Shining Light on Molecules: Electron Transfer Processes in Model Systems for Solar Energy Conversion Investigated by Transient Absorption Spectroscopy

Full text

(1)Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1645. Shining Light on Molecules Electron Transfer Processes in Model Systems for Solar Energy Conversion Investigated by Transient Absorption Spectroscopy JENS FÖHLINGER. ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2018. ISSN 1651-6214 ISBN 978-91-513-0273-7 urn:nbn:se:uu:diva-343443.

(2) Dissertation presented at Uppsala University to be publicly examined in Siegbahnsalen, Lägerhyddsvägen 1, Uppsala, Friday, 4 May 2018 at 09:15 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Professor Benjamin Dietzek (Jena University). Abstract Föhlinger, J. 2018. Shining Light on Molecules. Electron Transfer Processes in Model Systems for Solar Energy Conversion Investigated by Transient Absorption Spectroscopy. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1645. 74 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-513-0273-7. In the recent years, solar energy conversion has attracted a huge research interest due to the potential application for limiting the greenhouse effect. In many solar cells and solar fuel cells, understanding of charge transfer (CT) and recombination is important for future improvement of the overall efficiency. One important tool for that is transient absorption spectroscopy (TAS). Mesoporous nickel oxide films were investigated due to their potential application in ptype dye-sensitized solar cells (DSSCs), tandem DSSCs and dye sensitized solar fuel cells (DSSFC:s). Firstly, it was found that the hole generated by band gap excitation is trapped on an ultrafast time scale by Ni3+ states. It was possible to observe a direct signal from the holes by transient mid-IR absorption spectroscopy allowing for direct detection of hole injection and trapping. On a ns time scale, the trapped holes relaxed to much less reactive holes which favored long lived NiO-dye charge separation (CS). A series of perylene monoimide (PMI) dyes with different anchoring groups was studied. Differences in binding affinity and stability were found. Nevertheless, all PMIs showed ultrafast charge separation and similar recombination kinetics. Furthermore, the effect of MLCT localization of ruthenium polypyridyl complexes was investigated. All those dyes showed slow or no hole injection. At the same time, a self-quenching process was found for all compounds that limited the photoconversion efficiency. Furthermore, a new core-shell structure of p-type DSSCs was proposed and investigated. Here, the liquid electrolyte was replaced by a layer of TiO2. That system was found to undergo both injection and regeneration of the dye on an ultrafast time scale (below 1 ps). Furthermore, the CS state did not show any decay within 2 ns making this structure interesting for application in DSSCs. A pentad consisting of a known Ru-based (electro)chemical water oxidation catalyst (WOC) linked to two zinc-porphyrin-fullerene dyads (ZnP-C60) was investigated. The charge transfer processes leading to the first oxidation of the WOC were understood. Low levels of water oxidation were detected in presence of a sacrificial electron acceptor. The gained understanding of the CT dynamics and recombination processes thus allows new strategies to improve the efficiency in molecular systems for solar energy conversion. Keywords: photophysics, photoinduced electron transfer, transient absorption spectroscopy, laser spectroscopy, solar energy conversion, p-type DSSCs, Charge separation, recombination, mesoporous NiO Jens Föhlinger, Department of Chemistry - Ångström, Physical Chemistry, Box 523, Uppsala University, SE-75120 Uppsala, Sweden. © Jens Föhlinger 2018 ISSN 1651-6214 ISBN 978-91-513-0273-7 urn:nbn:se:uu:diva-343443 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-343443).

(3) To my family.

(4)

(5) List of papers. This thesis is based on the following papers, which are referred to in the text by their Roman numerals. I. Unveiling hole trapping and surface dynamics of NiO nanoparticles L. D’Amario∗ , J. Föhlinger∗ , G. Boschloo, and L. Hammarström Chem. Sci., 2018, 9, 223. II. Direct Spectroscopic Observation of Hole Trapping in Dye-Sensitized NiO Films by Transient mid-IR Absorption Spectroscopy J. Föhlinger, L. Antila, D. Narouzi, S. Ott, and L. Hammarström Manuscript in preparation. III. A Comparative Investigation about the Role of Anchoring group on Perylene Monoimide Dyes in NiO Based Dye-Sensitized Solar Cells Y. Farré, J. Föhlinger, A. Planchat, Y. Pellegrin, E. Blart, L. Hammarström, and F. Odobel Manuscript in preparation. IV. Self-quenching and Slow Hole Injection May Limit the Efficiency in NiO-based Dye-Sensitized Solar Cells J. Föhlinger, S Maji, A. Brown, E. Mijangos, S. Ott, and L. Hammarström Manuscript submitted. V. Ultrafast dye regeneration in a core–shell NiO–dye–TiO2 mesoporous film L. Tian∗ , J. Föhlinger∗ , P. B. Pati, Z. Zhang, J. Lin, W. Yang, M. Johansson, T. Kubart, J. Sun, G. Boschloo, L. Hammarström, and H. Tian Phys. Chem. Chem. Phys., 2018, 20, 36. VI. A Ruthenium Complex–Porphyrin–Fullerene–Linked Molecular Pentad as an Integrative Photosynthetic Model M. Yamamoto, J. Föhlinger, J. Petersson, L. Hammarström, and H. Imahori Angew. Chem. Int. Ed. 2017, 56, 3329..

(6) Papers not included in this thesis. During my PhD studies, I also contributed to the following papers which are not included in the thesis. VII. VIII. IX. Ultrafast Interligand Electron Transfer in [Ru(4,4’-dicarboxylate -2,2’-bipyridine)2 cis-(NCS)2 ]4- and Implications for Electron Injection Limitations in Dye-Sensitized Solar Cells B. Pettersson Rimgard, J. Föhlinger, J. Petersson, M.Lundberg, B. Zietz, A. M. Woys, S. A. Miller, M. R. Wasielewski, and L. Hammarström Manuscript submitted Light driven electron transfer in methyl-bipyridine phenol complexes is not proton coupled R. Tyburski, J. Föhlinger, and L. Hammarström Manuscript submitted Ultra long-lived electron-hole separation within water-soluble colloidal ZnO nanocrystals: Prospective Applications For Solar Energy Production A. M.Cie´slak et. al. Nano Energy 2016, 30, 187. X Soret fluorescence involved in Caryophyllales plants ultraviolet protection J. Sá, M. V. Pavliuk, A. M. El-Zohry, D. L. A. Fernandes, J. Föhlinger, and E. Mukhtar Sci. Lett. J. 2016, 5, 226 XI Solid State p-Type Dye-Sensitized NiO-dye-TiO2 Core-Shell Solar Cells L. Tian, J. Föhlinger, Z. Zhang, P. B. Pati, J. Lin, T. Kubart, Y. Hua, J. Sun, L. Kloo, G. Boschloo, L. Hammarström, and H. Tian Chem. Commun., accepted manuscript DOI: 10.1039/C8CC00505B Reprints were made with permission from the publishers. *. Shared first authorship.

