• No results found

Interactions in Dye-sensitized Solar Cells

N/A
N/A
Protected

Academic year: 2022

Share "Interactions in Dye-sensitized Solar Cells"

Copied!
68
0
0

Loading.... (view fulltext now)

Full text

(1)Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 901. Interactions in Dye-sensitized Solar Cells BY. HELENA GREIJER AGRELL. ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2003.

(2)  

(3) 

(4)     

(5)      

(6)  

(7)             !"!# $  % & $    $ '%  %( )%  

(8) 

(9) *  

(10)   

(11) +

(12) &%(   ,- .& /(  ( 0

(13)   

(14)  

(15) 1

(16)  2 3   4( . 

(17)     

(18) ( 

(19)  

(20)  

(21)  

(22)

(23)  

(24)            5!( #5 (    ( 03 5!1##1#6718 )% 

(25)   

(26)   *

(27) %    

(28)   

(29)  

(30) 1

(31)  2     *   * % 91903

(32)  0:      : 

(33)  

(34) & 

(35)   

(36)   

(37)  

(38)  

(39) (        $ %  ;     

(40) <1; % 

(41) < ;=1 

(42) 11    =1   <  %

(43) ;00< 

(44) %         % % 

(45)  

(46) &

(47)  *  $ 

(48)     $  % ( .  %  *      %

(49)  

(50)   1

(51)  2     

(52) & 

(53)

(54)  : 

(55)  

(56) & ;::3<( ::3   $     1

(57)  2     

(58) 

(59) &  

(60) 

(61)   %   

(62) %     

(63)   %   %

(64) & % 341 &

(65)  $ % ( 0 *    %

(66) 

(67)  >?@0 1 $  

(68)  *%%        

(69)  *

(70) 0 1

(71)  341 $   % %

(72) & $ %

(73) 

(74)  :;00< $ % ( )%   !1 %

(75) 2 2    % 341A0 1 &

(76)  %

(77) &  $ 

(78) &.    * % >?( 0

(79)    % & 

(80)   

(81)

(82)

(83)    )B $   * %    . %

(84) C *    $   

(85) 

(86) &   

(87) 

(88) & $ 

(89)  

(90)   

(91).  

(92)

(93)  $$    ( )* &

(94)  *  

(95) &% $  %  

(96)  *

(97) 

(98)   

(99)   

(100) 

(101) 

(102)  

(103) ( 0

(104) % $ &

(105) ;D!1  

(106)   )B.  < *%%   % &

(107) *% %

(108)

(109)    1

(110)  2       %   

(111)  $     

(112) * %  

(113) &A  

(114) &  % 

(115) &   $  % & 

(116)    

(117) * % 

(118)  

(119) 

(120)   (   % 

(121)  &

(122) ;E   

(123)   )B  < % & 

(124)     

(125)  

(126) % 

(127)  

(128) 

(129)    %      ( )% &% % * F 

(130)  $$  $  % 

(131)     *

(132) %      

(133) 

(134)  * ( )%  

(135)   % 2 %  

(136)  $ *1 

(137)  

(138)   

(139)  

(140) 1

(141)  2    (  

(142)      %    1

(143)  2   

(144)

(145)    : 

(146)  

(147) &   ! "  # $     % 

(148)   

(149) $ &' ()*$ 

(150)    

(151) $ +,)(-./ 

(152) $   G /

(153) ,- .&  033 !!1 8 03 5!1##1#6718 

(154) "

(155) 

(156) """ 1 6# ;% "AA

(157) (F(A H

(158) I

(159) "

(160) 

(161) """ 1 6#<.

(162)  

(163) .          

(164)

(165)                     .  !  " #   $ %&' "   ( )*++,'.   

(166)              & /          

(167)  *01 *2&03. .  !  (  "   4 "   ( 5         & /         67&7%   

(168) 

(169)   

(170) 

(171) . .  !  "   4 "   ( )*++,-            & /      8            

(172)    

(173)   ,+91 :0,;&:0*+. 7.  !  (  "   4 "   ( < .      & /      

(174)   

(175)        

(176)  ) -. 7.  !  (  " =  "   ( <.       >* 

(177)        

(178)    

(179)   ) -. 

(180)       

(181)   

(182)  .

(183)

(184) Contents. Abbreviations.................................................................................................vi 1 Introduction..................................................................................................1 1.1 The need for renewable energy ............................................................1 1.2 Solar cells .............................................................................................1 2 General semiconductor electrochemistry.....................................................4 2.1 Semiconductors ....................................................................................4 2.2 Electrolytes...........................................................................................9 2.3 Transition metal compounds ..............................................................12 2.4 Semiconductor electrolyte interfaces .................................................13 3 The dye-sensitized solar cell......................................................................17 3.1 Design ................................................................................................18 3.2 Operation principles ...........................................................................20 3.3 Performance .......................................................................................22 4 Coordinative interactions of the dye and the electrolyte............................31 4.1 Characteristic Raman lines -vibrational assignments.........................32 4.2 SCN- ligand loss correlated to [I3-].....................................................34 4.3 Interaction between Li+ and 1-methylbenzimidazole.........................36 4.4 Coordination of I3- ..............................................................................37 5 Conductivity of nanostructured TiO2 .........................................................39 5.1 Conductivity.......................................................................................40 5.2 Non-linear increase of conductivity with charge ...............................43 5.3 Linear increase of conductivity with charge ......................................45 5.4 Role of electrolyte cations..................................................................46 6 Methods .....................................................................................................48 6.1 Raman scattering................................................................................48 6.2 Conductivity and electron accumulation............................................50 7 Concluding remarks...................................................................................53 Acknowledgements.......................................................................................55 References.....................................................................................................56. v.

(185) Abbreviations. B CH CB CE c D D* D+ Di Dn Dp DSC dcbpy ∆φSC E EA Eag Eaκ Eaµ EaµC EaQ EC EC(0) EC(Q) ED EF Eg EOx ERed Eredox Eredox* EV e ε0 εr F. pre-exponential factor Helmholtz capacitance conduction band counter electrode ion concentration dye excited dye oxidized dye ion diffusion coefficient electron diffusion coefficient hole diffusion coefficient dye-sensitized solar cell dicarboxybipyridine potential difference within the space charge layer energy acceptor energy level activation energy of free electron generation activation energy of electron conduction activation energy of effective electron mobility activation energy of electron mobility in the CB activation energy of electron accumulation conduction band edge conduction band edge without accumulated electrons conduction band edge with accumulated electrons donor energy level Fermi energy bandgap energy of oxidized state energy of reduced state energy of redox couple redox energy of excited molecule valence band edge electron vacuum permittivity relative permittivity Faraday’s constant vi.

