Discovering Hidden Traps: in Nickel Oxide Nanoparticles for Dye-Sensitised Photocathodes

Full text

(1)Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1515. Discovering Hidden Traps in Nickel Oxide Nanoparticles for Dye-Sensitised Photocathodes LUCA D'AMARIO. ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2017. ISSN 1651-6214 ISBN 978-91-554-9911-2 urn:nbn:se:uu:diva-320187.

(2) Dissertation presented at Uppsala University to be publicly examined in Häggsalen, Ångströmlab, Uppsala, Wednesday, 7 June 2017 at 21:15 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Professor James Durrant (Faculty of Natural Sciences, Department of Chemistry, Imperial College London). Abstract D'Amario, L. 2017. Discovering Hidden Traps. in Nickel Oxide Nanoparticles for DyeSensitised Photocathodes. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1515. 95 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-554-9911-2. The finite nature of fossil fuels and their effect on the global climate, raised the need to find an alternative source of energy. This source should be environment compatible, cheap and abundant. The light coming from the Sun is a promising alternative. To be fruitful, the solar energy needs to be transformed in storable and transportable energy forms like electricityor fuels. Amongst the most studied techniques dye sensitised devices offer the possibility to be designed for both the scopes: solar-to-electricity and solar-to-fuel conversions. In these applications a photocathode and a photoanode, constructed by mesoporous semisconductor films sensitised with dyes, are placed in series with one another.It follows that the photocurrent generated by one electrode should be sustained by the photocurrent produced by the other electrode. At the moment there is a substantial difference between the conversion efficiencies and the photocurrent produced by photoanodes and photocathodes. In this thesis the reasons for this discrepancy are investigated. The main responsible of the bad performance is identified in the semiconductor normally used in photocathodes, Nickel Oxide (NiO). Electrochemical impedance spectroscopy was used to elucidate the electrical properties of mesoporous NiO films. The study revealed that NiO films are able to carry a large enough current to establish that conductivity is not a limiting factor. The recombination reactions were then accused as the cause of the power losses. A time resolved spectroscopic study revealed that NiO can host two kinds of holes. One of these holes is responsible for a fast dye-NiO recombination (100 ns) and the other one for a slow recombination (10 ms). A cell featuring only the slow dye-NiO recombination would possibly reach high efficiency. The characterisation of the species associated with these two holes was performed by density-of-state assisted spectroelectrochemistry. The holes were found to be trapped by Ni2+ and Ni3+ sites located on the NiO surface forming respectively Ni3+ and Ni4+ states. A study by fs and ns transient absorption spectroscopy revealed that Ni3+ sites can trap a hole in subpicosecond time scale and this hole relaxes into a Ni2+ trap in ns timescale. The control of the Ni2+/Ni3+ratio on the NiO surface was found to be crucial for a high cell photovoltage. In the thesis these results are discussed and used to propose an explanation and some solutions to the poor performance of NiO-based dye sensitised cells. Luca D'Amario, Department of Chemistry - Ångström, Physical Chemistry, Box 523, Uppsala University, SE-75120 Uppsala, Sweden. © Luca D'Amario 2017 ISSN 1651-6214 ISBN 978-91-554-9911-2 urn:nbn:se:uu:diva-320187 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-320187).

(3) Alla mia famiglia..

(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. Tuning of Conductivity and Density of States of NiO Mesoporous Films Used in p-Type DSSCs Luca D’Amario, Gerrit Boschloo, Anders Hagfeldt, and Leif Hammarström J. Phys. Chem. C, 2014, 118 (34), 19556-19564. II. Kinetic Evidence of Two Pathways for Charge Recombination in NiO-Based Dye-Sensitized Solar Cells Luca D’Amario, Liisa J. Antila, Belinda Pettersson Rimgard, Gerrit Boschloo, and Leif Hammarström J. Phys. Chem. Lett., 2015, 6 (5), 779-783. III. Chemical and Physical Reduction of High Valence Ni States in Mesoporous NiO Film for Solar Cell Application Luca D’Amario, Roger Jiang, Ute Cappel, Elizabeth A. Gibson, Gerrit Boschloo, Håkan Rensmo, Licheng Sun, Leif Hammarström and Haining Tian ACS Appl. Mater. Interfaces, accepted.. IV Unveiling Hole Trapping and Surface Dynamics of NiO Nanoparticles Luca D’Amario, Jens Föhlinger, Gerrit Boschloo and Leif Hammarström Manuscript ready for submission Reprints were made with permission from the publishers..

(6) Papers not in the thesis. During the Ph.D. studies the author contributed in other scientific works that are not reported in this thesis. They are listed in the following: A comprehensive comparison of dye-sensitized NiO photocathodes for solar energy conversion, Wood C. J., Summers G. H., Clark C. A., Kaeffer N., Braeutigam M., Carbone L. R., D’Amario L., Fan K., Farre Y., Narbey S., Oswald F., Stevens L. A., Parmenter C. D. J., Fay M. W.,La Torre A., Snap C. E., Dietzek B., Dini D., Hammarström L., Pellegrin Y., Odobel F., Sun L., Artero V., and Gibson, E. A.Phys. Chem. Chem. Phys., 2016, 18, 10727-10738; Supramolecular hemicage Cobalt mediators for dye-sensitized solar cells, M. Freitag, W. Yang, L. A. Fredin, L. D’Amario, K. M. Karlsson, A. Hagfeldt, G. Boschloo, ChemPhysChem 2016, 17, 3845; Ultra long-lived electron-hole separation within water-soluble colloidal ZnO nanocrystals: Prospective applications for solar energy production, Cieslak A.M., Pavliuk M.V., D’Amario L., Abdellah M., Sokolowski K., Rybinska U., Fernandes D.L.A., Leszczynski M.K., Mamedov F., El-Zhory A.M., Fohlinger J., Budinska A., Wolska-Pietkiewicz M., Hammarstrom L., Lewinski J., Nano Energy, 2016, 30, 187-192..

(7) Contribution to papers. Paper I: Main responsible for the design of the project, performed all the measurements, the analysis of the data and the interpretation. Wrote the first draft of the manuscript. Paper II: Participated in the design of the project and co-supervised the student that performed the measurements. Analysis of the major part of the data and main responsible for the interpretation. Wrote the first draft of the manuscript. Paper III: Participated in the design of the project, prepared the sample and performed all the spectroscopic measurements and their analysis. Main responsible for the interpretation of the results and wrote the first draft of the manuscript. Paper IV: Main responsible for the design of the project, performed the ns-transient absorption measurements and steady state measurements and their analysis. Main responsible for the interpretation of the results. Wrote the first draft of the manuscript..

(8) List of abbreviations. BG C343 CB CE DOS DSC DSFC FTO HEC HOMO LUMO NP OEC P1 RE Ru-NMI TA TAS VB WE C E E EF EFq e− FF fkww h+ Jsc R Voc β τ. band gap coumarin c343 dye ued in Paper I conduction band counter electrode Density of States (cm−3 eV−1 ) dye-sensitised solar cell dye-sensitised solar fuel cell Fluorine-doped Tin Oxide Hydrogen Evolving Catalyst highest occupied molecular orbital lowest unoccupied molecular orbital nanoparticle Oxygen Evolving Catalyst dye used in Paper III reference electrode ([Ru(dcb)2 (NMI-phen)](PF6 )2 ) dye transient absorption transient absorption spectroscopy valence band working electrode capacitance (in F) electrochemical potential (in J/mol) energy (in J or eV) Fermi level (in eV) quasi-Fermi level (in eV) electron fill-factor stretched exponential function hole, electron vacancy short circuit current density(in A/cm2 ) resistance (in Ω) open circuit voltage (in V) stretching parameter of the fkww time constant of the fkww.