(7) Contribution to the papers. My contributions to the papers included in this thesis are stated below: I. Performed the fs-transient absorption experiments,their data analysis, and participated in the discussion. Approved the manuscript II. Participated actively in the design of the study, performed the UV/Vis transient absorption experiments, and participated in the data analysis and discussion. I was lead for writing the manuscript. III. Performed the transient absorption measurements, contributed to the discussion and wrote that section in the paper. IV. Performed all measurements (except for spectroelectrochemistry) of the bis-tridentate compounds, main responsible for data analysis, discussion, and writing of the manuscript. V. Performed all transient absorption spectroscopy measurements and their data analysis. I participated actively in discussion and writing of the manuscript and approved the final version. VI. Participated in the fs-transient absorption measurements and their data analysis, performed the nanosecond timescale experiments and their data analysis, participated in writing the manuscript, and approved the final version..

(8)

(9) Contents. 1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Climate Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 World Energy Production Sustainability Problem . . . . . . . . . . . . . . . . . . . . . . . 1.3 Solar Energy Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.1 Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.2 Artificial Photosynthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.3 Importance of Electron-Transfer Processes in Solar Energy Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 13 13 13 14 14 15. 2. Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Photophysical Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Electron-Transfer Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Marcus Theory for Electron-Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2 Photoinduced Electron-Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Charge-Transfer Transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Dye-sensitized Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1 Interfacial Electron Transfer in DSSCs . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.2 Solid-State DSSCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Photocatalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 17 17 18 18 20 21 22 23 24 24. 3. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Steady-State Absorption and Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Time-Resolved Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 27 27 28 31. 4. Hole Trapping in Nickel Oxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Aim of the study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Implications for NiO-based p-type DSSCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 33 33 33 36 37. 5. Perylene Monoimides with Different Anchoring Groups . . . . . . . . . . . . . . . . . . . . . . . 38 5.1 The different Anchoring Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 5.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38. 6. Self-quenching limits the efficiency of NiO based DSSCs . . . . . . . . . . . . . . . . . . . . . 43 6.1 Aim of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 6.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44. 16.

(10) 7. Ultrafast regeneration in Core-Shell Structure DSSCs . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 7.1 Design of the Core-Shell Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 7.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48. 8. Macromolecular Pentad for Water Splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 8.1 Expected processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 8.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53. 9. Conclusions and Future work. 10 Zusammenfassung. .................................................................... 55. ...................................................................................... 57. 11 Populärvetenskaplig sammanfattning. ........................................................ 60. .................................................................................... 63. ......................................................................................................... 65. 12 Acknowledgements References.

(11) List of Abbreviations. ALD. Atomic layer deposition. BET. Back electron transfer. C60. Fullerene. CB. Conduction band. CS. Charge separated state. CSh. Charge shift. CT. Charge transfer. DAS. Decay associated spectra. DSSC. Dye-sensitized solar cell. DSSFC. Dye-sensitized solar fuel cell. ETM. Electron transporting material. GSB. Ground state bleach. HER. Hydrogen-evolving reaction. HOMO. Highest occupied molecular orbital. HTM. Hole transporting material. IA. Induced Absorption. IPCE. Incident photon-to-current efficiency. LC. Ligand centered. LUMO. Lowest unocupied molecular orbital. MLCT. Metal-to-ligand charge transfer. NDI. naphthalene diimide. OECD. Organization for Economic Co-operation and Development. OER. Oxygen-evolving reaction. PEC. Photoelectrochemical cell. PET. Photoinduced electron transfer. PMI. Perylene monoimide. PMI. perylene monoimide. PSI. Photosystem I.

(12) PSII. Photosystem II. RS. Reactant state. SC. Semiconductor. SE. Stimulated Emission. ss-DSSC. Solid-state dye-sensitized solar cell. TAS. transient absorption spectroscopy. VB. Valence band. WOC. Water oxidation catalyst. ZnP. Zinc porphyrin.

(13) 1. Motivation. 1.1 Climate Change In December 2015, the Paris Agreement was signed by 195 countries. In that treaty, the governments of these countries have agreed to the following (article 2 from reference 1): a) Holding the increase in the global average temperature to well below 2 ◦C above pre-industrial levels and to pursue efforts to limit the temperature increase to 1.5 ◦C above pre-industrial levels, recognizing that this would significantly reduce the risks and impacts of climate change; b) Increasing the ability to adapt to the adverse impacts of climate change and foster climate resilience and low greenhouse gas emissions development, in a manner that does not threaten food production; c) Making finance flows consistent with a pathway towards low greenhouse gas emissions and climate-resilient development. As can be seen from this text, there is a consensus on the relation between greenhouse gases and temperature increase. One of the most "famous" greenhouse gases is carbon dioxide (CO2 ). Its concentration in the atmosphere has increased significantly within the last decades with significant contribution from mankind. To meet the temperature increase, green house gas emissions have to be decreased. To reach this goal, CO2 emissions have to either be stored, reduced, or a combination of both.. 1.2 World Energy Production Sustainability Problem A major source for anthropogenic CO2 is the combustion of fossil fuels. This includes both electricity production and usage for transport. Today, the main global energy source are fossil fuels which contribute with more than 75 % to the global energy production[2]. Globally, energy consumption has increased exponentially and is expected to continue[3]. The largest contribution to future increases in energy consumption is expected to originate from developing countries, evolving towards life standards comparable to those in the Organization for Economic Co-operation and Development (OECD) countries. This implies a need for new energy 13.

(14) sources to meet present and future demands. One advantage of many developing countries is the absence of (modern) infrastructure, making it easier to employ new concepts such as decentralized grids. Furthermore, there is not yet a commitment to any energy source, allowing for choice in the adoption of optimal and sustainable energy alternatives. Employing renewable, greenhouse gas emission free technologies to meet additional energy demands in the future will at least help to not increase emissions. Simultaneously, there has to be a transition to renewable energies in the OECD. In contrast, in these countries, there is a higher need for implementation of renewable energy technologies that are compatible with the already existing infrastructure, which is to a large extend based on fossil fuels.. 1.3 Solar Energy Conversion One of promising renewable and sustainable energy sources is the sun, which delivers 120 000 TW to Earth[4]. Compared to the total energy consumption of 30 TW to 35 TW[2], one can calculate that the solar irradiation of a little bit more than two hours would be able to cover mankind’s annual energy consumption. To replace, for example, Sweden’s annual fossil fuel consumption (150 TW h [5]), it would be sufficient to cover 0.4 % of the Swedish land area with solar cells with 15 % efficiency and the average annual solar irradiation of 1000 kW h m−2 [6, 7]. This corresponds roughly to four times the land taken up by golf courses and ski slopes or ≈ 15 % of the built up land only[8]. When even aiming for the replacement of nuclear energy, this percentage of needed land increases to 0.7 %. Therefore, solar energy is an appealing source as future renewable and sustainable energy source.. 1.3.1 Solar Cells One very common and mature solar energy conversion technology is the generation of biomass from plants. This approach, however, faces low efficiency around 0.15 % solar energy to fuel[9]. This leads to a necessity of large land use. Another drawback is that the plants for biomass production compete with food crops and thus potentially increase prices for food. Hence, biomass alone, is not an (optimal) solution for meeting the future energy demand. Another well known alternative for solar energy conversion are solar cells. Their common general working principle is: 1. Absorption of sun light 2. Separation of charges 3. Extraction of charges.. 14.

