You may notice some images loading slow across the Open Collections website. Thank you for your patience as we rebuild the cache to make images load faster.

Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Scanning probe study of organic semiconducting molecules Tom, Gary Ka Wai 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


24-ubc_2020_may_tom_gary.pdf [ 9.49MB ]
JSON: 24-1.0390310.json
JSON-LD: 24-1.0390310-ld.json
RDF/XML (Pretty): 24-1.0390310-rdf.xml
RDF/JSON: 24-1.0390310-rdf.json
Turtle: 24-1.0390310-turtle.txt
N-Triples: 24-1.0390310-rdf-ntriples.txt
Original Record: 24-1.0390310-source.json
Full Text

Full Text

Scanning probe study of organic semiconductingmoleculesbyGary Ka Wai TomB.Sc., McGill University, 2017A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFMaster of ScienceinTHE FACULTY OF GRADUATE AND POSTDOCTORALSTUDIES(Physics)The University of British Columbia(Vancouver)April 2020© Gary Ka Wai Tom, 2020The following individuals certify that they have read, and recommend to the Fac-ulty of Graduate and Postdoctoral Studies for acceptance, the thesis entitled:Scanning probe study of organic semiconducting moleculessubmitted by Gary Ka Wai Tom in partial fulfillment of the requirements for thedegree of Master of Science in Physics.Examining Committee:Sarah A. Burke, Associate Professor, Physics and Astronomy; Chemistry, UBCSupervisorDouglas A. Bonn, Professor, Physics and Astronomy, UBCAdditional ExamineriiAbstractWhen compared to conventional inorganic semiconductors, organic semiconduc-tors are lightweight, flexible, and compatible with less expensive high-throughputmanufacturing techniques. Applications of organic semiconductors in power gen-eration and light emitting applications have been realized through the developmentof organic photovoltaic (OPV) and organic light emitting diode (OLED) devices.However, to optimize the performance and efficiency of these applications, themolecular orbital energy and the role of the exciton in charge generation and lumi-nescence in organic materials need to be further explored.In this work, scanning probe microscopy (SPM) techniques including scanningtunnelling microscopy (STM), scanning tunnelling spectroscopy (STS), and scan-ning tunnelling microscopy luminescence (STML) were used to probe the elec-tronic and optical properties of individual organic molecules deposited on insu-lating NaCl layers on a metallic substrate. Pixel-by-pixel STS energetically andspatially resolves molecular orbitals. Concurrent STML induces molecular lumi-nescence through electron tunnelling, giving spectral information of the excitonsand vibrational modes of the organic molecule on sub-nanometre length scales.The results presented here are the first signals of molecular emission obtainedfrom our microscope, demonstrating the capability of our system in detecting sin-gle molecule luminescence. Conventional organic molecules 3,4,9,10-perylenetetracarboxylic dianhydride (PTCDA) and zinc (II) phthalocyanine (ZnPc) werestudied and the results compared to those presented in the literature. SPM was alsoperformed on F8ZnPc to explore the effects of fluorination. Our results revealedthat electronic and structural changes due to the additional fluorine atoms and in-teractions with the substrate can affect the energy levels and luminescence of theiiimolecule.Organic donor-acceptor molecules based on the hexamethylazatriangulene(HMAT) complex were also studied. The effects of various acceptor groups onthe energetic gap and spatial distribution of molecular orbitals were explored fordifferent HMAT derivatives using STS. We demonstrate the effects of gap engi-neering at the sub-molecular level for this promising class of optoelectronic organicmaterials.ivLay SummaryOrganic semiconductors are a class of materials that can be used in electronic de-vices that interact with light, such as solar cells and imaging devices. These ma-terials have unique properties that allow these devices to be flexible, lightweight,transparent, and easier to manufacture. But to improve the performance of these de-vices, the interaction of light and electrons have to be understood at the nanometrelength scale. This is done using scanning probe techniques, which allow us to studythe electronic and optical properties of single organic semiconducting molecules.With this, we can understand how the nearby environment and the geometry of themolecule can affect its interaction with light and electricity. And by comparingdifferent materials, we can learn how to build new molecules that have propertiestuned for specific purposes. Understanding organic semiconductors at the molecu-lar level allows for more effective and efficient devices.vPrefaceExperiments outlined in Chapter 4 were performed by me with help from GiangNguyen and Erik Ma˚rsell. Experiments outlined in Chapter 5 were performed byme with help from Giang Nguyen and Jiabin Yu. I analyzed all data with customMATLAB scripts, and also performed the density functional theory calculations.The work presented is unpublished as of the writing of this thesis.viTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Excitons in organic semiconductors . . . . . . . . . . . . . . . . 31.3 Scanning probe techniques . . . . . . . . . . . . . . . . . . . . . 62 Experimental Techniques . . . . . . . . . . . . . . . . . . . . . . . . 82.1 Scanning tunnelling microscopy . . . . . . . . . . . . . . . . . . 82.2 Scanning tunnelling spectroscopy . . . . . . . . . . . . . . . . . 112.2.1 Tunnelling theory . . . . . . . . . . . . . . . . . . . . . . 122.2.2 Normalization of dI/dV . . . . . . . . . . . . . . . . . . 16vii2.3 Scanning tunnelling microscopy luminescence . . . . . . . . . . . 182.3.1 Plasmon emission . . . . . . . . . . . . . . . . . . . . . 182.3.2 Organic molecule emission . . . . . . . . . . . . . . . . . 192.3.3 Experimental factors . . . . . . . . . . . . . . . . . . . . 213 Experimental Setup and Simulation Methods . . . . . . . . . . . . . 233.1 The microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.1.1 Scanning probe with optical access . . . . . . . . . . . . 253.1.2 Tip preparation . . . . . . . . . . . . . . . . . . . . . . . 273.1.3 External optical setup . . . . . . . . . . . . . . . . . . . . 303.2 Sample preparation . . . . . . . . . . . . . . . . . . . . . . . . . 323.3 Simulation methods . . . . . . . . . . . . . . . . . . . . . . . . . 374 Luminescence from Organic Molecules . . . . . . . . . . . . . . . . 394.1 Plasmon emission from substrates . . . . . . . . . . . . . . . . . 404.2 Study of PTCDA . . . . . . . . . . . . . . . . . . . . . . . . . . 424.3 Study of ZnPc . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444.4 Study of F8ZnPc . . . . . . . . . . . . . . . . . . . . . . . . . . . 464.4.1 Comparison of STM/STS of ZnPc and F8ZnPc . . . . . . 464.4.2 Luminescence on F8ZnPc . . . . . . . . . . . . . . . . . 474.4.3 Adsorption geometry of F8ZnPc . . . . . . . . . . . . . . 545 Engineering Organic Molecular Energy Levels . . . . . . . . . . . . 585.1 Introduction to HMAT . . . . . . . . . . . . . . . . . . . . . . . 585.2 STM/STS study of HMAT derivatives . . . . . . . . . . . . . . . 595.3 STML study of HMAT derivatives . . . . . . . . . . . . . . . . . 646 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 666.1 Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . 68Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70viiiList of TablesTable 3.1 Deposition temperature and time for each molecule used in ex-periments. Deposition times vary depending on the desired cov-erage. The times listed are for sparse coverages. The depositionmay be repeated multiple times until coverage is sufficient. . . 37Table 4.1 Table of band and optical gaps of three types of F8ZnPc on(2ML)NaCl/Ag(111), extracted from STS and STML experi-ments respectively. . . . . . . . . . . . . . . . . . . . . . . . . 54Table 4.2 Table of band gaps of the four types of F8ZnPc on(2ML)NaCl/Ag(111), extracted from STS experiments. . . . . 55Table 5.1 Table of DFT calculated, and STS measured band gaps (directlyon Ag(111) substrate) of HMAT derivative molecules. The en-ergy of the gap tends to decrease with increasingly electroneg-ative acceptor groups attached. . . . . . . . . . . . . . . . . . 64ixList of FiguresFigure 1.1 Various applications of organic semiconducting materials. (a)Organic solar cell that absorbs in the infrared range, making ittransparent to visible light. (b) Flexible thin film OPV thatare fabricated in large rolls (Infinity PV). (c) Foldable dis-play made possible by flexible OLED technology (SamsungElectronics). (d) Biological imaging application of OLEDmolecules attached to nanoparticles. . . . . . . . . . . . . . . 2Figure 1.2 Schematic of energy levels involved in organic molecule ex-citations. (a) A molecule in ground state. The HOMO andLUMO energies and the band gap of the molecule are labelled.(b) The optical gap, and exciton binding energy for an excitedelectron-hole pair are indicated for a typical exciton. . . . . . 4Figure 1.3 Schemtic demonstrating the formation of the charge transferstate. (1) An incoming photon excites an electron in the donorsystem. The exciton forms due to Coulomb force between theelectron and hole. (2) Before recombination occurs, the exci-ton diffuses to the heterojunction. The charge transfer excitonform. (3) With the exciton delocalized, dissociation occurs andthe charge is transferred to the acceptor. . . . . . . . . . . . . 5xFigure 1.4 Franck-Condon diagram. Vibrational modes are discrete quan-tum harmonic oscillator levels. (1) Photon excites an electronfrom singlet ground state into the third mode of the singlet firstexcited state, the S0(ν = 0)→ S1(ν ′ = 3) absorption transi-tion. (2) Relaxation in the excited molecule into the lowest vi-brational mode S1(0), reducing the optical gap. (3) Exciton re-combination into S0(2) due to the wavefunction overlap. Thisdiagram demonstrates the S1(0)→ S0(2) fluorescence transition. 6Figure 2.1 Schematic of STM system. . . . . . . . . . . . . . . . . . . . 9Figure 2.2 Schematic of 1D tunnelling between tip and sample. Wavefunctions in the tip and sample are oscillatory, and exponen-tially decaying in the vacuum barrier. A positive bias allowselectrons under the tip Fermi energy εt to tunnel into statesabove the sample Fermi energy εs. A trapezoidal barrier isoften approximated with a constant potential based on the av-erage work function and the bias energy. . . . . . . . . . . . . 10Figure 2.3 Constant current STM image of bilayer NaCl on Ag(111)(7nm×7nm, Vb = −1V, It = 100pA). The atoms of theNaCl(001) lattice are resolved, with the bright spots corre-sponding to the Cl– ions. Various defects are observed on thelattice. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Figure 2.4 Schematic showing the tunnelling process as a function of biasvoltage. (a) For positive bias voltages, the sample Fermi en-ergy is below the tip Fermi energy. The resulting current arisesfrom electrons of the occupied tip LDOS tunnelling into thesample. (b) At negative bias voltages, the reverse process oc-curs, and electrons in the sample tunnel into the unoccupied tipLDOS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12xiFigure 2.5 Diagram of excitation of SPP through inelastic tunnelling. Theprocess is possible at negative biases as well. (a) Electrons tun-nel inelastically into the sample. The energy excites a surfaceplasmon which decays to give emission. (b) Energy from hotelectron convert into charge oscillations and associated elec-tromagnetic fields. Plasmons localized to the surface becomeSPPs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Figure 2.6 Energy diagrams to describe two different exciton formationmechanisms of an organic molecule. (a) Plasmon-mediatedand (b) charge injection exciton formation. . . . . . . . . . . 21Figure 3.1 Image of Omicron UHV LT SPM with optical access. . . . . . 24Figure 3.2 Image of the SPM head. . . . . . . . . . . . . . . . . . . . . 25Figure 3.3 Details of filters and windows. (a) IR and bandpass filters andrespective windows. The transmission curves are shown in (b)and (c), with region of interest, ∼500 nm–800 nm, highlighted. 26Figure 3.4 Diagram of Ag tip electrochemical etching setup. (a) Ag wireis suspended in acid with mesh electrode around it. An AC biasof 39 V is applied with the negative terminal attached to thewire. Telfon tubing protects the bottom portion of the wire. (b)Preferential etching at the meniscus and shifting of the menis-cus as etching progresses results in a tapered shape. (c) Thedropped tip is used. Etching immediately stops for dropped tiponce it is separated from the wire. . . . . . . . . . . . . . . . 28Figure 3.5 Ag tip imaged with optical microscope after etching. (a) is at10x magnification, and (b) is at 20x magnification. . . . . . . 29Figure 3.6 Plasmon emission with Ag tip on Ag(111) (Vb = 3V, It =200pA, tx = 10s). Emission intensity ∼ 0.5–1.0 a.u. is suf-ficient for STML experiments. . . . . . . . . . . . . . . . . . 30Figure 3.7 Diagram of external optics on optics table mounted on micro-scope. The red line represents the beam path. . . . . . . . . . 31xiiFigure 3.8 STM and STS of bare Ag(111) substrate. (a) STM topography(30nm×30nm, Vb = 1V, It = 10pA). (b) Surface state ofAg(111), indicated at −67 meV, seen in normalized dI/dV . . 33Figure 3.9 STM and STS of bare Au(111) substrate. (a) STM topogra-phy (30nm×30nm, Vb = 1V, It = 5pA). (b) Surface state ofAu(111), indicated at −100 meV, seen in normalized dI/dV . . 34Figure 3.10 STM image of 2 and 3ML NaCl on Ag(111) substrate(100nm×100nm, Vb = 1V, It = 10pA). NaCl islands areidentified by the step heights and surface state shifts. . . . . . 35Figure 3.11 Shifted surface state of substrates due to bilayer NaCl. (a)Surface state of Ag(111) shifted to 100 meV, and (b) Au(111)shifted to −250 meV. . . . . . . . . . . . . . . . . . . . . . . 36Figure 3.12 Chemical structures of all molecules presented in this thesis. . 36Figure 4.1 Plasmonic emission on Ag(111) and Au(111) thin film col-lected with the same Ag tip (Vb = 3V, It = 200pA, tx = 10s).The signal on the thin film is lower by an order of magnitude,and red-shifted. The Ag(111) emission peaks at 573 nm, whilethe Au(111) emission peaks at 633 nm. . . . . . . . . . . . . 40Figure 4.2 Plasmonic emission from Ag(111) and on(2ML)NaCl/Ag(111) (Vb = 2.75V, It = 250pA, tx = 10s).There is an enhancement on the bilayer NaCl with minimalshift in energy. . . . . . . . . . . . . . . . . . . . . . . . . . 41Figure 4.3 Luminescence on PTCDA on (2ML)NaCl/Au(111) at positivebias (Vb = 3V, It = 200pA, tx = 300s), with correspondingplasmon on substrate (Vb = 3V, It = 200pA, tx = 30s). Inset isthe corresponding STM image with tip position during STMLindicated (20nm×20nm, Vb = 2V, It = 10pA). . . . . . . . 43Figure 4.4 Luminescence on ZnPc on (2ML)NaCl/Ag(111) at positivebias, with corresponding plasmon on substrate. Both spectrawere taken at Vb = 3V, It = 300pA, tx = 300s. Inset is thecorresponding STM image with tip position during STML in-dicated (15nm×15nm, Vb = 2V, It = 10pA). . . . . . . . . 45xiiiFigure 4.5 Luminescence on ZnPc on (2ML)NaCl/Ag(111) at negativebias. Spectra was taken at Vb =−2.5V, It = 200pA, tx = 20s.The emission peak is at 653 nm, and the FWHM is 15 meV. . 46Figure 4.6 STM and STS data of ZnPc and F8ZnPc on(2ML)NaCl/Au(111). (a) STM images of ZnPc and F8ZnPc(3nm×3nm, Vb =−2.5V, It = 5pA). (b) Normalized dI/dVof ZnPc and F8ZnPc on (2ML)NaCl/Au(111). . . . . . . . . 47Figure 4.7 Luminescence on F8ZnPc on (2ML)NaCl/Ag(111) at positivebias (Vb = 3V, It = 250pA, tx = 60s), with corresponding plas-mon on substrate (Vb = 3V, It = 250pA, tx = 10s). A peakis observed at 675 nm. Inset is the corresponding STM im-age with tip position during STML indicated (20nm×20nm,Vb = 2V, It = 10pA). . . . . . . . . . . . . . . . . . . . . . 48Figure 4.8 Luminescence at negative bais detected on F8ZnPc on(2ML)NaCl/Ag(111) with 300 gr/mm grating (Vb = −2.5V,It = 200pA, tx = 120s). Peak detected at 640 nm, with lowerenergy vibrational peaks. Inset is the corresponding STM im-age with tip position during STML indicated (5nm×5nm,Vb =−2V, It = 10pA). . . . . . . . . . . . . . . . . . . . . . 49Figure 4.9 Changes in vibrational satellite peaks as the tip is positionedover different parts of F8ZnPc on (2ML)NaCl/Ag(111) with600 gr/mm grating. STML parameters were Vb = −2.5V,It = 100pA, tx = 300s for each point. Inset is the corre-sponding STM image with tip position during STML indicated(5nm×5nm, Vb =−2V, It = 10pA). . . . . . . . . . . . . . 51Figure 4.10 Same sample of F8ZnPc on (2ML)NaCl/Ag(111) scanned atpositive and negative biases. (a) STM image at 20nm×20nm,Vb = 2V, It = 5pA. All molecules appear similar. The stripedMoire´ pattern between NaCl(001) and Ag(111) lattices canbe observed. (b) STM image at 20nm×20nm, Vb = −2V,It = 5pA. Same sample of molecules showing diversity in to-pography at negative bias. . . . . . . . . . . . . . . . . . . . 