and then track the momentumas a function of time of the exciton’s electron.2Both polymer solar cells and (recently) small molecule solar cells.3Only small molecule solar cells.4We want to separate as many excitons as possible, and use as little energy as possibledoing so.51 femtosecond = 1× 10−15 seconds6We can use green, which is a doubled NIR pulse.31.4. Thesis ObjectiveWhere TR-ARPES monitors a material’s electronic dynamics after exci-tation, STM and STS provide local information on the electronic structure ofsingle-particle excitations (i.e. electrons and holes independently), includ-ing both occupied and unoccupied states. By setting a voltage differencebetween the (probing) tip and the sample of material, electrons will tunneleither from the sample to the tip or from the tip to the sample; this tun-neling allows us to probe the electronic states both above and below theFermi energy – including the lowest unoccupied molecular orbital (LUMO)or the highest occupied molecular orbital (HOMO), respectively. This typeof analysis more closely resembles the post-exciton, charge separated stateportion of charge dissociation. Combined, these measurements will facilitatean understanding of energy level alignment within the OPV material, theband structure of the system, and the evolution of the excited states; onlytogether do they create a spatial and temporal map of exciton dissociationand provide valuable insight into the charge transfer dynamics.1.4 Thesis ObjectiveAlthough the primary measurement tools for this larger project comprisea scanning tunneling microscope and a TR-ARPES machine, our organicmolecular-beam epitaxy growth chamber (OMBE) is an integral part of ouranalysis capabilities. The OMBE allows us to grow OPV films of inter-est, by controlled evaporation of the materials of choice (e.g. C60) onto asubstrate in ultra-high vacuum (UHV). The deposition rates achieved us-ing this technique facilitate epitaxial growth of the film (i.e. our depositedcrystalline molecular film will have a well-defined orientation with respectto the substrate). The combination of this deposition technique and theUHV environment (as well as our ability to control temperature) results inthe highest achievable purity of the grown films, with reproducible qual-ity and ordered structure, thereby enabling us to focus on the fundamentalphysics. Furthermore, our OMBE is attached to a TR-ARPES machine,facilitating direct UHV transfer from the growth chamber to the analysischamber. This is a prerequisite to our research, because our samples aresurface sensitive and become contaminated when they are exposed to theair. Additionally, our research requires multiple rounds of analysis; in orderto correlate results across analyses and samples, we must ensure the consis-tency of our samples. Finally, organic films are too fragile to withstand thecleaning techniques necessary to get an atomically clean surface (includingheating and sputtering with ions). It was critical, therefore, that we con-41.4. Thesis Objectivestruct a system which allows us to maintain a UHV environment for ourorganic thin films over the course of our TR-ARPES analysis. This thesisdiscusses the commissioning of the OMBE and the design of the attachmentchamber from the OMBE to the ARPES analyzer. It also covers our initialdata collection and characterization techniques.As seen in Chapter 1 of this thesis, our intended research of OPVs in-cludes multiple analysis techniques used to better understand the spatial andtemporal dynamics of electrons in OPV thin films, however a prerequisite tothis analysis is the ability to produce pure, crystalline, ordered, reproduciblethin films. Chapter 2 delves more deeply into OPV technology, describinghow it differs from conventional silicon solar cells and delineating the differ-ent types of OPVs. It also discusses the charge transfer process in greaterdetail. Chapter 3 of this thesis covers low energy electron diffraction (LEED)theory. LEED is used to characterize our OPV films. Chapter 4 can be con-sidered a manual for the OMBE. It covers the design and commissioningprocess. Sample preparation techniques and corresponding LEED imagescan be found in Chapter 5. Finally, Chapter 6 describes the vision for thisproject going forward.5Chapter 2Organic PhotovoltaicsThis Chapter covers organic photovoltaics (OPVs) in much greater depththan the introduction. It discusses how organic solar cells differ from con-ventional (silicon-based) ones, the nature of charge separation and transferin OPVs, and the different types of OPVs. A brief summary (see Section2.7) is provided at the end.2.1 HistoryThe field of organic photovoltaics can be dated back to Becquerel’s 1839paper noting the photoelectrochemical process; however, it wasn’t until the1950s that research into organic dyes marked the start of real investigationinto organic photovoltaic devices. The power conversion efficiencies (PCEs)of these initial monolithic devices was extremely low (< 0.1%[15]), but in1986,[16] Tang discovered that implementing a donor-acceptor heterojunc-tion increased the PCE to 1%. The first donor-acceptor heterojunctiondevices were bilayers (see Figure 2.5), designed to facilitate charge sepa-ration at the interface. In 1994,[17] Yu made the first bulk polymer/C60heterojunction device (see Figure 2.6) to render more of the heterojunc-tion interface within diffusion length of the excitons. In 2011,[18] Zhanget al., introduced the concept of dilute heterojunctions (bulk heterojunc-tions containing low concentrations of donor material) to minimize excitonrecombination at the interface. The physics behind these developments isdiscussed in greater detail throughout this Chapter, but suffice it to say thatTang’s discovery in 1986 paved the way for over 30 years of research focusedon donor-acceptor-based OPVs. Researchers have been striving to under-stand the physical dynamics behind this phenomenon – and to maximizeefficiency – ever since.62.2. Mathematical Description of an Excitonic Solar Cell2.2 Mathematical Description of an ExcitonicSolar CellMathematically, we can classify a material as a conventional semiconductoror an organic semiconductor using its dielectric constant and the Bohr radiusof the relevant charge carrier. The effect of these two parameters can bemodeled using the following equation, producing a quantity γ that when> 1 indicates the material is an excitonic semiconductor and, when < 1, aconventional semiconductor[19]:γ =rcrB≈ ( q24pi0kBr0me)(meff2T). (2.1)Here, rB is the Bohr radius of the charge carrier, defined byrB = r0memeff. (2.2)In conventional semiconductors, the effective mass of the electron inthe semiconductor (meff ) is less than the mass of an electron in vacuum(me): meff < me. In excitonic semiconductors, however, the reverse istrue: meff > me. This is because as the charge carrier becomes morelocalized (as in an excitonic semiconductor), the electron transport becomesless wave-like and more particle-like, and the effective mass increases.In order to generate free pairs of charges (free electron-hole pairs), theCoulomb attraction energy must be less than the average thermal energy ofthe charge carrier (kBT ):E = (q24pi0)(1rc) < kBT, (2.3)which can be reordered to give us rc, the critical distance between twocharges:rc =q24pi0kBT. (2.4)As seen above, the larger the dielectric of the semiconductor, the easierit is to for the charges to overcome their Coulomb attraction and dissoci-ate. Similarly, as the charge carrier become more delocalized (and meffdecreases) the average distance between paired charges (rB) increases, con-tributing to charge dissociation and the advent of free charges.72.3. Excitonic Solar Cells vs. Conventional Solar Cells2.3 Excitonic Solar Cells vs. Conventional SolarCellsAn excitonic solar cell (XSC) uses organic semiconductors, which are ex-citonic due to their low dielectric constants and non-covalent interactionsbetween molecules. The weaker intramolecular interactions implies that theelectron’s wave function is more localized, and therefore in closer proximity(on average) to its conjugate hole. Localized electrons with a higher effec-tive mass are unable to screen charges as effectively (due to their spatialrestrictions), causing materials with these localized electrons to have lowerdielectrics.[20] All of this results in a Coulomb binding energy that is greaterthan the average thermal energy of the charge carrier (E > kBT ), and there-fore excitons that require input energy on the order of a few hundred meVto dissociate.[21]In a conventional solar cell (CSC), however, the crystalline inorganic(or conventional) semiconductor such as a silicon p-n junction cell containsstrong interatomic electronic interactions due to the covalent bonding be-tween silicon atoms. These strong interatomic interactions lead the individ-ual atoms’ highest occupied orbital and lowest unoccupied orbital to formvalence and conduction bands – regions for delocalized free charges. Addi-tionally, conventional semiconductors have high dielectric constants: screen-ing due to the high dielectric constant results in excitons that are bound byonly a few meV of Coulomb energy, and thus require an input energy of onlya few meV for charge dissociation. This enables spontaneous disassociationof bound excitons into charge pairs.[21][20]The principal difference between excitonic solar cells (XSCs) and conven-tional solar cells (CSCs) is, therefore, deceptively simple: in a conventionalphotovoltaic device, photoabsorption leads to free electrons and holes in thematerial, whereas in OPVs, excitons (bound electron-hole pairs) are formed.Whether photoabsorption leads to free electron-hole pairs or excitons de-pends largely on two things: the Bohr radius of the relevant charge carrierand the dielectric constant of the material (as demonstrated in Section 2.2).82.3. Excitonic Solar Cells vs. Conventional Solar Cellsa.LUMOHOMOincident photone-b.LUMOHOMOe-h+LUMOHOMOcathodeanodec.e-h+LUMOHOMOd.e-e-h+BothCSCsandXSCsCSCs XSCsincident photonLUMOHOMOe.e-h+e-incident photonFigure 2.1: Description of charge generation and transfer process in a PV.a. Photoabsorption: an incident photon excites an electron from its groundstate in the highest occupied molecular orbital (HOMO) to the lowest un-occupied molecular orbital (LUMO). b. Exciton formation: electron andhole are now coulombically bound, forming an exciton. c. Free charges: inany PV, the electron and hole must become free charges able to reach thecathode and anode respectively. d. In a conventional solar cell (CSC), theexciton dissociates spontaneously, immediately generating free charges. e.In an excitonic solar cell (XSC), photoabsorption results in an exciton, thedissociation of which must be orchestrated.92.4. Charge Separation in Excitonic Solar CellsIn both XSCs and CSCs, absorption of a photon excites an electron,which leaves behind a hole. In an XSC, these charges are coulombicallybound and form an exciton. The exciton is able to move throughout thepartially-delocalized pi orbitals by hopping;[21][22] however, it can be local-ized to a single conjugated segment (or even a single molecule, if the organicsemiconductor does not contain conjugated segments). In an organic semi-conductor, an exciton can be considered a (semi-) mobile, electronically-neutral excited state.