(15) To date, most solar cells available are based on the concept of a p-n junction. Such solar cells consist of doped crystalline n-type (excess electrons) and p-type (depletion of electrons) semiconductors brought in contact with each other. This leads to the formation of an interface layer where positive and negative charges cancel each other. As a consequence, a smooth gradient of the bands is formed resulting in an intrinsic electric field within this layer. When absorbing light, an electron-hole pair is generated in the semiconductor. If this happens within the interface layer, the above mentioned electric field pulls electrons and holes in different directions, thus apart from each other. The separated charges are finally extracted from the solar cell by electrical contacts. The theoretical efficiency limit for a single junction solar cell is 33 %, calculated by Shockley and Queisser [10]. One way to circumvent this limit is the application of multi-junction solar cells[11–13]. The efficiency record to date is a solar cell with four junctions giving a total conversion efficiency of 46 % under illumination with concentrated sunlight[14, 15]. The most common commercial solar cells, however, show efficiencies of approximately 15 %[16]. Another emerging technology are the so called third generation solar cells. These include organic solar cells, perovskite solar cells, quantum dot solar cells, and dye-sensitized solar cells (DSSCs, also known as Grätzel cells). The main focus of this thesis will be on the latter and therefore, only those are described in more detail. DSSCs consist of a transparent mesoporous semiconductor sensitized by a dye. Especially the DSSCs have several possible advantages: by changing the type of dye and dye concentration, these solar cells can have basically any color and be transparent. This makes them attractive for possible build-in architecture. Furthermore, they can have a higher power conversion efficiency at low light levels such as ambient or stray light compared to p-n junction based cells[17]. A detailed description of device structure and working principle will be presented in chapter 2.4.. 1.3.2 Artificial Photosynthesis The production of solar electricity has the disadvantage of discontinuity. Especially in winter and during night, when the energy demand is high, there is little to no solar electricity generation, a fact that makes power storage necessary. One option to store electricity is the use of batteries. Their typical energy density ranges from 10 W h m−3 to 200 W h m−3 [18]. A more appealing energy storage alternative are fuels because of their even larger energy densities. Therefore, the conversion of solar energy directly into chemical energy in the form of fuels is an appealing approach for long term energy storage. Another positive effect is the potential usage in the already existing fuel based infrastructure. A well known example for this conversion is natural photosynthesis 15.

(16) in plants, which splits water into oxygen and utilizes the formed electrons and protons together with CO2 to form sugars as energy storage. In this process, plants utilize sunlight in a two step reaction (so called Zscheme). In Photosystem II (PS II), the energy of absorbed photons is used for charge separation, and subsequent splitting of water into molecular oxygen, protons and electrons. The electrons are then transported to Photosystem I by an electron transport chain. In a second step, when photons are absorbed in PS I, it gains sufficient energy to produce a reductant (NADPH) which later is used for carbon fixation to form biomass. The protons form a concentration gradient which then is used to produce ATP. This Z-scheme is mimicked in artificial photosynthesis. As there is no need for vital processes as in plants (e.g. respiration, metabolism), the theoretical maximum solar to bond conversion efficiency is 40 %[12], which is equal to that of a double junction solar cell. Even here, water is used as a cheap substrate and electron source. The reducing power at the second step is aimed to be used for either proton reduction to molecular hydrogen or for CO2 reduction to form CO or other higher value hydrocarbons such as e.g. formic acid. More details about these processes can be found in chapter 2.5.. 1.3.3 Importance of Electron-Transfer Processes in Solar Energy Conversion As discussed above, light absorption and charge separation processes are important for the functioning of solar energy conversion devices. Therefore, a deeper understanding and finding of unwanted and unproductive side reactions, especially of the initial steps after light absorption, can be crucial for further improvements of the devices. Hence, the main focus of this thesis is on the first charge-transfer processes following light absorption in different solar energy conversion systems which will be described more detailed in the following chapters.. 16.

(17) 2. Theory. 2.1 Photophysical Processes In the following sections, the interaction between molecules and light will be discussed. Light can be described as photons having both particle and wave character[19, 20]. Due to the wave character, they can be assigned a frequency and wavelength. Due to the particle character, every photon is attributed an energy which is proportional to the frequency E = hν. According to quantum mechanics, photons can only interact with a molecule if the

(18)

(19) photon energy corresponds to the energy dif

(20) ference between two states in the molecule as depicted by the horizontal bars in fig ure 2.1. Depending on the character of the states, there is different energy spacing, e.g. electronic and vibrational states. The electronic states are formed by different occupa tions of the molecular orbitals as exemplified in the right hand part of figure 2.1. In the state with lowest energy, the ground state (GS), often denoted S0 , normally the lowest molecular orbitals are occupied. Light ex cites an electron from one of those occupied orbitals into an unoccupied orbital to form an excited state (ES). For many dyes, the electronic transition with the lowest energy possible can be described as promotion of an Figure 2.1. Schematic of steady electron from the highest occupied molec- state absorption and emission. ular orbital (HOMO) to the lowest unoccu- Electronic and vibrational states pied molecular orbital (LUMO) to form a S1 (left) and the corresponding occustate. For other dyes, this transition has to pation of molecular orbitals (right) be described by excitation of multiple elec- For simplicity, the S2 state is iltrons. Despite the lowest energy difference, lustrated as a HOMO to LUMO+1 the electron can also be excited to higher un- transition. occupied molecular orbitals such as e.g. S0 – S2 transition. Higher transitions can also originate from energy levels lower than the HOMO (e.g. HOMO-1) to a LUMO orbital. The light inducing electronic transitions are mostly in the visible to UV region.. . 17.

(21) The electronic states often have a vibrational substructure as shown in figure 2.1. This substructure is due to different vibrational states of the molecule. A transition between those energy levels changes the amplitude of the molecular vibration. The corresponding light is in the infrared region. In the case of molecules, the IR signal consists of distinguished peaks that correspond to vibrational modes in the molecule. In the case of semi-conductors, there is especially a broad featureless IR signal if there are free charge carriers, e.g. electrons in the conduction band (CB) [21, 22]. After excitation to higher states than the S1 -excited state, most molecules in solution quickly dissipate the excess energy to the surrounding solvent and relax to the lowest vibrational state of the S1 -excited state[23]. Therefore, most transitions and reactions from the excited state happen from the S1,υ = 0 state where υ describes the vibrational quantum number of the corresponding electronic state (i.e. S1 ). Besides absorption, many molecules also can emit light. When going back into the ground state, the molecule can emit light as depicted by the right hand arrow in the left part of figure 2.1. This light can provide information about relaxation processes and the vibrational structure of the ground state. Possible decreases in its intensity upon interaction with other molecules or materials, i.e. quenching, can give information about the efficiency of those quenching mechanisms. Except for the singlet states as described above, there are also triplet states (not shown in figure 2.1) which differ by a spin flip of one electron. Because of the spin flip, singlet-triplet transitions are spin forbidden and suffer from low probabilities.. 2.2 Electron-Transfer Theory 2.2.1 Marcus Theory for Electron-Transfer The theory of electron-transfer was mainly developed by Marcus[24–26]. It predicts the rate at which an electron can be transferred from an electron donor (D) to an electron acceptor (A). For both reactant state (RS, D + A) and charge separated state (CS, D+ + A- ), the free energy surface is approximated as a parabola with respect to the reaction coordinate as shown in figure 2.2. The reaction coordinate is a projection of three dimensional system on a one dimensional axis indicating the ratio of RS and CS character. At the equilibrium configuration of the RS (req,D+A ), the ground state energy is lower than that of the CS. At the crossing of the parabolas on the other hand, the electronic configurations of both (distorted) reactant state and (distorted) CS have the same free energy, making it possible for the electron to tunnel from one state to the other. This nuclei-configuration is also called the transition state (TS). The energy difference between RS and TS is the activation energy ΔG‡ . The reorganization energy λ is the energy that would be released if the electron would 18.