52xivFigure 4.11 STM, STS and STML of 3 types of F8ZnPc observed on(2ML)NaCl/Ag(111). (a) Three molecules labelled in STMimage (7.5nm×7.5nm, Vb = −2.7V, It = 10pA). (b) Nor-malized dI/dV of each molecule. (c) Luminescence of eachmolecule taken at Vb =−2.5V, It = 100pA, tx = 120s. . . . 53Figure 4.12 The four identified types of F8ZnPc on (2ML)NaCl/Ag(111).(a) STM images of the four types (Vb = −2.5V and It =5pA). STM of types 1–3 are 2.5 nm×2.5 nm, while type 4is 3 nm×3 nm. (b) Normalized dI/dV from point STS takenon each of the four types. . . . . . . . . . . . . . . . . . . . 55Figure 4.13 STM images of four identified types of F8ZnPc on(2ML)NaCl/Ag(111) with underlying NaCl lattice superim-posed (Vb = −2.5V and It = 5pA). The circles indicate theCl– ion. Types 1–3 are 2.5 nm×2.5 nm, while type 4 is3 nm×3 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . 56Figure 5.1 Calculated density of states for HMAT in gas phase. HOMOand LUMO isosurfaces are pictured. The DFT band gap is4.45 eV. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Figure 5.2 STM and STS data on HMAT-O on Ag(111). All scale bars are1 nm. (a) Normalized dI/dV on the donor (HMAT) and accep-tor (oxadiazole) complex, indicated on the inset (6nm×6nm,Vb = 0.1V, It = 5pA). (b) STS grid image of HOMO andLUMO at respective biases. Band gap is 3.76 eV. (c) DFTcalculated HOMO and LUMO. . . . . . . . . . . . . . . . . 61Figure 5.3 STM and STS data on HMAT-TZ on Ag(111). All scale barsare 1 nm. (a) Normalized dI/dV on the donor (HMAT) and ac-ceptor (triazine) complex, indicated on the inset (4nm×4nm,Vb = −1V, It = 10pA). (b) STS grid image of HOMO andLUMO at respective biases. Band gap is 2.57 eV. (c) DFTcalculated HOMO and LUMO. . . . . . . . . . . . . . . . . . 62xvFigure 5.4 STM and STS data on HMAT-HZ on Ag(111). All scalebars are 1 nm. (a) Normalized dI/dV on the donor (HMAT)and acceptor (heptazine) complex, indicated on the inset(4nm×4nm, Vb = −1V, It = 30pA). (b) STS grid image ofHOMO, LUMO, and LUMO+1 at respective biases. Band gapis 2.41 eV. (c) DFT calculated HOMO, LUMO, and LUMO+1. 63Figure 6.1 STM image of dimerized F8ZnPc and ZnPc on (2ML)NaCl/-Au(111) (5nm×5nm, Vb =−2.5V, It = 5pA). The white andyellow arrows indicate the F8ZnPc and ZnPc, respectively. . . 69xviGlossary1D one-dimensional3D three-dimensionalAC alternating currentB3LYP Becke 3-parameter Lee-Yang-ParrCCD charge-coupled deviceCT charge transferDFT density functional theoryDOS density of statesETKDG experimental-torsion basic knowledge distance geometryFWHM full width at half maximumHMAT hexamethylazatrianguleneHMAT-O HMAT-oxadiazoleHMAT-HZ HMAT-heptazineHMAT-TZ HMAT-triazineHOMO highest occupied molecular orbitalIET inelastic tunnellingxviiIR infraredLDOS local density of statesLHe liquid heliumLN2 liquid nitrogenLT low temperatureLUMO lowest unoccupied molecular orbitalML monolayerNEGF non-equilibrium Green’s functionOLED organic light emitting diodeOMBE organic molecular beam epitaxyOPV organic photovoltaicPCE power conversion efficiencyPTCDA 3,4,9,10-perylene tetracarboxylic dianhydrideSPM scanning probe microscopySPP surface plasmon polaritonSTML scanning tunnelling microscopy luminescenceSTM scanning tunnelling microscopySTS scanning tunnelling spectroscopyUBC the University of British ColumbiaUFF Universal Force FieldUHV ultra-high vacuumWKB Wentzel-Kramers-BrillouinZnPc zinc (II) phthalocyaninexviiiAcknowledgmentsI want to thank my supervisor Sarah Burke for her guidance and instruction.Thank you for your support in my growth as a scientist and researcher.I want to thank Doug Bonn for being a fantastic unofficial co-supervisor.Thank you to Giang Nguyen for his advice and assistance in experiments.Your enthusiasm for science and research is infectious.Thank you to Erik Ma˚rsell for being an excellent teacher, and a reliableemergency contact for my experimental woes and blunders.Thank you to Miriam DeJong for being an awesome office mate, mentor,and friend. The music blasting sessions made the rough days more tolerable.Thank you to the former members of the Omi-homies: Katherine Cochrane,Bingkai Yuan, and Tanya Roussy. Thank you for all you have taught me.Thank you to the wonderful members of LAIR, my home away from home:James Day, Jisun Kim, Aaron Kraft, Amy Qu, Mohamed Oudah, Jiabin Yu,Alexandra Tully, Seokhwan Choi, Brandon Stuart, Graham Baker, Ashley NicoleWarner, Rysa Greenwood, Timothy Branch, and Dong Chen.And thank you to Mom and Dad, for all that you’ve done.xixChapter 1Introduction1.1 MotivationThe rapid development of silicon-based and inorganic semiconductors, startingfrom the early 1950s, has since led to the age of modern computing and elec-tronics. In parallel, the first conducting organic polymer, polyacetylene, was syn-thesized in 1977 [1, 2]. The discovery of non-insulating organic plastics estab-lished a new research area focused on the development of organic semiconductors,and led to the Chemistry Nobel Prize in 2000 being awarded to their discoverersHeeger, MacDiarmid, and Shirakawa. After several decades of research, the classof semiconducting organic molecules has grown substantially, and organic semi-conductors have found success in commercial applications such as organic lightemitting diodes (OLEDs) [3, 4], organic field effect transistors [5, 6], and organicphotovoltaics (OPVs) [7–10].The major advantage of organic semiconductors, when compared to their in-organic counterparts, is their relative ease of processing. Organic molecules canbe thermally deposited or solution processed at relatively low temperatures, allow-ing for cost-effective device fabrication [11, 12]. Additionally, by drawing from asuite of chemical synthesis techniques, new molecules can be created largely fromEarth-abundant elements, and engineered with different functional groups to havethe desired optical, electronic, and structural properties. This versatility allowsfor tunable light absorption and emission, transparency, and mechanical flexibility,1which have already been applied to niche commercial products, such as visiblytransparent solar cells or foldable displays (Figure 1.1).(a) (b)(c) (d)Figure 1.1: Various applications of organic semiconducting molecules. (a)Organic solar cell that absorbs in the infrared range, making it trans-parent to visible light (Zhao et al. [13]). (b) Flexible thin film OPVthat are fabricated in large rolls (Infinity PV [14]). (c) Foldable dis-play made possible by flexible OLED technology (Samsung Electronics[15]). (d) Biological imaging application of OLED molecules attachedto nanoparticles (Crossley et al. [16]).Organic materials possess low electronic screening, a result of their intrinsi-cally low dielectric constant (or low electric susceptibility), which presents chal-lenges in optimizing device performance [9]. Particularly, in the application ofphotovoltaics, the photoexcitation of OPVs results in excited electrons that arebound by the electric force to the positively charged holes left behind. The tightly-bound neutral electron-hole pairs, or excitons, make charge extraction difficult inOPVs, resulting in low power conversion efficiency (PCE). As of 2019, the record2PCE for an OPV cell is 17.4% [17, 18], while typical efficiencies are at ∼ 10%[19]. Inorganic photovoltaics consistently have PCE > 20% [20].Understanding the role of excitons is also important to the development ofOLED devices. Organic semiconductors emit energy in the form of light when ex-citons recombine—the excited electron “falls” back into the hole. The colour andintensity of the emitted light is dependent on the energy and recombination rate ofexcitons formed in the material. The goal in light emitting applications is to max-imize quantum efficiency, the number of photons emitted per charge carrier, whiletuning the optical properties of the exciton. Understanding the underlying physicsof exciton formation, dissociation, and recombination will not only optimize de-vice performance, but contribute to the understanding of excited state phenomenain organic molecules.1.2 Excitons in organic semiconductorsWhen looking at exciton physics in organic semiconductors, there are several en-ergy scales we need to consider (Figure 1.2). The band gap is measured as theenergy Egap from the highest occupied molecular orbital (HOMO) to the lowestunoccupied molecular orbital (LUMO), which are analogous to the valence andconductance bands inside a solid-state system, respectively. Due to the low dielec-tric constant in organic materials, electric fields between charges are stronger dueto lower screening. The attractive Coulomb force between the excited electron andthe hole favours the formation of an exciton with a lower energy, giving an opticalgap, Eopt , that differs from the HOMO-LUMO gap. The exciton binding energyEb is the difference between the band gap and the optical gap, and is typically onthe order of ∼ 1eV for organic semiconductors [21]. For excitons to dissociate,enough energy needs to be supplied to overcome the binding energy; thermal en-ergy, which is on the order of kBT ∼ 0.1meV (kB is the Boltzmann constant), isinsufficient for thermal dissociation of excitons in organic materials.Typically, OPV systems are composed of two species of organic semiconduct-ing materials: an electron donor and an electron acceptor molecule. In 1986, C.W.Tang demonstrated the importance of the acceptor-donor heterojunction betweenorganic semiconductors to the dissociation of excitons into free charge carriers3Figure 1.2: Schematic of energy levels involved in organic molecule excita-tions. (a) A molecule in ground state. The HOMO and LUMO energiesand the band gap of the molecule are labelled. (b) The optical gap, andexciton binding energy for an excited electron-hole pair are indicatedfor a typical exciton.[22]. In OPVs, electrons inside an organic semiconductor are excited from HOMOto LUMO by photons that have energies greater than the band gap of the molecules.Excitons generated in the system act as quasi-particles that can diffuse through thematerial. At the interface, mismatched electron energy levels between the acceptorand donor generate an electric field that pulls apart the electron-hole pair, similarto the process at p-n junctions in inorganic photovoltaic systems. Acceptor organicmolecules have high electron affinity, making it energetically favourable for theexcited electron to transfer into the acceptor molecule LUMO (Figure 1.3). Thisdelocalizes the exciton between two molecules, forming a charge transfer (CT)complex, and allows for exciton dissociation and charge generation in the system[23].Excitons in OPV materials typically have diffusion lengths of 1−30 nm [24],and have lifetimes ranging from attoseconds to microseconds [25]. There is a prob-ability of exciton recombination, in which the excited electron drops in energy andre-emits the Eopt as light, if the exciton is unable to diffuse to an interface to formthe CT state. Even after formation of the CT state, or the charge-separated CT state,recombination can still occur albeit with different probabilities and spectroscopicsignatures [26].In light emission applications, the reverse of the photoexcitation process in4Figure 1.3: Schemtic demonstrating the formation of the charge transferstate. (1) An incoming photon excites an electron in the donor system.The exciton forms due to Coulomb force between the electron and hole.(2) Before recombination occurs, the exciton diffuses to the heterojunc-tion. The charge transfer exciton form. (3) With the exciton delocalized,dissociation occurs and the charge is transferred to the acceptor.OPVs is observed. A year after the two-layer OPV publication, C.W. Tang andS.A. Van Slyke created the first practical OLED device [27]. Using electron and holetransport layers, charges are directed from the cathode and anode into an emissivelayer composed of organic semiconducting molecules, allowing for exciton forma-tion and subsequent recombination. However, complications arise due to the natureof excitons in organic molecules. Vibration modes within the molecule create dis-crete energy states for each molecular orbital, as described in the Franck-Condonprinciple (Figure 1.4). Relaxation between these discrete states can change the en-ergy of the exciton, thereby changing the colour of emitted light.1 Additionally,excited triplet states can form in organic semiconductors, which generally havelonger lifetimes and can decay non-radiatively, reducing the efficiency of OLEDdevices [28].1This phenomenon is described by Kascha’s rule. Differences in the absorption and emissionspectra can arise, known as the Stoke’s shift.5Figure 1.4: Franck-Condon diagram. Vibrational modes are discrete quan-tum harmonic oscillator levels. (1) Photon excites an electron from sin-glet ground state into the third mode of the singlet first excited state,the S0(ν = 0)→ S1(ν ′ = 3) absorption transition. (2) Relaxation in theexcited molecule into the lowest vibrational mode S1(0), reducing theoptical gap. (3) Exciton recombination into S0(2) due to the wavefunc-tion overlap. This diagram demonstrates the S1(0)→ S0(2) fluorescencetransition.1.3 Scanning probe techniquesScanning probe microscopy (SPM) provides a suite of experimental characteriza-tion techniques that involve measuring the interaction, as a function of experimentparameters, between an extremely sharp probe and the sample of interest. Theprobe can then be raster scanned, using precise piezoelectric motors, across thesample to give a local spatial map of the interaction, or other determinable quan-tities. Unlike conventional optical microscopy techniques, SPM resolution is de-termined by the sharpness of the probe which can be as small as a few picometres,allowing for real-space atomically resolved imaging that would not be possiblewith diffraction limited systems.The SPM techniques employed in this work include STM, STS, and STML,which are all based on the tunnelling of electrons between the metallic tip and the6sample. The samples investigated are organic semiconducting molecules supportedby a conducting substrate, often with an insulating spacer layer. These three tech-niques can map out, with atomic resolution, the local structural, electronic, andoptical properties of small assemblies of organic semiconducting molecules. Thedetails of the techniques will be discussed in Chapter 2.Extensive research on organic semiconducting molecules has been conducted,but often in the context of a device, or as a bulk ensemble of molecules. However,changes in heterojunction geometry, local electronic environment, and molecu-lar structure can drastically affect the optoelectronic properties which occur at thenanometre length scale. SPM can directly probe these effects at sub-molecularresolution, allowing for correlation between electron and exciton physics withchanges in molecular configuration.In all, six molecules were studied using SPM. Prototypical organic semicon-ductors 3,4,9,10-perylene tetracarboxylic dianhydride (PTCDA) and zinc (II) ph-thalocyanine (ZnPc) were first studied to verify the setup used for STML. PTCDAis a highly emissive dye molecule that is ideal for SPM study as it is easily ther-mally deposited and highly planar in structure. A limited number of publicationshave demonstrated STML on PTCDA [29–31]. Metal phthalocyanine molecules,such as ZnPc, are known emitters in STML and have been studied extensively us-ing SPM [29, 32–38]. We further studied the effects of fluorination by looking atF8ZnPc. Previous studies have shown that fluorination can tune the energy levelsof the occupied and unoccupied states of ZnPc [39].Novel OLED molecules were also studied; the molecules were derived fromhexamethylazatriangulene (HMAT), a highly planar electron donating molecule,which has a large photon absorption cross section and high quantum efficiency[40–42]. These molecules include HMAT-oxadiazole (HMAT-O), HMAT-triazine(HMAT-TZ), and HMAT-heptazine (HMAT-HZ). Thin insulating films of NaClwere used to decouple the molecules from the noble metal surfaces used [43].However, the molecules may not be stable on the NaCl, depending on the molecule-substrate interactions present.7Chapter 2Experimental TechniquesIn this chapter, three scanning probe microscopy (SPM) techniques will be dis-cussed in detail, including discussion of the underlying physical principles, meth-ods of operation, and analysis of experimental data.2.1 Scanning tunnelling microscopyScanning tunnelling microscopy (STM) was invented in 1981 by Gerd Binnig andHeinrich Ro¨hrer at IBM, and was the first example of a scanning probe microscope[44]. The STM gives atomic resolution topographic maps of the electronic struc-ture of a conductive sample by employing quantum tunnelling between the sampleand a conductive tip, controlled by sub-nanometre precision piezoelectric motors.As seen in Figure 2.1, a sharp tip is brought close to a sample, with tip-sample dis-tances (z) on the order of 1–10 A˚. A bias voltage difference (Vb) is in place betweenthe electrodes, driving net directional quantum tunnelling of electrons through thevacuum potential barrier between the tip and the sample. A detected tunnellingcurrent (It), typically on the order of picoamperes, goes through a preamplifier andis sent to the control system.An approximation of the tunnelling current could be gathered from a one-dimensional (1D) one particle model (Figure 2.2). The solutions of the Schro¨dingerequation in the tip and sample regions would give free-electron wave functions1,1Electrons in a metal are assumed behave like free particles.8Figure 2.1: Schematic of STM system.while inside the barrier—the vacuum gap between tip and sample—would givewave function ψ(z)∝ e−kz, with k2 = 2m(UB(z,Vb)−E)/h¯2, where UB(z,Vb) is thebarrier potential and E is the energy of the tunnelling electron. For electrons withenergies high enough and tip-sample distance z small enough, the wave functionwould be non-zero on the other side of the barrier. From the 1D Wentzel-Kramers-Brillouin (WKB) approximation, the probability of transmission isT (z,E,Vb) = exp(−2∫ z2z1kdz)= exp(−2∫ z2z1√2mh¯2(UB(z,Vb)−E)dz).