[23] The low dielectric permittivity in the exciton’senvironment implies low screening of the localized charges, and thus a high(> kBT ) level of Coulomb energy binding them together. In conjunction,this low dielectric permittivity along with the highly-localized excitons makeit difficult for the exciton to separate into a distinct electron and hole. Ina CSC, however, after photoexcitation the electron-hole pair spontaneouslydissociates due to high dielectric constants, forming free charges.When spatially separated, these electrons and holes constitute free charges.Accumulation of these charges by respective electrodes results in success-ful generation of current (see Figure 2.1). Any recombination of electronsand holes (either radiative recombination, or non-radiative recombinationat charge traps or defects), failure to separate the bound electrons and holes(excitons), or failure to excite an electron due to photoabsorption fails togenerate a current and thus constitutes a loss.2.4 Charge Separation in Excitonic Solar CellsOf course, the predominant goal for any photovoltaic (PV) device is ab-sorb photons that excite electrons, forming free charges that can (throughapplied voltages, electrodes, etc.) become directed current. As discussedabove, a defining characteristic of organic semiconductors is the localizedexcitonic structure and the lack of spontaneous charge dissociation. Rather,in excitonic solar cells, we must engineer the dissociation of these excitonsin order to generate free electrons and holes. Typically (and most relevantto this thesis), exciton dissociation is performed by adding a second molec-ular component with a higher electron affinity (such as a fullerene) to thecompound.In the context of OPVs, a higher electron affinity component is known asthe electron “acceptor,” whereas the region in which the photon is absorbedand excitons are formed (a region of lower electron affinity) is called the102.4. Charge Separation in Excitonic Solar Cells“donor.” If an exciton is formed in the donor region, but is close enoughto the acceptor region, it is attracted to the acceptor region by the freeenergy difference. As previously mentioned (see Section 2.1), in order forthe exciton to dissociate, the exciton binding energy must be compensatedfor; in this case, Tang[16] discovered the the difference in LUMO energylevels between the donor and the acceptor can serve as the driving force forinitial exciton dissociation (See Figure 2.2).These regions where an electron donor material lies next to an electronacceptor material is called a donor-acceptor (or D-A) heterojunction. Unlikein conventional solar cells, where free charges are typically generated withinthe bulk of the semiconductors,[24] charge carrier separation and recombi-nation in excitonic solar cells takes place at the heterojunction interface.[25]The Canonical ApproachExciton dissociation via a D-A heterojunction is canonically thought of asa four-step process (see Figure 2.2), though it is not without dispute:1. Photoabsorption occurs in the donor material resulting in an exciton.7Caveat: In general this is currently true. However, this phenomenonmay be largely due to the fact that fullerenes (an extremely commonacceptor material) are not good photon absorbers. At the very least,there is no theoretical reason that photoabsorption could not occur inthe acceptor material.2. The exciton diffuses to a donor-acceptor interface.3. The electron is attracted to the acceptor material and the electron andhole now form a geminate pair (still held by Coulombic attraction)across the heterojunction interface; the exciton is now in a “chargetransfer” state. Caveat: The charge transfer (CT) state may or maynot actually be a geminate pair. It is possible that the CT state is infact a charge trap, and ought to be avoided.7In an organic semiconductor, the optical band gap (the energy absorbed from aphoton) is less than the electrical band gap (the energy required to generate a free electronand hole). As discussed above, this is due to both the low dielectric constant found inorganic materials and the weak intermolecular electronic interaction characterized by thesmall Bohr radius of the charge carriers. Upon photoexcitation, each charge carrier isbound by Coulombic attraction to its conjugate charge carrier, forming a charge-neutral,bound exciton. The exciton binding energy is defined as the difference between the opticalband gap and the electrical band gap.[26] (See Figure 2.2)112.4. Charge Separation in Excitonic Solar Cells4. Once the electron and hole are within the acceptor and donor phases,respectively, they are free to move separately through the mediums.They must move away from one another in order to prevent recombi-nation – where the electron and hole recombine to form an exciton thatcan then decay. Typically, electrodes will be placed on opposite sidesof the OPV which, by a difference in work function, will attract elec-trons to one and holes to the other (though a similar functionality canbe achieved with hole-conducting electron-blocking and hole-blockingelectron-conducting layers, as can be seen in Figure 2.6).a.b.Donor MaterialDonor Material Acceptor Material122.4. Charge Separation in Excitonic Solar Cellsc.d.Donor Material Acceptor MaterialFigure 2.2: Exciton dissociation via a donor-acceptor heterojunction: a.Photoabsorption results in an exciton, straddling the HOMO and LUMOof the donor material. b. Exciton diffuses to D-A interface. c. Excitonenters the charge-transfer state. d. Exciton dissociates into free chargesthat provide charge transport to electrodes.It is important to note that in order for dissociation to occur, the ex-citon must be able to reach the D-A interface prior to decaying. In orderto ensure this, the domains of pure donor and acceptor material must besmaller than the maximum distance an exciton can diffuse within its lifetime(approximately 10nm) – i.e. any generated excitons must be within 10nm of132.4. Charge Separation in Excitonic Solar Cellsthe heterointerface.[27] To get efficient dissociation, an exciton must reacha D-A interface quickly. Optimizing for rapid exciton transport can leadto complications, however: efficient charge generation requires a quenchingrate on the order of the exciton’s residence time at the interface. The elec-tron transfer rate throughout the acceptor material (the interfacial electrontransfer or “quenching” rate) limits the efficiency of exciton dissociation byrequiring that the exciton remain at the interface for long enough that theacceptor material can diffuse the electron away. Rapid exciton transport,therefore, requires ultrafast quenching rates at the D-A interface for efficientcharge carrier generation.[26, 28]Problems with the Canonical ApproachThe above understanding of exciton dissociation implies that wherever alongan interface the exciton is separated, both donor and acceptor material musthave a continuous path of material to an electrode, to ensure the electronsand holes diffuse away from one another. There are two obvious methodsof doing so – a bilayer heterojunction where one entire side of the OPV isthe donor and the other side is the acceptor, and a continuous percolatingnetwork of both donor and acceptor material that each reaches the requisiteelectrode (known as a bulk heterojunction). Both of these methods will bediscussed in Section 2.6. However, no continuous path of donor materialis present in dilute heterojunctions. Instead, donor molecules are dispersedthroughout the bulk acceptor in a 5-6% concentration (designed to minimizerecombination at the interface – see Sections 2.1 and 2.5). Intuitively, thisshould result in charge build-up on the donor molecules. No such chargebuild-up occurs, however, leading some to hypothesize hopping-like trans-port on the order of 5-10 nm.[29]Another important outstanding question regarding charge transfer inOPVs is the role of hot CT excitons. Hot CT excitons are generated whenthe OPV absorbs a photon with excess energy in comparison to the opticalbandgap of the semiconductor (see Figure 2.2), leading to an exciton withexcess energy in a charge-transfer state at the heterojunction interface.[30]The excess energy in the charge-transfer state may lead to rapid chargeseparation. In the canonic example, charge transfer occurs when an excitonmoves through a bound (and relaxed) CT state, to become free charges;instead the hot exciton process indicates that excitons may skip the bound,relaxed CT state and convert directly into free charges. In 2010, Lee etal.[31] determined (in a study using PCBM) that hot exciton processes do142.4. Charge Separation in Excitonic Solar Cellsnot contribute to photogenerated charge carriers. However, in 2012, Bakulinet al.[32] came to the opposing conclusion, with both Jailaubekov et al.(2012)[33] and Grancini et al. (2012)[34] independently supporting Bakulinet al.’s position. Bakulin et al. showed that exciton separation in OPVs isfacilitated by the formation of excited, delocalized band states, claiming thatit is these short-lived (< 1 ps) delocalized band states (rather than energylevel offsets between the donor and acceptor) that overcome the Coulombinteraction between electron and hole.[32] Bakulin et al. used an IR opticalexcitation pulse to excite bound charge pairs (at the interface) to delocalizedband states, followed by an additional IR pulse to probe the excited states.“Pushing” these bound charge pairs into a delocalized band state resultedin increased photocurrent.[30]Jailaubekov et al. investigated the lifetime of these excited delocalizedbands, using second-harmonic generation spectroscopy to analyze the chargeseparation dynamics at the interface of a CuPc/C60 heterojunction.[33]Jailaubekov et al. found that the excitation lifetime was on the order of∼ 100 fs, indicating that any charge separation of hot CT excitions must oc-cur on an ultrafast timescale. Using two-photon photoemission spectroscopy(2PPE),8 Jailaubekov et al. determined that hot CT states are at ∼ 0.3 eVhigher energy than relaxed excitons, and that hot CT excitons relax within∼ 1 ps.Concurrently, Grancini et al. came to similar conclusion regarding thePCPDTBT/PCBM9 heterojunction, but go even further to argue that typ-ical CT states often act as charge traps. They determined that withouthigh-energy excitation, excitons will split at the D-A interface within thefirst 50 fs, generating interfacial CT states and polarons. Higher-energy ex-citation, however, results in hot interfacial CT states that help generate freepolarons. Grancini et al. conclude that the higher degree of delocalizationin the hot CT state results in a greater probability of charge dissociation inthe first 200 fs, leading to an overall increase in free charge generation.[34]All three of these studies in 2012 indicate the importance of hot CTstates to exciton dissociation in OPVs. However, this conclusion is by nomeans universally accepted. Suggested explanations of charge dissociation82PPE uses a probe pulse to excite electrons. Electrons in excited states are thenphotoemit by the second pulse, and their kinetic energy (and emission angle) is analyzedin an electron energy analyzer.9Another efficient polymer/fullerene OPV material.152.5. Efficiency in OPVsmechanisms include the following:1. Deibel et al.’s (2009) suggestion that charge delocalization occurs overmultiple polymer segments and/or fullerene molecules.[35]2. Lee (J.) et al.’s (2010) determination that hot exciton processes do notcontribute to photogenerated charge carriers at all.[31]3. Bakulin et al.’s and Grancini et al.’s (2012) idea (detailed above) thathot interfacial CT states with delocalized wave functions increase freepolaron generation.