(22) . .

(23) . . . . .

(24). . .

(25)

(26)

(27) . Figure 2.2. Schematic of Marcus parabolas. Left: Parabolas indicating reactant state (RS, D+A), charge separated state (CS, D+ +A- ), reorganization energy (λ ), driving force (ΔG0 ) and activation energy ΔG‡ according to Marcus theory. [25–27]. Right: Three different CS state equilibrium free energies result in different activation energies, if the driving force is larger than the reorganization energy, it is called inverted region due to the decrease in rate upon increase in driving force.. relax to the equilibrium configuration of the CS (req,D+ +A− ) after a transfer directly into the CS from the equilibrium conformation of the GS (req,D+A ). Using geometric considerations, the activation energy can be calculated to be ΔG‡ET = (λ + ΔG0 )2 /(4λ ). In the case of weak electronic coupling (non-adiabatic) rate expression for electron transfer takes a Arrhenius type form and is:. kET. 2π 2 (λ + ΔG0 )2 −1/2 H (4πλ kB T ) = exp − h¯ rp 4λ kB T. (2.1). where Hrp = ψP |Hrp |ψR is the energy of the electronic coupling between reactant and product states, h¯ the reduced Planck constant, kB the Boltzman constant, and T the temperature. As can be seen from equation (2.1), there is an increase in ET rate with increasing driving force that reaches a maximum when −ΔG0 = λ . At even larger driving forces, the ET rate diminishes again. This can be rationalized by the increase of activation energy in the right part of figure 2.2. This counterintuitive decrease in rate is the inverted region which was predicted by Marcus in 1956 [24]. However, it took until 30 years for the first clear and generally accepted experimental evidence obtained by Closs and Miller [28, 29]. A reason for the experimental difficulties were excited states of the CS. As those have less driving force, they also result in a lowered activation energy, which consequently leads to a constant observed rate in the predicted inverted region. 19.

(28) Distance Dependence of ET If there is weak electronic coupling between the donor and the acceptor and the driving force is constant, the rate depends on the square of the energy of the coupling HRP . The coupling includes the overlap integral of the electronic wave functions of donor and acceptor. As those decay exponentially in space, also the coupling is assumed to decay exponentially by R − R0 HRP = H0 exp −β (2.2) 2 where β is a constant determining the decrease in rate with increasing distance, R the donor-acceptor distance and R0 the van-der Waals contact radius and H0 the electronic coupling at van-der Waals contact. Hence, short distances and large electronic coupling between D and A are necessary for fast and efficient electron transfer. A common strategy to slow down (unwanted) electron transfer processes is the enlargement of the distance of the reactants as well as diminishing their electronic coupling. Both strategies are employed within this thesis (see below). Estimation of the Driving Force The driving force for ET can be estimated by the standard electrode potential of the donor (E ◦ (D+• /D)) and the acceptor (E ◦ (A/A−• )), and the electrostatic work terms that account for the Coulomb interaction of products and reactants, respectively (w(D+• A−• ) and w(DA)) by using . ΔG◦ET = NA e E ◦ (D+• /D) − E ◦ (A/A−• ) + w(D+• A−• ) − w(DA) (2.3) where NA is the Avogadro constant and e the elementary charge[30, 31]. As can be seen, it strongly depends on the redox potentials of the molecules involved. However, there is a correction term including the change in Coulomb interaction upon charge separation. The magnitude of this term is harder to measure, especially if there are interfaces involved (see below). Hence, the values for driving forces derived from equation (2.3) only can give a order of magnitude of the driving force. This means especially that a seemingly slightly uphill process (estimated from the electrochemical standard electrode potentials) may still take place.. 2.2.2 Photoinduced Electron-Transfer Instead from the ground state of the molecules, the electron transfer can also be initiated from an excited state of one of the ET partners. This photoinduced electron transfer (PET) process can be divided into oxidative and reductive quenching. The difference between these two processes is whether the electron donor or acceptor is excited, respectively (see figure 2.3). 20.

(29) . . .

(30). . . . . .

(31). . Figure 2.3. Scheme of photoinduced electron transfer (PET) including oxidative and reductive quenching. For simplification, the excited state is approximated as a HOMO-LUMO transition.. The driving force for PET is based on equation (2.3) with the exception that the electrochemical standard potential of the excited donor/acceptor is used. The formula for PET, ΔPET G◦ , thus transforms to ΔG◦PET = ΔG0ET − E00. (2.4). where ΔE00 is the vibrational zero electronic energy of the excited partner.[31, 32] The use of the vibrational zero energy implies the assumption that the electron transfer takes place from the vibrationally relaxed excited state. If the PET takes place before thermalization, there can be a higher driving force making PET possible from "hot states". In that case, the driving force is increased by the access energy above the vibrationally relaxed S1 -state. Despite this change in driving force, the Marcus theory for electron transfer remains unchanged. This includes especially the distance dependence for electron transfer. This implies that larger donor acceptor distances slow down electron transfer reactions. This is especially used in natural photosynthesis where back electron transfer (BET) is suppressed by increasing the distance of the electron-hole pair. A similar approach is often applied in artificial systems.. 2.3 Charge-Transfer Transitions Transitions do not necessarily take place within one (part of a) molecule. When an electron donor and acceptor are in close proximity to each other (either by covalent binding, coordination, or by electrostatic interactions), an electron can be transferred from the HOMO of the donor to the LUMO of the acceptor[19, 33]. Hence, the electron tunnels from the donor to the acceptor during the excitation process. In contrast to PET, there is no excited donor or 21.

(32) acceptor state involved in such a charge transfer (CT) transition. This additional optical process results in a new, absorption band, red shifted compared to those of the moieties/molecules alone. As a consequence, CT transitions are commonly utilized to improve the light harvesting properties of the photosensitizers. Many metal complexes show a metal-to-ligand charge-transfer (MLCT) band. Here, an electron is transferred from the metal center to the ligand that is the easiest to be reduced, which results in an oxidized metal center and a reduced ligand. By that, the localization of the electron on the ligands can be influenced by their reduction potential. The properties of these CT-bands are strongly solvent dependent (solvatochromic)[34]. Polar solvents generally stabilize charge separation and lower thus the transition energy. Hence, the CT bands are more red shifted the more polar the solvent is. This effect can be used to distinguish CT bands from π − π ∗ transitions of the ligands (ligand centered (LC) states).. 2.4 Dye-sensitized Solar Cells One solar energy converting system applying the principle of PET are dyesensitized solar cells (DSSCs). The concept of an n-type DSSC based on mesoporous titanium dioxide (TiO2 ) was invented by O’Reagan and Grätzel in 1991 [35]. The working principle of a tandem dye-sensitized solar cell consisting of a n-type and a p-type DSSC respectively is shown in figure 2.4. For individual n- or p-type DSSCs the respective counter electrode is not designed to be photoactive. Both are very similar with only minor differences. After excitation of the dye 1 , the photo-excited dye injects an electron into the ntype TiO2 or accepts an electron from p-type NiO (hole injection) 2 to form the oxidized/reduced dye respectively. The so oxidized/reduced photosensitizer is regenerated by the redox mediator M in the electrolyte, 3 resetting the dye back to the ground state. For this process to take place, the mediator has to diffuse close to the oxidized/reduced dye to perform the regeneration; it is therefore a diffusion limited process in most cases. To close the circuit, the electrons travel from the (photo)anode to the (photo)cathode through a load where they can perform work while the mediator shuttles between the two electrodes. However, there are also several unwanted side reactions that diminish the efficiency of the DSSCs. These processes are the deactivation of the excited state, 4 where the dye goes back into the ground state without an electron being transferred into/from the semiconductor. Another process is recombination where the oxidized/reduced dye recombines with the electron/hole in the respective semiconductor 5 . Especially in p-type DSSCs this process often takes place at the order of hundreds of picoseconds, thus competing with the regeneration of the photosensitizer [13, 36–40]. This is in contrast to n22.