(2.1)The barrier potential is unknown, and is often assumed be a trapezoidal barrier,as shown in Figure 2.2. For experimental analysis, the trapezoidal barrier is fur-ther approximated by a rectangular barrier of height UB(Vb) = (Φt +Φs + eVb)/2,which holds when bias voltages are low. For a constant bias voltage in STM, thetransmission probability in Equation 2.1 decays exponentially, giving tunnellingcurrent It ∝ e−2kz sensitive to changes in tip-sample distance. This sensitivity al-lows for picometre height resolution, while a sharp tip allows for sub-nanometrelateral resolution imaging.9Figure 2.2: Schematic of 1D tunnelling between tip and sample. Wave func-tions in the tip and sample are oscillatory, and exponentially decayingin the vacuum barrier. A positive bias allows electrons under the tipFermi energy εt to tunnel into states above the sample Fermi energyεs. A trapezoidal barrier is often approximated with a constant potentialbased on the average work function and the bias energy.There are two modes of STM operation: the constant height, and the constantcurrent mode. In constant height mode, the tip-sample distance is held constantwhile the tip is rastered across the sample. The corresponding variations in thetunnelling current is measured as a function of the spatial variables. A drawback ofthis mode is the possibility of crashing the tip into the sample due to tall features onthe surface, damaging both the tip and the sample. Alternatively, constant currentmode holds the tunnelling current constant by using a feedback loop to adjust thetip height as it scans across the sample. In this case, the tip height is recorded asa function of the lateral (x,y) positions of the tip, giving a tunnelling current iso-surface plot (Figure 2.3), typically referred to as the topography. All STM imagespresented in this thesis are obtained in constant current mode.An STM image should not be interpreted as a physical topography. The tun-nelling current is dependent on both the tip-sample height and the local density ofstates (LDOS) of the sample. More details on the tunnelling current are given inSection 2.2.10Figure 2.3: Constant current STM image of bilayer NaCl on Ag(111)(7nm×7nm, Vb = −1V, It = 100pA). The atoms of the NaCl(001)lattice are resolved, with the bright spots corresponding to the Cl– ions[45]. Four defects of various kinds are observed on the lattice.2.2 Scanning tunnelling spectroscopyScanning tunnelling spectroscopy (STS) is a technique similar to STM, but allowsfor energetic resolution of the LDOS of the sample. At a point over the sample,the tip is held at a constant height and the current feedback loop is turned off. Thetunnelling current is then measured as the bias between tip and sample is sweptacross a range of values. For studying organic molecules, the bias typically rangesfrom −3V to 3V, depending on the molecular stability and the energies at whichthe orbital features are found.The relationship between the current signal, I(V ), and the LDOS can be under-stood by returning to the 1D tunnelling model (Figure 2.4). The tip is assumed tohave a constant density of states. The bias voltage shifts the energy of the electronsin the tip relative to those of the sample. At positive biases, the Fermi level2 ofthe tip shifts up, allowing electrons to tunnel into the unoccupied states of the sam-ple. Conversely, at negative biases, the Fermi level of the tip shifts down, allowing2The Fermi level is the electrochemical potential of the material, or the energy at which there is50% state occupation at finite temperatures. This is not to be confused with the Fermi energy, whichis defined for T = 0K. The terms will be used interchangeably, as our experimental condition isclose to absolute zero.11electrons to tunnel from the occupied sample states into the tip. Changes in thesample density of states ρs(E), both occupied and unoccupied, will open new tun-nelling channels that are reflected in changes in the measured I(V ). More detailson the tunnelling current and the recovery of the density of states are described inthe below sections.Figure 2.4: Schematic showing the tunnelling process as a function of biasvoltage. (a) For positive bias voltages, the sample Fermi energy is belowthe tip Fermi energy. The resulting current arises from electrons of theoccupied tip LDOS tunnelling into the sample. (b) At negative biasvoltages, the reverse process occurs, and electrons in the sample tunnelinto the unoccupied tip LDOS.The technique can be further extended by performing STS pixel-by-pixel in agrid, giving a spatial map of the LDOS over an area of the sample. This is done inconjunction with STM, with the tip pausing at grid points for STS spectra, beforeresuming the STM scan. The added time for pixel-by-pixel spectra means thatSTM/STS grids can take hours or days to complete, during which changes in thesample or the tip can affect measurements.2.2.1 Tunnelling theoryIn order to extract quantitative information from STS, we must understand thetheory of quantum tunnelling, and its application to the tip-sample configuration ofthe experiment. The first theoretical formulation of many particle tunnelling wasdeveloped by Bardeen in 1960, using what was called a “transfer Hamiltonian”12[46]. It was then applied to STM/STS by Tersoff and Hamann [47, 48].Bardeen’s formulation does not take into account many-body effects, strongtip-sample interactions, and strong tunnelling. To date, the most rigorous analytictunnelling theory is formulated using non-equilibrium Green’s functions (NEGFs),first developed by Feuchtwang and later expanded upon by Pendry et al., whichcan address the short-comings of the transfer Hamiltonian [49, 50]. However, thetransfer Hamiltonian formulation is sufficient for extracting the LDOS from STSdata, with the conditions for validity approximated in experiment. Furthermore, theNEGF tunnelling theory has been demonstrated to be reducible to Bardeen’s tun-nelling theory, and thus, this section will discuss the transfer Hamiltonian approachto tunnelling [47].In assuming that the tip and sample are independent, with weak interactionsbetween them, we can define the sample and tip Hamiltonians separately asHsψ(r) =− h¯22m∇2ψ(r)+Vs(r)ψ(r), (2.2)Htφ(r) =− h¯22m∇2φ(r)+Vt(r)φ(r), (2.3)where Vs(r) and Vt(r) are the sample and tip potentials, respectively; ψ(r) and φ(r)are eigenstates bounded to the sample and tip, respectively. We can also define aHamiltonian of the entire system responsible for the transfer of electrons betweenthe tip and sample (the so called “transfer Hamiltonian”)HΨ=− h¯22m∇2Ψ+V (r)Ψ where V (r) =Vt(r) in tipVt(r)+Vs(r) in barrierVs(r) in sample, (2.4)where Ψ(r) are the eigenstates of the transfer Hamiltonian, and the potentials ofthe tip and sample only overlap in the vacuum barrier region.Consider an initial sample state3 ψ(t = 0) which satisfies the eigenvalue equa-tion Hsψ(0) = εψ(0). When tunnelling is weak enough, then for some small timet the sample states would be given by ψ(t) = e−itHs/h¯ψ(0) = e−itε/h¯ψ(0). For all3The spatial r dependence of the states are implied, and dropped from the notation.13t, assuming that tunnelling is weak, we expand over the tip states φk satisfying theeigenvalue equation Htφk = ξkφkψ(t) = e−itε/h¯ψ(0)+∑kak(t)φk, (2.5)where ak are coefficients of the expansion. Since there is no tunnelling initially,there would be no contribution from the tip state, and for all k the coefficientsvanish at t = 0. By approximating ak(t), we can take the time derivative to look atthe rate of “transfer” from the initial sample state to the tip state, and from that thetunnelling current.By plugging the state (Equation 2.5) into the time-dependent Schro¨dinger equa-tion of the transfer Hamiltonian (Equation 2.4), and separately taking the timederivative, we have two expressions for ∂ψ∂ t . Equating gives usih¯∑kddtak(t)φk = e−itε/h¯(H−Hs)ψ(0)+∑kak(t)(ξkφk +(H−Ht)φk), (2.6)where H is the transfer Hamiltonian, and ξk are the eignenenergies of the tip states.Taking the inner product with φ j, for any j-th tip stateih¯ddta j(t) = e−itε/h¯〈φ j∣∣H−Hs |ψ〉+ξ ja j(t)+∑kak(t)〈φ j∣∣H−Ht |φk〉 . (2.7)Due to weak tunnelling, for a little while (t  1), a j(t) is the only non-zerocoefficient, and the last term of Equation 2.7 vanishes. The resulting differentialequation has solutionsa j(t) =e−itε/h¯− e−itξ j/h¯ε−ξ〈φ j∣∣H−Hs |ψ〉 . (2.8)For rate of scattering from the sample into the tip as a result of the transfer Hamil-tonian (Equation 2.4), calculate the total ∑k|ak(t)|2 and take the time derivativeddt∑k|ak(t)|2 = ddt∑kt2h¯2sinc2(t(ξk− ε)2h¯)| 〈φk|H−Hs |ψ〉 |2. (2.9)14For sufficiently large t, the sinc function can be approximated as the Dirac δ -function. For a continuous density of states (DOS), the summation over k becomesan integral over all energies, and the integral can be evaluatedddt∑k|ak(t)|2 ≈ ddt∫ ∞−∞dE[pit2h¯2δ(t(E− ε)2h¯)| 〈φE |H−Hs |ψ〉 |2], (2.10)≈ 2pih¯ρt(ε)|M|2, (2.11)where ρt(ε) is the tip density of states near ε , the eigenenergy of the sample stateψ . The equation above is also known as Fermi’s Golden Rule. The matrix element|M|2 = |〈φε |H−Hs |ψ〉|2 is the transition probability of the sample state into a tipstate near ε , denoted as φε .To get the tunnelling current, we look to the work of Tersoff and Hamann,considering tunnelling from either side of the junction. We continue to assumeindependent sample and tip and weak tunnelling, such that the addition or removalof electrons does not greatly affect the chemical potentials of the tip and sample,µt and µs respectively. We will also assume a low bias so that band-bending isnegligible.In these limits, the statistics of the electrons are governed by the Fermi-Diracdistribution: at temperature T and chemical potential µ , the probability of findingan electron at energy E is Fµ,T (E) = (e(E−µ)/kBT +1)−1. The tunnelling of elec-trons from a filled sample state ψn with energy εn into an empty tip state wouldbeIs→t =2pieh¯ ∑nFµs,T (εn)(1−Fµt ,T (εn))ρt(εn)|M|2, (2.12)where e is the elementary charge. The tunnelling of electrons into an empty samplestate ψn from filled tip states near energy εn would beIt→s =2pieh¯ ∑n(1−Fµs,T (εn))Fµt ,T (ε)ρt(εn)|M|2. (2.13)The difference between the expressions in Equation 2.12 and 2.13 will give thetotal tunnelling current. Rewriting the total current by referencing the energy εn15with respect to the Fermi level of the sample, which is biased lower than the tipenergy by eVb, we haveI =2pieh¯ ∑nf (εn,Vb)ρt(εn)|M|2, (2.14)where f (εn,Vb) = Fµt=0,T (εn−eVb)−Fµs=0,T (εn). Note that the Fermi-Dirac equa-tions no longer depend on the chemical potentials of the sample and tip, as they arealready accounted for by the bias eVb = µt −µs.The entire expression can be approximated by an integralI ≈ 2pieh¯∫ ∞−∞dE f (E,Vb)ρt(E)ρs(E)T (E,Vb,z), (2.15)where the matrix elements are approximated by averaging over the sample states φnear the energy E, giving |M|2 ≈ T (z,E,Vb)ρs(E), where T (z,E,Vb) is the trans-mission probability in Equation 2.1 .In the low temperature limit, the Fermi-Dirac equations approach the Heavi-side functions, but are thermally broadened at temperatures above absolute zero,limiting the energy resolution of STS. Physically, only electrons with energies be-tween the Fermi levels of the sample and tip contribute to the tunnelling current;the expression for the tunnelling current becomesIT→0K ≈ 2pieh¯∫ eVb0dEρt(E)ρs(E)T (z,E,Vb). (2.16)It is important to remember the limits in which this equation is valid. FromBardeen’s formalism, we assume (1) independent tip and sample states, and (2)weak tunnelling. In Tersoff and Hamann’s analysis, we further assume (3) Fermi-Dirac statistics, (4) first-order WKB tunnelling, (5) low temperatures, and (6) neg-ligible band-bending effects. In general, these conditions hold when the tip andsample are sufficiently far apart, and bias between the two is sufficiently low.2.2.2 Normalization of dI/dVFrom the Tersoff and Hamann equation, we can extract the density of states of thesample from the STS spectra. In Equation 2.16, the current is a convolution of the16DOS of the tip and sample, and the transmission probability. Assuming that theDOS of the tip is featureless, and equal to unityI ∝∫ eV0dEρs(E)T (z,E,V ). (2.17)In taking the derivative with respect to the bias voltage4, we obtain the differentialconductancedIdV∝ eρs(E = eV )T (z,E = eV,V )+ e∫ eV0dEρs(E)dT (z,E,V )dV. (2.18)A commonly used form for the DOS of the sample is ρs(E) ∝ dI/dV , where,for small enough biases (∼ 10meV), the integral term in Equation 2.18 is ignored[51]. However, for studies of organic semiconductor molecules, features are oftenfound at larger biases (∼ 1–3eV), which would be obscured by the exponentialform of the transmission probability (Equation 2.1). To eliminate the exponentialdependence, the differential conductance can be “normalized” with the conduc-tance [52]dI/dVI/V=eρs(eV )+ e∫ eV0 dEρs(E)T (z,E=eV,V )dT (z,E,V )dV1V∫ eV0 dEρs(E)T (z,E,V )T (z,E=eV,V ). (2.19)In this form, the transmission probability appears in ratios with itself, qualitativelycancelling the exponential form. The normalized differential conductance givesthe ρs(eV ) with an additional term arising from interactions of the wave functionwith the electric field in the junction.However in Equation 2.19, divergences arise when we consider V = 0V orI = 0A. To mitigate this, a small constant term can be added to the conductancewithout significantly modifying the normalization term [53]I/V =√(I/V )2 + c2, (2.20)where for constant c I/V we would have I/V ≈ I/V to the first order. For STSdata analysed in this thesis, all DOS is approximated by (dI/dV )/(I/V ).4The subscript in Vb is implied.172.3 Scanning tunnelling microscopy luminescenceIn the STM setup, an optical system can be integrated to allow for detection oftunnelling induced luminescence. This mode of operation is known as scanningtunnelling microscopy luminescence (STML). The tip acts as a source of electronsand holes that can locally excite photon emission from the sample [54], thus al-lowing the near-field study of luminescence from nanoscale structures on metals,semiconductors, and molecules [55]. The intensity and spectral distribution of thephotons provide information on the optical properties of the material, as well asinelastic tunnelling processes which would otherwise be undetectable in STM orSTS [56]. Additionally, with organic semiconductor molecules, vibrational statescan be seen in the photoemission spectra, particularly in the visible and infraredranges [55, 57].To perform STML, there must be optical access to the tip-sample junction ofthe STM, which acts as a point source of emission. A lens mounted to the micro-scope head is focused at the junction, collimating the emission. An optical paththrough several windows and viewports allows the photons to exit the microscopeand be refocused by an exterior lens into a detector, such as a spectrometer.Emission from the tunnelling junction can be described by two main processes:(1) the excitation of the free electron gas in the metallic substrate, and (2) thefluorescence of a quantum confined system—such as an organic semiconductormolecule. The subsequent sections will describe both processes in further detail.2.3.1 Plasmon emissionThe first observations of tunnelling induced luminescence were seen in metal-oxide-metal junctions in 1976, by Lambe and McCarthy [58]. Later experimentson metal surfaces [59–61] demonstrated that the same effect can be seen in STM.While there are other proposed mechanisms, the emission is generally attributed tothe decay of plasmons excited by inelastic tunnelling (IET) processes [59].Inelastic processes in tunnelling typically account for about 1% of the totalcurrent [56]. In IET, the tunnelling electrons or holes lose a fraction of energyto an excitation, such as phonons or plasmons [62]. Plasmon excitation occurswhen the tunnelling electrons excite charge oscillations in the substrate. Plasmons18at the interface between the metallic substrate and the vacuum can couple to theelectromagnetic