[32][34]4. Jailaubekov et al.’s (2012) slightly different take that the hot interfa-cial CT exciton’s electron and/or hole wave functions are localized onmolecules far away from the interface.[33]5. Ge´linas et al.’s (2014) suggestion of a model that assumes delocalizedpi-electron states over the entire ordered fullerene domain.[36]6. Lee (M. H.) et al.’s (2015) use of a hybrid model to describe inter-facial CT state dynamics, in which semiclassical quantum dynamicsdescribes charge dissociation (on a shorter timescale) and Redfieldtheory describes relaxation (on a longer timescale).10[37]Not only are the structural details of interactions at the D-A interfacethat result in the delocalization of hot CT excitons still unknown,[30] buteven the mechanism that results in efficient charge separation at organicheterojunctions remains an open question.[38]2.5 Efficiency in OPVsIn OPVs, efficiency is defined by five values: open-circuit voltage (VOC),short-circuit current density (ISC), quantum efficiency (QE), fill factor (FF ),and power conversion efficiency (PCE). Just like in conventional solar cells,the maximum theoretical efficiency for an OPV is given by the SchockleyQuiesser limit, the only difference being that an electrical bandgap is con-sidered for CSCs and an optical bandgap for OPVs:[20]10They suggest that a CT state evolves into a charge-separated state due to quantumdiffusion, but these charge-separated states eventually relax to low-energy CT states as aresult of their interaction with the thermal bath.[37]162.5. Efficiency in OPVsI =11 + RSRP(I0[exp (V − IRSAnkTe)− 1]− (Iph − VRPA)), (2.5)where RS is the series resistance (i.e. resistance within semiconductingmaterial, electrode connections, etc.), RP is the shunt resistance (indicatesloss of charge carriers via leaks and unexpected recombination – at structuraldefects and impurities – RP ought to be maximized), I0 is the current densityin the dark at reverse bias, Iph is the photocurrent upon illumination, n isan ideality factor,11 A is the area of the cell, e is the elementary charge, kTis the thermal energy, V is the voltage, and I is the current.[39]From the Schockley equation, we can derive equations for the first twokey parameters: the open-circuit voltage, and the short-circuit current den-sity, which can be approximated in the following way when RS is sufficientlysmall or RP is sufficiently large that the effects of both can be ignored:[39]VOC = nkTeln (1 +IphI0[1− VOCIphRPA]) ≈ nkTeln (1 +IphI0) (2.6)ISC = − 11 + RSRP[Iph − I0(exp( |ISC |RSAnkTe)− 1)] ≈ −Iph. (2.7)With an OPV under illumination, both of these values can be determinedexperimentally. Otherwise, they must typically be solved numerically. Qual-itatively, these parameters can be understood in Figure 2.3.The open-circuit voltage VOC is the maximum voltage difference betweenelectrodes. As the electrons and holes build up charge on their respectiveelectrodes, however, they are limited by the difference in workfunctions ofthe electrodes and by the strength of the acceptor material. The maximumpossible photo-voltage Vmax, therefore, is less than VOC , as can be seenin Figure 2.3. Increasing an optical band gap (equivalent to an opticalabsorption gap) can increase VOC .11An ideality factor is a measure of how much a particular system deviates from theideal diode equation, which assumes that all charge recombination occurs via band to bandor charge traps in the bulk material (as opposed to at the junction). An ideal system hasan ideality factor of 1.172.5. Efficiency in OPVsThe short-circuit current ISC is determined by connecting the two elec-trodes together, thus setting the potential across the cell to zero, and thenmeasuring the current flow while illuminating the cell. It provides informa-tion regarding charge separation and transport efficiency. The maximumISC is dependent on the external quantum efficiency EQE (η) (or QE) ofthe solar cell and the photon flux density (Nph(λ)) at a wavelength λ.ISC =∫AM1.5eηEQE(λ)Nph(λ)dλ (2.8)Power = I*VI-V curve(specific to OPV device)FF = (Imax*Vmax)/(Isc*Voc)I,PVVmaxImaxPmax = Imax*VmaxVocIscPmaxFigure 2.3: I − V curve of an OPV under illumination, depicting the fillfactor (FF) and the power curve. VOC and ISC are determined by themakeup of the solar cell (the internal electric field). Vmax and Imax arechosen to maximize the power output (Pmanx = Imax×Vmax) depending onthe I − V curve. The I − V curve is determined by the internal resistanceof the OPV; in a cell with no internal resistance, Vmax = VOC , Imax = ISC ,and the I − V curve would be a right angle. The power curve is given byP = I × V .182.5. Efficiency in OPVsThe EQE of a solar cell describes how efficiently an incident photon leadsto an electron flowing in the external circuit (e.g. the number of electronsthat can be generated by each absorbed photon[15]). It can be understoodas the product of several efficiencies, including absorption efficiency, excitondiffusion, exciton dissociation, charge transport, and charge collection.[39]In order to determine the maximum possible ISC , one integrates the aboveequation from the high photon energy (short wavelength) side of the spec-trum to the wavelength corresponding to the optical band gap of the device.The smaller the optical band gap, therefore, the larger the maximum ISC(in direct contradiction to the criteria for maximizing VOC).Multiplying Vmax × Imax gives the maximum power output of the solarcell. Dividing this by VOC × ISC gives the device’s fill factor (FF ):FF =VmaxImaxVOCISC. (2.9)Dividing the maximum power output instead by the power of incidentlight gives the power conversion efficiency (PCE) of the device.η =VmaxImaxPinc(2.10)Although this Section thus far describes the quantifiable definition ofefficiency in OPVs, the underlying premise involves generating the mostseparable excitons for as little energy as possible (specifically avoiding ex-cess energy to whatever extent possible), and avoiding recombination. Aseparable exciton must be within diffusion length of a D-A heterojunction(hence the development of BHJs).12 Avoiding excess energy can be aided bytailoring the bandgap to match the energies of incoming photons (facilitateexciting electrons to the LUMO but not substantially over it). Exciton re-combination can occur in two ways. First, after dissociation, if the electronis localized and does not diffuse away from its hole, the geminate pair willrecombine at the D-A interface. Second, once an electron is a free chargewithin the bulk material, it can encounter a different hole at an interface,form a CT pair with this new hole, and then decay into an exciton. This lat-ter process is called non-geminate recombination, and it is the main processthat dilute heterojunctions (see Section 2.1) attempt to minimize.12Failure to achieve this results in non-radiative decay or fluorescence.192.6. Types of OPVAny deviation in the electronic structure of different molecules (whichcan be due to disorder in the solid or molecular packing, among other things)is an additional source of inefficiency. It can lead to variation in the highestoccupied molecular orbital (HOMO) and the lowest unoccupied molecularorbitals (LUMO) energies within the system, where states with the lowestLUMO levels and highest HOMO levels may act as charge traps.[27]2.6 Types of OPVIn this Chapter, I have discussed excitonic solar cells primarily in contrast toconventional solar cells – specifically, their common use of a donor-acceptorinterface for exciton dissociation. However, as a broader category, XSCsconsist of the following:[26]• dye-sensitized solar cells (DSSCs)[40–42]• conducting polymer solar cells (PSCs)[17, 43]• molecular semiconducting or small molecule solar cells (SMSCs)[16, 44]• (possibly the proposed quantum dot solar cells,[45] though those arebeyond the scope of this thesis)DSSCs are a hybrid organic-inorganic device constituting a single mono-layer of organic dye adsorbed to an electron-acceptor substrate like TiO2(see Figure 2.4).+ anode +redox electrolytedye— cathode —acceptor material (e.g. TiO2)Figure 2.4: Schematic of a basic dye-sensitized solar cell (DSSC) comprisedof a layer of organic dye adsorbed onto an electron acceptor along with aredox electrolyte to replenish electrons in the dye.202.6. Types of OPVIn DSSCs, the dye absorbs photons and, once excited, injects an electroninto the conduction band of the semiconducting acceptor material.[46] Sincethe excitons are created at the interface of the dye and the acceptor mate-rial, this eliminates concerns regarding exciton transport lengths. However,a redox electrolyte material is required to replenish electrons that are ex-cited in the dye and lost to the electron-acceptor substrate. This redoxelectrolyte can be liquid,[47] quasi-solid,[48] or solid state.[49] It must becapable of quickly replenishing the electrons in the dye, and it needs to haveboth high solubility and ionic mobility in an organic solvent, among othercomplicating criteria.[50] Although DSSCs constitute a promising subfieldof organic photovoltaic technology, search for the ideal redox electrolyte andother methods of maximizing efficiency is ongoing.Developed in the early 1990’s, polymer solar cells (PSCs) involve oneor more thin films of organic semiconducting material, typically comprisedof a fullerene-based acceptor and using a polymeric donor molecule, withelectrodes of different work potentials on either side.[15] Although the firstpolymer solar cell design took a monolayer structure (two electrodes withdifferent work potentials on either side of a single organic semiconductingmaterial), today there are several types of PSC, including inverted solarcells, tandem solar cells, bilayer solar cells, and – one of the most promisingtypes of PSC – the bulk heterojunction (BJH) solar cell. When discussingcharge separation in XCSs, this thesis introduced the concept of excitondissociation over a donor-acceptor interface; bilayer PSCs were an attemptto take advantage of that method by utilizing two layers of material – bothdonor and acceptor – between electrodes (see Figure 2.5).+ anode +acceptor materialdonor material— cathode —Figure 2.5: Schematic of a bilayer OPV device comprised of layered elec-tron donor and electron acceptor materials, designed to facilitate excitondissociation via the heterojunction interface.212.6. Types of OPVBHJs developed from bilayer PSCs, as a means of further maximizingthe D-A interface: BJHs have a large active layer comprised of a mix of theelectron-donor and electron-acceptor materials, sandwiched between elec-trodes with differing work potentials (see Figure 2.6). This design max-imises the surface area of the heterojunction, enabling more efficient chargedissociation.+ anode +acceptor materialdonor material— cathode —electron blocking buffer layer Figure 2.6: Schematic of a bilayer OPV device comprised of layered electrondonor and electron acceptor materials, designed to put an interface withinthe diffusion length of a greater percentage of excitons. The electron block-ing buffering layer prevents electrons from screening the anode; it collectsand transports holes to the anode while blocking electrons from reaching it.Figure 2.7: Two of the common acceptor molecules used in OPVs. C60 isthe molecule discussed most prominently in this thesis. Figure adapted fromK. A. Cochrane’s PhD Thesis.