(9) Contents. 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.1 Motivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.2 Solar energy conversion challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13. 2. Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.1 Fermi level and Density of states (DOS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.2 quasi-Fermi level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25. 3. Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1 Semiconductor: Nickel Oxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2 Sensitisers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Electrochemical Impedance Spectroscopy (EIS) . . . . . . . 3.2.2 ns-Transient Absorption Spectroscopy (ns-TAS) . . . . . .. 26 26 26 27 29 29 34. 4. Main 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8. 38 38 40 43 45 52 55 65 67. 5. Summary. 6. Sommario divulgativo. 7. Populärvetenskaplig sammanfattning. 8. Acknowledgments. findings and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DSC performance: n-type vs. p-type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NiO electrical conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The lithium effect on the DOS and conductivity . . . . . . . . . . . . . . . . . The recombination issue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NiO trap characterisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NiO surface trap dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The role of the NiO surface in DSCs operation . . . . . . . . . . . . . . . . . . . Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................................................. ................................................................. 71. ........................................... 74. ....................................................................... 77. ................................................. 80. ................................................................................... 82. Appendix A: Earth’s energy balance Bibliography. 69.

(10)

(11) 1. Introduction. This chapter aims to introduce the reader into my field of work, motivating the relevance of my science. The introduction has purposely been made so that only little prior knowledge of the basic concepts is required.. 1.1 Motivations In thermodynamics a “closed system” is an imaginary box that can exchange energy with the external world, but can not exchange material [1, 2]. The system where we live, the surface 1 of Earth together with the atmosphere, can be considered a closed system. I will refer to it as the surface-atmosphere system. In fact this system exchanges energy with the external space but its mass remains rather fixed and confined (at least during a human lifespan) [3, 4]. This last statement is true with some few exceptions: the atmosphere releases light gases (He and H2 ) into space, the volcanoes inject some minerals and gas from the internal part of Earth and finally humans introduce fossil hydrocarbons into the system [3, 5–8]. The latter has been ascribed, by the scientific community, as the cause of global warming [9–11]. There are several reasons at the basis of global warming, in Appendix A this phenomenon is described in more details. In a few words, the amount of carbon we are introducing in the surface-atmosphere system, turns up directly in the atmosphere in the form of CO2 , increasing the green house effect [4, 12]. This leads to a rise of the Earth’s average temperature that is potentially lethal for the ecosystems [12–15]. The most accepted origin of the fossil hydrocarbons is attributed to the decomposition of biological material that around 250 million years ago sedimented at the bottom of the oceans [16–18]. If this is true, it would mean that the entire mass of carbon stored in the fossil hydrocarbons reservoirs was previously contained in living materials, i.e. participating in the carbon cycle. In terms of flux of material we can see the fossil fuel extraction like the inverse process of petroleum formation. Petroleum formation removed material from the surface-atmosphere system transferring it inside the Earth’s crust and 250 millions years later fuel extraction brings it back. In these terms petroleum extraction does not seem so dangerous, but it 1. By surface is meant the whole material contained in the first few hundreds meters of the Earth’s crust.. 11.

(12) actually is. It is estimated that the amount of carbon contained in the fossil hydrocarbon resources is about half of the one contained in the oceans and ten times the one in the biosphere [19–21]. The surface-atmosphere system, in particular the biosphere, had 250 million years to adapt to the present amount of carbon. Now humans are resuming the carbon level to the one of 250 millions years ago in only 150 years. It is like deflating a balloon in 2 seconds and blowing it up again in 1 microsecond. The balloon would probably explode. The system we live in is much more complex than a balloon, predicting its behaviour upon such a quick perturbation is not easy, but a reaction is surely expected. Especially, if one considers that the perturbation affects the atmosphere, which is the main responsible for the mean temperature of our planet by the green house effect. In the worst case scenario the thermal shock will be lethal for some ecosystems, and we might observe the sixth big mass extinction in history of our planet, that for the first time would have been caused by a choice [22, 23]. The choice in fact is whether or not to replace our main energy source from the fossil hydrocarbons to a “renewable” kind of source. A renewable energy source is able to re-create itself in a reasonably short amount of time, it is thus not finite [24]. Additionally, this form of energy should keep the system closed to perturb the cycles of matter as little as possible. It follows that this form of energy cannot come from the system itself, and it should not be carried by matter (like fossil hydrocarbons). In fact the majority of renewable energies can be attributed to the main energy fluxes arriving on the surface of Earth [24]: Sun (direct light-energy conversion, wind, hydropower, ocean waves), Earth’s nucleus (geothermal) and gravitational forces (tidal energy). In particularly the Sun delivers about 150 PW (1PW=1015 W) to Earth. The human need for energy is 17 TW (1TW=1012 W) [25]. Its abundance makes this source of energy very profitable. Unfortunately, how to convert solar light into a transportable, storable, energy-dense and cheep form of energy remains unknown. One of the first who wondered why humans use “solar produced” fossil hydrocarbons instead of direct solar energy was G. Ciamician, an Italian chemist, who dreamed of “chimney and smoke free industries” in 1912. [26, 27] Inspired by similar dreams, I gladly dedicated my doctoral research in solar energy conversion.. 12.

(13) 1.2 Solar energy conversion challenges For solar energy conversion it is usually meant the conversion of electromagnetic energy into another form of energy that is either electrical or chemical. The discussion of all the possible ways of converting the solar energy is beyond the scope of this thesis. Thus I will just briefly summarise them (for a better overview see [28–30]). The most common strategies for solar energy conversion into electricity involves the use of a p-n junction as light harvesting site and a charge separation system to generate an electric current. The techniques based on this principle are: single junction cells (Si and Ge as most common semiconductors), multi-junction cells (world record of energy conversion) and thin film cells [31–33]. Organic photovoltaics use a similar strategy but the charge separation is carried out with donor and acceptor polymers [34, 35]. Other techniques involve the conversion of the solar energy into heat which is then used to either produce steam for turbines or perform a chemical reaction [32, 36]. The conversion in chemical energy is currently accomplished in the production of “biofuels” [37, 38]. Here, plants using photosynthesis transform CO2 and O2 into organic material, an oil, that is extracted by squeezing the plant and used as fuel. There is a big effort in the scientific community to genetically modify living organisms, like algae or bacteria, to be able to artificially direct the bio-synthesis for product excretion avoiding the destruction of the living organism [39–41]. The research I carried out in these years of doctoral studies is focused on the study and characterization of p-type NiO based photocathodes. These electrodes could be used in electrochemical devices to convert solar energy to electricity or chemical energy. These devices are called respectively dyesensitized solar cells (DSCs) and dye-sensitized solar fuel cells (DSFCs). Dye-sensitized solar cells As already mentioned, the most common way to convert light into electricity is to use a doped semiconductor that is able to absorb photons and create a charge separation. In these devices the semiconductor material is shaped in the form of a film or a foil. One of the two sides of the foil is pdoped, the other one is n-doped, thus a p − n junction is created. The sides are covered with transparent electrodes for charge collection. The light harvesting occurs by fundamental absorption (or band gap excitation) where an electron is promoted from the valence band (VB) to the conduction band (CB) of the semiconductor [32, 42]. The charge separation instead happens in the p-n junction. The depletion layer created by the junction builds a local electric field transversal to the semiconductor foil. The hole (h+ ) and electron (e− ) created in the band gap excitation can be separated by 13.