field, becoming surface plasmon polaritons (SPPs), which thendecay into photons.5Figure 2.5: Diagram of excitation of SPP through inelastic tunnelling. Theprocess is possible at negative biases as well. (a) Electrons tunnel in-elastically into the sample. The energy excites a surface plasmon whichdecays to give emission. (b) Energy from hot electron convert intocharge oscillations and associated electromagnetic fields. Plasmons lo-calized to the surface become SPPs.2.3.2 Organic molecule emissionIn a quantum-confined system such as an organic semiconductor molecule, lumi-nescence is seen due to the recombination of excitons formed in the semiconductorgap. Excitons are excited electrons that are bound by the Coulomb force to thepositively charged “holes” left behind. Recombination is the process in which theexcited electron falls back into the hole, emitting the energy as a photon. In or-ganic photovoltaics, recombination should be minimized in order to increase thepower conversion efficiency of the device. However, in light emitting applications,recombination is necessary, but the optical gap and the emission energy needs tobe finely tuned. Using STML, exciton physics of the organic semiconductors can5Although no plasmons are physically emitted, due to the origin of the emissions, I will refer tothe photons as “plasmon emission.”19be studied at atomic-resolution, in relation to the geometric configuration and elec-tronic environment of the molecule.There are two generally accepted mechanisms for exciton generation in or-ganic molecules (Figure 2.6): plasmon-mediated, and charge injection. Plasmon-mediated exciton formation involves the photoexcitation of electrons in the semi-conductor by plasmonic emission. In this mechanism, the energy of the plasmonemission needs to be greater than the HOMO-LUMO gap in an organic molecule.In charge injection exciton formation, electrons directly tunnel into the unoc-cupied state of the organic semiconductor, while an electron tunnels out of theoccupied state into the substrate, effectively injecting a hole into the molecule.The electron-hole pair can then form and recombine to emit a photon. The chargeinjection mechanism requires an insulating barrier between the molecule and thesubstrate to allow for a shift in the substrate Fermi level that is (relatively) inde-pendent of the molecular energy states. Additionally, the tip-sample bias needs tobe high enough that the tip and sample Fermi levels allow for simultaneous elec-tron and hole injection. The rate of exciton formation would depend on the rate ofcharge tunnelling through the vacuum and insulating layer barriers.Charge injection is generally accepted as the primary mechanism for lumines-cence of organic molecules in STML, however, evidence for plasmon-mediatedmechanisms has been seen in some experiments [63–65]. It is possible that theemission comes from a combination of both mechanisms.20Figure 2.6: Energy diagrams to describe two different exciton formationmechanisms of an organic molecule. (a) Plasmon-mediated and (b)charge injection exciton formation. Figure adapted from [66].2.3.3 Experimental factorsThe success of an STML experiment is dependent on many factors, such as thematerial of the tip and sample, the geometry of the tip, and the optical propertiesof the molecule.For strong luminescence from tunnelling electrons, the tip and sample shouldbe composed of materials with low dielectric loss in the wavelength region of in-terest. Organic semiconducting molecules typically have optical gaps of 1−3 eV,resulting in exciton emission in the visible/IR range. STML study of organicmolecules should be performed with tip and sample composed of noble metalssilver (Ag) [67] and gold (Au) [68]. Low dielectric loss means that the electricfield can more easily penetrate the material, and the conduction electrons in themetal can move more freely within the bulk, allowing for stronger plasmonic re-sponse [56]. Oxide layers or other contaminants can quench emissions from thetip-sample junction.The tip geometry is crucial to maximizing signal-to-noise ratio in STML. Lu-minescence signals can be enhanced by the plasmonic nano-cavity formed by the21tip and the sample, an effect known as Purcell enhancement [69]. Typical emis-sion rates from semiconducting molecules in STML are ∼ 10−5–10−3 photons perelectron [70, 71]. The coupling of the exciton and the plasmon can increase thestrength of the organic molecule STML signal by up to five orders of magnitude[72]. Theoretical simulations have determined that stronger plasmonic responseis seen in sharper tips due to the enhancement of electric fields at the tip apex bythe lightning rod effect [56, 73]. Experimentally, the exact tip geometry cannotbe determined, however, small changes to the tip apex can change the energy andstrength of the plasmon emission dramatically [74]. For our experiment, the tipwas pulsed and indented until the detected plasmon emission is strong, featureless,and broad. More details on tip preparation are presented in Section 3.1.2.The molecules should possess a high quantum efficiency—the ratio of pho-tons emitted per charge carrier—and high stability on the surface, especially underthe strong electromagnetic fields in the tip-sample junction. Typical STML condi-tions at Vb ∼ 2V and It ∼ 100pA would give tip-sample distances ∼ 5A˚ [55], andelectric fields∼ 1V/nm [75]. Regardless of the prevalent mechanism for STML onorganic molecules, molecule-substrate interactions should be minimized to preventquenching of luminescence due to additional energetic pathways for the excitationto relax [66, 76]. This is usually done by using a thin insulating layer between themolecule and substrate, such as metal-oxide or NaCl layers.22Chapter 3Experimental Setup andSimulation MethodsIn this chapter, I will discuss the instrument and optical setup used, along withdetails on sample and tip preparation. I will also discuss the density functionaltheory (DFT) details, including the software package, potential approximationmethod, and basis set used. The results of the DFT calculations are qualitativelycompared with the experimental results.3.1 The microscopeAll experimental measurements were made on an Omicron ultra-high vacuum(UHV) low temperature (LT) SPM (Figure 3.1). Experimental data was obtainedat liquid helium (LHe) temperatures (∼ 4.3K) and at pressures around 1× 10−11mbar.The entire microscope is inside of an ultra-low vibration facility (dubbed “thepod”), where it sits on a 36 tonne concrete inertia block that is resting on six pneu-matic isolators. To further decouple vibrations of the main building from the mi-croscope, the pod is surrounded by a double-walled concrete enclosure containingacoustically isolating material, and rests on a foundation separate from the mainbuilding. Small feed-through openings allow wires to be passed from the micro-scope to a control unit outside the pod. The microscope can then be remotely oper-23Figure 3.1: Image of Omicron UHV LT SPM with optical access.ated using a computer in a separate control room. More details on the performanceand design of the ultra-low vibration facility is given in reference [77].The microscope is composed of two main chambers: the preparation and thelow temperature (LT) chamber. The preparation chamber is typically at a base pres-sure of ∼ 1× 10−10 mbar, minimizing contamination during sample preparation.In this chamber, the metallic substrates were cleaned by repeated sputtering andannealing. Decoupling layers of NaCl were then deposited onto the surface usinga home-built Knudsen evaporator. The substrates were then transferred into the LTchamber for imaging and experimentation.The LT chamber is separated from the preparation chamber by a gate valve,and has a base pressure of ∼ 1× 10−11 mbar. Samples are often stored in thischamber on a rotating carousel. The samples can be loaded into the SPM, whichis kept at ∼ 4.3K by a cryostat of liquid helium (LHe) surrounded by a liquidnitrogen (LN2) bath. With the sample inside the SPM, molecules were thermally24deposited on the cold sample using an organic molecular beam epitaxy (OMBE)evaporator. At such low temperatures, the molecules are more stable on insulatingNaCl layers, diffusing less, preventing aggregation and self-assembly. For moreinformation on the equipment in the chambers, including the evaporators, and thesample design, refer to [78, 79].3.1.1 Scanning probe with optical accessThe imaging SPM head is shown in Figure 3.2. The entire apparatus is suspendedon damping springs, which can be clamped to the cryostat for sample transfer andfaster cooling. The tip sits on a scanner, which is controlled by a coarse motorand piezoelectric motors. To begin an experiment, the tip was roughly approachedto the sample using the coarse motor. An auto-approach function was then acti-vated, taking a coarse step and extending the piezoelectric motor repeatedly until atunnelling current was detected.Figure 3.2: Image of the SPM head.There are two ports for optical access to the SPM head. A camera used forcoarse tip approach occupies one, while a home-built lens tube used for STMLoccupies the other. A planoconvex sapphire lens (Melles-Griot) sits inside the lenstube, about 18 mm from the tip-sample junction, at an angle of 25° from the sample.This geometry allows for collection of photons in about 5% of the half solid angleabove the sample. In order to maximize photon collection, the lens was focused at25the tip-sample junction for the Ag(111) substrate, an 8 mm tall top-hat crystal. Thetip-sample junction is well approximated as a point source of light, and any shiftfrom the lens focal point can weaken spectral features. When using the Au(111)thin film substrate, the height of the sample is different, resulting in weaker STMLsignals.(a) Left: KG5 IR filter with 20 mm open-ing cover. Right: WG41050 bandpass filterwith 12 mm opening cover.(b) KG5 transmission curve. (c) WG41050 transmission curve.Figure 3.3: Details of filters and windows. (a) IR and bandpass filters andrespective windows. The transmission curves are shown in (b) and (c),with region of interest, ∼500 nm–800 nm, highlighted.The light from the lens then passes through ports in the two heat shields. Theseports have KG5 infrared (IR) filters (SCHOTT) situated inside to reduce radiativeheating of the SPM head that would result in reduced cryostat hold time. Photonemission from organic semiconductors is often in the visible to near-IR range, so26the filters were later replaced with a WG41050 bandpass filter (ThorLabs) thattransmits wavelengths from at least 300 nm–2 µm. The transmission curves of thefilters are shown in Figure 3.3. To counteract increased thermal energy enteringthe heat shields, the inner heat shield port opening was reduced from 20 mm to12 mm in diameter using a 0.4 mm thick gold-plated BeCu plate. After replacementof the filter, no noticeable changes were seen in the cryogenic hold time or basetemperature. All STML data was taken with the KG5 IR filters in place. Due to timeconstraints, STML measurements were not repeated after the new window systemwas installed. The photons finally exit the LT chamber through a Kodial vacuumviewport (VacGen), where they then enter an external optical setup mounted on themicroscope.3.1.2 Tip preparationIn this thesis, a platinum/iridium (Pt/Ir) and a silver (Ag) tip were used. Below is adescription of the tip preparation. The tips were then clamped into an Omicron tipholder and transferred into the microscope.The Pt/Ir tips were made by mechanically shearing a 0.38 mm diameter wire.The Pt/Ir tip oxidizes relatively slowly and is stiff, meaning it can be conditionedby poking into the metal substrates, resulting in a Ag or Au terminated Pt/Ir tip.27Figure 3.4: Diagram of Ag tip electrochemical etching setup. (a) Ag wire issuspended in acid with mesh electrode around it. An AC bias of 39 Vis applied with the negative terminal attached to the wire. Telfon tubingprotects the bottom portion of the wire. (b) Preferential etching at themeniscus and shifting of the meniscus as etching progresses results in atapered shape. (c) The dropped tip is used. Etching immediately stopsfor dropped tip once it is separated from the wire.Ag is a material with low dielectric loss, making it optimal for STML. Dueto the importance of tip geometry in enhancing the luminescence signal, Ag tipswere electrochemically etched, ensuring a well-defined and sharp tip shape. Theprocedures for Ag tip preparation are adapted from [71, 79].The etching setup is shown in Figure 3.4. A 0.404 mm diameter Ag wire wascleaned with acetone and isopropyl alcohol to give a clean etching surface. Teflontubing covered the bottom ∼ 3 mm of the wire to protect the region from chemicaletching. The wire was then clamped and suspended in a solution of 50% w/w citricacid in deionized water, with the meniscus about 1.5 mm above the teflon protectedregion. A cylindrical mesh electrode was submerged into the solution with the wirein the centre to provide uniform etching on all sides. The wire was then etchedusing an alternating current (AC) voltage source at 39 V, with the negative terminalconnected to the wire, and the ground terminal connected to the mesh electrode.Etching typically took 30 min. Surface tension effects and preferential etchingat the meniscus produces a sharp metallic tip from the submerged portion of thewire, which drops into the solution, breaking electrical contact with the negativeterminal and ending etching processes (Figure 3.4). The tip was removed fromthe solution with the sharp side facing away from the meniscus, as the surface28(a) (b)Figure 3.5: Ag tip imaged with optical microscope after etching. (a) is at 10xmagnification, and (b) is at 20x magnification.tension can bend the tip. The teflon tubing was removed, and the tip was rinsedin deionized water, followed by acetone and isopropyl alcohol. The tip was thenexamined with an optical microscope to ensure the tip was macroscopically sharp,and possessed a shiny metallic surface (Figure 3.5). With the tip exposed to air,the Ag is susceptible to oxidation which can quench luminescence signals. If thetip passed inspection, it was immediately mounted and placed under vacuum in themicroscope. We found the most success with Ag tips that were exposed to air forno longer than 15 min.29Figure 3.6: Plasmon emission with Ag tip on Ag(111) (Vb = 3V, It = 200pA,tx = 10s). Emission intensity ∼ 0.5–1.0 a.u. is sufficient for STMLexperiments.1Due to the softness of Ag, it is extremely important that the tip does not crashinto the surface. To prepare the tip for STML experiments, the tip was repeatedlypulsed with voltages ranging from 5− 10V in order to remove residue and oxidefrom the tip. The shape of the tip was modified by piezo-controlled indentationinto the metal substrate, ranging from 1−10nm. With the optical setup aligned(discussed in the next section), the tip should demonstrate strong and broad plas-monic emission around ∼ 600nm on Ag(111) at bias voltage Vb = 3V, tunnellingcurrent It = 200pA, and exposure time tx = 10s (Figure 3.6). These parametersoffer a baseline signal for later comparison.3.1.3 External optical setupLight collected by the lens inside the SPM head exited the LT chamber through aviewport. The external optical setup is shown in Figure 3.7. A mirror mounted on1Due to issues with intensity calibration on the spectrometer, the intensity of emission is given inarbitrary units rather than photon counts.30the viewport directed the photons toward an optical table installed on the micro-scope. On the optical table, another mirror directed the light into a lens mountedon an xyz micrometer stage, focusing the light into a spectrometer. The spectrom-eter energetically resolved the incoming photons, and the signal was detected witha LN2 cooled charge-coupled device (CCD). The external optical setup was de-signed by T. Roussy [79].Figure 3.7: Diagram of external optics on optics table mounted on micro-scope. The red line represents the beam path.Before starting any experiment, the CCD was calibrated for each spectrometergrating using a mountable Hg lamp. The provided software uses the characteris-tic emission peaks of Hg to relate CCD pixels to photon wavelength, with typicaluncertainty values of ±0.1 nm. Intensity calibration was not performed due to is-sues with the provided calibration lamp, hence the intensity of measured emissionsare on an arbitrary scale rather than photon counts. However, comparison betweenrelative peak intensities are still possible in the arbitrary scale.In this work, two spectrometer gratings were used, with grating densities 30031gratings (gr)/mm and 600 gr/mm.2 Higher grating density allows for higher resolu-tion at the expense of spectral range. Once calibrated, all light sources in the roomwere covered or turned off and a background spectrum was taken. This backgroundspectrum is subtracted from all experimental data, and accounts for thermal noiseand stray light in the room.The optical setup was then aligned to ensure that emission from the tip-samplejunction entered the spectrometer. To begin, the lens on the optical table was re-moved, and the tip was coarse approached to the sample. The face of the spectrom-eter was illuminated such that a shadow of the slit could be seen on the surface ofthe metallic substrate. The shadow of the slit and the tip-sample junction werealigned, either by moving the tip or changing the angle of the two mirrors. Withthe tip-sample junction and spectrometer slit aligned, the lens can be mounted ontothe micrometer stage. Shining light in through the camera port, the light reflectedoff the tip-sample junction can be focused into the spectrometer slit by adjustingthe position of the lens.During data acquisition, the tip was held at a point on the sample with a certainbias voltage (Vb) and tunnelling current (It). The spectrometer shutter was thenopened for an exposure time (tx). All measurements were done with the largestslit opening (3 mm) to maximize the amount of light entering the spectrometer.This comes at a cost: spectral features that are sharper than a full width at halfmaximum (FWHM) of 3 nm (for the 300 gr/mm) or 1 nm (for the 600 gr/mm) arebroadened and not resolved [81].3.2 Sample preparationSample preparation involves cleaning the substrate, depositing the NaCl film, anddepositing the molecules. In this thesis, five different organic semiconductingmolecules were studied, each with different deposition parameters. All samplepreparation procedures will be discussed in detail.2Some signal may be lost due to the imperfect reflectivity of the gratings. Reflectivity of thegratings at the regions of interest are around 70−80% [80]32Ag(111) substrateA top-hat silver crystal with (111) termination was the primary substrate used forexperiments. The crystal surface was cleaned by sputtering for 20 min using ion-ized Ar gas, with ionizing potential ∼ 1kV and preparation chamber pressures atPprep ≈ 3×10−6 mbar. The substrate was then annealed using an e-beam heater atTsub = 420 ◦C for another 20 min. This was repeated 2–3 times depending on thestatus of the crystal surface.A typical scan of a clean Ag(111) surface is seen in Figure 3.8. When scannedwith a sharp metallic tip, step heights are typically ∼ 2A˚, and the STS point spec-tra has a “kink” at −67 mV. This corresponds to a step in the density of states,approximated by the normalized differential conductance, which is a result of theonset of the surface state of Ag(111) [82].