(186) FF f(E) HOMO h η I I ID Imax IP ISC IR ielec iF1 and iF2 ifilm i1,corr and i2,corr k κ LCA LUMO MBI 3MPN me* µeff µn µp µ+ µNA NC ND N(E) NHE n nC npart ν0 {Ox} Pin Pmax pV. fill factor Fermi-Dirac distribution highest occupied molecular orbital Planck’s constant efficiency total current intensity of vibrational band dark current current at the working point Pmax photogenerated current short-circuit current infra-red leak current through the electrolyte Faradaic currents through the conducting glass of electrodes 1 and 2, respectively current through the TiO2 film, a mean value of i1,corr and -i2,corr equilibrium currents measured at electrodes 1 and 2, respectively, corrected for a small error in the bipotentiostat Boltzmann’s constant conductivity life cycle assessment lowest unoccupied molecular orbital 1-methylbenzimidazole 3-methoxypropionitrile effective mass of electron effective electron mobility electron mobility hole mobility mobility of positively charged ion mobility of negatively charged ion Avogadro constant effective density of states within few kT above the conduction band edge effective donor density number of allowed states per unit volume at energy E normal hydrogen electrode electron concentration electron concentration in the CB number of electrons per TiO2 particle fundamental vibration frequency chemical activity of the oxidized species total radiation power incident on the solar cell maximal electrical power hole concentration in the valence band vii.

(187) Q QE QD QH QSC qe R Relec Rfilm RSnO2 RCT RE {Red} r0 ρ T TBA+ 4TBP UOx URed Uredox $ U redox. UV UV-VIS VOC Vmax VB v v% Wp WE xD xIH xOH xSC z zj. electron accumulation excess charge in the electrolyte excess charge in the electrolyte diffuse layer excess charge in Helmholtz layer space charge in the semiconductor elementary charge gas constant resistance of the electrolyte resistance of the TiO2 film sheet resistance in the SnO2 layer resistance of electrochemical charge-transfer reference electrode chemical activity of the reduced species particle radius resistivity temperature tetrabutylammonium ion 4-tert-butylpyridine potential of oxidized state potential of reduced state solution redox potential standard redox potential ultra-violet ultra-violet and visible open-circuit voltage voltage at the working point Pmax valence band vibrational state volume percent peak watts, a dimension of the solar cell system at 1000 W/m2 and 25°C (AM1.5) working electrode diffuse layer thickness distance to inner Helmholtz plane distance to outer Helmholtz plane width of the space charge layer number of electrons transferred in redox reaction charge of ion j. viii.

(188) 1 Introduction. 1.1 The need for renewable energy It is assumed that the world energy demand and CO2 emissions will both increase by about 70% between 2000 and 2030 [1]. Fossil fuels, supplying 80% of all energy consumed worldwide, are facing rapid resource depletion [2]. The resource reserves of fossil fuels in the whole world in 2002 were projected to last 40 years for oil, 60 years for natural gas and 200 years for coal [3]. Because of a growing demand for energy, combined with the depletion of fossil resources, global warming and its associated climate change, there is an urgent need for environmentally sustainable energy technologies. Renewable energy is the production of electricity, transport fuel or process heat from sources that, for all practical purposes, do not run out. Renewable energy technologies, which include photovoltaic, solar thermal, wind turbines, hydropower, wave and tidal power, biomass-derived liquid fuels and biomass-fired electricity generation, supply to date only 14% of all energy consumed worldwide [2].. 1.2 Solar cells A photovoltaic system consists of solar cells and ancillary components such as power-conditioning equipment and support structures. Solar cells utilize the energy from the sun by converting solar radiation directly into electricity. The potential of the sun as an energy resource is enormous as about 1×1018 kWh strike the earth’s surface yearly [4]. This can be compared to the total world energy consumption of 1×1014 kWh in 2000 [5]. Today, the barrier for market penetration is directly related to the high cost of the electricity generated by photovoltaics. The price for a complete gridconnected photovoltaic system is about 5.5 USD/Wp [4], which with 12.5% efficiency, 1200 kWh/m2year insolation and a 20-year lifetime, corresponds to an electricity price of 0.3 USD/kWh. In 2001, crystalline silicon 1.

(189) accounted for nearly 74% of the solar cell production [4]. The high costs associated with crystalline silicon solar cell modules are mainly due to the high-energy demand for the purification of silicon dioxide from quartz or sand to silicon and to low material yield during the fabrication process. At present, photovoltaics are competitive mainly in remote areas. Markets here are small in the developed world and potentially enormous in the developing world, but largely inaccessible due to lack of access to finance. Fortunately, a market-pull mechanism seems to have recently been established that seems likely to provide growth of photovoltaic applications and cost reductions due to higher production volumes. This is by the subsidization of rooftopmounted systems in urban areas of the developed world [6]. The potential for on-going cost reduction is the key reason for confidence in a significant role for photovoltaics in the future [6]. The future of photovoltaics is believed to belong to thin film solar cells. For thin film solar cells the cost reduction is achieved by reduced material needs compared to crystalline solar cells and the possibility of producing large-area modules in a continuous process. Thin film approaches are usually thought of as those utilizing active semiconductor material of approximately 10 µm or less in thickness. There are mainly three thin film modules on the market: devices based on amorphous Si, and polycrystalline materials based on the semiconductor compounds CdTe or Cu(In,Ga)Se2. Amorphous silicon devices show stabilized efficiency of 4-8% [7]. Modules of CdTe give 910% efficiency [8]. Modules of Cu(In,Ga)Se2 give 10-12% efficiency, which is not far behind that achieved using standard crystalline silicon wafer modules (15-18%) [8]. It could be noted that the thin film group at Ångström Solar Center has reached 16.6% efficiency on so called minimodules. In the dye-sensitized solar cell (DSC), which is relatively new, a completely different thin film approach is used, based on electrochemistry at the interface between a dye adsorbed onto a porous network of nanometer-sized titanium dioxide particles (nanostructured) and an electrolyte. At present the certified cell efficiency is 10.4%. One unique characteristic of this solar cell is the ease with which it is produced. Low energy-demanding processing is attractive from both a cost and an environmental perspective. Life cycle assessment (LCA) is a useful tool for evaluating the environmental impact from a new technology. In an LCA of a DSC system (Paper I), it was found that the DSC system would result in about 20-50 g of CO2 emission per kWh electricity generated, compared to 450 g CO2/kWh for a natural gas power plant. The most significant activity contributing to the environmental impact over the life cycle for the DSC system is the process energy required for the solar cell module fabrication. One improvement of the current process 2.

(190) technology from an environmental (and cost) point of view would be the use of a new low-temperature procedure using compression [9] instead of sintering at increased temperature. As the dye-sensitized solar cell will have to compete with the photovoltaic technology that is currently on the market, a similar performance (efficiency and stability), a similar environmental impact and significantly lower production costs must be achieved. The research in this thesis is aimed at a better fundamental understanding of the operation principles of the DSC with the long-term goal of improving efficiency and stability.. 3.