(33) . . . . . . . . . . . . . .

(34).

(35). . .

(36) . . . . . . . . . . . . . . Figure 2.4. Scheme of a tandem dye-sensitized solar cell. For individual p- or ntype DSSCs the respective counter electrode is not designed to be photoactive. Left: Structure and schematic electron flow through the cell. Right: Energy diagram of the semiconductor-dye interfaces and wanted forward reactions (blue) and unwanted side reactions (red). For more details, see text.. type DSSCs where recombination is taking place at the micro- to millisecond time-scale and is therefore suppressed by regeneration [41–43]. Additional to the electron/hole-dye recombination, recombination can also take place between the electron/hole in the TiO2 and the NiO respectively and the redox mediator 6 . All of these processes 1 to 6 follow the theory of PET as discussed above. The kinetics are, however, complicated to understand as the exact nature of the states in the semiconductors are not known. The concept of DSSCs is employed in papers II to V. Despite their possible application in tandem cells and in dye-sensitized solar fuel cells, p-type DSSCs are not as widely investigated as their n-type counterparts. The until now most common p-type semiconductor is nickel oxide (NiO). The efficiency of p-type DSSCs (2.5 %)[44] is much lower than that of n-type (14 %)[45]. For both p-type and n-type DSSCs, charge separation mostly takes place on the sub picosecond time scale. The main difference between these two different types of DSSCs is believed to be recombination kinetics, see below.. 2.4.1 Interfacial Electron Transfer in DSSCs In most cases, DSSCs employ ultrafast, i.e. at a sub-picosecond time scale, electron or hole injection upon excitation[36, 46–48]. The reason for this is the high density of accepting states due to the band structure of the semiconductors. Consequently there are many acceptor states, allowing for barrierless electron transfer. However, there have been reports about a slower injection phase (at picosecond timescale) which mostly have been attributed to localization of charges with respect to the semiconductor[49–52]. As seen from 23.

(37) equation 2.2, a localization of the MLCT state on the ligand close to a NiO surface potentially can slow down ultrafast hole injection. Similarly, an electron further away from the TiO2 surface would hinder injection in n-type DSSCs. The major difference between n-type and p-type DSSCs are the time scales of recombination processes. For n-type DSSCs, recombination is excitation intensity dependent[41]. As soon as there is more than one electron in one TiO2 particle, recombination is accelerated due to higher probability of charges to meet and recombine. In addition to the intensity dependence, charge recombination in TiO2 also is very sensitive to applied voltage and electrolyte composition. This is attributed to different occupation ratios of electrons in the conduction band and in trap states within the bandgap. The recombination kinetics of NiO-based p-type DSSCs differs significantly from those of their counterparts. Firstly, it was found that the recombination is not light intensity-dependent [53, 54]. Furthermore, investigation of the dependence on an applied voltage showed a bi-phasic decay with constant time scale for both phases but an increasing ratio of the fast component at positive potentials[55]. The different recombination phases were attributed to different trapping states. A deeper understanding of hole trapping in NiO would thus allow for development of new strategies to slow down recombination and consequently increase the overall efficiency of p-type DSSCs.. 2.4.2 Solid-State DSSCs Originally, DSSCs employed solution phase electrolytes based on organic solvents and a redox mediator[35, 43]. Potential leaking of the electrolyte poses a threat to the environment, which makes it important to seal the cells tightly. Furthermore, the regeneration of the oxidized/reduced dye relies on diffusion of the redox mediator through the electrolyte, which limits the regeneration time to the diffusion limit from the nano- to micro second time scale[56, 57]. For both n-type and p-type DSSCs, there have been recent advances replacing the electrolyte by a hole/electron conducting material (HTM/ETM) such as spiro-OMe-TAD [58, 59] and PCBM [60]. Due to the proximity of the HTM/ETM and the rigidity of the system, regeneration is not diffusion limited any more. In the case of p-type DSSCs, the application of PCBM as ETM shortened the regeneration time to ≈ 50 ps, thus much shorter than for most electrolyte based DSSCs [60]. An extension of this concept was applied in paper V and will be explained in more detail in chapter 7.1. 2.5 Photocatalysis Except for generating electricity from sun light, its energy can also be converted to fuels by photoelectrochemical cells (PEC). A common strategy is to split water into molecular oxygen and hydrogen, a reaction that stores 1.23 eV 24.

(38) per oxygen molecule. In a PEC, the principle of natural photosynthesis is applied. Water splitting takes place in two half reactions to avoid the necessity of high energy photons for water splitting: water oxidation (oxygen-evolving reaction, OER) and proton reduction (hydrogen-evolving reaction, HER, which corresponds to carbon fixation in plants). [4, 61–64] Except for hydrogen, also other fuels with higher energy densities per weight, such as formic acid from CO2 reduction are aimed for. The general working principle on the example of water splitting is shown in figure 2.5. Both reactions are initiated independently by the absorption of a photon, in most cases by a photosensitizer. At the photoanode, the catalyst is oxidized by the photosensitizer. After uptake of in total four oxidation equivalents by the catalyst and release of four protons, molecular oxygen is evolved. The protons diffuse through a membrane towards the photocathode while the reduced photosensitizer is re-oxidized by the anode. At the cathode, the catalyst is reduced by the photosensitizer. After the uptake of two reduction equivalents and two protons by the catalyst, molecular hydrogen is released. There are many different architectures for photoelectrochemical cells. In the homogeneous case, catalyst and photosensitizer are in solution. In most cases, they are linked chemically to each other to facilitate the first charge separation. This principle is also applied in paper VI. An alternative strategy is the utilization of heterogeneous catalysis, e.g. by linking sensitizer and catalyst on a semiconductor surface which results in similar charge-transfer processes to DSSCs[64]. In these dyesensitized solar fuel cells (DSSFCs) the reduction/oxidation of the pho tosensitizer is done by the semicon ductor followed by a charge shift or electron hopping between photosen sitizer and catalyst [65, 66]. As the initial CT dynamics in DSSCs and DSSFCs are similar, improvements Figure 2.5. Principle of a photoelectroof DSSCs also gain DSSFCs. In other cases, the electrode ma- chemical cell. At the photoanode (left), terial acts simultaneously as photon water is oxidized to protons and molecuabsorber and catalyst. For those lar oxygen. The protons are later reduced at the photocathode to form molecular hyelectrodes, metal oxides are often drogen. The electrodes can either be housed[67]. To date, it is not clear mogeneous (i.e. in solution phase) as dewhich design is the most promising picted in this figure or heterogeneous (i.e. for solar energy conversion to fuels. the active material is part of the electrode). 25.