(14) the local electric field and flow away from each other and finally reach the electrodes. This mechanism is used in all the photovoltaic devices based on the p-n junction. It is possible to calculate the maximum theoretical conversion efficiency of the p-n junction cell. The conversion efficiency is the ratio between the power obtained from the system after the conversion and the power delivered to the system. The power of a solar cell is proportional to the product of the voltage and current that the cell can deliver at a certain time [42, 43]. This is intuitive, the voltage account for the potential difference of the separated charges, i.e. the energy stored in each h+ -e− couple, while the current account for their flux. The potential of the h+ -e− couple cannot be higher than the photon that generated it and it is assumed to be the energy of the band gap.2 With this assumption the voltage of the cell becomes strictly related with the band gap of the semiconductor. Now it is easy to see that the power delivered by a p-n junction solar cell has a theoretical maximum. In fact, a material with a large band gap will show a high voltage but will suffer from a small current since the material will not be able to absorb the low energy photons of the solar spectrum. Contrarily a material with low band gap will cover the entire solar spectrum, thus will have a large current but it will have a low potential. The theoretical maximum conversion efficiency3 of a single p-n junction solar cell is called “Shockley-Queisser” limit and is equal to 34% with an optimal band gap of 1.34 eV [44]. Devices that exceed this limit have been built by coupling in series several p-n junctions with different band gaps [45]. Currently a multi-junction devices holds the record of solar energy conversion efficiency of 46% [46]. A different approach respect to the p-n junction device was adopted in developing the dye-sensitized solar cell. In this kind of device the light is absorbed by a molecular light harvester, a dye, and the charge separation occurs at the interface between the dye and the mesoporous semiconductor where the dye is linked [43, 47]. A schematic representation of the working principle of a DSC is presented in Figure 1.1. DSCs can be divided in two main kinds, p-type and n-type, based on the kind of semiconductor used in the cell. The vast majority of scientific work regarding DSC has been made on n-type, mainly using TiO2 as semiconductor. Most of the concepts and characterization methods used in this thesis have been developed studying TiO2 -based DSCs. The working principle of the two kinds of DSCs is similar. p-type DSCs is the 2. The absorbed photons with higher energy than the band gap are considered to loose the excess of energy by thermalization. 3 This is done assuming that every photon contributes to the generation of a charge pair, i.e. a quantum efficiency = 1.. 14.

(15) main subject of this thesis, therefore, in the following, only the working mechanism of p-type DSCs is presented. A p-type DSC consists of two transparent electrodes, a “photocathode” and a counter electrode, see Figure 1.1 left side. The photocathode is covered by the active material, a mesoporous film (∼1μm thick) of a wide band gap p-type semiconductor. Normally the film consists of nanoparticles (NP) but DSC based on nano-rods, nano-wires, nano-leafs, etc. have also been reported [48–50]. The nanoparticles are sensitized with a suitable dye that can inject a hole in the semiconductor VB upon photon absorption. The photocathode faces the counter electrode typically by a distance of few tens of micrometer. The electrodes are kept in electric contact by an electrolyte solution of a redox couple that wets both of them.. Figure 1.1. Schematic representation of a p-type DSC. Left scheme: from the left, the photocathode, an FTO (Fluorine-doped Tin Oxide) glass with semiconductor NPs (NiO) deposited on it; the dye linked to the surface of the NPs is in contact with the redox couple, this closes the circuit at the platinized counter electrode (FTO) on the right. In the right scheme: 1, photo-excitation of the dye; 2, hole injection to the valence band of NiO; 3, regeneration of the reduced dye by the redox couple; 4, recombination between the reduced dye and the VB hole; 5, recombination redox couple-VB hole.. The right side of Figure 1.1 describes the processes occurring in the DSC. The absorption of a photon brings the dye in an excited state, reaction 1. The excited state of the dye is oxidative enough to inject a hole in the VB of the semiconductor, reaction 2. This creates a charge separation; the hole, h+ , resides in the nanoparticle while the electron, e− , is located in the dye that in this phase is reduced. If the charge separation lives long enough the dye is regenerated by the redox couple that accepts the e− , reaction 3. 15.

(16) The redox couple carries the e− to the counter electrode while the semiconductor transports the h+ to the photocathode closing the circuit. It follows that the output voltage of the cell is given by the difference of the potential of the holes, the quasi-Fermi level (EFq )4 , and the one of the redox couple. The beauty of this mechanism is comparable only to the frustration of the scientist working on p-type DSC trying to make it work well, since it does not. In fact, each of the steps discussed above can be suppressed by a competitive process that ends up recombining the h+ and the e− created by hole injection. Every time this happens the energy accumulated in the charge separation is lost. Normally just two processes are considered to be the main source of power losses, reaction 4 and 5 in the scheme, respectively dye-hole recombination and electrolyte recombination. In the dye-hole recombination the h+ recombines with the e− in the reduced dye, while in the electrolyte recombination the h+ reacts with the reduced part of the redox couple. There are also other factors that can cause power losses. Two examples are the resistance to the transport of the charges in the electrolyte or the semiconductor, and the resistance to the charge transfer between two parts of the cell (i.e. redox couple-counterelectrode) [43, 51, 52]. In terms of power losses there is a substantial difference between p-type and n-type DSCs. One of the important points of this thesis is that the main differences between n-type and p-type DSC should be addressed to the particular material, TiO2 vs. NiO. The concept of DSC has been present in the scientific literature from 1991 when B. O’Regan and M. Grätzel published their work on a n-type TiO2 based DSC [47]. In n-type DSCs the working mechanism follows the reaction steps described above with the difference that, instead of a hole, an electron is injected in the conduction band of the semiconductor while the dye oxidises. So far the field of n-type DSC has developed and has become one of the main research fields of chemical science. Moreover, the record of solar conversion efficiency of a TiO2 based DSC reached 14% [53], meaning that many concepts regarding the mechanism introduced above have been understood and well applied. From the first work by Grätzel the field of DSC has evolved and transformed. New branches of research were born and the two most popular ones are the Solid-State-DSC and the perovskite solar cell [54, 55]. In the Solid-State-DSC the liquid contact between the two electrodes has been substituted with a solid conductor. Perovskites on the other hand are generated from the attempt of replacing the molecular sensitizer with an inorganic sensitizer. Perovskite-based solar cells became the new research fashion in the solar energy conversion community replacing DSC in just three years from the first publication. The reason for this is the high con4. For more information about the quasi-Fermi level see Section 2.1 and 2.2. 16.

(17) version efficiency of ∼20% obtained even in early works. In contrast to n-type DSC, the field of p-type DSC, mainly based on NiO as semiconductor, is rather new (see Lindquist, 1999 [56]) and it did not have the same success as that of its n-type counterpart. Despite years of research and huge knowledge inherited from the n-type field there are only three cases where the conversion efficiency exceeds 0.5% [57–59]. Many attempts have been made to improve the performances of the p-type cell: different sensitizers were used [60–67], different linker groups [68–72] and many redox couples [56, 58, 59, 73–75]. Instead, the semiconductor was modified very few times [76–88]. It seems that the secret of the poor performances is hidden in some property of the semiconductor material. The interest for p-type DSC is not purely conceptual. A working ptype DSC could be coupled to a n-type DSC to build the so called tandem DSC [56, 89]. In a tandem DSC the counter electrode of an n-type DSC is substituted with a p-type photocathode. The two electrodes have complementary absorptions covering together the entire visible range. Similarly to a multi-junction solar cell the theoretical maximum efficiency of a tandem DSC exceeds the Shockley-Queisser limit.. Solar fuel cells By photosynthesis nature has built the entire biosphere using light, inorganic carbon (CO2 ) and molecular oxygen to produce carbohydrates and other highly energetic molecules. The branch of science that tries to emulate this process using chemistry principles is called artificial photosynthesis. At the moment its main scope is restricted to water splitting or CO2 reduction [90–92]. The first process would use solar energy to decompose H2 O in its two components: H2 O(l) −−→ 12 O2(g) + H2(g) hν. This is a highly endoergonic reaction, 141.8 kJ/g(H2 ) , which makes hydrogen a powerful combustible. In comparison, the combustion energy of propane is 50.3 kcal/g(C3 H8 ) (all the combustion heat in the thesis are retrieved by CRC Handbook of Chemistry and Physics [93]). Moreover, burning hydrogen produces mainly water that solves the problem of smog. Unfortunately, hydrogen has a very low energy density which is inconvenient for transportation. The energy density, dE , of a system is the energy stored per unit of volume [94, 95]. For example, the dE of a fuel is the energy released by burning a litre of fuel. Hydrogen, as a gas, has an energy density of about 0.01 MJ/L at 1 atm (5.6 MJ/L if compressed at 700 atm) while LPG/propane has 25.3 mJ/L. The scientific community is putting a 17.