(191) 2 General semiconductor electrochemistry. There is a complex interaction between a semiconductor, a dye and an electrolyte in the nanostructured dye-sensitized solar cell. In this chapter the general properties of the different components in a DSC device, i.e. semiconductors, transition metal compounds, electrolyte solutions and semiconductor electrolyte interfaces are reviewed [10-16].. 2.1 Semiconductors 2.1.1 The band structure When forming a solid material, N atomic orbitals combine to N molecular orbitals with a spacing between the discrete energy levels, which decreases with increasing number of atoms, i.e. increasing the number of molecular orbitals. When the number of atoms is large enough, the energy levels will form a continuum, since the spacing between the discrete energy levels becomes infinitely small. The ensembles of energy levels are called energy bands. The electronic structure may have a band gap, a range of energies in which no orbital states exist for the electrons. Due to differences in the electronic structure, solid materials can be divided into conductors (metals) and dielectrics (semiconductors and insulators). For a semiconductor or insulator, as opposed to a metal, there is a gap (bandgap, Eg) between the highest filled energy band (valence band, VB) and the lowest empty band (conduction band, CB) (Fig. 1). The valence band is occupied by valence electrons and the conduction band is associated with excited state energy levels. The highest filled energy level and the lowest empty energy level are called the valence band edge (EV) and conduction band edge (EC), respectively. In the ground state at zero Kelvin, the ½N orbitals in the valence band are completely filled with N electrons. Electrons in a completely filled band cannot carry current. At temperatures above zero Kelvin, electrons can be excited by the thermal motion of the atoms, resulting in electrons populating orbitals in the conduction band. The electrons are now mobile, and the solid is an electric conductor. There will 4.

(192) now be a vacant position in the valence band (a hole) and this vacancy makes motion in the valence band possible. Hence, the current flow in a semiconductor can be regarded as being due to the sum of the motion of electrons in the conduction band and holes in the valence band.. E (eV) 0. E (eV). Vacuum. 0. Conduction Band EC Eg. E (eV). Vacuum. Conduction Band. EF Valence Band. EV. (a). 0. ED EF. Vacuum. Conduction Band. Valence Band. Valence Band. (b). (c). EF EA. Figure 1. Energy level diagrams of (a) intrinsic, (b) n-type and (c) p-type semiconductors in vacuum.. Thus, electronic conductivity in semiconductors requires electrons to be excited to the conduction band or holes present in the valence band. This may be done thermally or optically. If the gap is large, however, very few electrons will be promoted thermally at ordinary temperatures and the conductivity will remain close to zero, giving an insulator. Materials with band gap energies between 0 to 3-4 eV, i.e. between a conductor and an insulator, are called semiconductors.. 2.1.2 Energetic distribution of electrons The number of electrons per unit volume occupying levels in the conduction band is given by integration EC max. nC =. ³ N ( E ) f ( E )dE. (1). EC. N(E) is the number of allowed states per unit volume at energy E. 5.

(193) At an energy near the conduction band edge, N(E) is given by. 8 2πme*3 / 2 (E − EC )1 / 2 N (E) = 3 h. (2). with an effective mass of one electron, me* and Planck’s constant, h. An effective mass needs to be introduced since the mass of an electron in a crystal lattice differs from that of an electron in free space. The Fermi-Dirac distribution f(E), is the fractional occupation of the available orbitals at energy E and temperature T.. f (E) =. 1 1+ e. (3). ( E − E F ) / kT. where k is the Boltzmann constant. The Fermi energy, EF, is the energy level where one half of the orbitals are occupied (f(E)=1/2). The fractional occupation of the available orbitals decreases rapidly for levels above EF and increases rapidly for levels below EF under standard conditions, and consequently most of the electrons in the conduction band are clustered near the conduction band edge. For energies well above EF, the exponential term in equation 3 is very much larger than unity, and a Boltzmann-like distribution may replace the FermiDirac equation. f ( E ) ≈ e − ( E − EF ) / kT. (4). At the energy of the conduction band edge this is valid (EC-EF>>kT) and the concentration of electrons in the conduction band is obtained by 32. § 2πme* kT · ¸ e −( EC − EF ) / kT = N C e −( EC − EF ) / kT nC = 2¨¨ 2 ¸ ¹ © h. (5). where NC is a constant at fixed T, known as the effective density of states within few kT above the conduction band edge. As a rough order of magnitude, NC at room temperature is around 1020 per cm3, or one state per hundred atoms. The concentration of holes in the valence band (pV) can be calculated from a similar treatment. For an intrinsic semiconductor the concentration of 6.

(194) electrons is equal to the concentration of holes and EF is located in the middle of the band gap, see Figure 1a.. 2.1.3 Perturbation of the electronic structure of semiconductors Perturbation of the electronic structure is often caused by defects in the crystal structure of the semiconductor. These perturbations can give rise to new energy levels, which change the Fermi level and alter the electrical properties of the semiconductor. The relative positions of the additional energy levels depend on the type of defects. The deliberate introduction of impurities into the lattice in order to create charge carriers is called doping. A semiconductor can be either n-doped (Fig. 1b) or p-doped (Fig. 1c) by the introduction of donor- or acceptor states, respectively. As an example of a localized donor impurity state, we may consider a phosphorous atom (with five valence electrons), which replaces a silicon atom in an otherwise perfect silicon crystal. The fifth electron of phosphorous has no valence band to occupy. The electron is still weakly bonded to the phosphorous nucleus, so it is not a free conduction band electron. The fifth electron has a localized orbital with energy ED in the band gap. The energy required to excite the electron from the impurity state into the conduction band is known as the donor ionization energy. The donor energy level should be close to the conduction band so the donor electrons can easily be thermally excited into the conduction band, increasing the conductivity of the semiconductor. The electron concentration in the conduction band and the Fermi level for an n-type semiconductor can be approximated according to equations 6 and 7, respectively.. nC = N D = N C e − ( EC − EF ) / kT §N E F = EC + kT ln¨¨ D © NC. (6). · ¸¸ ¹. (7). where ND is the effective donor density. From equation 7 and typical values for NC=1020 cm-3 and ND=1017 cm-3, the difference between EC and EF can be calculated to 0.2 eV. The hole concentration in the valence band and the Fermi level for a p-type semiconductor can be approximated in a similar way. Crystal surfaces and grain boundaries are two other examples of defects that cause energy levels in the band gap. Energy levels in the band gap are often 7.

(195) referred to as traps or recombination centres, depending on the electron lifetime in the state [10]. The band gap states can play an important role in the charge transport and recombination dynamics [17]. Depending on the energy distance from the conduction band, traps are divided into shallow and deep traps. The probability for an electron in a shallow trap to be thermally excited to the conduction band is relatively large whereas the probability is low for an electron in a deep trap.. 2.1.4 Carrier transport In a semiconductor, electrons in the conduction band and holes in the valence band (electron vacancies) are the charge carriers. In an n-type semiconductor, electrons dominate the carrier transport (majority carrier). In a p-type semiconductor, holes are the majority carrier. Drift Under the influence of an applied electric field, a randomly moving free electron would accelerate in a direction opposite to the field. Due to collisions with atoms an electron in a crystal lattice would not continue to accelerate for very long. The velocity of electrons between collisions caused by the electric field is called the drift velocity. The electron mobility, µn is defined as this electron drift velocity in an electric field of unit strength and is inversely proportional to the effective mass of the electron. The mobility is a material constant. Electron mobilities are typically in the range of 1001000 cm2V-1s-1 for semiconductors. The conductivity κ of electrons is proportional to the electron mobility and electron concentration in the conduction band. The conductivity κ of holes is proportional to the hole mobility (µp) and hole concentration in the valence band. Hole mobilities are typically in the range 1-1000 cm2V-1s-1 for semiconductors. For semiconductors with both electrons and holes as carriers, the conductivity is determined by. κ = qe (µ n nC + µ p pV ). (8). in which qe is the elementary charge. Diffusion Apart from motion by drift, carriers in semiconductors can also flow down a concentration gradient, i.e. by diffusion. Drift and diffusion processes are related. The diffusion coefficient of electrons, Dn and holes, Dp, can be. 8.