(a) (b)Figure 3.8: STM and STS of bare Ag(111) substrate. (a) STM topography(30nm×30nm, Vb = 1V, It = 10pA). (b) Surface state of Ag(111),indicated at −67 meV, seen in normalized dI/dV .Au(111) on Mica substrateAs an alternate substrate, a thin film of (111) terminated gold on mica was used.The cleaning is similar to the procedures for Ag(111), but with sputtering potential∼ 0.75kV, and annealing temperature Tsub = 345 ◦C.33An STM image of the Au(111) thin film (Figure 3.9) shows a herringbonestructure as a result of the reconstruction of the surface into regions of face centredcubic and hexagonal close packed atomic arrangements [83]. Step heights on thesurface are ∼ 2A˚. The STS spectra shows that the onset of the surface state isaround −450mV.(a) (b)Figure 3.9: STM and STS of bare Au(111) substrate. (a) STM topography(30nm×30nm, Vb = 1V, It = 5pA). (b) Surface state of Au(111), in-dicated at −100 meV, seen in normalized dI/dV .NaCl depositionFilms of NaCl(001) were thermally deposited onto the metallic substrates. Thisinsulating film partially decouples the molecule from the metallic substrate, whilestill allowing for tunnelling between the tip and sample [43]. The decoupling isalso necessary for STML experiments, as emission from excitons in the moleculeare quenched by electronic pathways between the molecule and the metal substrate.34Figure 3.10: STM image of 2 and 3ML NaCl on Ag(111) substrate(100nm×100nm, Vb = 1V, It = 10pA). NaCl islands are identifiedby the step heights and surface state shifts.NaCl was thermally deposited using a home-built Knudsen cell. The thicknessof the film can be controlled by deposition temperature, time, and substrate tem-perature. With the substrate held at Tsub = 100 ◦C, NaCl was deposited for 12 minwith deposition temperature at approximately 550 ◦C.STM of bilayer and trilayer (2-monolayer (ML) and 3ML) NaCl on Ag(111) isshown in Figure 3.10, seen as rectangular terraces. The step height of 2ML NaClis around 3 A˚, while 3ML is at 4.5 A˚. Experiments were carried out on the bilayer(2ML) for better molecule and tip stability.The presence of the NaCl layers energetically shifts the surface state of themetallic substrates (Figure 3.11). This signature in the STS spectra is useful fordetermining whether a surface is metallic or insulating NaCl.35(a) (2ML)NaCl/Ag(111) (b) (2ML)NaCl/Au(111)Figure 3.11: Shifted surface state of substrates due to bilayer NaCl. (a) Sur-face state of Ag(111) shifted to 100 meV, and (b) Au(111) shifted to−250 meV.Molecule depositionFive different organic semiconducting molecules were used in our experiments(Figure 3.12). All molecules were deposited on a cold sample in the LT chamber,with substrate temperature between 4.3 K and 4.5 K. The deposition parametersare summarized in Table 3.1.Figure 3.12: Chemical structures of all molecules presented in this thesis.36The HMAT derivative molecules were synthesized by C. Tonge in the Hudsongroup in the chemistry department of the University of British Columbia (UBC).To obtain the deposition temperature for these molecules, thermogravimetric analy-sis was performed to give an approximate temperature at which the molecule beginsto degrade. The molecules were then deposited onto the sample at approximately50 ◦C below the degradation temperature, and the sample scanned for presence ofthe molecules. The deposition temperature was ramped up in 5 ◦C increments untilthe molecule was found on the surface.MoleculeDepositiontemperature (◦C)TimePTCDA 325 ∼ 30sZnPc 370 ∼ 30sF8ZnPc 370 ∼ 30sHMAT-O 160 5 minHMAT-TZ 330 5 minHMAT-HZ 330 10 minTable 3.1: Deposition temperature and time for each molecule used in exper-iments. Deposition times vary depending on the desired coverage. Thetimes listed are for sparse coverages. The deposition may be repeatedmultiple times until coverage is sufficient.Unfortunately, the HMAT derivatives were not stable at the sublimation tem-perature. Repeated degassing to remove impurities in the crucible resulted in con-tinued deposition of molecular fragments. In order to preserve the integrity of themolecules, the deposition temperature was lowered, and the deposition time in-creased to the order of minutes. Final successful parameters are listed in Table Simulation methodsDensity functional theory (DFT) calculations were performed for the molecules,giving the molecular orbitals and energy levels of the isolated molecules. The re-sults can be qualitatively compared to the experimental results. First, the chemical37structures were drawn with the Avogadro software [84], which generated a filewith the three-dimensional (3D) coordinates of the atoms. As organic moleculescan be flexible, different conformations were generated using the knowledge-basedexperimental-torsion basic knowledge distance geometry (ETKDG) algorithm [85]built into the Python package RDKit [86]. The ETKDG algorithm generates molec-ular structures based on empirically defined libraries of torsional angles, ring con-formations, and atomic distances. Molecular dynamics with the Universal ForceField (UFF) [87] was used to calculate the energy of the conformations, and thelowest energy conformation was selected for DFT calculations. Due to the manyaromatic rings in the molecules we studied, the possible conformations were lim-ited to planar structures, comparable to the planar configuration of the moleculeson our metallic substrate.Molecular orbitals and energy levels of free molecules were calculated usingDFT Gaussian 16 software package [88]. The Becke 3-parameter Lee-Yang-Parr(B3LYP) functional [89, 90] and the 6-31G(d) [91] basis set were used for allcalculations. A geometry optimization was performed on the conformer generatedfrom molecular dynamics until the average force on all atoms were below 3×10−4 Ha/rbohr. Finally, electronic structures of the fully relaxed molecules werecalculated. The molecular orbitals were plotted with Avogadro, with an electrondensity isosurface value of 0.02 electron/r3Bohr. The DOS curves were extractedwith the Multiwfn software, with molecular states broadened by a Gaussian with astandard deviation of 0.25 eV.It is important to reiterate that the DFT results can only be qualitativelycompared with the experimental results. The DFT calculations on the isolatedmolecules do not account for the interactions between the substrate and themolecule such as hybridization or van der Waals interactions. Additionally, ge-ometrical changes to the molecule occur when they are adsorbed onto a surface,and a variety of stable conformations are possible.38Chapter 4Luminescence from OrganicMoleculesPrior to the work done in this thesis, STML from a single organic molecule hasnot yet been detected on our system. The system was designed by T. Roussy, andthe plasmonic emission had been detected [79]. To optimize and ensure the vi-ability of the experimental setup, STML was performed on prototypical organicsemiconducting molecules PTCDA [30, 31] and ZnPc [32–35, 92, 93], both ofwhich have had successful reports of STML experiments. Through the SPM ofindividual molecules, we can explore the influence of the local environment onthe optoelectronic properties of the organic semiconductor. Fluorination of ph-thalocyanine molecules has previously been used to tune the electronic and opticalenergies of the molecule in bulk [39, 94]. Further experiments were carried outon F8ZnPc to explore the effects of fluorination on the structural, electronic andoptical properties of ZnPc.During STML experiments, the tip was parked on top of the molecule with acertain bias, and setpoint current. The shutter to the spectrometer was then openedfor a certain exposure time (tx).394.1 Plasmon emission from substratesPlasmon emission from the bare metallic surface can vary in energy and intensitydepending on the geometry of the tip. Aside from attaining a sharp metallic tipfor STM and STS, the tip needs to be poked and pulsed until it produces a strongplasmonic response in the energy region of interest. Representative spectra of theplasmon on Ag(111) and Au(111) with a Ag tip is presented in Figure 4.1. Thesame Ag tip was used to acquire both spectra, allowing for comparison. Due tothe shift in tip-sample junction on Au(111), a result of the thinner substrate, theplasmon emission intensity was decreased by about an order of magnitude whencompared to the Ag(111) substrate. The red-shift in photon energy was caused bythe inherent dielectric response of the gold substrate [67, 68].Figure 4.1: Plasmonic emission on Ag(111) and Au(111) thin film collectedwith the same Ag tip (Vb = 3V, It = 200pA, tx = 10s). The signal onthe thin film is lower by an order of magnitude, and red-shifted. TheAg(111) emission peaks at 573 nm, while the Au(111) emission peaksat 633 nm.When STML experiments were carried out on molecular species, the moleculeswere decoupled by a bilayer (2ML) NaCl spacer, which could modify the plasmon40emission. Using the same Ag tip, modification of the plasmon emission on Ag(111)by layers of NaCl can be seen in Figure 4.2. Overall, there was an enhancement inthe detected photons when compared to the bare metal substrate.Figure 4.2: Plasmonic emission from Ag(111) and on (2ML)NaCl/Ag(111)(Vb = 2.75V, It = 250pA, tx = 10s). There is an enhancement on thebilayer NaCl with minimal shift in energy.Thicker layers of NaCl further dissociate the molecule from the metallic sub-strate, allowing for a more accurate study of the intrinsic properties of the molecule[43]. The thickness of the spacer layer also adjusts the tunnelling rate between themolecule and substrate, changing the rate of charge injection involved in excitonformation. Molecules deposited on thicker layers of NaCl have also demonstratedenhanced molecular luminescence [92, 95]. However, due to the insulating effectsof NaCl, the tip-sample distance is smaller for the same bias and setpoint current,resulting in stronger interactions between the tip and the molecule, and affectingthe stability of the molecule on the NaCl. Bilayer NaCl gave the most consistentand stable configurations for STML.414.2 Study of PTCDAIsolated PTCDA molecules on (2ML)NaCl/Au(111) were studied with STML dueto our familiarity with this system. As discussed before, the Au(111) thin filmsample is slightly out of focus of the lens, resulting in weaker signals. However,PTCDA has a high electron affinity, and the (2ML)NaCl/Ag(111) substrate has asufficiently low work function that the LUMO of PTCDA sits below the Fermienergy of the substrate, resulting in negatively charged PTCDA [36, 78]. By us-ing Au(111), which has a higher work function, the PTCDA molecule was notcharged on the surface, giving a simpler system to probe. Additionally, previ-ous preliminary work in our group has demonstrated luminescence quenching onthe PTCDA/(2ML)NaCl/Ag(111) system [79], although a recent publication hasshown luminescence at higher biases than had been explored [31].Luminescence from a single PTCDA on (2ML)NaCl/Au(111) was detected,with an emission peak at 670 nm or 1.85 eV, at Vb = 3V, It = 200pA, tx = 300s(Figure 4.3). The FWHM of the peak is about 90 meV, quite broad for singlemolecule emission. During the acquisition, the PTCDA shifted away, resulting inadditional photon counts from the plasmonic emission without the molecule signalpresent.42Figure 4.3: Luminescence on PTCDA on (2ML)NaCl/Au(111) at positivebias (Vb = 3V, It = 200pA, tx = 300s), with corresponding plasmon onsubstrate (Vb = 3V, It = 200pA, tx = 30s). Inset is the correspondingSTM image with tip position during STML indicated (20nm×20nm,Vb = 2V, It = 10pA).We will compare our results to references [30] and [31]. In the former,Rzez´nicka et al. studied self-assembled bilayers of PTCDA on Au(111), with thefirst monolayer acting as the decoupling layer. The authors reported broad emis-sion peaks at 1.8 eV and 2.35 eV, with FWHMs of approximately 100 meV, at abias voltage of 3 V. In the latter, Kimura et al. studied single negatively chargedPTCDA on (3ML)NaCl/Ag(111), detecting a sharp emission peak at 2.45 eV at abias of −3.4 V.The peak observed in Figure 4.3 most closely resembles the lower energy1.8 eV emission seen by Rzez´nicka et al., who attributed the emission to the CTexciton formed between the first and second monolayers of PTCDA. However, oursystem only consisted of a single isolated PTCDA molecule, with no neighbouringmolecules to form a CT complex. Additionally, the broadness seen by Rzez´nickaet al. could be explained by the effects of adjacent molecules, but our isolated43PTCDA system exhibits similar breadth in emissions. Regarding the higher energyphoton emission at ∼2.4 eV, both publications attributed the emission peak to theS1(0)→ S0(0) exciton recombination. However, we do not see this peak in ourexperiment, possibly due to the weak plasmonic enhancement in that region of thespectrum.While the previous references are useful as comparisons, the PTCDA on(2ML)NaCl/Au(111) system studied here was isolated and uncharged. The lu-minescence observed on this system may be the S1(0)→ S0(0) exciton transitionfor a neutral PTCDA molecule. Additionally, the luminescence was observed at ahigh positive bias of 3 V. The molecules were noticeably more unstable at positivebias, shifting and rotating during acquisition. It is possible that over the exposuretime interval, different vibrational transitions were suppressed and enhanced dueto the conformational changes in the molecule, resulting in a shifted and broadenedemission peak.It is also possible that an alternative mechanism contributed to the emissionseen in Figure 4.3, such as the plasmon-mediated exciton formation, or chargeinjection into LUMO+1 of PTCDA. While there remain open questions about theemission seen on PTCDA, due to stability issues, further experiments were notperformed. A more thorough investigation of the shifted and broadened emissionat positive biases is required to fully understand the signal observed.4.3 Study of ZnPcZnPc is a relatively planar organic semiconducting molecule that can be thermallydeposited, making it optimal for SPM study. To maximize STML signal, ZnPc wasdeposited on (2ML)NaCl/Ag(111) and probed with the Ag tip. With parametersVb = 3V, It = 300pA, tx = 300s, we were able to detect an emission peak fromZnPc at 630 nm or 1.96 eV (Figure 4.4).44Figure 4.4: Luminescence on ZnPc on (2ML)NaCl/Ag(111) at positive bias,with corresponding plasmon on substrate. Both spectra were taken atVb = 3V, It = 300pA, tx = 300s. Inset is the corresponding STM imagewith tip position during STML indicated (15nm×15nm, Vb = 2V, It =10pA).There are notable differences between