(39) Recently, a life cycle analysis for large scale hydrogen production based on nanorod silicon solar cells was made[68]. There, it was found that there is a positive energy balance with an energy payback time of 8 years. The main part of the energy for their production was, however, used for production of the nano structured silicon cells. Here, DSSFCs potentially have the advantage of using cheap mesoporous TiO2 and NiO semiconductors that do not require a lot of energy to be produced. The usage of molecules also has the advantage of tunability. In natural photosynthesis, the structure of the catalytically active sites are well defined by the protein backbone. In principle it is possible to fine tune the structure of a molecular catalysts in a similar way. As complete PECs and DSSFCs are rather complicated and difficult to investigate, it is common to use sacrificial electron donors or acceptors and only investigate and optimize one of the two half reactions[69]. This was also done in paper VI where only the OER was investigated. As there are four electrons involved in this process, it is more complicated that the other half reaction which often only is a two electron process. Under solar light illumination, a single dye in DSSCs absorbs light with a rate of approximately 1 s−1 [70]. Because of this, the CS lifetime in dyecatalyst systems has to be on the same order of magnitude to avoid recombination losses. The strategy to circumvent this problem is to increase the ratio of dyes to catalysts to match charge separation with catalytic activity.. 26.

(40) 3. Methods. 3.1 Steady-State Absorption and Emission As described in chapter 2.1, dyes have distinct energy levels and consequently different light absorption properties making it possible to identify molecular species. To measure absorption spectra of molecules, spectrophotometers are used. A typical setup is shown in the left part of figure 3.1. The light of a lamp is directed through a monochromator to allow measurements at single wavelengths. The light is then split into two beams. One is directed through the sample to measure the intensity of light it transmitted (I1 ) while the other beam is a reference beam to access the light intensity in absence of the absorbing materials (I0 ). Knowing these two intensities, the absorption is then calculated as. I0 = ελ · c · l A = log10 I1 where ελ is the wavelength dependent, dye specific molar extinction coefficient of the dye at wavelength λ , c the dye concentration, and l the path length of the light through the sample. The shape of the absorption spectrum also gives information about the electronic properties of the molecule and aggregation or degradation. Fluorescence and phosphorescence are measured with fluorometers whose general working principle is described on the right part of figure 3.1[71]. The sample is excited by monochromatic light from a xenon lamp where the wavelength is selected with help of a monochromator. To account for possible fluctuations in the excitation intensity, a small fraction of the excitation light is reflected on a photodiode. To diminish the influence of stray light hitting. Figure 3.1. Scheme of the working principle of an absorption spectrometer (left) and a fluorescence spectrometer (right). BS: beam splitter, PD: photodiode, PMT: photomultiplier tube.. 27.

(41) the detector, the emission from the sample is generally detected at right angle from the excitation. Additionally long pass filters are used in the emission path to suppress the second order of the monochromator and scattering from the excitation light.. 3.2 Time-Resolved Absorption To investigate the kinetic processes after excitation of molecules, transient absorption spectroscopy (TAS) is used. The basic principle is described in figure 3.2. The sample is excited by a monochromatic Detector laser pulse (pump). After a certain time delay, a polychromatic laser pulse (probe) pump is used to measure the transmission of the Monot=0 chromator sample. By varying the time delay between t=0 t>0 pump and probe, kinetics of the processes inprobe volved can be measured. Figure 3.2. General scheme of a The pump pulse (often hundreds of nJ) pump-probe experiment. excites typically only a small fraction of the molecules. Therefore, the absorption of the sample does not change a lot upon excitation (see figure 3.3). The (few) excited molecules, however, have a different absorption spectrum compared to the molecules in the ground state. In order to eliminate the contribution of unexcited molecules to the absorption, the difference between pumped and unpumped sample is calculated to result in a difference spectrum (right part of figure 3.3). A typical TA spectrum shows in most cases three features: • Ground-state bleach (GSB) • Stimulated emission (SE) • Induced absorption (IA) The GSB arises because the number of molecules in the ground state is diminished after excitation as those in the excited state do not contribute to the ground state absorption any more. Hence light is less absorbed and results in a negative absorption change upon excitation. The GSB has the same shape as the negative GS absorption spectrum as indicated by the gray dotted line in figure 3.3. The other process leading to a negative TA signal is stimulated emission. In that case a photon from the probe pulse, which has the same energy as the S1 state can make a molecule emit one photon in the same direction while relaxing to the ground state. This process results in more photons than in the unpumped sample leading to a an apparent decrease in absorption and hence a negative feature as well. It can be noted that this process only applies 28.

(42) . .

(43)

(44) .

(45) . . .

(46) . . . . . . .

(47)

(48) . . . .

(49) . Figure 3.3. General scheme of transient absorption spectroscopy. The TA spectrum results from subtracting the absorption spectrum of the unpumped sample from that one of the pumped sample. The green spectrum is an excited state difference spectrum and the yellow one is that of the reduced state. The gray curves are negative steady state absorption (dotted line) and emission (dashed line) for illustration of the origin of the bleach.. for fluorescence, hence it is an indicator for the presence of excited singlet states. The corresponding feature resembles the negative emission spectrum of the compound (dashed gray line figure 3.3). The most important TA feature is the IA, which is the only one that is positive. Here, the absorption that arises from molecules that were excited by the pump is measured. Hence it contains information about excited states and possible reactions occurring in excited states. Because of that, it will be called induced absorption (IA), even though the name excited-state absorption is commonly used in the literature. As most dyes show spectral changes upon oxidation or reduction, IA allows one to follow the processes after excitation. In the example of figure 3.3, the TA spectra of the excited (green) and the reduced (yellow) state of a molecule are shown. As expected, both spectra show similar GSB features while the reduced species has a pronounced peak where the excited state shows SE. One special case of induced absorption is the Stark effect. Here, the dye is in a changed electric field. If this field has a component parallel to the transition dipole moment of the molecule, the wavelength of this transition is red or blue shifted depending on whether the components are parallel or antiparallel respectively. Due to the small shift, the TA signal is proportional to the derivative of the steady-state absorption of the molecule as given in equation (3.1). ΔA =. dA ΔμΔE · dν h. (3.1) 29.

(50) Figure 3.4. Scheme over the TA-setup used for measurements in this thesis. BS: Beam splitter, WL: White light generation, Ch: Chopper, PD: Photodiode, DS: Delay state. where ΔA is the transient absorption signal, dA /dν , the first derivative of the absorption, Δμ and ΔE the change in dipole moment and electric field upon excitation, respectively. To obtain information about kinetics of the sample, TA spectra at different times after excitation were recorded by changing the delay between probe and pump pulse. This is typically done by directing one of the beams through an optical delay line which changes the path length without affecting the position and direction of the beam. The laser setup for the UV/Vis TA measurements in this thesis is illustrated in figure 3.4[72]. Ultrashort 800 nm pulses are generated in the seed laser. With the help of a pump laser, these are amplified in a Ti:Sapphire amplifier. Its output beam is split into two fractions by a beam splitter. One fraction is converted to the desired excitation wavelength in a optical parametric amplifier (TOPAS). Before being directed on the sample, it passes through a chopper blocking every second pump pulse, which allows for shot-to-shot comparison of pumped and unpumped sample. The other fraction of the 800 nm light is first directed through a delay line to allow for adjustment of the time delay between pump and probe. Afterwards, it is converted to a super continuum (white light) by focussing into a CaF2 or sapphire crystal. The white light is split by a beam splitter into a probe and a reference beam accounting for fluctuations in probe light intensity. Both pump and reference beams are directed through a grating to the detector consisting of diode arrays. For the measurements in the mid-IR, a similar setup is used[73]. The only differences are that the infrared probe is generated by a TOPAS with subsequent difference generation. Furthermore, the pump is delayed to avoid changes in absorption of the mid-IR probe light by air due to changes in the path length. 30.