(196) calculated from the electron and hole mobilities, respectively, according to the Einstein equations. Dn =. kT µn qe. Dp =. kT µp qe. (9). From now on electrons are assumed to dominate the carrier transport. The variation of conductivity with temperature can be interpreted using the Arrhenius equation.. κ = Be ( − E κ / RT ) a. (10). where R is the gas constant, the parameter B is called a pre-exponential factor (related to the frequency of electron movement) and Eaκ is the activation energy (the minimum kinetic energy) for electron conduction. The fraction of electron movement with a kinetic energy in excess of energy Eaκ, hence leading to conduction, is given by the Boltzmann distribution (exponential part of Eq. 10). The variation of electron mobility with temperature can be interpreted in a similar treatment of the Arrhenius equation. The overall activation energy for electron conduction may be a sum of two terms. E aκ = E ag + E aµ C. (11). where EaµC and Eag are the energies associated with electron mobility in the conduction band and free electron generation, respectively. The mobility is usually not or only weakly activated in a semiconductor.. 2.2 Electrolytes In contrast to solid electronic conductors such as metals and semiconductors, the current in liquid electrolytes is carried by ions. The ions are formed by the dissociation of salts in polar solvents. In contrast to the situation with semiconductors, where there are also two types of charge carriers (electrons and holes) but one type usually dominates because of doping, both types of carriers (positively and negatively charged ions) are always present in equal concentrations in the electrolyte. The electrolyte in the dye-sensitized solar cell consists of a redox couple dissolved in a solvent. Redox couples are. 9.

(197) characterized by molecules or ions (in solution) which can be in a reduced state (Red) and an oxidized state (Ox) in electron transfer reactions.. 2.2.1 Energetic distribution of electrons Like the VB and CB in a semiconductor, the dissolved redox system has empty (Ox) and occupied (Red) energy levels. The position of the energy levels associated with the redox couple will determine the possibility of donating or accepting electrons when the redox molecules approach the semiconductor. The difference in energy of the reduced and oxidized state is due to the reorganization energy (solvent and redox molecule reorganization) i.e. the change in energy if the reduced state were to distort to the equilibrium configuration of the oxidized state. Due to fluctuation in the solvation shell surrounding the redox molecules, the energy states of the redox couple are distributed over a certain energy range and can be described by a Gaussian function, see Figure 2. The intersection of the oxidized and reduced distributions represents the solution redox energy, Eredox = qeUredox. This point represents a state where one half of the orbitals are occupied (f(E)=1/2), as was EF for the semiconductor. Energy Empty levels qeUOx qeUredox 0. qeU redox qeURed. Occupied levels Density of states Figure 2. Gaussian distributions of occupied and empty electronic states in a redox electrolyte. Solid lines indicate an electrolyte with {Red}={Ox} and the intersection of the two curves corresponds to the standard redox potential, U0redox. An electrolyte with a ratio of 9/1 for {Red}/{Ox} is indicated by dotted lines with the solution redox potential Uredox.. 10.

(198) The potential of a redox couple, Uredox can be measured in a cell against a reference electrode and can be calculated using the Nernst equation, 0 U redox = U redox +. {O x} RT ln zF {Red }. (12). where z is the number of electrons transferred in the redox reaction, F is the Faraday constant and {Ox} and {Red} are the activities of the oxidized and reduced species of the redox couple, respectively. Due to interactions between the species, an effective concentration, i.e. activity, is introduced. The potentials URed and UOx involved in redox reactions can be measured by cyclic voltammetry.. 2.2.2 Ionic carrier transport Drift The conductivity for one type of dissociated molecules is in principle the same equation as that used for electronic conductivity in solids.. κ = qeNAc(|z1|µ+ + |z2|µ-). (13). in which z1 and z2 are the charges of the ions, c is the concentration of ions, NA is the Avogadro constant and µ+ and µ- are the mobilities of the positively and negatively charged ions, respectively. The mobility is the velocity of the ion in an electric field of unit strength. The mobility of most ions is around 10-4 cm2 V-1s-1. These values are about 4-7 orders of magnitude smaller than the mobility of electrons and holes in a conventional semiconductor. In order to achieve electrolyte conductivities larger than 10-2 Ω-1cm-1 in an electrochemical cell, ion concentrations larger than 10-1 M are required. It is important that the solution is made sufficiently conductive by the addition of ions, which are not involved in the electrode reaction. This is usually called a supporting electrolyte. Diffusion In the absence of an applied electric field, ions and molecules in solutions can also move down a concentration gradient, by diffusion. If the rate of an electrode process becomes large then the concentration of the reacting species decreases and that of the generated species increases near the electrode, which leads to a concentration profile. The diffusion process is characterized by the ion diffusion coefficient, Di. Similarly to electrons and. 11.

(199) holes, the diffusion coefficient can be calculated from the mobility through Eq. 9. Since the mobilities of ions and molecules in solution are very low, the diffusion coefficients are also low. With µi ≈ 5×10-4 cm2 V-1s-1, one obtains Di ≈ 2×10-6 cm2s-1.. 2.3 Transition metal compounds The dye used in the DSC is a transition metal compound. Some common properties of transition metal compounds are their absorption of visible light and variable oxidation states. A transition metal compound consists of a central transition metal ion bonded to neutral and/or anionic ligands. The geometry of a compound depends on the number of ligand atoms (coordination number) bonded directly to the central metal ion. The octahedral shape is the most common geometry (six ligands). The bonding and the properties of a transition metal compound can be described by crystal field theory and molecular orbital theory. The crystal field model explains the fact that the properties of compounds result from the splitting of d-orbital energies, which arise from electrostatic interactions between metal ion and ligands. The model assumes that an ion compound forms as a result of electrostatic attractions between the metal cation and the negative charge of the ligands. π-bonding and the presence of other than crystal field based electronic transitions cannot be explained by the crystal field model and are best described by molecular orbital theory. In the molecular orbital theory the interaction between the metal and ligands is described by linear combinations of atomic orbitals and allows for charge transfer between the central metal atom and its ligands. The interaction between orbitals depends on their energy differences and their orbital overlaps. For transition metal compounds, the frontier orbitals involved in the interactions are the s, p and d-orbitals on the metal and orbitals of σ-type and π-type with respect to the metal-ligand axis on the ligand. The reduction and oxidation of a molecule involve an empty state and an occupied state, respectively and therefore these two processes occur at two different energies, determined by the electronic structure of the molecule. The energy levels of the highest occupied and lowest unoccupied molecular orbital is usually called HOMO and LUMO, respectively. The ground state redox potential of a transition metal compound can be determined in a similar way as for electrolytes, described in Section 2.2. The energy levels of a reduced state (ERed= qeURed), an oxidized state (EOx = qeUOx) and the redox 12.