our results and the results of past reportsof STML on ZnPc [33, 34, 92, 93]. An example of ZnPc STML spectra from theliterature is provided in Figure 4.5. The previously reported main emission peakwas at 652 nm or 1.9 eV. This peak was attributed to the S1(0)→ S0(0) transitionin ZnPc. The peak energy in Figure 4.4 is higher in energy, and about four timesbroader, with FWHM of about 60 meV while previously reported emissions haveFWHM of about 15 meV.In contrast to past reported results taken at −2.5 V, we detected luminescenceat a tip-sample bias of 3 V. ZnPc has two equivalently stable adsorption angleson NaCl [96], and rapid shuttling between these geometries has been observed.Similar to PTCDA, the ZnPc was noticeably more unstable at high positive bi-ases, rotating and shifting more frequently, seen as sub-A˚ngstro¨m jumps in the tip45Figure 4.5: Luminescence on ZnPc on (2ML)NaCl/Ag(111) at negative bias.Spectra was taken at Vb =−2.5V, It = 200pA, tx = 20s. The emissionpeak is at 653 nm, and the FWHM is 15 meV. Figure taken from [38].height. This instability can change the vibrational transitions of the ZnPc observedin the luminescence, shifting and broadening the emission signal in Figure 4.4.Another factor that may explain the discrepancies observed is the presence of analternative mechanism of molecular luminescence. The luminescence seen at posi-tive bias may be a result of plasmon-mediated exciton recombination or some othermolecular luminescence mechanism.4.4 Study of F8ZnPcThe energy of molecular orbitals can be tuned through halogenation, such as the in-troduction of fluorine into a molecular structure. Studies of fluorination on organicmolecule films have demonstrated a rigid downward shift in energy of the HOMOand LUMO states, effectively raising the Fermi level [39, 94]. Using SPM, theeffects of fluorination could be understood at the single molecule level.4.4.1 Comparison of STM/STS of ZnPc and F8ZnPcZnPc and F8ZnPc were co-deposited on (2ML)NaCl/Au(111), and STM and STSwas conducted using a Pt/Ir tip (Figure 4.6). In the topography, the effects offluorination on the adsorption geometry of the molecule is immediately obvious.ZnPc on bilayer NaCl appears 16-lobed, as it shuttles between two equally stableadsorption angles, agreeing with past STM studies of ZnPc [32, 96]. The fluorinesin F8ZnPc “anchor” and stabilize the molecule on the NaCl, giving an 8-lobed46structure in the STM topography. The effects of the fluorination on the adsorptiongeometry of the molecule will be discussed in detail in Section 4.4.3.Figure 4.6: STM and STS data of ZnPc and F8ZnPc on (2ML)NaCl/Au(111).(a) STM images of ZnPc and F8ZnPc (3nm×3nm, Vb = −2.5V,It = 5pA). (b) Normalized dI/dV of ZnPc and F8ZnPc on(2ML)NaCl/Au(111).STS reveals a downward shift of about 0.49 eV in the HOMO state onset andabout 0.41 eV in the LUMO state onset of F8ZnPc relative to the states of ZnPc(Figure 4.6b), resulting in a 0.09 eV widening of the band gap due to fluorinationof ZnPc. This result differs slightly from the rigid shift observed in mixed films offluorinated and non-fluorinated ZnPc [39], likely due to the differences in configu-ration and local environment of the molecules between the experiments.4.4.2 Luminescence on F8ZnPcF8ZnPc was thermally deposited onto (2ML)NaCl/Ag(111) and probed with a Agtip. With positive bias Vb = 3V, It = 250pA, tx = 60s, a weak emission peak was47detected at 675 nm or 1.83 eV, with FWHM of approximately 40 meV (Figure 4.7).Figure 4.7: Luminescence on F8ZnPc on (2ML)NaCl/Ag(111) at positivebias (Vb = 3V, It = 250pA, tx = 60s), with corresponding plasmonon substrate (Vb = 3V, It = 250pA, tx = 10s). A peak is observed at675 nm. Inset is the corresponding STM image with tip position duringSTML indicated (20nm×20nm, Vb = 2V, It = 10pA).When compared to the positive emission on ZnPc (Figure 4.4), the emission inFigure 4.7 is slightly sharper, possibly due to the stabilizing effect of the fluorineatoms in F8ZnPc. However, shifts and rotations of F8ZnPc were observed duringpositive bias STML, broadening the emission relative to negative bias emissionsseen in previous studies on ZnPc (Figure 4.5). Additionally, the positive bias emis-sion peak on F8ZnPc is red-shifted by about 45 nm when compared to that of theZnPc, but without further knowledge of the vibrational transitions involved in thepositive bias emission, quantitative comparisons between the emissions cannot bemade.STML was also conducted at negative bias for F8ZnPc, and luminescence wasdetected at Vb = −2.5V, It = 200pA, tx = 120s. A strong and sharp S1(0)→48S0(0) emission peak was observed at 640 nm or 1.93 eV, along with well-resolvedsatellite peaks corresponding to vibrational transitions (Figure 4.8). The FWHMis approximately 20 meV, comparable to the negative bias emissions observed onZnPc.Figure 4.8: Luminescence at negative bais detected on F8ZnPc on(2ML)NaCl/Ag(111) with 300 gr/mm grating (Vb = −2.5V, It =200pA, tx = 120s). Peak detected at 640 nm, with lower energy vi-brational peaks. Inset is the corresponding STM image with tip positionduring STML indicated (5nm×5nm, Vb =−2V, It = 10pA).When compared to the positive bias emission in Figure 4.7, the negative biasemission is blue-shifted by 35 nm. A broad vibrational peak in the negative biasemission in Figure 4.8 coincides at the energy of around 670 nm—the same energyas the positive bias emission peak in Figure 4.7—providing supporting evidencethat lower energy vibrational transitions were responsible for the observations inthe positive bias STML experiments. Due to complications involving the shiftingand broadening of spectral features due to different adsorption geometries and sta-bility of the molecules at positive bias, further positive bias STML investigation is49required to determine the source and mechanism of the emission.When compared to published negative bias STML data on ZnPc (Figure 4.5),the emission peak on F8ZnPc in Figure 4.8 is slightly higher in energy by about0.03 eV. This suggests a widening of the optical gap, corresponding to an asym-metric shift of the HOMO and LUMO states due to fluorination of ZnPc, wideningthe band gap of the molecule. The change in the optical gap is in accordance withthe change in the band gap seen in STS of the ZnPc and F8ZnPc (Figure 4.6). Localenvironment, adsorption geometry, and molecule-substrate interactions may affectthe energy levels of F8ZnPc; this is explored in detail in Section 4.4.3.Higher resolution spectra were obtained with the 600 gr/mm grating to betterresolve the vibrational emission peaks (Figure 4.9). With the tip positioned abovedifferent parts of the molecule, changes in the lower energy vibrational satellitepeaks were observed due to different preferred transitions between the excited S1state and the vibrational states of S0. Sub-molecular location dependent emissionfrom vibrational transitions has been reported before on ZnPc [33, 34]. Shiftsin the energy of the main S1(0)→ S0(0) transition were minimal (∼±1nm). Thestrongest emission was observed with the tip positioned above the lobes of F8ZnPc,and so all further experiments were performed at this location for maximal signalstrength.50Figure 4.9: Changes in vibrational satellite peaks as the tip is positioned overdifferent parts of F8ZnPc on (2ML)NaCl/Ag(111) with 600 gr/mm grat-ing. STML parameters were Vb = −2.5V, It = 100pA, tx = 300s foreach point. Inset is the corresponding STM image with tip position dur-ing STML indicated (5nm×5nm, Vb =−2V, It = 10pA).Diversity in molecular typesSurprisingly, the STM scans at negative biases revealed a variety of topographi-cally different molecules on the surface that were not previously evident at positivebiases. To eliminate the possibility of fragmentation or degradation of molecules,a new sample of F8ZnPc was loaded into the evaporator crucible, and the samplewas prepared again. However, the same diversity was still present (Figure 4.10).Additionally, we were able to induce switching from one type to another with thetip, suggesting differences in the interaction between molecule and substrate as thecause for the distinct types. Possible reasons include the presence of defects orcontaminants, and adsorption geometry of F8ZnPc on the substrate.51Figure 4.10: Same sample of F8ZnPc on (2ML)NaCl/Ag(111) scanned atpositive and negative biases. (a) STM image at 20nm×20nm, Vb =2V, It = 5pA. All molecules appear similar. The striped Moire´ pat-tern between NaCl(001) and Ag(111) lattices can be observed. (b)STM image at 20nm×20nm, Vb = −2V, It = 5pA. Same sample ofmolecules showing diversity in topography at negative bias.Further STML experiments at negative biases were conducted on the differenttypes of F8ZnPc, with S1(0)→ S0(0) transition peaks varying in wavelengths from630 nm to 643 nm. In order to relate the emission peaks to the electronic statesof the molecules, STML along with STS was performed on three of the observedtypes (Figure 4.11).52Figure 4.11: STM, STS and STML of 3 types of F8ZnPc observed on(2ML)NaCl/Ag(111). (a) Three molecules labelled in STM image(7.5nm×7.5nm, Vb = −2.7V, It = 10pA). (b) Normalized dI/dVof each molecule. (c) Luminescence of each molecule taken at Vb =−2.5V, It = 100pA, tx = 120s.With increasing band gaps, measured as the difference in onset energy ofHOMO and LUMO, there is a corresponding increase in the optical gaps, observedas higher energy S1(0)→ S0(0) emission peaks from exciton recombination. Theresults are tabulated in Table 4.1. Further experimentation on this sample was notattempted due to the unstable nature of the Ag tip, making it unsuitable for highresolution STM or STS, and also the wide variety of types of F8ZnPc observed on53the surface.Type STS band gap (eV) STML optical gap (eV)Molecule 1 2.00 1.94Molecule 2 2.32 1.96Molecule 3 2.75 1.97Table 4.1: Table of band and optical gaps of three types of F8ZnPc on(2ML)NaCl/Ag(111), extracted from STS and STML experiments re-spectively.4.4.3 Adsorption geometry of F8ZnPcTo further investigate the adsorption geometry of F8ZnPc on the substrate, the samesample was prepared after a full system vent and bake-out to ensure all contami-nants in the chambers were purged.1 The Pt/Ir tip was used for consistent STM andSTS of the F8ZnPc on (2ML)NaCl/Ag(111). Despite the same sample preparationconditions, there were clear differences between the sample used for STML exper-iments, and the sample used for investigating the adsorption geometry of F8ZnPc.The insulating NaCl layers had a lower density of defects, and there were fewertypes of F8ZnPc after deposition, making classification more manageable. Thissuggests that some of the molecular variety may have been the result of underlyingdefects.1It was during this time that the new windows were installed (Figure 3.3).54Figure 4.12: The four identified types of F8ZnPc on (2ML)NaCl/Ag(111).(a) STM images of the four types (Vb = −2.5V and It = 5pA). STMof types 1–3 are 2.5 nm×2.5 nm, while type 4 is 3 nm×3 nm. (b)Normalized dI/dV from point STS taken on each of the four types.Nevertheless, we identified four main types of F8ZnPc, each with distinct to-pography and STS resonances, as seen in Figure 4.12. The types were named inorder of most to least abundant: T1, T2, T3, and T4. The band gaps changed witheach molecule type, as observed in the sample used for STML. The band gaps,measured from the onsets of HOMO and LUMO states, for each molecule typeare summarized in Table 4.2. There were additional types, but, due to rarity andmolecular instability, they were not classified or examined in depth.Type STS band gap (eV)T1 2.75T2 2.58T3 2.55T4 3.01Table 4.2: Table of band gaps of the four types of F8ZnPc on(2ML)NaCl/Ag(111), extracted from STS experiments.Furthermore, the configuration of each type of F8ZnPc relative to the under-55lying NaCl lattice was examined. For each molecule type, the surrounding NaClwas scanned with high current and low bias to give atomic resolution of the NaCllattice, as seen in Figure 2.3. The scanning parameters were then restored whenscanning over the molecule. The lattice was then extended for the entire image,with the bright atoms corresponding to Cl– ions [45]. The results for each type ofF8ZnPc are summarized in Figure 4.13.Figure 4.13: STM images of four identified types of F8ZnPc on(2ML)NaCl/Ag(111) with underlying NaCl lattice superimposed(Vb = −2.5V and It = 5pA). The circles indicate the Cl– ion. Types1–3 are 2.5 nm×2.5 nm, while type 4 is 3 nm×3 nm.In the T1 configuration, F8ZnPc has a 4-fold rotationally symmetric 8-lobedstructure. It also has a relatively low height of 2.56 A˚ in STM, appearing as adarker molecule in the topography. In this configuration, the centre and lobes ofthe molecule sit on the chloride ions, similar to adsorption geometries of othermetal-phthalocyanine molecules on NaCl(001) films [96]. The normalized dI/dVspectra of T1 also corresponds to that of Molecule 3 identified in Figure 4.11,which also has a band gap of 2.75 eV, and demonstrated a strong S1(0)→ S0(0)emission peak at 630 nm, or 1.97 eV.Configuration T2 is also 4-fold rotationally symmetric and 8-lobed, but themolecule is brighter in the centre. The T2 F8ZnPc has the centre and lobes adsorbedthe Na+ ion. During high bias and current scans, T2 F8ZnPc can be changed intoT1 through tip manipulation, demonstrating that T2 is a meta-stable state. TheT3 F8ZnPc appears similar to T2, but the lobe pairs are closer together, and themolecule appears to be rotated relative to the NaCl lattice. T3 also has the centreand lobes on the Na+ ion, and can be converted into T1 through tip manipulation.T4 is 2-fold rotationally symmetric, with the axis of the molecule along the [100]56direction on the NaCl. The centre of T4 also sits on Na+, and the darker lobesobserved may be due to interactions of the fluorine atoms with the sodium ions ofthe substrate.Overall, adsorption geometries T2–4 were all found with the centre and lobeson the sodium ions of the NaCl film. This was not observed for ZnPc, and indicatesthat the peripheral fluorine atoms can create meta-stable adsorption configurations,which in turn change the electronic and optical properties of the molecule. Byscanning at high bias and current over types 2–4, the molecules were convertedinto T1, indicating that T1 was the most stable configuration of F8ZnPc on NaCl.The brighter molecules observed in the STML sample (Figure 4.11) likely corre-sponded to the meta-stable Na+ adsorbed configurations (T2–4) of F8ZnPc.Comparing the results on ZnPc and F8ZnPc, there are two processes involvedin the changes in electronic and optical properties due to fluorination of phthalo-cyanine molecules: the doping effects of the additional fluorine atoms, and thestabilization of new adsorption configurations of the molecule on the surface.The data seen in Figures 4.12 and 4.13 confirm that the effects of adsorptiongeometry resulted in the diversity of molecules with differing orbital energy levelsobserved in the previous sample. And from the STML data in Figure 4.11, there isa correlation between the change in electronic structure and the observed lumines-cence. But due to the differences in observed types of F8ZnPc on our two samples,it is difficult to draw a direct relationship between adsorption geometry and emis-sion energy. Additional STML and STS experiments could rectify this, althoughidentification of classified types may be complicated by the instability of the Agtip.57Chapter 5Engineering Organic MolecularEnergy LevelsThe energy levels of organic semiconducting molecules determine the electronicand optical properties of the molecule. For different applications, there are differ-ent optimal energy level alignments. Through variation in the design of organicmolecules, possible through organic synthesis techniques, the energy levels of theorganic semiconductor can be engineered. This is realized through the addition offunctional groups to an organic molecule or polymer [39, 97]. In particular, thischapter will discuss the functionalization of HMAT with various acceptor com-plexes, with HMAT acting as the donor, and the resulting molecular orbitals andtheir energy levels. I will also discuss the attempts of STML experiments on theseHMAT derivative molecules.5.1 Introduction to HMATHexamethylazatriangulene (HMAT) is a highly stable and planar molecule withinteresting optoelectronic properties. With a theoretical HOMO-LUMO gap of4.45 eV, HMAT is a good electron donor molecule and a deep blue-violet fluo-rophore [41] (Figure 5.1). With a rigid planar structure, electrons in HMAT haveenhanced pi-conjugation, giving it a high quantum efficiency, along with a highphoton absorption cross-section [98]. Additionally, functionalized HMAT deriva-58tives have demonstrated optical effects such as thermally activated triplet exic-ton formation [42], and two-photon excited fluorescence [40, 99], making thesemolecules candidates for OLED and biological imaging applications.Figure 5.1: Calculated density of states for HMAT in gas phase. HOMO andLUMO isosurfaces are pictured. The DFT band