[51]222.6. Types of OPVA typical bulk heterojunction OPV uses glass coated in a tin oxide com-pound as the substrate for the donor and acceptor materials which are grownas a thin film. As mentioned earlier, the acceptor material is often fullerene-based (such as C60[52] or PCBM[53], see Figure 2.7), while the polymericdonor materials span quite a range.[54, 55] Fullerene derivatives as a re-placement for n-type molecules became common in OPVs following experi-ments in the early 1990s showing extremely fast (on the order of 50-100 fs)photo-induced electron transfer between a conjugated polymer and fullerenederivatives.[56, 57] PSCs today have many advantages, including high ab-sorption coefficients, the ease and efficiency of manufacturing them throughsolution processes (PSCs are generally solution-processed using organic sol-vents), and being comprised of tunable materials that also have mechanicalflexibility, high transparency, and low specific weight.[58] As of 2018, PSC-based tandem OPVs have achieved PCEs of 17.3%, and and are on the cuspof commercial viability.[5]Although PSCs continue to be an active research field due to their im-pressive efficiencies and distinct advantages as listed above, small moleculeOPVs are a strong competitor. Small molecule OPVs can be engineeredthrough vacuum-deposition[59] or solution processing,[60] and they haveseveral advantages over their polymer counterparts. Small molecules donot experience the chain kinks and chain-end defects (which can lead tostructural disorder and charge traps[61]) of macromolecules. Unlike the dis-tribution of molecular weights contained in polymers, molecular weights insmall molecules are discrete and well-defined, allowing for better purifica-tion and consistency in the material properties[62] and keeping them cost-efficient.[63] Like polymer solar cells, small molecule solar cells (SMSCs)are often bulk heterojunctions; in both cases, the donor material can befullerene- or non-fullerene-based, and the acceptor material will be eithera polymer or a small molecule (thus dictating whether the organic photo-voltaic device is a PSC or an SMSC). Small molecule BHJs are responsiblefor one of the outstanding mysteries in the field: although the initial premiseof bulk-heterojunction solar cells was to incorporate equal amounts of donorand acceptor material designed to maximize the D-A interface that facili-tates charge transfer, more recent studies with dilute heterojunctions haveshown that a decreased donor concentration of only 5-6% in a fullerene-basedacceptor material leads to an increased open-circuit voltage.[18]It is worth noting that multilayer, hybrid organic tandem solar cells232.7. Summaryare currently being researched, in which both polymer:fullerene and smallmolecule (with a fullerene-based donor) active layers exist in a single OPVdevice. These small molecule–polymer hybrid organic tandem solar cellshave VOCs that nearly equal the sum of the individual VOCs of the subcells.[64]Although the maximum PCE of these devices (at 6.26% as of 2017) havenot yet matched efficiencies of either polymer solar cells (17.3%) in a tan-dem structure)[5] or small molecule solar cells (13.20 ± 0.25%),[65] smallmolecule–polymer hybrid organic tandem solar cells are an intriguing newstep in the subfield.2.7 SummaryDespite significant progress, excitonic solar cells remain less efficient thanconventional solar cells. This is in large part due to the difficulty of excitondissociation as a result of the large binding energy of Frenkel excitons andthe low dielectric constant of many organic materials. The materials chosenfor an OPV affect the electronic structure of the device, which is key increating an efficient and viable solar cell. In order to optimize OPVs basedon our current understanding, the following considerations must be takeninto account:[66]• Long exciton diffusion lengths: As mentioned earlier, efficient excitondissociation requires the exciton to reach a D-A interface; therefore, amaterial must be chosen that allows for long exciton diffusion lengths.• High electron and hole mobility: Efficient charge transport requiresmaterials with high electron and hole mobilities. Additionally, highcharge mobility allows for a thicker active layer (comprised of donorand acceptor materials), which leads to both increased light harvestingand reduced charge recombination.• Low optical bandgap: A low optical bandgap is required for a broadabsorption range in the solar spectrum (and a high extinction coeffi-cient).• Appropriate HOMO/LUMO levels: HOMO and LUMO levels mustbe tailored to facilitate exciton dissociation and ensure a large VOC .However, in order to realize a vision of OPVs permeating everyday life,we must better understand the fundamental process of charge transfer dy-namics in excitonic solar cells. I touched upon the differing explanations242.7. Summaryfor exciton dissocaiton and the CT state in Section 2.4, and the variousmeans of exciton recombination that limit efficiency in Section 2.5. All ofthis indicates the need for more insight into these basic dynamics, and con-tributing to the understanding of these processes is the overarching goal ofour ongoing research.25Chapter 3Experimental TechniquesThis section covers the two principal experimental techniques used withinthe OMBE system: molecular beam epitaxy and low energy electron diffrac-tion. Molecular beam epitaxy is the method by which we grow controlledfilms; low energy electron diffraction (LEED) allows us to characterize thestructure of the films, specifically the order and orientation relative to thesubstrate. Together, these techniques enable us to prepare samples of ade-quate quality for further analysis, including time- and angle-resolved ARPES.3.1 Organic Molecular Beam Epitaxy3.1.1 HistoryMolecular beam epitaxy (MBE) is a growth technique that incorporatescollimated, collision-free, molecular beams directed towards a substrate ina UHV environment, facilitating epitaxial growth13 of a sample.[67] Devel-oped in the late 1960s,[68] MBE quickly became integral to progress in laserand transistor technology, among others,[67] and continues to make majorcontributions to both physics and device technology today.[69] Semiconduc-tors, for example, require perfect and pure semiconductor crystals; prior tothe 1970s, semiconductor thin films were grown from vapour[70] and werenot structurally equivalent to the bulk material. The ability to produceatomically flat, ordered layers of extreme purity (complete with real-timediagnostic tools such as reflectance high energy electron diffraction!) wasindispensable to the study of semiconductor physics.13Epitaxial growth refers to the ordered growth of a crystalline sample, oriented withrespect to the substrate. Epitaxial growth can be either homoepitaxial (where the sub-strate is of the same material as the sample) or heteroepitaxial (where the substrate is adifferent material than the sample).263.1. Organic Molecular Beam Epitaxy3.1.2 MBEMBE combines low-growth rate, temperature control of the substrate, andin-situ characterization techniques in a UHV environment to grow films ofa specific atomic thickness and high purity. To borrow a simile from JohnArthur,14 MBE can, at its simplest, be understood as the ability to spraypaint your deposition molecules onto a substrate, controlling each atomiclayer’s composition and impurity level as you go.[70]In order to achieve purity and ordered structure, MBE requires a specificenvironment. Localized beams of atoms (or molecules) are emitted fromsource crucibles (typically quartz or PBN, depending on the temperaturesrequired to sublime the deposition material), and must travel on a nearlycollision-free path to reach the substrate. The substrate must be exposedto those beams for precise amounts of time, with beams being “turned on”or “turned off” nearly instantaneously. The substrate must be kept at aconsistent temperature, often moderately high, to tune the diffusion rate15of atoms relative to their arrival rate in order to control the overall structure.A basic MBE chamber, therefore, contains the following:• a temperature-controlled sample holder where the substrate can beplaced• multiple cell evaporation cells for source materials16• mechanical (often pneumatic or electric) shutters for blocking cellsources to control growth, and for protecting the sample• crystal growth monitoring14One of the pioneers of molecular beam epitaxy due to his work with Alfred Cho atBell Laboratories on the epitaxial growth of GaAs in the late 1960s.15The elevated substrate temperature provides sufficient thermal energy that the inci-dent molecules can migrate over the substrate surface to lattice sites.16For our low temperature evaporation materials, we use sources containing crucibleswrapped with tungsten filaments for heating purposes.273.1. Organic Molecular Beam EpitaxyFigure 3.1: Basic MBE chamber; no built-in characterization method.The time it takes to grow a sample in such a controlled manner can becalculated, assuming a Knudsen cell source, with equation 3.1. This equationdetermines the flux impinging on the substrate surface to be proportional tothe equilibrium vapour pressure of the deposition material in the Knudsencell.[71]F =P (T )apiL2√2pimkBT1s · cm2 (3.1)Here, F is flux, P (T ) is the equilibrium pressure in the cell, a is the areaof the cell aperture, L is the distance to the substrate, m is the mass of thedeposition material, kB is the Boltzmann constant, and T is the temperaturein Kelvin.Foremost, however, facilitating this process requires UHV. Although thespecifics of achieving UHV will be addressed in greater detail in the ex-perimental setup chapter, it is worth noting the importance of UHV as itpertains to MBE – at a chamber pressure of ∼ 1.33 × 10−5, held at roomtemperature and assuming an average molecular weight of 28, the numberof ambient molecules necessary to build a ML of adsorbate material strikesthe substrate every second.[71]283.1. Organic Molecular Beam Epitaxy3.1.3 Measurement TechniquesMost MBEs use reflection high energy electron diffraction (RHEED) to bothmonitor in-situ thin film growth and characterize the crystalline structure,as well as quartz crystal monitors (QCMs) to monitor film growth. QCMselectronically track the frequency response of the quartz crystals duringdeposition. As the quartz crystals are coated in the deposition material, thefrequency will change; this frequency change can be related to film thickness.RHEED provides information regarding the crystal structure as it forms, byshowing if the growth process forms independent full layers or if it formsmulti-layer islands. However, organic materials are damaged by the highenergy electron beam needed for this technique. Instead, our OMBE systemuses a combination of QCMs to monitor the growth of our films, and a post-growth analysis technique called low energy electron diffraction (LEED) tocharacterize our films.293.2. Low Energy Electron Diffraction3.2 Low Energy Electron Diffraction3.2.1 HistoryLow energy electron diffraction, or LEED, is a technique used to character-ize films by determining the surface structure of single-crystalline materials.LEED crystallography was first demonstrated by Davisson and Germer atBell Laboratories in 1927.[72, 73] Although Louis de Broglie had alreadyproposed the possibility of electron diffraction as a consequence of wave me-chanics in 1923,[74] Davisson and Germer’s experiment definitively showedthat electrons diffract off of crystalline surfaces and can thus be treated asboth particles and waves.17 Though LEED was developed in 1927, it wasn’tuntil the 1970s – with the advent of the requisite experimental and theoret-ical techniques18 – that LEED became the powerful and nearly ubiquitoustool it is today.