(200) couple (Eredox = qeUredox) were shown in Figure 2. The redox energy level of the excited molecule (Eredox*) can be estimated by adding the excitation energy to the energy level of the redox couple in the ground state. The excitation energy is given by the electronic transition between the lowest lying vibrational levels in the ground state and the excited state. The absorption and redox properties of a dye complex can be changed by the use of different ligands [18]. In order to increase the absorption in the red wavelength region, the HOMO and LUMO levels can be tuned to higher and lower energy levels, respectively.. 2.4 Semiconductor electrolyte interfaces 2.4.1 Energy and electrochemical potential The Normal Hydrogen Electrode (NHE) is the reference scale in electrochemical measurements, defined as zero point at certain standard conditions [19]. It is useful to note that in semiconductor physics the vacuum level is used as zero. The relation between these energy and potential scales has been established according to [20-21]. EF = -4.7eV-qeUredox. (14). If two phases, with electroactive species, which have different electrochemical potentials are brought into contact, they tend to equilibrate by exchanging charges (Fig. 3). Electrons will flow from the material with the highest Fermi level until EF ≡ Eredox. Charged planes or layers are formed at the interface. They lead to electrical double layers, consisting of layers of positive charge, layers of negative charge, and regions of high electric field. Such double layers are dominant in controlling the electrical and chemical properties of the surface. Three double layers appearing at a semiconductor /liquid interface are the space charge double layer, a Helmholtz double layer and the Gouy-Chapman double layer. The space charge double layer is a charged region, region of “space charge” between the bulk and the surface of the semiconductor. The space charge can be in the form of immobile trapped carriers or immobile charged impurities near the surface of the semiconductor, or can be in the form of mobile electrons or holes in the conduction or valence bands of the semiconductor. If majority carriers are extracted in moderate amounts, the surface region is 13.

(201) assumed to be depleted of mobile carriers, i.e. the surface region is completely ionised. A depletion layer is formed and the space charge (QSC) is the immobile charge of ionised donors. The density of such a charge is usually assumed to be constant in the space charge region. The excess charge can be distributed over a considerable distance below the interface (xSC). The potential difference between the surface and the bulk of the semiconductor results in an electric field represented by the so-called band bending (Fig. 3a). The Helmholtz double layer is considered to be formed between two planes of charge. One plane is due to charges at the surface of the solid. The charge can arise in three forms: an accumulation of free charge, free charge trapped from the solid onto surface states, or adsorbed ions. The other plane is due to ions in solution attracted by the charged surface. Solvated ions can only approach the solid to a distance xOH and the locus of centres of these nearest solvated ions is called the outer Helmholtz plane. One layer of specifically adsorbed solution species at the solid surface is usually called the inner Helmholtz plane. The excess charge in the Helmholtz layer is QH. The possibility of storing charge in the Helmholtz planes is related to electrical capacitance. The Gouy-Capman theory describes a region in the solution near the electrode in which there is a space charge due to an excess of free ions of one sign. The ions attracted to the outer Helmholtz plane do not suffice to compensate all the charges on the electrode and the residual electric field results in a charged Gouy-Chapman double layer (diffusive layer). The thickness of the diffuse layer (xD) depends on the total ionic concentration in the solution. If the solution is concentrated, the layer will be extremely thin and can be considered to have melted into the outer Helmholtz plane. The excess charge in the diffuse layer is QD, so that the total excess charge on the electrolyte side of the double layer, QE, is given by QE =QH + QD = -QSC. (15). The interfacial region of an n-type semiconductor and an electrolyte, where EF > Eredox, is shown in Figure 3. If the Fermi level of the semiconductor lies above the redox energy of the electrolyte, conduction band electrons will be withdrawn from the semiconductor to the solution. If there is an excess of redox compounds in the solution compared to the number of electrons transferred from the semiconductor, the net change in the potential of the solution will be negligible. At equilibrium, the Fermi level of the 14.

(202) semiconductor will then shift to the position of the redox potential in the electrolyte, and thus the electrolyte plays an important role in these systems in determining the electrochemical equilibrium potential. Energy (eV) 0. Charge. Semiconductor QSC. Vacuum. Electrolyte. 0. Conduction band. QE. xOH qeUredox. EF. xIH Valence band Distance. xSC 0. xSC Distance. xD 0. (a). (b). Figure 3. (a) Energy level diagram for an n-type semiconductor in contact with an electrolyte and (b) a schematic representation for charge (Q) distribution at the semiconductor/electrolyte interface.. The total potential difference, ∆φSC for planar electrodes within the width of the space charge layer, is given by the Schottky equation.. ∆φ SC =. q e N D x SC 2ε r ε 0. 2. (16). ε0 is the vacuum permittivity and εr is the relative permittivity of the semiconductor. The width of the space charge layer is estimated to 2×10-7 m for a 0.5 V potential difference, an impurity density of 1017 cm-3 and a relative permittivity of 50 [22]. The potential difference in the space charge layer is the driving force for charge transport in the planar semiconductor electrode.. The situation is different in a small particle in contact with an electrolyte. A nanometer-sized particle (10-9 m) may be too small to develop the whole potential difference inside the particle so that EF ≡ Eredox. The maximum potential difference within such a small semiconductor particle with the radius r0 was derived [23]. 15.

(203) ∆φ SC =. q e N D r0 3ε r ε 0. 2. (17). For example, in order to obtain a 50 mV potential difference in a nanoparticle with r0 = 6 nm, a concentration of 1019 cm-3 of ionized donor impurities is necessary [17]. Undoped TiO2 particles have a much smaller carrier concentration and the band bending within the particles is therefore negligibly small.. 16.

(204) 3 The dye-sensitized solar cell. The dye-sensitization concept was invented in order to find a photoelectrochemical solar cell based on a semiconductor, which is stable against photocorrosion and yet absorbs light in the visible region. Many metal oxides fulfil the former requirement, however, most only absorb UVlight. A way to extend their spectral response is to adsorb dye molecules absorbing visible light on the semiconductor surface: dye-sensitization. A breakthrough by Grätzel and co-workers with sunlight to electricity conversion efficiency of 7.1% came with the use of nanostructured TiO2 electrodes [24]. By the use of a porous network of nanometer-sized crystals of TiO2 deposited on a conducting substrate (Fig. 4), the TiO2 surface can become 100 times larger than the macroscopic area, per micrometer film thickness. The enlarged surface area of TiO2 leads to increased dye adsorption (by the same factor) resulting in an increased harvesting of sunlight and an increased interface between the dye-sensitized semiconductor and the electrolyte. After a brief description of the design and the physical principles underlying the operation of a dye-sensitized nanostructured solar cell, the performance will be presented.. -. hv. +. e d c. eI3- ID+-. e. b a. Figure 4. Schematic drawing of a dye-sensitized solar cell. (a) transparent substrate coated with (b) transparent conducting oxide, (c) nanostructured dye-sensitized TiO2 film, (d) I-/I3- redox electrolyte and (e) counter electrode, consisting of conducting oxide with a small amount of Pt.. 17. hv.