gap is 4.45 eV.Being relatively difficult to synthesize, the functionalized HMAT derivativeswere not available commercially and were provided by our collaborators in theHudson group at UBC Chemistry. We examined three different HMAT derivativemolecules: HMAT-O, HMAT-TZ, and HMAT-HZ. All prior studies on HMATand its functionalized derivatives were in chemical ensembles with conventionalanalytical techniques. With SPM, for each functional group, we can visualize thesub-molecular spatial distribution of the orbitals, and measure the energy levels ofthe single molecule.5.2 STM/STS study of HMAT derivativesThe molecules were deposited onto the bare Ag(111) surface, and probed with aPt/Ir tip dipped in silver. While the HMAT molecule is highly stable, the attachedfunctional groups were fragmented at high deposition temperatures. Large areaSTM scans showed an abundance of fragments on the surface for each moleculedeposition, even after repeated degassing.Fortunately, intact molecules were found with still attached functional groups.59The molecules were scanned, and a pixel-by-pixel STS grid was performed on themolecules on Ag(111). Taking the normalized differential conductance, the LDOSwas plotted as a function of energy, and the HOMO and LUMO orbitals were vi-sualized spatially. As the molecules were studied directly on the Ag(111) surface,the resonances in the normalized dI/dV were broadened due to hybridization withthe metal, and so the peaks in the STS spectra, rather than the onsets, were usedto identify the molecular states. DFT calculations were performed for each of theHMAT derivative molecules and used as qualitative comparisons to the experi-mental results. The calculations were performed for the isolated molecules in gasphase, and do not account for the effects of the substrate. The results for HMAT-O,HMAT-TZ, and HMAT-HZ are shown in Figures 5.2, 5.3, and 5.4, respectively.60Figure 5.2: STM and STS data on HMAT-O on Ag(111). All scale barsare 1 nm. (a) Normalized dI/dV on the donor (HMAT) and acceptor(oxadiazole) complex, indicated on the inset (6nm×6nm, Vb = 0.1V,It = 5pA). (b) STS grid image of HOMO and LUMO at respectivebiases. Band gap is 3.76 eV. (c) DFT calculated HOMO and LUMO.61Figure 5.3: STM and STS data on HMAT-TZ on Ag(111). All scale barsare 1 nm. (a) Normalized dI/dV on the donor (HMAT) and acceptor(triazine) complex, indicated on the inset (4nm×4nm, Vb =−1V, It =10pA). (b) STS grid image of HOMO and LUMO at respective biases.Band gap is 2.57 eV. (c) DFT calculated HOMO and LUMO.62Figure 5.4: STM and STS data on HMAT-HZ on Ag(111). All scale barsare 1 nm. (a) Normalized dI/dV on the donor (HMAT) and acceptor(heptazine) complex, indicated on the inset (4nm×4nm, Vb = −1V,It = 30pA). (b) STS grid image of HOMO, LUMO, and LUMO+1 atrespective biases. Band gap is 2.41 eV. (c) DFT calculated HOMO,LUMO, and LUMO+1.For each of the molecules, there is qualitative agreement in the molecular or-bital distribution between the STS and the DFT—the HOMO state is localized onthe HMAT, while the LUMO state is localized on the acceptor group. For bothexperimental and DFT results, there is little to no overlap observed in the spatialdistribution of the occupied and unoccupied orbitals. In the case of HMAT-HZ, theLUMO and the LUMO+1 were both imaged in STS, and, as seen in the DFT andSTS results, both were localized primarily on the heptazine.The experimental and theoretical band gaps for each molecule are summarizedin Table 5.1. The exact band gap energies do not correspond between DFT andSTS. This is expected, as the DFT calculations do not account for the interactions63between the molecule and substrate. The STS experiments were performed withoutan insulating NaCl layer, so the molecular orbital energies experience significantshifts, and substrate-induced conformation changes can also influence orbital ener-gies.Molecule DFT band gap (eV) STS band gap (eV)HMAT 4.45 —HMAT-O 3.45 3.76HMAT-TZ 3.28 2.57HMAT-HZ 2.54 2.41Table 5.1: Table of DFT calculated, and STS measured band gaps (directlyon Ag(111) substrate) of HMAT derivative molecules. The energy of thegap tends to decrease with increasingly electronegative acceptor groupsattached.Nevertheless, in both the DFT and STS, there is a clear trend in the HOMO-LUMO gap of each of the HMAT derivatives. With the addition of increasinglyelectronegative species (with increasing number of O and N atoms), the LUMOof the molecule is lowered in energy. The HOMO experiences no shift in energyrelative to the Fermi level. The consistent positions of the HMAT HOMO and thetunable acceptor LUMO indicate a highly predictable acceptor-donor strategy fortuning band gaps of HMAT derivatives.Attempts were made to perform STM/STS on HMAT-O on an insulating NaClbilayer. The HOMO state of the molecule on (2ML)NaCl/Ag(111) was found at∼ 3.2V. But due to the fragility and instability of the molecules at |Vb| > 2.5V,no successful STS grids were attained for HMAT-O on (2ML)NaCl/Ag(111). Nofurther experiments were conducted for HMAT-TZ and HMAT-HZ on insulatingNaCl on metal.5.3 STML study of HMAT derivativesWhile the molecules were not decoupled from the substrate, STML was attemptedfor HMAT-O and HMAT-HZ. With successfully resolved molecular states in theSTS, we hypothesized that the bulky methyl groups (six on each HMAT) may have64sterically dissociated the molecule from the underlying Ag(111).However, with the Ag tip parked on the molecule at parameters up to Vb =3V, It = 200pA, tx = 360s, no molecular luminescence was detected, only theplasmonic emission. We also attempted to obtain an absorption spectra by excitingplasmonic photons near the molecule, a technique demonstrated in the past onZnPc [38], but again, only the plasmonic emission was seen with no discernibleabsorption. The same experiments were repeated on Au(111), in an attempt tored-shift the plasmon emission so as to observe the two-photon absorption of themolecules. In all our experiments, no observable STML molecular signatures werefound from the HMAT-O or HMAT-HZ directly adsorbed on either Ag(111) orAu(111).These failed attempts illustrate the importance of the decoupling of themolecule from the substrate in STML. Metal-molecule coupling results in addi-tional energy pathways that rapidly quench excited states, and any emission signalfrom the molecule. Additionally, the instability and large band gaps of HMATderivatives make it difficult to inject charges into the electronic orbitals, or to gen-erate high energy plasmonic emissions to form excitons in the material, without themolecules breaking or moving. Future studies of novel OLED materials will needto take these factors into account.65Chapter 6ConclusionMolecular emission was successfully detected, demonstrating that our system iscapable of STML experiments. In our studies of PTCDA, broad molecular lu-minescence signals were observed at positive biases. These are the first reportedsingle molecule emission signals on uncharged PTCDA. In our study of ZnPc andF8ZnPc, broad emissions signals were again observed at positive biases. The sig-nals differ from STML on the molecules at negative biases. As the molecules weremore mobile when scanned at high positive bias, it is possible that small shifts androtations during STML resulted in changes to the observed exciton recombinationvibrational transitions, giving broadened and shifted luminescence peaks. Otherexplanations include the presence of an alternate pathway for exciton generation,such as photon-mediated, or the tunnelling into higher energy unoccupied orbitals,such as LUMO+1.The effects of fluorination on the luminescence of ZnPc were studied throughthe SPM of F8ZnPc. Single molecule STS showed that there was a widening ofthe band gap, and an overall downward shift of both the HOMO and LUMO statesof the fluorinated ZnPc. STML signals were detected at a bias of −2.5 V, similarto parameters of previously reported STML on ZnPc. The main S1(0)→ S0(0)transition was blue-shifted for the F8ZnPc relative to ZnPc, which is consistentwith the larger band gap. However, photon emission detected on F8ZnPc variedin wavelength from 630–643 nm for different F8ZnPc on the surface. STM scansat negative biases revealed a variety of topographically different molecules, and66point STS on the molecules showed differing tunnelling resonances correspond-ing to shifted molecular states for the molecules in different local environments.Repeated sample preparation and replacing of molecules ruled out the presenceof degraded molecules, and the switching between molecular species through tipmanipulation seemed to indicate that the variety was due to changes in surface ad-sorption of the molecules. Molecules with smaller band gaps in the STS spectraproduced lower energy emission peaks in the STML spectra, demonstrating thecorrelation between the shifts in electronic states and the optical gaps of formedexcitons.After a full system bake-out, a new sample was produced, and the variety ofF8ZnPc was observed again. However, the sample was noticeably different, withfewer defects on the NaCl bilayers. Regardless, the different types of moleculeswere categorized by STS spectral features, STM topographic differences, and rel-ative positions on the NaCl lattice. We conclude that the presence of the fluorinesin F8ZnPc promotes meta-stable adsorption geometries on the Na+ cation whichis not observed for ZnPc. STML emission was detected at 630 nm for the stableadsorption of F8ZnPc on the Cl– anion. The meta-stable configurations gave riseto emissions ranging from 631–643 nm. These changes in conformation and in-teractions with the substrate can affect the electronic and optical properties of themolecules.In our study of HMAT derivatives, using pixel-by-pixel STS, sub-molecularimages of the orbitals for each HMAT molecule were generated. We find thatthe HOMO was localized around the donor complexes, the HMAT groups, of themolecules, while the LUMO was localized around the acceptor complexes. Withthe functionalization of increasingly electronegative acceptor groups, the band gapsobtained through STS were observed to decrease, demonstrating the tuning of elec-tron energy levels in molecules through chemical design. The results observed inour experiments qualitatively agree with DFT calculations of the HMAT deriva-tives in gas phase. STML experiments were attempted for the HMAT derivatives,however, due to the large band gap, high biases would be required to generateexcitons in the material, which could break or move the molecules on the surface.676.1 Future directionsWith the installation of wider bandwidth filter windows in our SPM, reproduc-tion of experiments described in the thesis may be worthwhile, as the cutoff of theIR filter windows would be eliminated in future STML experiments. For furtherimprovements, all samples used for STML should be at the correct height for max-imum emission collection by the in situ lens. Additionally, a photomultiplier tubecan be installed in place of the spectrometer for photon mapping. With softwareimprovements, the position of the tip can be synchronized with the photomultipliertube signal to give sub-molecular photon maps of molecular luminescence.The broad emission peaks in our positive bias STML studies remain unex-plained. In the thesis, several explanations have been provided, but additionalexperimentation may reveal the exact emission mechanism, and explain the diver-gences from previously reported results. In particular, the hypothesis of excitonsformed by tunnelling into the LUMO+1 of uncharged PTCDA may be tested bymapping out the LUMO+1 using bias dependent STML. To test whether the sta-bility of molecules at higher biases have an effect on the observed emission, STMLexperiments can be carried out on ZnPc or PTCDA “anchored” to features such asdefects or step edges on the surface.Future experiments can be directed to the study of systems of organic semi-conductors on the surface. By annealing the sample, molecules can self-assembleinto dimers. The effects of molecule-molecule interactions, such as polarizationand charge transfer, on the formed exciton can be examined by STML on thedimerized molecules. Additionally, heterodimers of different molecules, such asPTCDA-ZnPc or ZnPc-F8ZnPc, can be studied as simplified models of deviceheterojunctions. Successful formation of ZnPc-F8ZnPc dimers have already beendemonstrated on (2ML)NaCl/Au(111), as seen in Figure 6.1.68Figure 6.1: STM image of dimerized F8ZnPc and ZnPc on (2ML)NaCl/-Au(111) (5nm×5nm, Vb = −2.5V, It = 5pA). The white and yellowarrows indicate the F8ZnPc and ZnPc, respectively.In our study of F8ZnPc, we have demonstrated that fluorination allows formeta-stable adsorption geometries of molecules on NaCl films. The different ad-sorption geometries can change the electronic structure of the molecule. Sepa-rately, we have demonstrated that the change in electronic structure correlates withchange in the exciton optical gap. Due to variations in the samples, we cannot makedirect correlations between the various types of F8ZnPc observed and the STML. Afuture direction would be to examine the molecules on cleaner NaCl bilayers withthe Ag tip, giving simultaneous STS and STML measurements. Furthermore, the-oretical simulations of adsorption geometries of F8ZnPc on (2ML)NaCl/Ag(111)may give information on the role of the fluorine atoms on the conformation of themolecule, and the resulting changes in molecular orbitals.Future experiments on the HMAT derivatives are limited due to the incom-patibility of the molecules to SPM on insulating layers on metallic substrates.Nonetheless, experiments on the HMAT molecules have revealed the necessaryproperties of candidate molecules for further STML experiments, such as a planarstructure, and HOMO/LUMO states that do not lie further than 3 V from the Fermilevel. With an appropriate molecule, the effects of chemical design on the excitonicemission can be studied in future STML experiments.69Bibliography[1] Hideki Shirakawa, Edwin J Louis, Alan G MacDiarmid, Chwan K Chiang,and Alan J Heeger. Synthesis of electrically conducting organic polymers:halogen derivatives of polyacetylene (CH)x. Journal of the ChemicalSociety, Chemical Communications, (16):578–580, 1977. → page 1[2] Chwan K Chiang, CR Fincher Jr, Yung W Park, Alan J Heeger, HidekiShirakawa, Edwin J Louis, Shek C Gau, and Alan G MacDiarmid. Electricalconductivity in doped polyacetylene. Physical Review Letters, 39(17):1098,1977. → page 1[3] Xu-Hui Zhu, Junbiao Peng, Yong Cao, and Jean Roncali.Solution-processable single-material molecular emitters for organiclight-emitting devices. Chemical Society Reviews, 40(7):3509–3524, 2011.→ page 1[4] Lei Ying, Cheuk-Lam Ho, Hongbin Wu, Yong Cao, and Wai-Yeung Wong.White polymer light-emitting devices for solid-state lighting: materials,devices, and recent progress. Advanced Materials, 26(16):2459–2473, 2014.→ page 1[5] Luisa Torsi, Maria Magliulo, Kyriaki Manoli, and Gerardo Palazzo. Organicfield-effect transistor sensors: a tutorial review. Chemical Society Reviews,42(22):8612–8628, 2013. → page 1[6] Wesley K Tatum and Christine K Luscombe. pi-conjugated polymernanowires: advances and perspectives toward effective commercialimplementation. Polymer Journal, 50(8):659–669, 2018. → page 1[7] Sarah Holliday, Yilin Li, and Christine K Luscombe. Recent advances inhigh performance donor-acceptor polymers for organic photovoltaics.Progress in Polymer Science, 70:34–51, 2017. → page 170[8] Nieves Espinosa, Lucı´a Serrano-Luja´n, Antonio Urbina, and Frederik CKrebs. Solution and vapour deposited lead perovskite solar cells:Ecotoxicity from a life cycle assessment perspective. Solar EnergyMaterials and Solar Cells, 137:303–310, 2015.