[77, 78]3.2.2 BasicsAt a basic level, LEED constitutes firing a 10-300 eV19 beam of electronsnormal to a crystalline surface that is at least a few atomic layers thick.[77]The electrons diffract off of the surface structure of the sample, and thoseelectrons which backscatter towards the fluorescent screen are detected (seeFigure 3.2). These photo-detected electrons produce images of regular (pat-terned) spots, called LEED patterns (see Figure 5.1), which enable us tocharacterize and confirm the surface structure of a crystalline thin film.17The experimental discovery of electron diffraction actually resulted from a luckymistake: Davisson and Germer were conducting scattering experiments in 1925, when theglass vacuum tube enclosing their set-up cracked. The resulting leak had led their nickelsample to badly oxidize. In order to clean the sample, they subjected it to prolongedheating in vacuum (i.e. they annealed it). When they resumed their experiment, Davissonand Germer saw unprecedented intensity peaks. After observing the sample with the helpof a microscopist, they discovered that the polycrystalline nickel surface had reformed tobecome a series of 10 distinct crystal facets. Davisson and Germer thus surmised that itwas not the structure of the atoms themselves that caused the intensity peaks, but ratherthe atomic arrangement within the crystal that did so. Therefore, in their subsequent andground-breaking experiment, they chose to use a single crystal nickel sample. Davisson wasa recipient of the 1937 Nobel Prize for his work on electron diffraction by crystals.[75, 76]18Specifically, UHV technology and models for multiple scattering theory.19What energy range of eV constitutes “low-energy” electrons is a matter of somedispute. Definitions given in highly-cited reviews of LEED include 20-500 eV[77], 20-150eV[79], and 0-200 eV[80], as well as the 10-300 eV range given the above paragraph[78].This thesis will use a 10-300 eV range as the defined range for LEED because it overlapswith most of the literature and it contains any energies we use in our lab.303.2. Low Energy Electron DiffractionLEED detection screengrid (retarding field)electron gun samplecameraFigure 3.2: A basic LEED setup: the electrons are emitted from an electrongun and scatter off of the sample, the grid provides a retarding field thatfilters out in-elastically scattered electrons, elastically scattered electronspass through the retarding field and hit the fluorescent detector screen, theimage is then captured by the camera. This process will be described ingreater detail later in this section.313.2. Low Energy Electron Diffractiona. b.Figure 3.3: Example of a LEED spot pattern: Au(001) with electron beamenergies of a. 80 eV and b. 150 eV. (MCP set at 480 V in both images.See Subsection 3.2.4 for explanation of an MCP.)Because LEED patterns are images created by the interference patternsof electrons that have diffracted off of the first few layers of the sample’scrystal structure, understanding the specific nature of the electron scatteringprocess is paramount.3.2.3 Electron Scattering in LEEDLEED analysis can be broken into two sub-techniques, which tell us fun-damentally different things: the position of the spots and the intensity ofthe spots themselves. Spot positions tell us about the size and shape ofthe surface cell, and intensities (as a function of electron energy) providesinformation about the makeup of the surface cell itself – the positions of theatoms, etc.We can relate the atomic structure of our surface crystal to the LEEDspot positions using Bragg’s Law:[81]nλ = d sin θ, (3.2)323.2. Low Energy Electron Diffractionwhere λ is the wavelength of the incident electron, d is the atomic rowspacing, θ is angle of diffraction relative to surface normal, and n is aninteger. Using the de Broglie equationλ =hmv, (3.3)(where m is the electron mass and v is its velocity), and the followingequation for kinetic energyKE =12mv2, (3.4)we can substitute to get an expression for the wavelength of an electronas a function of energy:λ =h√2mEA˚ (3.5)where E is in eV.Using the above equations and some simple trigonometry, we can see inFigure 3.4 that (in addition to the specifics of the sample lattice) the LEEDpattern depends on the radius of the LEED optics, the diffraction order ofthe spots (n), and the kinetic energy of the diffracted electrons.Substitutingsin θ =xR(3.6)andλ =h√2mEA˚ (3.7)intonλ = d sin θ, (3.8)we getnh√2mE= dxR. (3.9)333.2. Low Energy Electron DiffractionLEED detection screenelectron gun samplecameraxRϴFigure 3.4: Illustration of how the observed LEED pattern depends on theconfiguration of the LEED optics. Figure adapted from Arthur T. Hubbard’sThe Handbook of Surface Imaging and Visualization, p. 291.[81]Rearranging equation 3.9, we see thatd =nRhx√2mEA˚. (3.10)Spot intensities correspond to the number of electrons hitting the phos-phorescent screen at a particular location. Because electrons are scatter-ing off of atoms throughout the surface, the electron wavefunction at thedetector screen is a sum of possible scattering events. The phase of thewavefunction depends on path length and the wavelength (or energy) of theelectron. Changing the electron beam energy directly affects the energyof the diffracted electrons, leading to varying constructive and destructiveinterference (see Figure 3.5) and oscillations in intensity I(E).343.2. Low Energy Electron DiffractionLEED detection screenelectron gun samplexRϴdϴϴϴ ddsinϴFigure 3.5: As incident electrons strike the sample surface (illustrated com-ing in from the left and reflecting off to the right), the electron that strikesthe second layer of atoms travels 2d sin θ A˚ further than the electron thatstrikes the top layer of atoms. When 2d sin θ is an integer multiple of the elec-tron wavelength (nλ), constructive interference leads to Bragg peaks. When2d sin θ is not an integer multiple of the wavelength, destructive interferencemeans we detect nothing more than noise. Additionally, it’s important tonote that because λ is a function of the incident electron’s kinetic energy(λ = h√2mE), changing the energy of the electrons affects the diffraction spotpositions.Although Bragg’s law accurately predicts a subset of LEED spots (asdescribed in Figure 3.5), it fails to predict a number of them. The reasonfor this is the same reason that interpreting the intensity data is difficult:atoms are strongly scattering object for electrons, meaning, in a crystal,electrons scatter off of multiple atoms, rather than just one.[82]A key advancement in LEED technology was the development of multiplescattering theory and computers of sufficient power to model it. Althoughmultiple scattering theory is beyond the scope of this thesis, I will brieflytouch on the principle considerations:[83]We need intensities of 1% to see clear, defined dots in our LEED pattern. Ifour rate of backscattering is given by the matrix element |Tb|2h¯ , and τ is thelifetime of the electron in our sample material, then the rate of backscatter-ing multiplied by the lifetime of the electron gives us the intensity:353.2. Low Energy Electron Diffraction|Tb|2h¯τ = .01. (3.11)Substituting in δE ≥ h¯τ , we have|Tb|2 = .01δE. (3.12)We have already specified that a narrow intensity peak can be approxi-mated as 8 eV; thus:|Tb|2 ≈ .08eV. (3.13)We also know that electron scattering is not isotropic. Because ourbackscattering matrix element is small, we expect multiple scattering togenerally be due to forward scattering, with a forward scattering matrixelement|Tf |2 ≈ δE ≈ 8eV. (3.14)In the above paragraph I refer to the lifetime of the electron in oursample material. This is a useful quantity to know because it correspondsto the probing depth of your LEED apparatus. When electrons scatter offof a crystal lattice, they scatter off of the atoms present in the lattice inthe first few atomic layers of the crystal structure. If they penetrate furtherthan that, they are absorbed by the lattice into the sea of electrons. Theseelectrons can be absorbed in the first few atomic layers of the material aswell, but if they are not absorbed, electrons in this region can backscatterout of the material, giving us LEED spots.[77]We can estimate[83] the penetration depth of these scattered electronswithin the sample by calculating the lifetime and velocity of the electronswithin the sample. The spots on our LEED patterns are created by backscat-tered electrons that strike a fluorescent screen. We can track the intensityof these spots as a function of electron beam energy and generate an I-Vcurve. Taking the narrowest of intensity peaks I(E), will result in a widthof approximatelyδE ≈ 8eV.[83] (3.15)The uncertainty principle tells us that363.2. Low Energy Electron DiffractionδE ≥ h¯τ. (3.16)Substituting in for δE, we have8eV ≥ h¯τ, (3.17)giving us an electron lifetime ofτ ≈ 8.23× 10−17s (3.18)within our material.From the lifetime of these electrons, we can determine their velocity(we expect an electron velocity of approximately .02c, so we can do a non-relativistic calculation):12mv2 = eV. (3.19)Assuming an electron gun voltage of 100eV, and substituting in for themass m and charge e of an electron, we havev ≈ 5.966m/s. (3.20)It is now straightforward to calculate the distance an electron wouldtravel within our material during its lifetime:vτ = d. (3.21)Substituting in our results from equations 3.18 and 3.20 gives usd ≈ 5A˚. (3.22)3.2.4 LEED Imaging and ApparatusIn order to achieve a useful LEED image, multiple steps must be taken:First, our crystalline sample surface must be clean, ordered, and prop-erly oriented with respect to the electron beam. In this case, clean refers toatomically cleaned and kept within a UHV environment. For LEED, it issufficient to follow the standard procedure for noble metal surfaces of two373.2. Low Energy Electron Diffractionto three rounds of sputtering the surface with argon ions and then anneal-ing it.20 Because irregularities in the crystal structure (including kinks21and point defects), step edges in the surface, as well as domain sizes andorientations all contribute to background noise and LEED spot distortion,the crystal lattice of our sample must be regular, ordered, and lacking indefects.[77] Specifically, irregularities and defects in the crystal contributeto general background noise, whereas step edges and crystal domain sizesaffect the shape of the LEED spots themselves.[84] Similarly, the numberand position of atoms within a unit cell only affect the spot intensity; itis the arrangement of the unit cells – the lattice structure – and the beamintensity that dictate spot position.