(18) lot of effort searching for ways to store hydrogen in an efficient and safe way [95–97]. A few examples of recent progresses in this direction are: cryo-compression, intercalation in clathrate or carbon nanotubes, as metalhydride and adsorbed into metal-organic frameworks. One more handicap in the use of hydrogen as fuel is that today’s transportation infrastructure is based on hydrocarbon thus it would need a complete reconstruction. Despite the disadvantages, using hydrogen as fuel seems to solve most of the environmental problems. This perspective has attracted the attention of the researchers to find the way to perform the water splitting reaction in the most efficient way. Along with water splitting, CO2 reduction is considered the main alternative for replacing fossil fuels [98–100]. The research in CO2 reduction aims to efficiently reduce CO2 in one of these products: methanol, formic acid, carbon monoxide. These three substances are aimed because each of them can be used in the present industrial infrastructure to produce any other carbon based chemical from fuel to plastic. This would facilitate the conversion to a fossil fuel free society since it would keep most of the present hydrocarbon based technology intact. Burning hydrocarbons derived from artificial photosynthesis would not increase the net amount of carbon in the surface-atmosphere system, this is thus considered a “zero emission” strategy. Though, it would probably not improve the smog issue. There are several ways to convert solar light into chemical energy. Usually the conversion into high energy chemicals occurs by a redox reaction at an electrode. A simple way to achieve this transformation is to convert light into electricity with a photovoltaic device and then use an electrochemical cell to perform the chemical reaction. This is not considered here as a solar fuel cell. A solar fuel cell performs the light harvesting and the energy conversion in the same device. The light harvesters and the catalysts, which allow the chemical transformations, are deposited together on the electrodes. In the case of the water splitting cell these two catalysts are the oxygen evolution catalyst (OEC) and the hydrogen evolution catalyst (HEC) [101]. The light harvester can also be of the same material as that of the catalyst. For example one of the first cases of artificial photosynthesis was the discovery that the band gap excitation of TiO2 initiates water oxidation [102]. After promoting an electron in the CB, the hole left in the VB of TiO2 is so oxidising that it can oxidise water in its proximity. The majority of the research in this field is done with inorganic systems, like in this case TiO2 , which can function like a light harvester as well as a catalyst. The choice of such a material is then made on the basis of the reaction that needs to be performed. In the case of water splitting, for example, the energetics of the reaction requires at least a photon of 1.23 eV to occur. Moreover the position of the VB of the semiconductor that performs the water oxidation needs to be more positive than +1.23 V vs. 18.

(19) NHE in the reduction potential scale. In the same way the reduction potential of the CB of the material that realises the water reduction needs to be more negative than 0 V vs. NHE. Despite the complexity there are already working devices that can execute photosynthesis, one example is the “artificial leaf” which uses a silicon solar cell as a light harvester and a cobalt OEC and a nickel HEC [103]. Another approach is to use molecules as catalysts and light harvesters immobilized on transparent electrodes. These cells are called dye-sensitized solar fuel cells (DSFCs) [104]. As shown in Figure 1.2 the DSFC has many features similar to a tandem DSC. The DSFC is composed by two transparent electrodes: the photocathode where the reduction takes place and a photoanode where the oxidation occurs. As in the tandem DSC, on top of each of the electrodes a mesoporous film of a semiconductor is sintered. At the photocathode a p-type material is used while an n-type one is applied at the photoanode. The two mesoporous films are sensitized with different dyes with complementary absorption. Moreover at the photocathode a catalyst for the reduction reaction is co-sensitised with the dye. Likewise, a catalyst for the oxidation reaction is co-sensitized at the photoanode. In the case of water splitting cell the photocathode is co-sensitised with a HEC, and the photoanode with OEC, see Figure 1.2. The two electrodes are immerse in the media that contains the substrate for the photosynthesis. In the water splitting case it is buffered water.. Figure 1.2. Schematic representation of a solar fuel cell. On the left the spatial configuration of the various components. On the right the schematic of the mechanism: the arrows indicate the movement of an electron.. 19.

(20) In right of Figure 1.2 each step of the electron cycle across the cell is represented. Starting from the NiO side, the excitation of the dye causes the hole injection into the VB which reduces the dye. The reduced dye transfers an electron to the HEC which can perform water reduction. In the photoanode, TiO2 side, after the excitation an electron is injected into the CB resulting in the oxidation of the dye. The oxidised dye is regenerated by the OEC which now can perform water oxidation. The electron in the TiO2 CB reaches the electrode and recombines with the hole in the NiO VB by travelling in the external short circuit. The charge neutrality of the cell is assured a flow of ions between the two electrodes. Similar to DSCs there are several issues that need to be understood and solved before this mechanism can fully work. Here the recombination issue can be considered even worse since the dye is regenerated by a species, the catalyst, that stays on the surface of the nanoparticle. This enhances the probability of recombination in DSFCs compared to DSCs where the regeneration occurs by a species that after the reaction is released in the bulk, far away from the nanoparticle. One of the major issues for DSFCs is the design of the water oxidation catalyst and the related problem of the charge accumulation. In fact, contrarily to DSC, there is a need of accumulation of charges for the redox reaction. For the water oxidation reaction, for example, the catalyst needs to transfer 4 electrons per molecule of oxygen. The catalyst then needs to store those electrons or catalyse a multi-step process. The importance of the semiconductor In both types of cells, DSC and DSFC, the semiconductor covers a fundamental role. In the first prototypes of DSCs, the dye was deposited on a flat surface [105]. The amount of dye that a flat surface can host might be thousands of times less than that of a mesoporous surface. In DSCs this is essential since only the dyes attached to the surface of the semiconductor are active in electron transfer. Due to the difference in the amount of adsorbed dyes, the mesoporous film enhances the optical density of the electrode by thousands, in some cases resulting in an absorption of more than 99% of the light. In DSCs wide band gap semiconductors are used since they need to be transparent to allow the light harvesting of the dye. Moreover, the bands of the semiconductor need to be properly aligned with the HOMO/LUMO of the dye to allow the photoinduced electron transfer. For example ZnO has the VB potential so positive that most of the p-type dyes cannot inject a hole in it, thus it can not be used in a p-type DSC. One of the most important differences between the DSC and the DSFC is the relative positions of the semiconductor bands. In the tandem DSC the output voltage of the cell is given by the difference between the CB 20.

(21) potential of the n-type semiconductor and the VB potential of the p-type semiconductor. Thus it is important, for a tandem DSC, that the bands of the two semiconductors express the largest difference in potential. While for DSFC the difference in potential of the two bands should ideally be zero. In the DSFC, in fact, the converted energy should be as much as possible in the chemical products, while a difference in potential between the two bands would convert energy into electricity. This is why the commonly used TiO2 and NiO are a good couple of semiconductors for a tandem DSC but not for a DSFC: the potential difference between the TiO2 -CB and NiO-VB is about 1 V. The semiconductor is also very important for the charge dynamics. The semiconductor is the carrier of the charges from the NP surface to the FTO electrode. It is important that the semiconductor has a good electrical conductivity in the working condition of the cell. Moreover, the dynamics of charge transfer and charge stabilization occurring in the interface dye/semiconductor or electrolyte/semiconductor or catalyst/semiconductor is crucial for preventing charge recombination events. An important drawback is that the majority of knowledge about semiconductors, especially NiO, regards bulk properties while most of the mechanisms occurring in these complex systems are happening on the surface. Considering the nanoscopic dimension of these semiconductors (NP ∅=4-50 nm), the surface is probably affecting the properties of the entire material. A better understanding of these concepts is needed to interpret the behaviour of the present materials and for the design of new ones. The research I carried out in these years was sparked by the mystery surrounding the poor performances of NiO based DSCs. At the beginning of my Ph.D. studies the maximum efficiency obtained from a p-type DSC was 0.5%. The research in this particular field had then already existed for more than 10 years. The first question I wanted to answer was whether the NiO is capable of driving enough current density to allow a high conversion efficiency. At that time, it was a common belief that NiO hole mobility was too poor to sustain a large current density [106]. The conductivity of NiO was studied in Paper I which showed the contrary, NiO can result in sufficiently large current density to allow a high efficiency. The research on NiO conductivity raised new questions on the nature of the NiO-hole and its role in the recombination reaction. The dye-NiO recombination was studied in Paper II where the presence of two kinds of holes was discovered. One of the two kinds of holes was found responsible for a rapid dye-NiO recombination which in turn is believed to cause the power losses in NiO DSCs. The chemical nature of these two holes was revealed in Paper III where the high valence Ni states, Ni3+ and Ni4+ , were spectroscopically characterized. Finally in Paper IV the surface dynamics of the holes were resolved using transient absorption spectroscopy. 21.