(51) Figure 3.5. Schematic of the calculation of decay associated spectra (DAS). Left: Spectra at different time delays evolving from purple to dark red. Middle: Kinetic traces, at different wavelengths indicated in the spectra by the colored lines. Right: Decay associated spectra, i.e. amplitude of the components at the fitted wavelengths. 3.3 Data Analysis The transient absorption data recorded according to the procedure described above were analyzed with a home-written matlab routine. From the recorded TA spectra at different time delays, kinetic traces were extracted by plotting the TA signal at certain wavelengths against time (see figure 3.5, middle). The so obtained kinetic traces were fitted to a sum of exponential decays and an offset according to equation (3.2) −. ΔA(t)λ = ∑ ci (λ ) · e. t−t0 (λ ) τi. (3.2). i. Because of the refractive index, the speed of light in media is wavelength dependent. Therefore, light with different wavelengths reaches the sample at different times, hence the maximal overlap between pump and probe (time zero, t0 (λ )) is wavelength dependent as well. This phenomenon is referred to as chirp. To account for this, t0 (λ ) is used as a floating parameter during the fitting procedure for all traces. To correct the spectra from the chirp, the fitted t0 (λ ) are interpolated and subtracted from the measured time for every wavelength accordingly. The rise of the components was calculated by convoluting it with a Gaussian shaped excitation pulse whose full width at half maximum (FWHM) was also a free fitting parameter. Except for the rise of the signal around time zero, there is also an artifact[74–76]. This originates from cross phase modulation which is the non-linear interaction between pump- and probe pulse and can be described according to equation (3.3)[74]. −. ca · e. (t−t0 )2 2σ 2.

(52). · sin δ − β ∗ (t − t0 − Δt)2. (3.3). The kinetic traces are then fitted according to equation (3.2) convoluted with a Gaussian response and the artifact according to equation (3.3). Hereby, the 31.

(53) lifetimes of the components are fitted globally, i.e. wavelength independent. Decay associated spectra (DAS) are the wavelength dependent coefficients of the different components. Figure 3.5 shows the schematic calculation of the decay associated spectra from the global fitting. These contain information about the spectral changes associated with the corresponding lifetime. From this, conclusions about the ongoing processes can be drawn by comparison between the DAS and those expected for these processes if reference spectra (e.g. of the reduced state) are known. A point that does not include any spectral change, like 532 nm in figure 3.5, is called isobestic point. Its position and shifts are indicators for present species and the number of different processes. Compared to differences in peak positions, changes of the isobestic points are easier to distinguish and facilitate therefore data analysis.. 32.

(54) 4. Hole Trapping in Nickel Oxide. 4.1 Aim of the study The fast recombination in p-type DSSCs is believed to be the main reason for the difference in solar conversion efficiency compared to their n-type counterparts[36–40]. As described in section 2.4, recombination in sensitized NiO films was found to occur in a biphasic fashion, which was attributed to two different kinds of holes[55]. Understanding of the nature of the holes is important for designing cells with slower recombination. To date, a commonly used strategy is the application of dyads. These are, however, synthetically demanding[77–79]. To investigate hole kinetics in NiO, UV/Vis transient absorption upon bandgap excitation (paper I) of bare NiO and transient mid-IR of sensitized films (paper II) were performed. After bandgap excitation, which leads to the creation of electron-hole pairs, the spectra obtained in the UV/Vis region were compared to reference spectra of different nickel states that were recorded previously[80]. Transient IR spectrosopy of (dye-sensitized) TiO2 films revealed a featureless broad band signal which was attributed to free charge carriers (electrons in the conduction band)[21, 22]. For NiO, similar signals might arise from free holes in the valence band (VB) as those also are free charge carriers. Therefore, these are two complementary measurements for investigation of the hole trapping process in NiO. To be able to attribute the possible mid-IR signal to holes, transient mid-IR absorption was performed on NiO films with two different dyes: a perylene monoimide (PMI), which exhibits typical recombination within 10 ps to 100 ps and a perylene monoimide - naphthalene diimide dyad (PMI-NDI) having a much longer excited state lifetime due the charge shift from the PMI to the NDI moiety resulting in a longer electronhole distance[77]. Hole kinetics are expected to follow the same trend, unless hole trapping leads to a decreased or changed transient signal. However, such a disappearance of the free charge carrier signal is not observed for electrons in TiO2 [21, 81] upon electron trapping[42, 46, 82].. 4.2 Results and Discussion Firstly, an unsensitized NiO-film was excited at 355 nm. The subsequent TA spectrum in the fs-setup was globally fitted by a biexponential decay and an 33.

(55) Figure 4.1. Decay associated spectra of an unsensitized NiO film upon bandgap excitation at 355 nm. The TA data were fitted from 1 ps with a biexponential decay and an offset. For comparison, the 126 ps DAS is also shown enlarged by a factor of 10. Figure reprinted from paper I. offset. The corresponding decay associated spectra (DAS) are shown in figure 4.1. The 2.9 ps component and the long offset component (τ3 >> 2 ns) show similar spectral shapes with IA at wavelengths longer 500 nm. The 126 ps component has a DAS displaying a broad band with a maximum peak around 470 nm. The DAS could be compared to previously obtained reference spectra[80] (see figure 4.2). The infinite component shows large similarity with Ni4+ states, thus trapped holes. As there is no pronounced rise in the DAS of the two faster components, this spectrum contributes to the TA from the beginning. Because of this, hole trapping at Ni3+ states to form Ni4+ h is concluded to take place before the start of the fitting, thus 1 ps. The DAS of the 2.9 ps component exhibits a broad absorption absorption in the below 500 nm. This feature often is assigned to electrons in the conduction band of TiO2 [83, 84]. Assuming that the electrons in the valence band of NiO give similar signals, this process can be assigned to electron trapping. The 126 ps component shows good agreement with the steady state absorption of Ni3+ -states. Due to the similarity of DAS, this process can be attributed to a change from Ni3+ -states to a non absorbing state, i.e (Ni2+ ). Hence, this process can be attributed to electron trapping in deep trap states. The hole trapped in a NiO4+ can perform hole relaxation while undergoing the following reaction 2+ 3+ → Ni3+ Ni4+ h + Ni h + Ni. The relaxed hole in a Ni3+ h state has a lower reactivity than its counterpart in 4+ a Nih state. This is an explanation for the above mentioned biphasic recombi34.