[80] I will discuss this distinction (andthereby how to interpret LEED patterns) in the following section on electronscattering in LEED.Second, we must discard inelastic scattering and only allow elasticallyscattered electrons to be photo-detected; otherwise, we will detect too highan intensity of electrons backscattered in every direction and practicallywill see a uniformly lit image with no clear diffraction spots. Inelasticallyscattered electrons can be filtered out by applying a negative potential toconcentric hemispherical grids that lie between the sample and the screen(see Figure 3.6). In practice, multiple (often four) grids are used in orderto provide multiple fields: the first grid is connected to ground in order toprovide an essentially field-free region between it and the sample, the secondand third grids have different negative potentials, and the fourth grid onceagain is connected to ground. A field-free region throughout the majority ofthe electrons’ trajectory from the sample to the screen is required to preventthe unintentional deflection of elastically-scattered electrons (though a largepositive voltage is applied to the detection screen to accelerate the electrons).The two different negative potentials create two fields that only allow a de-sired (and small) energy-range of electrons to pass through, thereby allowingonly the elastically scattered electrons and filtering out the inelastically scat-tered ones.[80]20See sample preparation procedure section for further explanation of cleaning proce-dure.21A kinked surface with one with step edges that themselves form a high-Miller-indexface.[80]383.2. Low Energy Electron DiffractionLEED detection screenelectron gunmulti-channel plate (MCP)samplecamera-V1-V2grids x4(retarding field)vacuum+VFigure 3.6: Model of LEED setup for organics.Note that conservation of momentum and of energy dictate that electronscoming from a specified angle, elastically scattering off of atoms in a crys-tal lattice, will rebound at a commensurate angle with equal energy to theincident electrons. It is also worth noting that only approximately 2-5% elec-trons are elastically scattered, while up to 98% percent of incident electronsare either absorbed or inelastically scattered.[77]Third, when dealing with organic molecules, as in this project, thefragility of the thin films requires the use of a multi-channel plate (MCP).An MCP is a thin (≈ 2mm) plate of highly resistive material (often leadglass) that functions as an electron multiplier while maintaining positional393.2. Low Energy Electron Diffraction(special) information. It contains many tiny tubes that travel from one faceto the other, with diameters of 5-10 micrometers and approximately 15 mi-crometers between them. The channels are parallel to one another, but areall oriented at a slight angle to the face of the MCP. When electrons enterthe tubes, therefore, they are practically guaranteed to strike the wall of thetube. An electric field is applied across the MCP, turning each tube into acontinuous dynode electron multiplier. The electron gain of an MCP can be104 − 107. Note that because the standard projection off of a hemisphericalLEED screen gives an undistorted reciprocal lattice, and an MCP screen is aflat screen between the grids and the detection screen of the LEED system,the inclusion of MCPs will distort the image and must be compensated for.22LEED software will often come with an option to compensate for an MCP.MCPs are fragile and must not be exposed to air for prolonged duration, orfrequently.3.2.5 The Ewald SphereA useful apparatus for understanding LEED patterns, Ewald spheres enableus to predict and qualitatively envision diffraction spot locations. The Ewaldsphere is a geometric construction that is simplest when applied to a 1-Dlattice. In that case, the Ewald sphere is actually a circle with a radius 1/λ,where λ is the electron wavelength, with its origin centered on the incidentwave vector. The circle intersects an infinite series of parallel lines withspacing 1a (a is the lattice constant) (see Figure 3.7).22It is common for the LEED detection screen to also be flat when the system includesan MCP, but the LEED software still expects a hemispherical detection screen.403.2. Low Energy Electron Diffraction1/ar𝞅 r𝛂𝞅1/𝝺1/a1/𝝺𝛂2/asin𝞅 = (n/a)/(1/λ) asin𝞅 = nλr = 1/𝝺Figure 3.7: An Ewald circle (a 1-D Ewald sphere). The circle has a radiusof 1/λ, where λ is the electron wavelength; its origin is centered on theincident wave vector. The circle intersects an infinite series of parallel lineswith spacing 1a , where a is the lattice constant of the crystalline sample.Using this construction, every wave vector that extends from the originof the Ewald circle to the intersection of the circle and a line representsa diffracted beam allowed by the Bragg condition. We can see the Braggcondition is satisfied in the following manner:1/ar𝞅 r𝛂𝞅1/𝝺1/a1/𝝺𝛂2/asin𝞅 = (n/a)/(1/λ) asin𝞅 = nλr = 1/𝝺Figure 3.8: Confirmation that the Ewald circle satisfies the Bragg condition.Triangles are from the example Ewald circle in Figure 3.7.sinφ =λa→ nλ = a sinφ (3.23)413.2. Low Energy Electron DiffractionWhen we extrapolate to apply this technique to a 2-D lattice, the Ewaldcircle becomes a sphere and the lines become an infinite 2-D array of rods(see Figure 3.9).Figure 3.9: A 2-D Ewald sphere, intersecting with 2-D array of rods.As previously discussed, despite predominantly characterizing the sur-face of a material, LEED does penetrate several A˚ngstroms into the crystal.It can therefore be considered a “quasi-3-D”[81] analysis technique. If wewere to model a fully 3-dimensional technique, such as X-ray diffraction, the2-D array of rods would collapse to a 3-D array of dots interspersed through-out the space surrounding the Ewald sphere (see Figure 3.10). These dotswould be at regular intervals along what used to be each of our rods, andthey are known as Bragg or kinematic peaks.423.2. Low Energy Electron DiffractionFigure 3.10: A 3-D Ewald sphere, where the rods have collapsed to a 3-Darray of dots.In our quasi-3-D system, however, rather than entirely collapsing, each rodwithin the array has varying intensity along its length. The high intensityregions along the rod correspond to the dots in the purely 3-D case, but theregions between the dots are now low intensity regions of the rods ratherthan zero-intensity regions (see Figure 3.11).433.2. Low Energy Electron DiffractionFigure 3.11: A quasi-3-D Ewald sphere, with rods of varying intensity.The Ewald sphere model also enables us to understand how varying theelectron beam energy in the LEED apparatus affects the observed diffractionpattern: increasing the beam energy is represented by expanding the Ewaldsphere while holding the rods fixed, and decreasing the beam energy cancorrespondingly be represented by shrinking the Ewald sphere. One canunderstand this by remembering that the electron wavelength is directlyproportional to the beam energy (λ = h√2mE),[81] and that the electronwavelength determines the radius of the Ewald sphere r = 1λ . As you changethe beam energy (and thus the radius of the Ewald sphere), the sphereexpands or shrinks and intersects with the rods in different places. TheEwald sphere “slides” up and down the rods, with varying beam energy (seeFigure 3.12).443.2. Low Energy Electron DiffractionFigure 3.12: Ewald sphere representation of the effect of changing beamenergy on LEED spots. Changing the beam energy changes the radius ofthe sphere (recall that λ = h√2mEand r = 1λ); this results in a differentnumber of intensity spots intersecting the sphere, which causes a differentscale of LEED pattern.As the sphere expands and contracts, it illustrates the number and loca-tion of diffraction spots as a function of beam energy. Because the rods arenot uniform in intensity, this process also gives us intensity as a function ofenergy – the I-V curve.23 Analyzing the I-V curve quantitatively can providedetails of adsorbate structure. The theory assumes an ideal, semi-infinitecrystal, with translational symmetry in each layer, and a “muffin-tin”24scattering potential.[85] This quantitative analysis can only be performedaccurately when the model takes into account multiple scattering, a modelknown as dynamical analysis (as opposed to kinematic analysis).The model of an Ewald sphere and rods of regularly varying intensityserves as an accurate model only for single-scattering processes. As dis-23Here, V stands for the accelerating voltage applied to the electrons.24A “muffin-tin” scattering potential is spherical around each atom and flat betweenthem.453.2. Low Energy Electron Diffractioncussed previously, LEED is not a single-scattering process; it involves multi-scattering. Incorporating multiple scattering into this model also requiresvarying the intensity, but in an irregular manner rather than a regular one.Importantly, however, the Ewald sphere model with rods of regularly vary-ing intensity provides an accurate qualitative description of a LEED patternwithout having to take into account multiple scattering.Ewald spheres can also be used to model scattering from stepped surfacesand surfaces containing defects, however that application is beyond the scopeof this thesis.3.2.6 Kinematic TheoryKinematic theory enables one to quantitatively calculate scattered wavefields in the Born approximation. It does not include multiple scattering,and therefore is an imperfect model for LEED, however it does predict bothLEED patterns and the kinematic peak positions and widths in intensitycurves.[80] The following derivation relies heavily on texts by V. Bortolaniet. al.[83] and Van Hove et. al.[80]Beginning with the Schro¨dinger equation,− h¯22m∇2Ψ(~r) + V (~r)Ψ(~r) = EΨ(~r), (3.24)we separate our potential into its constant and position-dependent parts:V (~r) = V0 + Va(~r). (3.25)We initially consider the constant potential: it is, in fact, constant onlywithin the surface of the material – it smoothly goes to zero outside thesurface. However, for our purposes we are interested in the fact that itchanges the wavevector of the incident electron inside the surface. We canbreak our constant potential V0 into a real and imaginary part:V0 = V0R + iV0I . (3.26)The real component represents the shift in electron energies within themedium, compared to in vacuum. The imaginary component models the life-time of an electron in the solid (a decay in Ψ with time, as I will presentlyshow). Taking only our constant potential into account, our simplifiedSchro¨dinger equation463.2. Low Energy Electron Diffraction− h¯22m∇2Ψ(~r) + (V0R + iV0I)Ψ(~r) = EΨ(~r) (3.27)can be solved such that (setting h¯ = m = 1 for simplicity)Ψ = ei~k·~re−iEt, (3.28)with|~k|2 + V0R + iV0I = E. (3.29)To show that V0i models the decay of an electron in a solid, we’ll take ~kto be real. In that case, E must be complex.