(205) 3.1 Design The dye-sensitized solar cell consists of a working electrode, a counter electrode and an electrolyte. The DSC can be made from a variety of different materials. The most conventional one will be described here. A 510 µm thick nanostructured titanium dioxide film is deposited onto a transparent conducting substrate (back contact). The dye-sensitized TiO2 film forms the working electrode. A platinized conducting substrate constitutes the counter electrode of the solar cell (Fig. 4). Titanium dioxide is a transition-metal oxide (d0), with the valence band and the conduction band mainly having O2p and Ti3d character, respectively. With a band gap of 3.2 eV (anatase), TiO2 is transparent to visible light. The bonding type is intermediate between ionic and covalent. Point defects such as oxygen vacancies give rise to extra electrons filling band gap states. Titanium has several stable oxidation states, Ti4+, 3+, 2+, and the extra electrons and band gap states are interpreted as one or two electrons occupying a metal site, i.e. in the upper half of the band gap region, therefore reducing the initial Ti4+ state. Due to the oxygen deficiencies, TiO2 crystals are found to be n-doped. An effective electron mass value of about me*=20me, is found in the literature for TiO2 with oxygen deficiency. Such an effective mass leads to an electron mobility of about 0.1 cm2V1s-1 [11]. For an effective mass between 9me-20me the density of states (NC) within a few kT above the conduction band edge is 1×1020-1×1021 cm-3. In nanostructured TiO2, electronic contact between the nanoparticles is achieved in a sintering process. Because of the large amount of grain boundary and the large interface between the TiO2 surface, dye and the electrolyte in the solar cell, further defects in the crystal structure are expected. The position of the conduction band edge of nanostructured TiO2 has been determined to -3.8 eV (versus vacuum) with photoelectron spectroscopy [25]. The position of the valence band edge was calculated to -7 eV (versus vacuum) from the value of the conduction band edge and the anatase band gap of 3.2 eV. The dye bis(tetrabutylammonium)cis-bis(thiocyanato)bis(2,2’-bipyridine-4carboxylicacid,4’-carboxylate)ruthenium(II) (abbreviated N719) (Fig. 5) is adsorbed on the surface of the nanostructured TiO2 film by soaking the film in an ethanolic solution of N719. The bonding between the dye and the TiO2 should give high stability, dense packing and efficient charge injection. The dye N719 is so far one of the most efficient and most studied dyes of nanostructured solar cells.. 18.

(206) -. O TBA+. -. O TBA+. O. O. N. N. O. O Ru. N. N OH. HO N. N. C. C. S. S. Figure 5. The molecular structure of the dye N719. The tetrabutylammonium ion is denoted TBA+.. IR studies indicate a bridging or bidentate chelate coordination (Fig. 6) to the TiO2 surface via two carboxylate groups per dye molecule [26-27]. A monolayer is formed with about 1×102 dye molecules per TiO2 particle (r0= 6 nm). The dye exists in two redox states in the solar cell (Ru2+ and Ru3+). The valence level of neutral ruthenium (Ru4d) contains eight electrons (d8) and therefore Ru2+ and Ru3+ have six and five d-electrons, respectively. The redox potential for the dye (Ru2+/Ru3+) dissolved in ethanol is 1.1 V versus NHE [28], corresponding to the energy -5.6 eV versus vacuum. The highest occupied molecular orbitals are almost entirely of Ru and NCS character and the lowest unoccupied molecular orbitals are mainly of dicarboxybipyridine (dcbpy) ligand character [29]. The dye absorbs light between 300-800 nm. Two absorption maxima in the visible wavelength region, 400 nm and 550 nm (when adsorbed onto the TiO2 film) are due to metal to ligand charge transfer (MLCT) transitions, in which an electron, localized mainly on the Ru-NCS center in the initial state, is transferred to the orbital set on the dcbpy ligands [30]. R. R. O. O. Ti. Ti. O. O Ti. (a). (b). Figure 6. Two different adsorption geometries of a carboxylate group (a) bridge and (b) bidentate chelate on a titanium dioxide surface.. 19.

(207) An electrolyte is attracted into the inter-electrode space (~50% porosity) of the nanostructured dye-sensitized film by the capillary force. The electrolyte should have high conductivity and be able to penetrate the nanostructured system. These requirements are best matched by a liquid electrolyte. The redox system iodide/triiodide dissolved in an organic solvent is the basis of the electrolyte. Organic solvents are used instead of water-based ones, since the dye is unstable in water. The conductivity of the acetonitrile-based electrolytes is about 0.1 Ω-1m-1 [31]. The diffusion coefficient of I3- in nanostructured TiO2 has been determined to about 3×10-6 cm2s-1 [32]. The mixing of iodine and a salt of iodide forms the iodide/triiodide. The choice of counter ion of the iodide salt affects the performance of the solar cell and will be discussed in more detail in Section 3.3 and Chapters 4-5. The redox potential of an electrolyte based on 0.5 M LiI and 0.05 M I2 is about +0.3 V versus NHE, corresponding to the energy -5.0 eV versus vacuum. Characteristic electronic absorption bands for the I3- ion are at 360 nm and 290 nm [33]. Additives such as 4-tert-butylpyridine (4TBP), 1methylbenzimidazole (MBI) and carboxylic acids are usually added to the electrolyte in order to improve the performance. The bases, 4TBP and MBI, are rather bulky molecules.. 3.2 Operation principles When the different components of the dye-sensitized solar cell, which have different chemical potentials, are brought into contact, they will equilibrate by exchanging charges. At dark, the Fermi levels of the working electrode and the counter electrode are in equilibrium with that of the electrolyte redox couple. When the working electrode is illuminated under open-circuit condition, the Fermi level of the counter electrode and electrolyte is unchanged, whereas the Fermi level of the working electrode changes, producing the open-circuit voltage (VOC). If the energy levels of the separated components are used to describe an energy level diagram of an illuminated DSC, it will resemble Figure 7. Dye-sensitized solar cells differ from conventional devices in that they separate the function of light absorption from charge carrier transport. The initial charge separation occurs at the dye/TiO2 interface and the role of the semiconductor in the DSC is mainly to transport electrons from the semiconductor toward the back contact.. 20.

(208) -3.8 -5.0. 4. CB 3 e-. EF 9. -5.6. 2. Counter electrode. Electrolyte. Dye. Semiconductor. 0. Back-contact conducting glass. Energy vs vacuum (eV). D*. 7 hν 1 D/D+. 8 e5. ∆E. 6 e- E redox. I-/I3-. -7.0. VB e-. e-. Figure 7. Energy level diagram summarizing the possible processes in a dyesensitized solar cell.. The initial charge separation is driven by photoexcitation of the dye (1), followed by electron injection into the TiO2 conduction band (2), resulting in a positively charged dye (D+) and a negatively charged TiO2 particle (e-). The electron injection occurs through the carboxylate-titanium linkage [34]. The linkage constitutes a strong coupling to the Ti3d conduction band orbital. After injection the electron is transported through the network of nanostructured TiO2 (3) to the back contact (4) to perform electrical work in the outer circuit. The original state of the dye (D) is subsequently restored by electron donation usually from iodide in the electrolyte (5). The current in the electrolyte is carried by the redox couple to the counter electrode, at which the iodide is regenerated, in turn, by reduction of triiodide (6), the circuit being completed. The counter electrode should be efficient in transferring electrons to the electrolyte. The small amount of platinum on the counter electrode improves the short-circuit current through a catalytic effect of the reduction of the triiodide. From the energy levels of the excited state of the dye (D*), the TiO2 conduction band edge, the redox couple of the electrolyte (I-/I3-) and the redox couple of the dye (D/D+), it is realized that 21.