[9] Brian A Gregg and Mark C Hanna. Comparing organic to inorganicphotovoltaic cells: Theory, experiment, and simulation. Journal of AppliedPhysics, 93(6):3605–3614, 2003. → page 2[10] Nathan S Lewis. Toward cost-effective solar energy use. Science, 315(5813):798–801, 2007. → page 1[11] Hin-Lap Yip and Alex K-Y Jen. Recent advances in solution-processedinterfacial materials for efficient and stable polymer solar cells. Energy &Environmental Science, 5(3):5994–6011, 2012. → page 1[12] Yunping Huang, Delwin L Elder, Alvin L Kwiram, Samson A Jenekhe,Alex KY Jen, Larry R Dalton, and Christine K Luscombe. Organicsemiconductors at the University of Washington: Advancements in materialsdesign and synthesis and toward industrial scale production. AdvancedMaterials, page 1904239, 2019. → page 1[13] Yimu Zhao, Garrett A Meek, Benjamin G Levine, and Richard R Lunt.Near-infrared harvesting transparent luminescent solar concentrators.Advanced Optical Materials, 2(7):606–611, 2014. → page 2[14] Mitch Jacoby. The future of low-cost solar cells, 2016. URL → page2[15] Ron Mertens. Foldable oleds: introduction and market status, 2019. URL → page 2[16] Daniel L Crossley, Laura Urbano, Robert Neumann, Struan Bourke, JenniferJones, Lea Ann Dailey, Mark Green, Martin J Humphries, Simon M King,Michael L Turner, et al. Post-polymerization C–H borylation ofdonor–acceptor materials gives highly efficient solid state near-infraredemitters for near-IR-OLEDs and effective biological imaging. ACS appliedmaterials & interfaces, 9(34):28243–28249, 2017. → page 2[17] Lingxian Meng, Yamin Zhang, Xiangjian Wan, Chenxi Li, Xin Zhang,Yanbo Wang, Xin Ke, Zuo Xiao, Liming Ding, Ruoxi Xia, Hin Lap Yip,Yong Cao, and Yongsheng Chen. Organic and solution-processed tandem71solar cells with 17.3% efficiency. Science, 361(6407):1094–1098, 2018. →page 3[18] Yong Cui, Huifeng Yao, Jianqi Zhang, Tao Zhang, Yuming Wang, LingHong, Kaihu Xian, Bowei Xu, Shaoqing Zhang, Jing Peng, Zhixiang Wei,Feng Gao, and Jianhui Hou. Over 16% efficiency organic photovoltaic cellsenabled by a chlorinated acceptor with increased open-circuit voltages.Nature Communications, 10(1), 2019. → page 3[19] National Renewable Energy Laboratory (NREL). Champion PhotovoltaicModule Efficiency Chart, 2019. URL → page 3[20] National Renewable Energy Laboratory (NREL). Best Research-CellEfficiency Chart, 2019. URL →page 3[21] Martin Knupfer. Exciton binding energies in organic semiconductors.Applied Physics A, 77(5):623–626, 2003. → page 3[22] Ching W Tang. Two-layer organic photovoltaic cell. Applied PhysicsLetters, 48(2):183–185, 1986. → page 4[23] B Bernardo, D Cheyns, B Verreet, Richard Daniel Schaller, Barry P Rand,and Noel Christopher Giebink. Delocalization and dielectric screening ofcharge transfer states in organic photovoltaic cells. Nature Communications,5(1):1–7, 2014. → page 4[24] Christopher M Proctor, Martijn Kuik, and Thuc-Quyen Nguyen. Chargecarrier recombination in organic solar cells. Progress in Polymer Science, 38(12):1941–1960, 2013. → page 4[25] Yasunari Tamai, Hideo Ohkita, Hiroaki Benten, and Shinzaburo Ito. Excitondiffusion in conjugated polymers: from fundamental understanding toimprovement in photovoltaic conversion efficiency. The Journal of PhysicalChemistry Letters, 6(17):3417–3428, 2015. → page 4[26] Carsten Deibel, Thomas Strobel, and Vladimir Dyakonov. Role of thecharge transfer state in organic donor–acceptor solar cells. Advancedmaterials, 22(37):4097–4111, 2010. → page 4[27] C. W. Tang and S. A. Vanslyke. Organic electroluminescent diodes. AppliedPhysics Letters, 51(12):913–915, 1987. → page 572[28] Anna Ko¨hler and Heinz Ba¨ssler. Triplet states in organic semiconductors.Materials Science and Engineering R: Reports, 66(4-6):71–109, 2009. →page 5[29] M. C. Cottin, E. Ekici, and C. A. Bobisch. STM-induced light emissionenhanced by weakly coupled organic ad-layers. Applied Physics Letters, 112(10):101602, 2018. → page 7[30] Izabela I. Rzez´nicka, Taro Yamada, and Maki Kawai. Light emission fromPTCDA bilayer on Au(111) induced in scanning tunneling microscopy.Surface Science, 605(23-24):2032–2037, 2011. → pages 39, 43[31] Kensuke Kimura, Kuniyuki Miwa, Hiroshi Imada, Miyabi Imai-Imada,Shota Kawahara, Jun Takeya, Maki Kawai, Michael Galperin, and YousooKim. Selective triplet exciton formation in a single molecule. Nature, 570(7760):210–213, 2019. → pages 7, 39, 42, 43[32] Hiroshi Imada, Kuniyuki Miwa, Miyabi Imai-Imada, Shota Kawahara,Kensuke Kimura, and Yousoo Kim. Real-space investigation of energytransfer in heterogeneous molecular dimers. Nature, 538, 2016. → pages7, 39, 46[33] Yang Zhang, Yang Luo, Yao Zhang, Yun-Jie Yu, Yan-Min Kuang, Li Zhang,Qiu-Shi Meng, Yi Luo, Jin-Long Yang, Zhen-Chao Dong, and J G Hou.Visualizing coherent intermolecular dipole-dipole coupling in real space.Nature, 531, 2016. → pages 45, 50[34] Benjamin Doppagne, Michael C. Chong, Etienne Lorchat, Ste´phaneBerciaud, Michelangelo Romeo, Herve´ Bulou, Alex Boeglin, FabriceScheurer, and Guillaume Schull. Vibronic spectroscopy with submolecularresolution from stm-induced electroluminescence. Physical Review Letters,118(12), 2017. → pages 45, 50[35] Kuniyuki Miwa, Hiroshi Imada, Miyabi Imai-Imada, Kensuke Kimura,Michael Galperin, and Yousoo Kim. Many-body state description ofsingle-molecule electroluminescence driven by a scanning tunnelingmicroscope. Nano Letters, 19(5):2803–2811, 2019. → page 39[36] Katherine A Cochrane, Tanya S Roussy, Bingkai Yuan, Gary Tom, ErikMa˚rsell, and Sarah A Burke. Molecularly resolved electronic landscapes ofdiffering acceptor–donor interface geometries. The Journal of PhysicalChemistry C, 122(15):8437–8444, 2018. → page 4273[37] Katharina Kaiser, Leo Gross, and Fabian Schulz. A single-moleculechemical reaction studied by high-resolution atomic force microscopy andscanning tunneling microscopy induced light emission. ACS Nano, 13:47,2019.[38] Yao Zhang, Qiu-Shi Meng, Li Zhang, Yang Luo, Yun-Jie Yu, Ben Yang,Yang Zhang, Ruben Esteban, Javier Aizpurua, Yi Luo, et al. Sub-nanometrecontrol of the coherent interaction between a single molecule and aplasmonic nanocavity. Nature Communications, 8(1):1–7, 2017. → pages7, 46, 65[39] Martin Schwarze, Wolfgang Tress, Beatrice Beyer, Feng Gao, ReinhardScholz, Carl Poelking, Katrin Ortstein, Alrun A Gu¨nther, Daniel Kasemann,Denis Andrienko, et al. Band structure engineering in organicsemiconductors. Science, 352(6292):1446–1449, 2016. → pages7, 39, 46, 47, 58[40] Nathan R. Paisley, Christopher M. Tonge, Don M. Mayder, Kyle A.Thompson, and Zachary M. Hudson. Tunable benzothiadiazole-baseddonor-acceptor materials for two-photon excited fluorescence. MaterialsChemistry Frontiers, 4(2):555–566, 2020. → pages 7, 59[41] Christopher M Tonge, Jiajie Zeng, Zujin Zhao, Ben Zhong, and Zachary MHudson. Bis(hexamethylazatriangulene)sulfone: a high-stability deepblue-violet fluorophore with 100% quantum yield and CIEy > 0.07 . Journalof Materials Chemistry C, 2020. → page 58[42] Xian Kai Chen, Youichi Tsuchiya, Yuma Ishikawa, Cheng Zhong, ChihayaAdachi, and Jean Luc Bre´das. A new design strategy for efficient thermallyactivated delayed fluorescence organic emitters: From twisted to planarstructures. Advanced Materials, 29(46):1702767, 2017. → pages 7, 59[43] Jascha Repp, Gerhard Meyer, Sladjana M Stojkovic´, Andre´ Gourdon, andChristian Joachim. Molecules on insulating films: scanning-tunnelingmicroscopy imaging of individual molecular orbitals. Physical ReviewLetters, 94(2):026803, 2005. → pages 7, 34, 41[44] Gerd Binnig, Heinrich Rohrer, Christoph Gerber, and Edmund Weibel.Surface studies by scanning tunneling microscopy. Physical Review Letters,49(1):57, 1982. → page 8[45] S Heidorn, C Bertram, J Koch, K Boom, F Matthaei, A Safiei, J Henzl, andK Morgenstern. Influence of substrate surface-induced defects on the74interface state between NaCl(100) and Ag(111). The Journal of PhysicalChemistry C, 117(31):16095–16103, 2013. → pages 11, 56[46] J. Bardeen. Tunnelling from a many-particle point of view. Physical ReviewLetter, 6:57–59, 1961. → page 13[47] Alex D Gottlieb and Lisa Wesoloski. Bardeen’s tunnelling theory as appliedto scanning tunnelling microscopy: a technical guide to the traditionalinterpretation. Nanotechnology, 17(8):R57, 2006. → page 13[48] J Tersoff and DR Hamann. Theory of the scanning tunneling microscope.Physical Review B, 31(2):59–67, 1985. → page 13[49] TE Feuchtwang. Tunneling theory without the transfer-hamiltonianformalism. I. Physical Review B, 10(10):4121, 1974. → page 13[50] J Pendry, B, AB Preˆtre, and BCH Krutzen. Theory of the scanningtunnelling microscope. Journal of Physics: Condensed Matter, 3(24):4313,1991. → page 13[51] RM Feenstra, JA Stroscio, WJ Kaiser, and JA Stroscio. Methods ofTunneling Spectroscopy. Academic Press, Inc.: San Diego, 1993. → page 17[52] RM Feenstra, Joseph A Stroscio, J Tersoff, and AP Fein. Atom-selectiveimaging of the GaAs(110) surface. Physical Review Letters, 58(12):1192,1987. → page 17[53] M Prietsch, A Samsavar, and R Ludeke. Structural and electronic propertiesof the Bi/GaP(110) interface. Physical Review B, 43(14):11850, 1991. →page 17[54] JK Gimzewski, B Reihl, JH Coombs, and RR Schlittler. Photon emissionwith the scanning tunneling microscope. Zeitschrift fu¨r Physik B CondensedMatter, 72(4):497–501, 1988. → page 18[55] X H Qiu, G V Nazin, and W Ho. Vibrationally resolved fluorescence excitedwith submolecular precision. Science, 299(5606):542–546, 2003. → pages18, 22[56] Lukas Novotny and Bert Hecht. Principles of nano-optics. CambridgeUniversity Press, 2012. → pages 18, 21, 22[57] S. W. Wu, G. V. Nazin, and W. Ho. Intramolecular photon emission from asingle molecule in a scanning tunneling microscope. Physical Review B, 77(20), 2008. → page 1875[58] John Lambe and SL McCarthy. Light emission from inelastic electrontunneling. Physical Review Letters, 37(14):923, 1976. → page 18[59] Richard Berndt, James K Gimzewski, and Peter Johansson. Inelastictunneling excitation of tip-induced plasmon modes on noble-metal surfaces.Physical Review Letters, 67(27):3796, 1991. → page 18[60] S Ushioda, Y Uehara, and M Kuwahara. STM light emission spectroscopyof Au film. Applied Surface Science, 60:448–453, 1992.[61] Germar Hoffmann, Jo¨rg Kro¨ger, and Richard Berndt. Color imaging with alow temperature scanning tunneling microscope. Review of ScientificInstruments, 73(2):305–309, 2002. → page 18[62] Lucia Vitali, MA Schneider, Klaus Kern, Ludger Wirtz, and Angel Rubio.Phonon and plasmon excitation in inelastic electron tunneling spectroscopyof graphite. Physical Review B, 69(12):121414, 2004. → page 18[63] T Uemura, M Furumoto, T Nakano, M Akai-Kasaya, A Saito, M Aono, andY Kuwahara. Local-plasmon-enhanced up-conversion fluorescence fromcopper phthalocyanine. Chemical Physics Letters, 448(4-6):232–236, 2007.→ page 20[64] Z. C. Dong, X. L. Zhang, H. Y. Gao, Y. Luo, C. Zhang, L. G. Chen,R. Zhang, X. Tao, Y. Zhang, J. L. Yang, and J. G. Hou. Generation ofmolecular hot electroluminescence by resonant nanocavity plasmons.Nature Photonics, 4(1):50–54, 2010.[65] C Zhang, R Zhang, S Jiang, L Zhang, HY Gao, XL Zhang, LG Chen,Y Liao, and ZC Dong. Tip-plasmon mediated molecularelectroluminescence on the highly oriented pyrolytic graphite substrate.Applied Physics Letters, 100(7):073111, 2012. → page 20[66] Klaus Kuhnke, Christoph Große, Pablo Merino, and Klaus Kern.Atomic-scale imaging and spectroscopy of electroluminescence at molecularinterfaces. Chemical Reviews, 117(7):5174–5222, 2017. → pages 21, 22[67] Honghua U Yang, Jeffrey D’Archangel, Michael L Sundheimer, Eric Tucker,Glenn D Boreman, and Markus B Raschke. Optical dielectric function ofsilver. Physical Review B, 91(23):235137, 2015. → pages 21, 40[68] Robert L Olmon, Brian Slovick, Timothy W Johnson, David Shelton,Sang-Hyun Oh, Glenn D Boreman, and Markus B Raschke. Optical76dielectric function of gold. Physical Review B, 86(23):235147, 2012. →pages 21, 40[69] E. M. Purcell. Spontaneous emission probabilities at radio frequencies.Physical Review, 69:681, 1946. → page 22[70] Peter Johansson. Light emission from a scanning tunneling microscope:Fully retarded calculation. Physical Review B, 58(16):10823, 1998. → page22[71] C Zhang, B Gao, LG Chen, QS Meng, H Yang, R Zhang, X Tao, HY Gao,Y Liao, and ZC Dong. Fabrication of silver tips for scanning tunnelingmicroscope induced luminescence. Review of Scientific Instruments, 82(8):083101, 2011. → pages 22, 28[72] Gong Chen, Xiao-Guang Li, Zhen-Yu Zhang, and Zhen-Chao Dong.Molecular hot electroluminescence due to strongly enhanced spontaneousemission rates in a plasmonic nanocavity. Nanoscale, 7(6):2442–2449,2015. → page 22[73] Javier Aizpurua, S Peter Apell, and Richard Berndt. Role of tip shape inlight emission from the scanning tunneling microscope. Physical Review B,62(3):2065, 2000. → page 22[74] Kazuyuki Meguro, Kenji Sakamoto, Ryuichi Arafune, Minoru Satoh, andSukekatsu Ushioda. Origin of multiple peaks in the light emission spectra ofa Au(111) surface induced by the scanning tunneling microscope. PhysicalReview B, 65(16):165405, 2002. → page 22[75] Christoph Große, Alexander Kabakchiev, Theresa Lutz, Romain Froidevaux,Frank Schramm, Mario Ruben, Markus Etzkorn, Uta Schlickum, KlausKuhnke, and Klaus Kern. Dynamic control of plasmon generation by anindividual quantum system. Nano Letters, 14(10):5693–5697, 2014. → page22[76] Frdric Rossel, Marina Pivetta, and Wolf Dieter Schneider. Luminescenceexperiments on supported molecules with the scanning tunnelingmicroscope. Surface Science Reports, 65(5):129–144, 2010. → page 22[77] Benjamin MacLeod. An ultra-low-vibration facility for housing a dilutiontemperature scanning tunneling microscope. PhD thesis, University ofBritish Columbia, 2015. → page 2477[78] Katherine Anne Cochrane. Single molecule perspectives of model organicsemiconductors: energy level mapping by high-resolution scanning probemicroscopy. PhD thesis, University of British Columbia, 2017. → pages25, 42[79] Tanya Roussy. Coupling spatially and spectrally resolved opticalmeasurements with a scanning probe system. PhD thesis, University ofBritish Columbia, 2016. → pages 25, 28, 31, 39, 42[80] Princeton Instruments. Products: Gratings, 2020. URL → page 32[81] Princeton Instruments. Grating dispersion calculator, 2020. URL→ page 32[82] Heinz Ho¨vel, B Grimm, and B Reihl. Modification of the Shockley-typesurface state on Ag(111) by an adsorbed xenon layer. Surface Science, 477(1):43–49, 2001. → page 33[83] JV Barth, Harald Brune, G Ertl, and RJ Behm. Scanning tunnelingmicroscopy observations on the reconstructed Au(111) surface: Atomicstructure, long-range superstructure, rotational domains, and surface defects.Physical Review B, 42(15):9307, 1990. → page 34[84] Marcus D Hanwell, Donald E Curtis, David C Lonie, Tim Vandermeersch,Eva Zurek, and Geoffrey R Hutchison. Avogadro: an advanced semanticchemical editor, visualization, and analysis platform. Journal ofCheminformatics, 4(1):17, 2012. → page 38[85] Sereina Riniker and Gregory A Landrum. Better informed distancegeometry: using what we know to improve conformation generation.Journal of Chemical Information and Modeling, 55(12):2562–2574, 2015.→ page 38[86] Greg Landrum. Rdkit: Open-source cheminformatics, 2020. URL → page 38[87] Anthony K Rappe´, Carla J Casewit, KS Colwell, William A Goddard III,and W Mason Skiff. UFF, a full periodic table force field for molecularmechanics and molecular dynamics simulations. Journal of the AmericanChemical Society, 114(25):10024–10035, 1992. → page 3878[88] MJ Frisch, GW Trucks, HB Schlegel, GE Scuseria, MA Robb,JR Cheeseman, G Scalmani, V Barone, GA Petersson, HJRA Nakatsuji,et al. Gaussian 16, 2016. → page 38[89] Chengteh Lee, Weitao Yang, and Robert G Parr. Development of thecolle-salvetti correlation-energy formula into a functional of the electrondensity. Physical Review B, 37(2):785, 1988. → page 38[90] Axel D Becke. Becke’s three parameter hybrid method using the LYPcorrelation functional. The Journal of Chemical Physics, 98(492):5648–5652, 1993. → page 38[91] Michael J Frisch, John A Pople, and J Stephen Binkley. Self-consistentmolecular orbital methods 25. Supplementary functions for Gaussian basissets. The Journal of Chemical Physics, 80(7):3265–3269, 1984. → page 38[92] Li Zhang, Yun-Jie Yu, Liu-Guo Chen, Yang Luo, Ben Yang, Fan-Fang Kong,Gong Chen, Yang Zhang, Qiang Zhang, Yi Luo, Jin-Long Yang, Zhen-ChaoDong, and J. G. Hou. Electrically driven single-photon emission from anisolated single molecule. Nature Communications, 8(1):580, 2017. → pages39, 41, 45[93] Benjamin Doppagne, Michael C Chong, Herve´ Bulou, Alex Boeglin,Fabrice Scheurer, and Guillaume Schull. Electrofluorochromism at thesingle-molecule level. Science, 361:251–255, 2018. → pages 39, 45[94] Ross Warren, Alberto Privitera, Pascal Kaienburg, Andreas E Lauritzen,Oliver Thimm, Jenny Nelson, and Moritz K Riede. Controlling energy levelsand fermi level en route to fully tailored energetics in organicsemiconductors. Nature Communications, 10(1):1–7, 2019. → pages 39, 46[95] Jo¨rg Kro¨ger, Benjamin Doppagne, Fabrice Scheurer, and Guillaume Schull.Fano description of single-hydrocarbon fluorescence excited by a scanningtunneling microscope. Nano Letters, 2018. → page 41[96] Kuniyuki Miwa, Hiroshi Imada, Shota Kawahara, and Yousoo Kim. Effectsof molecule-insulator interaction on geometric property of a singlephthalocyanine molecule adsorbed on an ultrathin NaCl film. PhysicalReview B, 93:165419, 2016. → pages 45, 46, 56[97] H. A.M. Van Mullekom, J. A.J.M. Vekemans, E. E. Havinga, and E. W.Meijer. Developments in the chemistry and band gap engineering ofdonor-acceptor substituted conjugated polymers. Materials Science andEngineering R: Reports, 32(1):1–40, 2001. → page 5879[98] Nikolay S. Makarov, Sukrit Mukhopadhyay, Kada Yesudas, Jean LucBre´das, Joseph W. Perry, Agnieszka Pron, Milan Kivala, and Klaus Mu¨llen.Impact of electronic coupling, symmetry, and planarization on one-andtwo-photon properties of triarylamines with one, two, or three diarylborylacceptors. Journal of Physical Chemistry A, 116(15):3781–3793, 2012. →page 58[99] Zhen Fang, Vijila Chellappan, Richard D. Webster, Lin Ke, Tianfu Zhang,Bin Liu, and Yee Hing Lai. Bridged-triarylamine starburst oligomers as holetransporting materials for electroluminescent devices. Journal of MaterialsChemistry, 22(30):15397–15404, 2012. → page 5980


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items