|Ψ(t)|2 = |e(i~k·~r)e(−iEt)|2= e(i~k·~r)e(−i~k·~r)e(−iEt)e(iEt)(3.30)Note that|eix|2 = ei(a+ib)−i(a−ib) = e−2b; (3.31)therefore, with E being complex,|Ψ(t)|2 = e(2×Im(E)) = e(2V0I). (3.32)Typically, V0I < 0, therefore Ψ exponentially decays with time.If we require instead for E to be real (as is the case in LEED, whereE is determined by the voltage applied in the electron gun), then we letcomponents of ~k be complex. Note that we can only consider ~kz to becomplex.25 If we chose either ~kx or ~ky to be complex, the wave functionwould grow to ∞ at ±x =∞, which is not physical. Choosing ~kz to have apositive imaginary part, however, leads to a wave function that decays awayinto the sample’s surface. Mathematically, taking ~kz to have a positiveimaginary component results in~kz = +√2E–2V0R–2iV0I − ~kx2 − ~ky2 (3.33)(where25Assuming the z direction is perpendicular to and into the surface of the sample.473.2. Low Energy Electron Diffraction− 12∇2Ψ(~r) + (V0R + iV0I)Ψ(~r) = EΨ(~r) (3.34)can be rearranged as∇2Ψ = 2EΨ− 2V0R − 2iV0I , (3.35)giving us~kz2+ ~kx2+ ~ky2= 2E − 2V0R − 2iV0I (3.36)and therefore~kz2= 2E − 2V0R − 2iV0I − ~kx2 − ~ky2). (3.37)Let us now consider our position-dependent atomic potential Va(~r). Thispotential is typically modeled as spherically symmetric, and it scatters Ψ intoa different state. We must account for this by correcting our wavefunctionin the following manner for scattering by an atom at ~Rj :~Ψ′ = e(i~k·~r) +f(θ) exp (i~k · ~Rj + i|~k||~r − ~Rj |)|~k||~r − ~Rj |, (3.38)wheref(θ) = 4pi|~k|−1∞∑l=0+1∑m=−1e(iδl) sin δl(−1)mYlm(kˆ)Yl−m(rˆ). (3.39)Here, θ is the angle between ~k and ~r, and rˆ is the angular coordinate of ~r.f(θ) is the scattering factor that uses spherical harmonics to represents anoutgoing spherical wave, centered on the atom. In order to fit experimentalresults, f(θ) contains phase shifts such that its amplitude depends on thedirection in which it is scattered – it will be large for forward scattering andsmall for backward scattering.To extend this model to represent scattering off of a full layer of surfacematerial, let us first assume a perfect layer of identical atoms, spaced on aBravais lattice, with~Rj = mj~a+ nj~b. (3.40)483.2. Low Energy Electron DiffractionThenΨ′′ = e(i~k·~r) +∑gM(~k±g ,~k) exp (i~k · ~Rn + i~k±g · (~r − ~Rn)) (3.41)withM(~k±g ,~k) =2piif(θg)[A|~k|| ~kgz|]. (3.42)Here, A is the area occupied by each atom in the layer, θ is the anglethrough which the gth beam is scattered, and ± represents scattering in the±z direction (of the layer). It is worth a logic check here: since we are nowscattering off of a periodic object, our scattered beams should be (and are)discrete and defined by the reciprocal lattice vectors.We have now modeled scattering off of a single layer of a perfect Bravaislattice. However, in LEED we often want to model scattering from multiple(the first few) layers of our material. In order to modify our model, wesimply sum the scattering from several layers, assuming them to be regularlyspaced, with~Rn = n~c. (3.43)Then,Ψ′′′ = e(i~k·~r) +∞∑n=0∑~gM(~k~g,~k) exp (i~k · ~Rn + i~k~g · (~r · ~Rn), (3.44)orΨ′′′ = e(i~k·~r) +∑~gΨ~ge(i~k~g ·~r) (3.45)whereΨ~g =M(~k~g,~k)1− exp (i(~k − ~k~g) · ~c); (3.46)otherwise known as the kinematic formula for diffracted amplitudes.493.2. Low Energy Electron DiffractionThough a useful and often accurate model of LEED analysis, kinematictheory does not include multiple scattering. Multiple scattering which, aspreviously discussed, is pertinent to LEED, mathematically enters the modelas corrections to M. However, any detailed analysis of multiple scatteringis beyond the scope of this thesis. If the reader is interested in furtherunderstanding of the subject, including the more complex dynamical LEEDtheory, I would suggest the following texts: Van Hove et. al’s 1986 text Low-Energy Electron Diffraction, [80] and Bortolani et. al’s 1990 text Interactionof Atoms and Molecules with Solid Surfaces.[83].50Chapter 4Experimental SetupIn the following chapter I will discuss the practical side of building and usingthe OMBE chamber. Specifically, I will describe our experimental setup,including our customized Scienta Omicron Lab10 MBE system, the home-built characterization chamber we added to facilitate LEED analysis on oursamples, as well as the modifications we have made so that our sample holdersystem is compatible with the custom Scienta ARPES system to which theOMBE is attached.4.1 The OMBE4.1.1 Scienta Omicron SystemOur customized Scienta Omicron Lab10 MBE system for surface scienceis an organic molecular beam epitaxy growth chamber (OMBE), used forepitaxially growing organic thin films for advanced spectroscopic charac-terization. It is a UHV system capable of preparing, temporarily storing,and (thanks to a home-built additional chamber) characterizing samples.This system is attached to an angle-resolved photoemission spectroscopy(ARPES) system to couple to different characterization techniques, thoughit is an independent system and could be moved elsewhere. The OMBEinitially comprised 2 chambers, and with the addition of our home-builtcharacterization chamber it now comprises 3 chambers: the main chamber(for substrate and sample preparation) (see Figure 4.1), the fast entry lock(FEL) chamber (for loading and storing our substrates) (see Figure 4.1),and the LEED characterization chamber (see Figure 4.2).514.1. The OMBEMain ChamberFEL ChamberManipulatorConnection toCharacterization ChamberTransfer ArmTransfer ArmTransfer Armfor Sample DockWater ManifoldSample ShutterQCMTSP/ EvaporatorsEffusion CellsFigure 4.1: Scienta Omicron customized Lab10 OMBE system. Does notinclude our home-built characterization chamber. Figure adapted from Sci-enta Omicron Lab10 system manual.524.1. The OMBELEED LocationConnectionOMBE ChamberZ-Shift for SampleManipulatorConnectionARPES ChamberGate ValveTurbo PumpFigure 4.2: Solidworks mock-up of characterization chamber.In addition to the 3 chambers, our system includes additional machineryand mechanisms to facilitate maintaining UHV, sample preparation andgrowth, and characterization. I will discuss this additional machinery in thecontext of the chamber in which it resides.All powered devices (e.g. pumps, ion gauges, evaporators, etc.) associ-ated with the unmodified Lab10 system are controlled through an integratedsystem called MISTRAL that accompanied the system. Additional powereddevices have separate power supplies.Main ChamberThe main chamber of the OMBE (see figure 4.1) is where both substratepreparation and sample growth occurs in our system. This chamber is avertical cylinder (305 mm diameter by 506 mm tall) with a sample holderand manipulator descending into the center of the chamber from the topflange. The chamber also contains 2 effusion cells, a pneumatic shutter,2 quartz crystal monitors (QCMs), an ion gauge, a sputter gun, an argonline with a leak valve, multiple viewports, and a long transfer arm thatenables the transfer of samples into both the characterization chamber andthe ARPES system. Additionally, our main chamber has an integratedcryopanel (a liquid nitrogen cooling system) that acts as a cooling shroud.534.1. The OMBEIt It’s maximum capacity is 5 litres and has an expected operation time of afew hours. This serves to help minimize any organic material contaminationof the ARPES chamber.Manipulator (see Figure 4.1): The sample manipulator consists of asample holder and a heater (maximum base plate temperature of 880 Kwhich translates roughly to a 1100 K sample plate temperature). Due toliquid nitrogen sample cooling capabilities, our sample baseplate can alsoreach temperatures as low as 140 K, with our sample itself reaching 160 K.Figure 4.3: Temperature of sample relative to temperature of baseplate(which is the temperature reported by MISTRAL). Graph provided by Sci-enta Omicron (in Lab10 MBE system manual)The sample manipulator in the main chamber can move along x, y, and zaxes, and can rotate around the vertical axis.Effusion Cells: The OMBE has 2 effusion cells, or evaporators, fordepositing material onto our substrates. These evaporators are MBE Kom-ponenten’s Dual Cluster Source (DCS) and Quad Cluster Source (QCS).2626DCS 40-2x1-14-S-2109538, QCS 40-4x1-12-S-2109438544.1. The OMBEThese effusion cells enable the controlled evaporation of molecules from theevaporators’ various crucibles onto the substrate. The DCS contains 2 cru-cibles and the QCS contains 4 crucibles. All crucibles are made of quartzdue to quartz’s low outgassing properties and the temperature profile of ourintended experiments (we do not need to exceed 800 C – if we did, PBN isa common alternative that can comfortably reach 1500 C depending on thetype of thermocouple). Our crucibles contain tungsten filaments for heatingpurposes and type K thermocouples.Figure 4.4: Sketch of QCS evaporator (effusion cell). Figure provided byMBE Komponenten (in QCS operating instructions manual).Both effusion cells are water-cooled to keep the other crucibles cool whenheating a particular crucible for degassing or evaporation purposes. Notethat in order to heat a crucible the cooling water system must be turnedon, with a Mistral-registered coolant pressure of 3-5 bar. With our currentset-up, Mistral typically reads our coolant flow as 1.3 L/min. The OMBEsystem is equipped with an overpressure valve with a threshold of 6 bar.Each effusion cell has a motor attached that manipulates the built-inshutter. Each cell has a number of pre-set, allowed shutter angles (seeFigure 4.5) that enable the exposure of various configurations of crucibles(see Figure 4.6), which allows for co-deposition from the same evaporator.554.1. The OMBEFigure 4.5: Drawing of allowed shutter positions for our QCS. Figure pro-vided by MBE Komponenten (in QCS operating instructions manual).564.1. The OMBEConfigurations 1 2 3 4all open x x x x1, 2, 4 x x x1, 2, 3 x x x1, 2 x x2, 3 x x1 xnoneCruciblesAllowed Shutter Configurations (Crucibles Exposed)Figure 4.6: Figure showing allowed shutter configurations for evaporators.Configurations represent crucibles open and exposed to the chamber.Because the motor attachments are not bakeable, note that upon re-attachment, the shutter feedthrough must be at 0 or “closed” position (whereit fully covers all crucibles). It is extremely important to note that when in-serting or retracting either of the effusion cell sources from the chamber, theshutter feedthrough must be manually set to 340 in order to avoid damagingthe shutter (see Figure 4.5).Pneumatic Shutter: The pneumatic shutter is a sheet of (metal) at-tached to a pneumatic actuator. It is designed to protect our sample fromany unexpected outgassing that might occur when heating the evapora-tor crucibles and, by opening and closing quickly, it enables more accurategrowth capabilities by exposing the substrate to molecular deposition for amore exact amount of time. In addition, it allows us to grow more evenfilms: a manual shutter takes longer to move across your entire sample sur-face, and since one side of your surface will be exposed to the depositionstream for longer than the other, using manual shutters can result in unevensample growth.QCMs: The OMBE chamber uses EDFelectronics’ QM20 Quartz Mon-itors to monitor film thickness during growth. As the quartz crystals arecoated in our deposition material, the frequency changes in a manner thatcan be related to film growth. QCMs electronically track the frequency re-sponse of the quartz crystals during deposition, enabling them to determinefilm thickness. Most QCMs require what’s called a Z-ratio or Z-factor of the574.1. The OMBEdeposition material to calculate film thickness; the Z-factor is an empiricallydetermined constant that relates the acoustic properties of the depositionmaterial and the quartz sensor crystal. When a Z-factor is unknown, it iscommon to use a Z-factor of 1. The manual for our QCMs does include atable of Z-factors, but it does not include the Z-factor for C60.Ion Gauge: Our hot-cathode gauge consists of a heated filament thatemits an electron current towards a helical grid or cage with a positivevoltage applied to it. Electrons pass through the cage and collide withgas molecules within the helical cage. This ionizes the gas molecules. Thenegative gas ions are then attracted to a collector wire in the center of thecage (the collector has a negative voltage applied to it). In a range from∼ 1×10−4 to 5×10−10 mbar, the ion current is proportional to the moleculardensity, enabling us to discern a pressure reading.Sputter Gun: See subsection 4.2.2.Argon Line and Leak Valve: See subsection 4.2.2.Magprobe Transfer Arm: The magprobe transfer arms are magnet-ically coupled transporters that provide rotary and linear motion for thesample. We have two transfer arms on our chamber: the first runs along thecentral axis of the FEL (our x-axis), providing the ability to remove samplesfrom the docking station (the multi-sample acceptor stage, i.e. the sampleloading, removal, and storage rack) and to move the sample from the FELinto the main chamber. The second transfer arm is in our main chamber,perpendicular to the FEL transfer arm (our y-axis), and provides the abil-ity to transfer our sample from the main chamber into the characterizationchamber or into the ARPES prep chamber (on the neighbouring machine).The magprobe arms work through a large black magnet visible on the ex-terior of the transfer arms being magnetically coupled to the internal shaft.Thus, moving the external magnet results in moving the sample inside thechamber.584.1. The OMBERotatable Internal ShaftClamp (in Closed Position)Sample PlateGuiding ProngsFigure 4.7: Drawing of the transfer arm head, showing the process of therotation of the internal shaft (controlled by rotation of the external magnet)leading to the opening and closing of the clamp on the sample plate. Figureprovided by Scienta Omicron (in Lab10 MBE system manual).The mechanism by which the transfer arm clamps onto a sample is notnecessarily intuitive. Two prongs extend from the head of the transfer arm.When they are free and not restrained by or in contact with anything else,rotating the external magnet will result in rotation of the magprobe head.When the prongs are restrained to a particular plane, rotating the externalmagnet will result in the clamp on the head of the magprobe arm openingand closing via an oblong cam, with a locking notch in the closed position.A full revolution of the external magnet corresponds to a complete cycle ofthe clamp (from closed) opening fully and then closing fully. See Figure 4.7for details.Pumping System: This chamber is pumped on by a 685 litre/secondturbo pump, backed by a scroll pump that it shares with the FEL chamber’sturbo pump. Our turbo pumps are air cooled with optional water cooling aswell. Our main chamber also contains a titanium sublimation pump (TSP).A TSP works by putting a periodic high current through a titanium fila-ment, which heats the titanium to its sublimation point. The surroundingchamber is thus coated in clean titanium. Titanium is extremely reactiveand as gas molecules bounce into the chamber walls, they are likely to ad-594.1. The OMBEsorb, decreasing the overall pressure in the chamber. Our main chamberalso contains a cryopumping system: a liquid nitrogen cooling system com-prised of an integrated cryopanel that acts as a cooling shroud for our mainchamber. Its maximum capacity is 5 litres and has an expected operationtime of a few hours. See Figure 4.18 for a schematic of the pumping systemfor both the main chamber and the FEL chamber.Fast-Entry Load LockThe FEL chamber allows us to load new substrates and samples into thesystem without needing to vent the main chamber. It contains the following:• a fast entry port for loading our substrates• a maneuverable sample acceptor stage• multiple viewports• a short transfer arm that enables the transfer of samples into the mainchamber• an ion gauge• an 80 L/s turbo pump, backed by the same scroll pump that backsthe main chamber’s turbo pump.604.1. The OMBEsample slotssampleFigure 4.8: Our sample acceptor stage, or dock. This is our sample loadingand storage dock, and it can hold up to 5 samples. Images provided byScienta Omicron (in Lab10 MBE system manual) showing4.1.2 Characterization ChamberOnce we have grown our films in the OMBE chamber, and prior to analyz-ing them, we must determine their configuration and quality. We want toensure that our films are crystalline, ordered, and contain as few defects aspossible. The typical characterization technique in MBE is reflective highenergy electron diffraction(RHEED). Due to their fragility, however, organicfilms are often characterized by Low Energy Electron Diffraction, or LEED.In order to perform LEED on our films, we needed to construct an indepen-dent though connected chamber, to keep our LEED protected during samplegrowth. Since we needed to connect the OMBE to the ARPES system tofacilitate sample transfer in a UHV environment, we chose to construct an614.1. The OMBEintermediary chamber that connects the OMBE to the ARPES system, pro-viding an in-situ UHV chamber for characterization.The defining constraints on our characterization chamber were as follows:1. facilitate LEED characterization2. attain UHV (approximately 5× 10−10 mbar in our case)3. meet the physical restrictions of the space between the OMBE and theARPES system4. be capable of receiving both typical Scienta Omicron sample platesand our modified sample plate via magprobe arm transfer.In order to meet constraints 1-3 (constraint 4 will be discussed momen-tarily), we designed and built the following chamber:OMBE Main ChamberCharacterization ChamberCharacterization ChamberSupport Structure forARPES SystemConnects HereFigure 4.9: Solidworks mock-up of characterization chamber with supportstructure, attached to OMBE.The characterization chamber contains the following components:Devices• LEED (the BDL600IR by OCI Vacuum Microengineering Inc)624.1. The OMBE• ion gauge• getter (see Figure 4.10)• z-shift (to facilitate translation and rotation of the sample stage in thecharacterization chamber)• turbo pump (304 L/s)• roughing pumpN50, N100, N200, N300, N400SERIES NON EVAPORABLEGETTERS (NEG)USER GUIDEPN 900030, Rev BISO 9001:2015 Certif edFigure 4.10: The N300 non-evaporable getter – or NEG – is a pump thatfunctions by presenting a large, coated surface area into the chamber and us-ing it to adsorb contaminant molecules. Image provided by Gamma Vacuum(in NEG manual).Construction Pieces• 6” viewport (to facilitate transfers and LEED alignment)• 6” cube• zero-length reducing flange• nipple (a nipple is a tube with CF flanges on either side, used tocombine segments of a vacuum chamber; the CF flanges can be varioussizes – i.e. the nipple can function as an adapter – and can also berotatable)• bellows634.2. Design and Construction• T (a vacuum chamber component in the shape of a lower-case T or across)• mesh (100 mm by 120 mm and inserted in cube, it covers the flange tothe T, preventing dropped samples from falling into the turbo pump)In order to meet constraint 4 (that the characterization chamber becapable of receiving both typical Scienta Omicron sample plates and ourmodified sample plate via magprobe arm transfer), our chamber required acustom manipulator stage.a.b.c.Holes forGuiding ProngsSample SlotFigure 4.11: Solidworks drawing of manipulator in characterization cham-ber, with both a. side-on view and b. top-down view.The manipulator stage is constructed of a mix of 316 stainless steel andsome aluminum pieces, bolts, nuts, and washers. Although our initial designcalled for tapped holes in the 316 stainless steel manipulator rod, we revisedour design to limit fine tapping to aluminum and use nuts instead whererequired.4.2 Design and Construction4.2.1 Support Structure for Characterization ChamberDue to the weight of the combined components of our characterization cham-ber, we needed to design and construct a support structure for the charac-terization chamber.644.2. Design and ConstructionAdjustableAdjustableSupport ClampFigure 4.12: Solidworks mock-up of support structure for characterizationchamber, with adjustable means of fixture to OMBE frame. Note that thebracket on the side of the support with 4 holes is actually missing from ourfinal frame; the vertical pieces of 80-20 it joined were replaced with a singlepiece on each side of the structure.The support structure comprises the following:1. 6 pieces of 80-20, cut to length2. a clamp cut from a single block of 316 stainless steel using a water-jet cutter, with tapped holes and polished using a combination of agrinding stone and a Dremel3. 2 custom aluminum brackets cut with a waterjet4. standard brackets (ordered from McMaster-Carr).The support structure was designed to be adjustable in the locationsmarked in Figure 4.12, in order to impose as few restrictions as possible.Ensuring the connection chamber could be aligned properly with both theOMBE and ARPES systems was paramount and not trivial – our supportstructure needed to be as flexible as possible.654.2. Design and Construction4.2.2 Argon LineIn order to prepare our sample substrates, we sputter them with argon(Ar+) ions before annealing. Our sputtering system contains the followingcomponents:1. Sputter gun2. Leak valve3. Argon line with pump-down capabilities4. Bottle of compressed argonanode cagecathoderepellerhole in repellerFigure 4.13: Sputter gun schematic. In our sputter gun, electrons are emit-ted from the cathode filament, which is kept at a -100 V potential withrespect to the anode cage. Argon is leaked into the anode cage, where it isionized by the electrons. The repeller that surrounds the anode cage reflectselectrons that have passed through the anode cage without ionizing an argonatom back into the anode cage, thereby increasing the ionization efficiencyof the sputter gun. The hole in the repeller allows argon ions to escape intothe chamber. This figure is adapted from SPECS’ IQE 11/35 manual.We use SPECS’ IQE 11/35 as our ion source for sputtering. The ionsource contains a filament assembly kept at a -100 V potential with respect tothe anode cage. The electrons emitted from the cathode are thus acceleratedinto the anode cage. Argon is leaked into the anode cage and is ionized bythe electrons. The anode cage is surrounded by a repeller which reflects anyelectrons that have passed through the anode cage without ionizing an argonatom back into the anode cage, thereby increasing the ionization efficiency664.2. Design and Constructionof the source. A hole is left in the repeller in order to allow the argon ionsto escape into the chamber, in a beam focused on the sample. Becauseour sample is at ground, when the sample is being bombarded by ions, acurrent can be measured between the sample and ground. The beam energyis determined by the difference in potential between the sample and the ionsconcentrated at the source.Our SPECS’ IQE 11/35 source came with an MDC leak valve, which weattached to our argon line, constructed out of a combination of 1/4” and1/8” Swagelock tubing, with custom fasteners to secure the argon line toour system’s frame. Custom plates securely mount our argon line valves andsupport our pressurized bottle of argon gas.Figure 4.14: Photograph of the argon line and valve setup, with gas bottle.4.2.3 Sample PlateDesignOne of the principal design considerations with this system was how to createa sample holder that is compatible with both our Scienta Omicron OMBEsample plate acceptors, and the sample holders used within the ARPESsystem that our system is attached to. These two systems are not designedto hold, manipulate, and analyze samples in the same way; consider thedefault sample holders for each system:674.2. Design and Constructiona. b.B2UNLESS OTHERWISE SPECIFIED:CHECKEDSIZETITLE:INTERPRET GEOMETRICTOLERANCING PER:Q.A.DWG NODATEDIMENSIONS ARE IN INCHESTOLERANCES:FRACTIONALANGULAR: MACH BENDTWO PLACE DECIMALTHREE PLACE DECIMALMATERIALNAMEENG APPR.COMMENTS:DRAWNMFG APPR.1UNLESS OTHERWISE SPECIFIED:CHECKEDSIZEAPPLICATIONTITLE:INTERPRET GEOMETRICTOLERANCING PER:Q.A.FINISHDWG.DATEUSED ON ADIMENSIONS ARE IN INCHESTOLERANCES:FRACTIONALANGULAR: MACH BENDTWO PLACE DECIMALTHREE PLACE DECIMALNEXT SSYMATERIALNAMEDO NOT SCALE DRAWING SCALE: 1:1AD PTEENG APPR.YOLECOMMENTS:DRAWNMFG APPR.2 112