(209) there is a “down hill path” for the electron in the complete circuit, i.e. the reaction path is thermodynamically possible. Overall there is no net chemical change in the system, thus optical energy is converted into electrical energy. The charge separation in the DSC is thought not to depend on a built-in electrical field, but mainly relies on competition between forward and backward electron transfer kinetics at the semiconductor/dye/electrolyte interface [17]. The electron injection from the excited dye into the conduction band of TiO2 (2) is in the femtosecond time regime [35-36]. The regeneration of the oxidized dye (D+) by the electrolyte redox couple (5) occurs in the nanosecond time regime [37] and competes with back-electron transfer from conduction band electrons (9) occurring in the nanosecond to millisecond time regime [39]. The back electron transfer from the conduction band to the oxidized dye was shown to depend strongly on applied potential [37], light intensity and the electrolyte [40]. The backelectron transfer between conduction band electrons and triiodide (8) has been found to occur in the millisecond time regime at 1 sun [38].. 3.3 Performance The solar cell performance is determined by its overall energy conversion efficiency (η) and stability. The solar cell stability is usually determined from the measurement of the overall energy conversion efficiency over time.. 3.3.1 Efficiency When an external load is connected to the illuminated solar cell, the total current (I) is the result of two counteracting components. I = IP – ID. (18). where IP is the photogenerated current and ID is the light independent recombination or “dark” current. The highest total current value ISC is obtained under short-circuit conditions, ∆E=V=0 as seen from the currentvoltage (I-V) curve under illumination in Figure 8. The open-circuit voltage (VOC) corresponds to the energy difference between the Fermi level of the semiconductor and the energy level of the electrolyte redox couple. The maximum power which can be delivered, Pmax= Imax × Vmax, is indicated by the area of the rectangle in Figure 8. From the maximum power point, the fill. 22.

(210) factor (FF) can be calculated, which is a quantitative measure of the device quality defined by the squareness of the I-V curve. FF =. I maxVmax I SCVOC. (19). The energy conversion efficiency of a solar cell is defined as. η=. I maxVmax I SCVOC FF = Pin Pin. (20). where Pin is the total radiation power incident on the solar cell. I ISC Imax. Vmax. VOC. V. Figure 8. I-V curve of a solar cell under illumination, displaying the current and voltage Imax and Vmax, respectively, at maximum power.. Photocurrent-voltage characteristics of 20.5 mA/cm2 (ISC), 0.72 V (VOC), 10.4% efficiency and a fill factor of 0.7 are already attained for small dyesensitized solar cells with a ruthenium-based dye [41]. The use of the additives 4TBP and MBI in the electrolyte, has improved the voltage output from the solar cell. An increase of the Fermi energy level by a decreased back-electron transfer from the TiO2 surface by the addition of 4TBP is one model discussed in the literature [28]. The additive would then block the TiO2 surface. Other mechanisms being discussed are a conduction band edge shift to higher energies by the adsorption of 4TBP on the TiO2 surface [42] or deprotonation (by 4TBP) of the TiO2 surface which is partially protonated during adsorption of the acidic dye [38, 42]. A lower current is usually observed for solar cells with the additives 4TBP or MBI in 23.

(211) the whole spectral range, with a specific decrease in the red region. The current decrease in the red region can be understood from a lower number of injected electrons in the red part of the spectrum, due to less energetic overlap between the dye and the conduction band edge. A decreased energetic overlap is attributed to a conduction band edge shift of the nanostructured TiO2 [42]. The observed photovoltage increase and photocurrent decrease with increased cation radius was also explained by a conduction band shift mechanism [43]. The finding of the complex formation between Li+ and MBI (Paper IV) supports this mechanism. In this mechanism it is the cation of the electrolyte salt, and possibly some effect by protons from the adsorbed dye, which influences the shift of the conduction band edge [44]. Small ions such as Li+ and H+ which have high affinity for the TiO2 surface would shift the conduction band edge to lower energies compared to larger cations having less affinity, such as Cs+, (Li-MBI)+, TBA+ (tetrabutylammonium ion) and hexylmethylimidazolium cations. Thus, the role here for additives such as 4TBP and MBI would be to decrease the affinity of Li+ for the TiO2 surface.. 3.3.2 Stability A commercial crystalline silicon solar cell has today a lifetime of at least 20 years. Regarding the stability of DSC there are many possible routes of degradation. The stability of the components (conducting substrate, the nanostructured TiO2, dye, electrolyte, additive and counter electrode) and how they affect the durability of the entire system are discussed in a review article [45]. Considerable efforts have been made to realize devices with high efficiencies that meet the stability criteria for outdoor use. In this context, new counter electrode materials, alternative redox electrolyte couples and dyes have been screened. The stability of a low-efficiency (2%) dye-sensitized solar cell up to 8300 hours, as seen from I-V curves, has been demonstrated under visible light soaking with 2.5 sun equivalent intensity at 20°C, corresponding to 10 years outdoor illumination [46]. Very recently, stable performance was shown for a quasi-solid state DSC under both thermal stress and soaking with light, matching the durability criteria applied to silicon solar cells for outdoor applications [47]. This is an important result since one of the factors that have hampered widespread practical use of the DSC is the problem with thermostability of these devices. Two main components were altered compared to the conventional DSC. The dye N719 was modified by replacing one of the 4,4’ dicarboxylic acid-2,2’-bipyridine ligands by the hydrophobic ligand 4,4’-dinonyl-2,2’-bipyridine (Z907). A quasi-solid state 24.

(212) gel electrolyte, consisting of 3MPN mixed with a photochemically stable fluorine polymer (polyvinylidenefluoride-co-hexafluoropropylene), was used in order to decrease solvent leakage under thermal stress. To be competitive with other photovoltaic technologies it is still a challenge to improve the efficiency of this quasi-solid state DSC from 6% to about 10%. In Paper II, we focused on possible degradation mechanisms of the dye N719 with the components presented in Section 3.1. In order to study factors and mechanisms of degradation, the dye adsorbed on TiO2 was deliberately treated harshly enough to cause chemical degradation and was studied with UV-VIS and IR spectroscopy. All treatments and measurements were made under simulated open-circuit conditions, which have been observed to be the most severe conditions for a dye-sensitized solar cell [37]. At open-circuit all photoinjected electrons recombine by one mechanism or another [48]. It was found that visible light soaking alone is not a dominant stress factor but is a stress factor in combination with air. Other stress factors for the solar cell are increased temperature, UV-light, water, and an incorrect composition of electrolyte. Factors of instability of solar cell devices will be related to the factors and mechanisms of degradation studied in Paper II. Temperature Instability observed at elevated temperatures is strongly correlated to the chemical composition of the electrolyte [46-47] and possibly to the dye. As a guideline for the required thermal stress, 80°C for 1000 hours should at least be targeted for mid-European locations [46]. The quasi-solid state DSC with the dye Z907 sustained this treatment, maintaining 94% of its initial performance compared to 65% with the dye N719 [47]. The device using the liquid electrolyte with the dye Z907 retained only 88% of its initial performance. The quasi-solid state DSC with the dye Z907 also maintained 94% of its initial performance under light soaking at 55°C for 1000 hours in a solar simulator (1000 W/m2) equipped with an ultra-violet (UV) blocking filter. Thus a stabilized effect by both the hydrophobic dye and the polymer gel electrolyte is suggested [47]. As presented earlier, the efficiency of solar cells with the dye N719 and organic or gelled organic electrolytes decreases immediately due to solvent evaporation. Solar cells with molten salt electrolytes keep 70% of their efficiency after 1000 hours in 85°C whereas solar cells with gelled molten salt do not decrease at all [49]. Based on the high stability of the quasi-solid state DSC with the dye Z907 as compared to N719, the latter dye is suggested to desorb at high temperatures [47]. No clear evidence for desorption of the dye was observed with either IR or UVVIS spectroscopy at the temperatures 100°C to 200°C (Paper II). Degradation of the dye by SCN- ligand loss was, however, clearly observed. 25.