(56) Figure 4.2. Comparison of the 126 ps and the infinite component with previously published reference spectra of the nickel states[80]. Figures reprinted from paper I. nation kinetics as reported in reference 55. This process could be observed for a NiO film sensitized with a Ru-NMI dye (for molecular structure, see figure 6.1). 20 ns after excitation, the TAS showed contribution from both Ni4+ states and the Ru-NMI dye (see figure 9 in paper I). At 5 μs, the Ni4+ features have disappeared while the Ru-NMI features still remain. Due to that, the hole relaxation time can be assigned to take place on a time scale of tens of nano seconds, even though this depends on the concentration of Ni2+ impurities. Summarizing the findings above, Ni3+ -states can act as efficient electron and hole traps in NiO. In the case of hole trapping, a reactive Ni4+ h -state is formed. This state relaxes within tens of nano seconds to a less reactive Ni3+ h hole. This process is the explanation for the biphasic recombination dynamics reported previously. This is also likely the explanation for the large difference in CS lifetime of dyads compared to single dyes only as e.g. reported by reference 77. The observed TA spectra upon band gap excitation have, however, only small signals (around 1 mOD). This is in many cases smaller than the TA signal from the dyes obtained with our setup (order of 10 mOD see below). Thus, hole kinetics are often shaded by those of the dye, even though they are possible to detect as discussed above. Mid-IR is a potential alternative handle for direct spectroscopic observation of holes. Performing transient mid-IR absorption spectroscopy on sensitized NiO films, there is indeed a broad band signal (see paper II, figure 3). Because of the absence of spectral features, the kinetics were constructed by taking the average over the whole spectral range in order to diminish noise. For the kinetics of the sensitized NiO films, there is a clear signal within the first 100 ps (see paper II, figure 4). For the unsensitized NiO and sensitized ZrO2 , there is only a very short-lived signal (within 0.5 ps) which is attributed to an artifact due to cross phase modulation. In another study performed in our group, a NiO film sensitized with a dye and a catalyst was investigated by transient mid-IR absorption spectroscopy. There, only the short-lived signal was observed, which was attributed to the absence of hole injection due to energy transfer from the dye to the catalyst[85]. There35.

(57)

(58).

(59) . .

(60). . .

(61) . . . . . .

(62) . . . .

(63) . Figure 4.3. Comparison of the kinetic traces of PMI and PMI-NDI in the UV/Vis and mid-IR upon excitation at 540 nm with an intensity of 125 nJ/pulse.. fore, the signal on the sensitized films can be attributed to the hole in the nickel oxide. When comparing the decay kinetics of the hole signal with that of the PMIanion, there is a good agreement between the two (see figure 4.3). This is, however, not true for the PMI-NDI dyad, which exhibits a much longer lifetime of the CS compared to PMI. The decay of the hole signal is unchanged for PMI and PMI-NDI, thus indicating that this is due to hole trapping. This is at a slightly slower time scale compared to the results from bandgap excitation in paper I. This difference in hole trapping lifetime could either be explained by different nature of the holes generated by bandgap excitation and injection, respectively. Furthermore, the sensitization dyes could also influence the nature of the hole traps, thus changing trapping kinetics. Despite the slight difference in timescale, hole trapping was found to take place at the sub-ns time scale. This is potentially an explanation of the usually found fast recombination in NiO based DSSCs.. 4.3 Future work To investigate whether hole trapping takes also place at a timescale longer than 1 ps upon bandgap excitation, transient mid-IR absorption might be conclusive. To be able to distinguish between electron and hole dynamics, an electron or hole scavenger might be useful. Unfortunately, such an experiment is not possible to date with the setup used, as it does not allow for UV excitation. However, an upgrade of that measurement setup is about to come, which will allow such experiments to be performed. 36.

(64) 4.4 Implications for NiO-based p-type DSSCs The understanding of hole trapping in dye sensitized NiO films leads to the possibility to develop strategies to improve the performance of NiO based DSSCs. Ni3+ -states were found to be trapping sites that form reactive holes. Therefore, their reduction by physical or chemical means might slow down the recombination kinetics[80]. Another possible strategy is the prolongation of the CS lifetime (often 100s of ps[40, 86–90]) to exceed hole relaxation (tens of ns as discussed above). After the hole relaxation to form a less reactive hole (Ni3+ h ), the CS lifetime might be on the ms to s time scale as seen for many dyads[40, 77, 79, 91]. Hence, it is only a formal increase in lifetime of one the order of magnitude to obtain a prolonged charge separation by 2 to 3 orders. The resulting recombination time scale would then be at the same order of magnitude as that for dye-sensitized TiO2 -films. By that, the most efficient loss process in p-type DSSCs would be suppressed.. 37.

(65) 5. Perylene Monoimides with Different Anchoring Groups. 5.1 The different Anchoring Groups In paper III, the influence of the binding group of a dye on the performance of NiO-based p-type DSSCs was reported. To date, the most commonly used anchoring groups are carboxylate and phosphonate groups[92]. The basic idea of this study was to change the binding group of a perylene monoimide (PMI) dye and investigate influence of the binding group on the photovoltaic performance. The structure of the dye and the anchoring groups used are shown in figure 5.1. The binding properties of the different groups such as dye loading and desorption were investigated. The photovoltaic performance in NiO based DSSCs were measured and related to charge separation, recombination and regeneration investigated by TAS.. t-Bu t-Bu O. N O. N. O. t-Bu. t-Bu O. =. N OH. OH. PMI-CO2H. PMI-Py. NH2. PMI-HQ. PMI-NH2. O. O. PMI-PO3H2. N OH. PO3H2 O. O. PMI-acac. O OH. PMI-DPA. Figure 5.1. Molecular structure of the perylene monoimides with different anchoring groups investigated in paper III.. 5.2 Results and discussion All PMIs of the series show similar steady-state absorption and emission spectra: The absorption has a maximum peak around 540 nm and a weaker vibronic shoulder at ca. 500 nm, which is typical for a π − π ∗ transition of PMI-based dyes. Both PMI-NH2 and PMI-HQ show a slightly broadened and red shifted absorption compared to the other dyes (see figure 1 in paper III). This red shift is attributed to the electron donating character of these anchoring groups. 38.

(66) Figure 5.2. Steady-state absorption and photographs of the PMIs coloring the NiO films (left) and dye loading obtained from desorption experiments (right) Reprinted from paper III. Dye loading studies were performed by soaking NiO films in DMF solution of these dyes. PMI-Py and PMI-NH2 did not attach to the film in a detectable amount. In contrast, dye soaking with the other dyes led to colored films (see figure 5.2). Investigating their absorption spectra, a change in intensity of the two absorption bands is observed, namely the 500 nm band has a higher absorption than the 540 nm one. (A comparison of the absorption of PMI-CO2 H in solution and on NiO can be found int the supporting information of paper III). Similar changes in intensity have previously been attributed to aggregation[93, 94]. Hence, all investigated PMIs show signs of aggregation independently from the anchoring group. Dye loading was measured through desorption experiments by soaking the sensitized NiO films in a DMF solution of phenyl phosphonic acid. Both absorption of the sensitized film and the dye loading measurements suggest the following order of affinity for the anchoring groups PMI-PO3 H2 > PMI-DPA > PMI-CO2 H > PMI-HQ > PMI-acac. . .

(67) . . Figure 5.3. Transient absorption spectra of PMI-CO2 H (left) and PMI-HQ (right) in THF solution after excitation at 540 nm. For comparison the negative steady state absorption (dotted lines) and emission spectra (dashed lines) are displayed as well.. 39.

(68) . . . . .

(69) . . . . . . . . . . . . .

No results found