(22) It turned out that Ni3+ sites, that are known to be on the NiO surface, can trap the injected hole very rapidly (subpicosecond time scale) in a Ni4+ state. This brings the hole to the surface of the nanoparticle increasing the rate of dye-NiO and electrolyte recombinations. The Chapter 4 of this thesis is intended to be a discussion about the findings of my papers, presenting the major results, their connections and meaning. Instead Chapter 2 and 3 introduce the reader to some important concepts and methods used in the thesis.. 22.

(23) 2. Fundamentals. In this section I am going to briefly describe some fundamental concepts that are essential in the understanding of the treated matters. The concepts of Fermi level, quasi-Fermi level and density of states (DOS) are going to be used often in this thesis. It is useful to have a reference on the physical meaning of these concepts without exploring all the details of how these can be estimated or calculated by principles. A more extensive introduction can be found in ref. [107] and [108].. 2.1 Fermi level and Density of states (DOS) In thermodynamics it is useful to define a quantity that can measure the contribution of the amount of matter to the balance of an equilibrium. This quantity is called chemical potential [1]. This concept is clear to a chemist for systems like solutions or gasses, but it might be unknown for solid-state materials. Briefly, given an arbitrary number of substances N, they are in equilibrium with each other, at constant pressure and temperature, when the Gibbs free energy (G) of the system reaches the minimum: dG = 0. G = f (P, T, n1 , n2 , ..., nN ),. where. P is the pressure, T is the temperature, ni is the molar quantity of substance i. When P and T are constant dG is only a function of the composition of the system: dG =. N dG i=1. dni. where,. . μi =. dni = (P,T,n2 ,..nN ). dG dni. N . μi dni. (2.1). i=1. (P,T,n2 ,..nN ). μi is the chemical potential of the species i. Basically, the chemical potential tells about the contribution of a certain quantity of a chemical species to the overall Gibbs free energy. The chemical system tries to reach a minimum of the free energy. Thus, if there is a way to convert a high chemical potential species to a low chemical potential species, this conversion happens following equation 2.1. In this way the concept of chemical potential 23.

(24) can explain all the basic chemical phenomena like reactions, diffusion, phase changes, etc. The chemical potential can easily be related with the electric potential defining the electrochemical potential [107]: μ ¯i = μi + zi F ΦE where zi is the net charge of the species i, F is the Faraday constant and ΦE is the electrostatic potential in the point where the electrochemical potential is considered. In a sample in equilibrium, the electrochemical potential is constant across the entire sample, this allows to calculate the concentration profile of the species contained in it. The chemical potential, especially in electron transfer reactions, is associated to the energy level of the species. Thus for a molecule, where the energy levels are discrete, it has a well defined value. In other words it is easy to sum the energy of the system counting every single contribution. For solid-state systems, where the energy levels are a continuum, the definition of a chemical potential is more complicated. Since the levels are not discrete, there is the need to define a quantity that allows for counting the energy levels in the system at a specific energy. This quantity is the density of states (DOS) that counts the number of energy states available for a certain particle per volume of the solid [108]. The DOS is a function of the energy and it is the quantity that, in the case of the electron, shapes the valence band and conduction band of a solid. Intuitively, if the electrons were a fluid, the VB were a tank then the DOS is the shape of the tank. At the thermal equilibrium, the probability of finding an electron in the VB at a certain energy is given by the Fermi-Dirac distribution:. p() =. 1 (e(−EF )/kT. + 1). where is the energy, k the Boltzmann constant, T the temperature, EF the Fermi level. The center of this distribution, i.e. the energy where the probability of finding an electron is 0.5, is called the Fermi level, EF , and it is defined to be the chemical potential of the electrons of the material. At room temperature, the Fermi-Dirac distribution is practically 1 at energies below the EF and 0 at higher energies. The region where 0.1<p<0.9 covers about 150 meV. With these concepts, the phenomena described in this thesis, occurring at the interface of a semiconductor, can be rationalized.. 24.

(25) 2.2 quasi-Fermi level In a system where the semiconductor is an electrode in contact with a solution, the Fermi level of the material is at the potential of the electrode. This means that if we set the potential of the electrode to a specific value vs. a reference, the entire Fermi level will shift. If the system is at equilibrium the electrochemical potential is constant across the entire sample, i.e. the Fermi level of the material matches the chemical potential of the solution. If this is not true, the system is not in equilibrium and a reaction can occur like an electron transfer from the solution to the semiconductor or vice versa. This builds an electrostatic potential that will move the electrochemical potential, eventually bringing the system to the equilibrium. A DSC in the darkness should be in equilibrium thus no voltage should be read at the electrodes. Under illumination the DSC is at working condition, the voltage read-out at the electrodes is the difference in potential between electrons in the semiconductor and the species in contact to the counter-electrode. The potential of the electrons of the material cannot be considered as the chemical potential or the Fermi level because the system is not at equilibrium. The system is instead at a stationary state. This means that the electrons in the semiconductor might occupy the bands with a different distribution than the Fermi-Dirac. In general there might exist several local Fermi levels where the electron occupation probability is 0.5, they are called quasi-Fermi levels (EFq ) [107]. Usually, in n-type DSC research, the term quasi-Fermi level is used for the one created in the conduction band by the injected electrons [43]. This is considered the location of the potential of the electrons.. 25.

(26) 3. Materials and methods. In this chapter the materials and the techniques relevant to my research are discussed, The sensitizer and the preparation methods for the NiO films are shortly described. A brief introduction to the two main techniques, electrochemical impedance spectroscopy and ns-transient absorption spectroscopy, is presented.. 3.1 Materials 3.1.1 Semiconductor: Nickel Oxide The semiconductor material concerned in this thesis is mesoporous nickel oxide. Nickel oxide (NiO) is a 3.5 eV indirect band-gap semiconductor. Indirect BG differs from direct BG by the mechanism of absorption of light, i.e. the way that an electron can be excited from the VB to the CB by the means of photon absorption [108]. Direct BG semiconductors feature the maximum of the VB aligned with the minimum of the CB in the energymomentum space, see Figure 3.1. Thus the excitation of the electron occurs with the absorption of a photon with energy equal to the BG. In indirect BG semiconductors the two bands are not aligned respect to the momentum, see Figure 3.1. Thus the excitation from VB to CB needs to occur with a photon together with a particle that carries momentum, like a phonon. For probability reasons, this two-particles-mechanism makes the extinction coefficient of indirect BG much lower than the one of direct BG. For the same reason, indirect BG have a very low fluorescence quantum yield, ΦF , like that found for NiO (ΦF 0). This fact implies that after BG excitation the possible electron-hole recombination dissipates the energy entirely in heat, which was observed and discussed in Paper III. NiO has a p-type character. This is mainly due to the presence of Ni3+ impurities in the crystal structure due to Ni2+ vacancies [109–111]. When NiO contains no Ni3+ defects it is a mixed Mott/charge-transfer insulator [112, 113]. NiO RT bulk conductivity is about 10−13 Ω−1 cm−1 [114], but it can be increased up to 10−1 Ω−1 cm−1 by introducing Ni2+ vacancies [115, 116]. The concentration of these vacancies is oxygen partial pressure dependent and affects the NiO conductivity [111].. 26.

(27) Figure 3.1. Direct (left) and indirect (right) band-gap excitations. “Pht.” stands for photon and “phn.” for phonon.. The mesoporous structure is obtained by aggregation of NiO nanoparticles. The nanoparticles can be created by precipitation, spray pyrolysis, and sol-gel sintering [117, 118]. Usually nanoparticles are made from a precursor (NiOH, Ni(NO3 )2 , NiCl2 , Ni(AcO)2 ) and subsequently sintered to form the oxide. The NiO mesoporous films used in this thesis were prepared by sintering a NiCl2 sol-gel. The sol-gel is prepared mixing 1 g of NiCl2 anydrous, 1 g of triblock polymer F108, 3 g of H2 O and 6 g of ethanol (99.5%). After complete dissolution of the solids (overnight sonication), the sol-gel is spread on the transparent substrate by doctor-blading. The ethanol is allowed to evaporate for 2 min. The films are then sintered in a closed oven at 450◦ C for 30 min (with 30 min ramping). In the oven the solvents evaporate and the polymer burns leaving the NiCl2 nanoparticles that are transformed into NiO by calcination. The mesoporous film was prepared in this way on conductive FTO, CaF2 windows or fused silica. The prepared films were used in the construction of the DSC or used in the electrochemical or spectroscopic measurements. The exact preparation of the cells varied depending on their sought purpose and it is reported in the papers.. 3.1.2 Sensitisers In Paper I-III NiO was sensitized with a dye to test the photocathode in working condition or to trigger a hole injection to study the recombination process. Three different dyes were used: coumarin C343 in Paper I, Ru-NMI in Paper II and P1 in Paper III. The formula of these dyes are reported in Figure 3.2. The summary of the properties of these dyes is reported in Table 3.1. 27.

(28) Figure 3.2. Molecular structure of the dyes used in this work. From left to right: coumarin C343, Ru-NMI, and P1. Table 3.1. Spectroscopic and electrochemical characteristics of the dyes used in this thesis.. Dye. λmax (nm). C343 Ru-NMI P1. 422 470 468. . 1 ( cm·M. HOMO. LUMO. ). (V vs.NHE). (V vs.NHE). 4. 1.4 1.52 1.38. -1.2 -1.12 -0.87. 4.4·10 1.2·104 5.8·104. ref. [119] [60] [65]. Coumarin C343. It has for many years been the reference dye for NiO-based DSC [120, 121]. It is a yellow green laser-dye; the performances are very low but are reproducible. Therefore it was taken as standard. The injection and recombination dynamics is clarified by ultrafast spectroscopy [119, 122]. The fast recombination kinetics (ns-timescale) makes it impossible to use it with a slow diffusing redox couple like Co-based complexes. Hence, it is used with I– /I3– redox couple. Ru-NMI. It is one of the few Ru-based dyes for p-type DSC, it was synthesized in reference [60]. It offers the possibility to be used with slow diffusing redox couple since its recombination kinetics is very slow (10−6 -0.1 s). The reduced dye spectrum, showed in Figure 3.3, is well characterized [60]. Ru-NMI is then a good choice for recombination studies, it was in fact used in Paper II.. 28.

(29) Figure 3.3. TAS of Ru-NMI− obtained by exciting Ru-NMI in presence of Ferrocene (FeCp2 ) as electron donor. Exc. wavelength 480 nm, [Ru-NMI]=5 mM FeCp2 =100 mM. Data reprinted from my master thesis [123].. P1. It is another standard dye for NiO-based DSC [117, 120]. Even though it features a fast self recombination, ∼200 ps, the I– /I3– redox couple can regenerate the dye quite efficiently, probably thanks to a pre-associated complex formed with the dye [124, 125]. In fact it is used as a reference sensitizer for DSC and DSSFC and was used in Paper III.. 3.2 Techniques 3.2.1 Electrochemical Impedance Spectroscopy (EIS) Any electrochemical cell can be rationalized as an electric circuit. An electrochemical cell, as any electric circuit, can be analysed with principles proper of electrodynamics [126]. First of all, the concepts of the electrochemical system and cell need an explanation to avoid misunderstanding. An electrochemical system is anything that can be connected to two electrodes to be able to apply an electric potential between them and measure the current of electrons that flows through them. Thus, it can be almost anything from a water solution of NaCl to a lithium battery. The electrochemical system together with electrodes constitute the electrochemical cell. Two kinds of cells are commonly used in electrochemistry: the three electrode cell and the two electrode cell. In the first one, the electrochemical system is connected to three electrodes: a working electrode (WE) which is the probe into the 29.

(30) electrochemical system; the reference electrode (RE) which constitutes the reference for the potential applied at the WE; the counter electrode (CE) where the electrons collected (or expelled) from the WE come from (or end up to), see Figure 3.4. In other words the potential of the WE is applied versus the RE while the current flows to the CE. There are many characteristics that RE and CE need to fulfil to be able to work as described. Without entering into details, the most important ones are: the RE needs to have very low polarizability, i.e. its electrochemical potential should not change when electrons are exchanged from the electrode; the CE instead should be able to deliver or store a large quantity of electrons changing its potential as little as possible. In the two electrode cell the RE and the CE are the same electrode. This can be crucial during some electrochemical analysis since the function of one electrode can influence the other. In general, the scope of the electrochemical investigation is to analyse the response of the system to electrical stimuli (usually a potential variation) to extract thermodynamic and kinetic information. There are several ways to test an electrochemical system. The simplest is to apply a potential at the WE varying its magnitude linearly with time across an interval of potential, the current that flows in the WE is then measured. The analysis of the behaviour of the current respect to the applied potential gives information about the chemical or physical processes occurring at the WE. This technique is called cyclic voltammetry. Another way to test an electrochemical system is to apply an oscillating potential to the WE. Usually this potential has a sinusoidal shape, i.e. it is a wave, and the operator can control its frequency (f, f= 1/τ ), amplitude (A) and potential offset respect to the RE, see Figure 3.4. The current generated from this potential modulation is recorded. This current has a sinusoidal shape too. The analysis of its amplitude and phase shift respect to the potential wave gives information about the system. The physical concept that is used here is the impedance. The easiest way to describe it is to use Ohm’s law. For direct current circuits, DC, the well known Ohm’s law V = R · I describes the relation of the current I across a resistance R caused by a constant potential difference V . The same concept is applied when the voltage is modulated in an alternate current circuit, AC. V has the form:1 V (t) = |V | · ei(ωt+φV ) 1. This was given in complex form, it can be transformed in real form by Euler’s formula cos(ωt + φ) = [ei(ωt+φ) + e−i(ωt+φ) ]. 30.

(31) Figure 3.4. Schematic representation a EIS measurement to a three electrode cell. On the left: the schematic graph of the shape of the applied potential with respect to time. On the right: the registered current vs time, where the change in amplitude and phase shift is emphasised.. where t is the time, i is the imaginary number, ω is the frequency of the oscillation and φV is the phase shift of the oscillation (the translation of the entire wave with respect to the time). The current that is created across the circuit assumes an oscillatory form as well: I(t) = |I| · ei(ωt+φI ) the variables definitions are analogue to the voltage ones. The Ohm’s law is valid any time in this system and it is written as V = Z · I, where Z is the impedance. Here the form of the impedance is found substituting the expressions for V and I: Z=. |V | · ei(ωt+φV ) = |Z| · ei(φV −φI ) = |Z| · eiθ |I| · ei(ωt+φI ). The impedance of a circuit is then found quite easily knowing the amplitudes of the potential and measuring the current and the phase shift of the current respect to the potential: |Z| =. |V | |I|. and. θ = φV − φI. Impedance is a complex number, it is then represented in a two-dimensional space, Re(Z) vs. Im(Z), the complex plane. In an EIS measurement the voltage modulation is applied to the WE in a three electrode cell, the frequency of the modulation is varied and the impedance is measured in function of the frequency. There are two common ways to represent such measurement: the Nyquist plot and the Bode phase plot. The Nyquist plot simply reports the measured impedances in a complex plane. The 31.

(32) Bode phase plot reports the phase shift vs. the frequency (in a logarithmic scale). In Figure 3.5 an example of how these two plots are used is given.. Figure 3.5. Schematic representation a EIS measurement to an interface (left) and its representation in Nyquist plot (center) and Bode phase plot (right).. EIS is often used to analyse charge dynamics at interfaces. In the left plot of Figure 3.5 an interface is represented by the contact of light and dark gray areas, e.g. the electrode in contact with an electrolyte. Usually, an interface is rationalised as a capacitor in parallel with a resistor, which resemble the non faradaic and faradaic behaviour of the interface, respectively. At really high frequencies (ω = ∞ limit) the impedance is 0 since the current pass entirely through the capacitor, see Nyquist plot, and the phase shift is π/2, see Bode phase plot. At really low frequencies (ω = 0 limit) the impedance is R since the current pass entirely through the resistor and the phase shift is 0.2 At frequencies in between this range the Nyquist plot assumes a semicircular behaviour. This is due to the impedance of the system becoming the sum of the contributions from the capacitor, purely imaginary, and from the resistor, purely real. The frequency top of the Nyquist semicircle is given by the characteristic frequency of the RC circuit, ωRC = 1/RC. Usually, the Nyquist plot shows more than one semicircle, this is due to more than one interface present in the system. This can happen only if the two interfaces have different enough characteristic frequencies associated with their RC circuits.. Equivalent circuit The way to perform an EIS analysis is based on the formulation of the so called equivalent circuit [126–128]. The equivalent circuit is an electric model of the system that should resemble the behaviour of the electrochemical cell under test. The equivalent circuit can then be simulated and the 2. Intuitively it is easy to see that the phase shift of a capacitor is π/2 while the one of a resistor is 0. The alternate current measured across a capacitor will be at the max when the applied voltage is transiting around zero while it will be zero at the potential peaks, i.e. θ = π/2. While for a resistor, the current is at the max when the potential is at its max, i.e. θ = 0.. 32.

(33) Nyquist plot and the Bode phase plot can be fitted, extracting the parameters of the equivalent circuit. It is important to wisely design the equivalent circuit to extract the correct information. In Paper I EIS was used to analyse the electrical properties of a bare NiO film. EIS was performed in a three electrode cell using as WE a NiO mesoporous film (0.7 μm) sintered on a FTO glass. The supporting electrolyte was LiClO4 0.5 M in acetornitrile. The Nyquist plot produced by such a system is represented in Figure 3.6. The equivalent circuit used to fit the data is also reported on the right.. Figure 3.6. Left: representation of the Nyquist plot obtained in the experiment of Paper I. Right: quivalent circuit associated to the system.. The analysed system produces three different regions in the Nyquist plot. The semicircle is due to the FTO-electrolyte interface created by the space on the surface of FTO left free from the semiconductor. The 45◦ angle straight line is due to the charge transport in the film and it is associated with a so called “Warburg resistance”. The vertical straight line is associated with the capacitance of the double layer formed at the film/electrolyte interface. The equivalent circuit was formulated following the transmission line model (for more information see ref. [127]). The Rs is the infinite frequency resistance of the cell. The Wf t indicates the Warburg resistance related to the film charge transport. The Rf to and CP Ef to are the resistance and the capacitance related with the FTO-electrolyte interface. The Rdl and CP Edl are the resistance and the capacitance related to the film-electrolyte interface.. 33.

(34) 3.2.2 ns-Transient Absorption Spectroscopy (ns-TAS) The light-matter interaction, besides being one of the most charming aspects of nature, can be used as a powerful tool for scientific investigation [129]. In the everyday experimental work, spectroscopic techniques like IR, UV-Vis absorption spectroscopy and fluorescence spectroscopy are used to characterise substances and investigate chemical processes. The ones just mentioned are called steady state techniques. Namely a beam of light passes through a sample (the spectrum of the light is measured prior and after the passage) and the absorbed light is then analysed. The light absorbed by the sample can trigger chemical reactions that can modify the absorption of the sample. This is not “seen” from a steady state technique since the signal detected is averaged over a time (0.1 s - 1 s) much longer than most chemical reaction kinetics. In other words the absorption is detected when the system is in a steady state produced by the light itself. This is usually not important since the intensity of light used in these techniques is not high enough to modify the sample noticeably. In studies of reaction dynamics or mechanism it is often needed to involve techniques which can time-resolve the property changes. In transient absorption spectroscopy (TAS) the sample is excited, usually with a light pulse, and the following changes in absorption is monitored. Obviously, the sample excitation should be as short as possible in relation to the studied process. For this purpose laser pulses are used since they can be produced in a wide range of durations from attoseconds (10−18 s) to continuous. The time resolved techniques working in the range as-ps are called ultrafastTAS, they work in a substantially different way from ns-TAS and will not be discussed further in this thesis. Usually, in a ns-TAS system a laser pulse of about 5-10 ns long is produced from a Q-switched Nd:YAG laser (in our case a Spectra-Physics Quanta-Ray Pro-230). The fundamental emission of such a laser is in the IR, 1064 nm. The energy of each pulse, after being amplified, is about 1 J/pulse, which is enough to be used in non-linear optics to undergo double harmonic generation (532 nm) and triple harmonic generation transforming the fundamental to 355 nm light. The 355 nm beam can be directed into an optical parametric oscillator that can convert it to light of any wavelength in the range of 440-780 nm. The light pulse (∅ ∼ 6 mm) is directed in the sample chamber and hits the sample, usually a 1 cm cuvette, with a 90◦ angle respect to the probe beam. The probe beam is generated by a Xe arc lamp that produces a strong and well collimated white light. The probe light hits the sample and is then directed into the detector chamber where it is analysed by a charge-coupled-device (CCD) camera or a photo multiplier tube (PMT). 34.

(35) Figure 3.7. Upper schemes: intensity vs. time of the pump and probe beam. The rainbow area indicates that the spectrum is measured as a whole in the TA, instead the turquoise area represents the use of monochromtic light in the transient kinetics. Bottom schemes: the transient absorption sectrum (left) and kinetic trace (right).. The CCD camera can analyse the entire spectrum of the probe beam at a specific delay after excitation, t0 , see Figure 3.7. The PMT instead is coupled with a monochromator so it can monitor one single wavelength of the evolving spectrum in a range of time. The CCD camera measures the spectrum of the probe beam in two times, prior excitation and after the excitation at a given time. The difference of these two spectra is called transient absorption or delta absorption, TA. A transient absorption spectrum can have negative and positive features, see Figure 3.7. The negative bands are called bleaches and arise from the disappearance of a chemical species after the excitation, such as the ground state molecules that were excited by the pump pulse. The positive bands are due to the spectrum of the population of species that were created upon excitation. The formation and evolution of these bands can be monitored by the PMT, recording the so called transient kinetics. After excitation the molecule can evolve by itself or react with other molecule. These processes can be followed by transient kinetics and basic principles of chemical kinetics can be used to extract information about the occurring mechanisms. 35.

(36) Stretched exponential function The kinetic informations are usually extracted performing fitting of the kinetic traces. The fitting functions are normally set by the kinetic law of the process in study. Usually this is enough to discriminate between first order or second order kinetics. Though in processes happening at an interface or on the surface of mesoporous materials, as in the case of DSC, the kinetics do not follow simple laws of bulk chemistry. There is not a very good physical model for describing interface kinetics and so far various strategies have been adopted to compare kinetics of this kind. The simplest way to treat non-ordinary decay is to fit the trace with a sum of exponential functions with different time constants and then compare the weighted times. The sum of exponentials is a concept that is at the base of another function used to fit non-ordinary decays: the stretched exponential or Kolraush-Williams-Watt function (kww) [130, 131]. The stretched exponential was used in Paper IIto describe the dye-NiO recombination kinetics. The kww-function can be written as: fkww (t) = e−( /τ ) t. β. where t is the time, τ is the observed time constant and β is a parameter that can assume values from 0 to 1. The influence of β on the function is shown in Figure 3.8. When β is 1, the function is a pure exponential and as β approaches 0 the function behaves faster for t < τ and slower for t > τ , see Figure 3.8 a.. Figure 3.8. The fkww function in: a linear scale and b in logarithmic scale. fkww is represented using the same τ but with different β.. Representing the fkww function in the logarithmic scale, the role of β becomes more clear, see Figure 3.8 b. The β sets the slope of the straight line found around τ . If one imagines the fkww function as a sum of exponentials then the more the β value approaches 0 the more exponentials are needed to describe the fkww function. In other words the kinetics becomes more disperse as the β value decreases. 36.

(37) The kinetics constant τ does not describe very well the behaviour of the function since much depends on the β value. For this reason the value of τ is normally given together with an “averaged” τ , so called τ which is calculated as: τ = τ · Γ(1 + β −1 ) . This gives unreliable results when β becomes very small, i.e. at β<0.1 τ assumes values larger than the lifetime of 98% of the species. Decay Associated Spectra (DAS) In Paper IV fs-TAS has been used and the data are presented in form of decay associated spectra (DAS), which are introduced in the following. In a fs-TAS a series of transient absorption spectra are collected at different delay times. Often the TA spectra can be several tens or hundreds, this allows to analyse the entire set of data with a particular algorithm called global fitting. The global fitting tries to reproduce the TA spectra (ΔAbs) by the following sum: ΔAbs(λ, t) =. n . αi (λ) · e− /τi t. i. where αi (λ) is a spectral component of the TA spectrum evolving with time constant τi . n is the number of components used for the fitting and it should be the smallest possible. Figure 3.9 shows schematically the result of a global fitting. On the left the TA spectra recoded at three times, the two bands on the left decay together while the band on the left decays more slowly. The DAS on the right shows the two DAS associated to these TAs.. Figure 3.9. Left: TA spectra of an hypothetical experiment. Right: DAS related to the TA of the same experiment.. 37.

(38) 4. Main findings and Discussion. In this chapter the most important findings of my research are presented. I will discuss their interrelation and the connections with some other results from the DSC community. Reading note: from now on the Figure n of paper M will be referred as Figure n(M) with M in Roman numbers.. 4.1 DSC performance: n-type vs. p-type As already mentioned in the introduction, the performance of the tandemDSC is limited by the photocathode. The conversion efficiency of a solar cell is given by the ratio between the maximum electrical power gained from the cell, Pmax , and the power delivered to it, Pin [43]. In formula: η=. Pmax Voc · Jsc · F F = Pin Pin. where Voc is the voltage at open circuit, Jsc is the current density at short circuit and F F is the fill-factor, a factor that accounts for the nonideality of the cell.1 In Figure 4.1 the typical result of a J-V test of an n-type and a p-type DSC are shown. The difference in performance between the photocathode and the photoanode is evident. The p-type DSC shown here shows weaknesses in all the three factors important for high efficiency: Voc , Jsc and F F . The efficiency is around 0.10.5% and for many years this was the normal result that was achieved by dye sensitised photocathodes. The two cases where the efficiency exceeds unity (1.3% and 2% [58, 59]), report a great improvement of the Voc , while the Jsc and the F F remain prohibitive. A photocathode featuring low Jsc and F F is a limiting factor not only for the single cell but also for the tandem DSC and DSFC, where the currents of the photocathode and photoanode must be comparable. Since no improvement was gained with modification of the dye and/or redox couple the attention of the community shifted on the electrical properties of NiO. Investigations on the mobility of the charges in the bulk of a working p-type DSC revealed that the mobility of the holes in 1. The F F is the ratio between the product V · J measured at the maximum power point, Pmax , and the product Voc · Jsc , see Figure 4.1.. 38.

(39) Figure 4.1. Typical J-V curve of an n-type DSC and a p-type DSC. The light gray area indicates the maximum power delivered by the n-type cell. One can compare this area with the teoretical one constructed by Jsc and Voc .. NiO VB is about three order of magnitude less than the electron mobility in the TiO2 CB (charge diffusion coefficients respectively 10−8 -10−7 cm2 /s vs. 10−5 -10−4 cm2 /s [106, 132, 133]). This result could explain the poor Jsc and the apparent high cell resistance.2 The poor Jsc was also attributed to a high rate of recombination between dye and NiO that was proven to be important in most of the cases [132, 134, 135]. Moreover, in the case where the rate of the recombination is a function of the electrode potential the recombination losses can be associated to the low F F , see Paper IIand [136]. With such a rapid recombination it is possible that the holes that were probed for the charge mobility are only a part of the original pool of injected holes. The holes that are lost in the recombination are probably the more reactive ones and most likely the highly reactive hole are also the most mobile. It follows that the holes that are measured at the electrode might be the ones less reactive and less mobile. The main purpose of Paper I of this thesis was to measure the electric properties of mesoporous NiO in conditions where the recombination issue can be ignored. 2. The cell resistance can be easily estimated from the J-V plot. It is the slope of the straight line tangent to the J-V curve in the Voc point. It is easy to see that in the n-type is relatively low (few Ohms) while it is very high in the p-type case (hundreds Ohms).. 39.

(40) 4.2 NiO electrical conductivity In Paper I the NiO film and NiO-electrolyte interface were characterised in absence of a dye. A NiO coated electrode film was prepared with the same procedure used for the DSCs, a 0.7 μm mesoporous film on FTO glass. The NiO electrode was used as WE in a device-like (two electrodes) cell.3 The electric contact between the two electrodes was achieved by an electrolyte, I– /I3– in acetonitrile, which has been the standard electrolyte for DSCs. The cyclic voltammetry of such a cell is reported in Figure 4.2, black curve. .

(41) . .

(42) !

(43) ! "#$

(44) % % % % . . . . . . . . . . . . .

(45) Figure 4.2. Cyclic voltammetry plot of Li doped NiO film on FTO glass with device-like cell (2 electrode cell, with a platinised counter electrode): orange FTO without NiO film, black 0% Li, red 0.1% Li, blue 0.5% Li green 1% Li. The measurement was performed in darkness and with a scan rate of 100 mV/s, the redox couple was I− /I− 3 100/100 mM in acetonitrile. –Reprinted from Figure 1 of Paper I. The electrode exhibits oxidation and reduction currents. The inert window between the two currents is remarkably small, ∼200 mV. Considering that the I− /I− 3 redox potential is believed to be ∼180 mV above the NiO VB, this is not the behaviour expected from this electrochemical system. In Figure 4.3 the scheme of two semiconductor-redox couple configurations are considered. As depicted in Figure 1.1, the redox couple should have a potential as negative as possible in comparison to the VB of the semiconductor. This will ensure a high Voc . This is the case 1 reported in Figure 4.3. A cyclic voltammetry resulting from this system is also represented in Figure 4.3 right. The voltammogram shows only the oxidation current. This is due to the position of the redox couple with respect to the VB. Since 3. The cell was composed by a platinised counter electrode separated by 50 μm Surlyn spacer.. 40.

No results found