(213) from exposure of the nanostructured dye-sensitized TiO2 film to 135°C for 24 hours. UV-light Modules with UV-blocking filter show higher stability than those without a filter [50]. The TiO2, the dye and the redox couple I-/I3- absorb UV-light and reactive species may be photogenerated in these processes. With electrolyte additives such as MgI2 and CaI2, a dramatic increase of UV stability has been shown [46]. Weak degradation of the SCN- ligands from UV radiation is possible, as was observed from UV-VIS and IR spectroscopy of the dyesensitized TiO2 film (Paper II). Water Illuminated solar cells containing water show instability, the higher the water content the lower the stability [50]. Together with water in the electrolyte (5v%) in the presence of air, the SCN- ligand is probably exchanged with H2O or OH- and this exchange is accelerated under visual radiation (Paper II). This was rationalized from UV-VIS and IR spectroscopy of electrodes, i.e dye-sensitized TiO2 films (Fig. 9). IR. 0,6 0,4. 1470. Absorbance (a.u.). 1434 1384 1470. 1543. 0,0. 1604. 0,2. 1575. 0,4. 550 0.5 1.5. B. 840h Dark TiO2 N719 (50mM I2 500mM LiI 5v% H2O in 3MPN). 1907. 0,6. 1.0. A. TiO2 N719. 2133 2103. Absorbance (a.u.). 0,0. A. TiO2 N719. 1436. 1906. 0,2. 1545. SCN 2100. -. COO 1380. B. 840h Dark TiO2 N719 (50mM I2 500mM LiI 5v% H2O in 3MPN). 1.0 529 0.5 1.5. C. 840h VIS TiO2 N719 (50mM I2 500mM LiI 5v% H2O in 3MPN) 1903. 0,2. 1613 1543 1474 1427 1388. 0,4. COO COOH 1610 1717. UV-VIS. 1.5. -. 0,6. 500. 1.0. 0.5. 0,0. C. 840h VIS TiO2 N719 (50mM I2 500mM LiI 5v% H2O in 3MPN). 350. 2200. 2000 1800 1600 -1 Wavenumber (cm ). 1400. 400. 450. 500. 550. 600. 650. 700. Wavelength (nm). Figure 9. IR and UV-VIS spectra of (A) non-treated working electrodes in argon atmosphere as references and working electrodes immersed in electrolyte with 5v% Milli-Q water for 840 hours (in air), (B) in darkness and (C) exposed to visual light. The IR spectra of the treated samples were normalised to the titanium dioxide absorption maximum at 832 cm-1 of the reference sample.. 26.

(214) A broadening and decreased intensity from the SCN- ligand absorption around 2100 cm-1 was observed as compared to a reference sample. At 1575 cm-1 a small absorption peak was found, possibly from symmetric bending of H2O, bound to the dye or adsorbed in the nanostructured TiO2 film. The UVVIS spectra of these films showed blue-shifted absorption maxima; 529 and 500 nm for the electrodes in darkness and exposed to visual light, respectively, compared to 551 nm for a reference. The dye cis-bis(aquo) bis(2,2’-bipyridine-4,4’-dicarboxylate) ruthenium(II) complex in ethanolic solution has one of its absorption maxima at 500 nm [28]. A red shift of the absorption band from 500 nm to 528 nm was observed upon going from neutral to basic solution (pH 12) and is assigned to deprotonation of the H2O ligand [51]. Based on the above results the ligand exchange between SCNand H2O/OH- seems to be a likely way of degradation. Surprisingly, even though 5v% water is present in the electrolyte, the solar cells have shown to be very stable after an equilibrium period, as long as the additive 4TBP is present [52]. Air The dye also degrades in air. The dye adsorbed onto TiO2, exposed to air either in darkness or exposed to visual light for 840 hours lost thiocyanate ion ligands, initially coordinated to the dye, as observed from the diminishing CN stretching band at 2100 cm-1 with IR spectroscopy (Fig. 10). For the electrode exposed to air in darkness, a small absorption peak at 1578 cm-1 is found, possibly from a symmetric bending band of H2O either bound to the dye or adsorbed in the nanostructured TiO2 film. These electrodes also showed blue-shifted absorbance maxima with UV-VIS spectroscopy (Fig. 10). For the electrode exposed to air in darkness, the absorption maxima were 380 nm and 530 nm compared to 400 nm and 550 nm for a newly prepared electrode as the reference. After illumination with visual light in air, both absorption maxima were suppressed and rather featureless. Degradation of the dye in air is probably due to the oxygen and/or water content in air. The resemblance between the IR and UV-VIS spectra of samples exposed to air and to an electrolyte with 5v% water, indicates that some degradation occurs by the exchange of SCN- to the H2O/OH- ligand.. 27.

(215) IR. 0.5 0.4 0.3. COO 1381. -. COO 1610. SCN 2100 COOH 1738. A. TiO2 N719 1.5 400. 0.5. 1382. Absorbance (a.u.). 1578. 1903. 0.2 0.1. 1438. 1788 1730. 0.3. 1607 1546 1470. 2.0. B. 840h Dark TiO2 N719 air. 2100. Absorbance (a.u.). 0.0 0.6 0.5 0.4. 0.0 0.6. B. 840h Dark TiO2 N719 air 1.5 530. 1.0 0.5. 2073. 0.3 0.2 0.1 0.0. C. 840h VIS TiO2 N719 air. 1388. 1633 1550 1470. 2.0. C. 840h VIS TiO2 N719 air 1902. 0.5 0.4. 550. 1.0. 1435. 0.2 0.1. UV-VIS. 2.0. -. A. TiO2 N719. 1543 1470. 0.6. 1.5 1.0 0.5. 2200. 2000. 1800. 1600. 350. 1400. 400. 450. 500. 550. 600. 650. 700. Wavelength (nm). -1. Wavenumber (cm ). Figure 10. IR and UV-VIS spectra of (A) non-treated working electrodes in argon atmosphere as references, and working electrodes exposed to air for 840 hours (B) in darkness and (C) exposed to visual light. The IR spectra of the samples exposed to air were normalised to the titanium dioxide absorption maximum at 832 cm-1 of the reference sample.. Photodegradation of the dye N719 may occur according to a mechanism proposed for cyanidin adsorbed onto nanostructured TiO2 by Tennakone et al. [53]. Photoexcited dye molecules (D*) inject electrons into the conduction band of TiO2, forming the dye cation D+. Electrons from the conduction band are readily accepted by oxygen to form the superoxide ion (O2-). Dye degradation is the result of a reaction between D+ and O2-. The presence of water vapour enhances the rate of degradation of the dye due to a reaction between D+ and OH- ions. This mechanism does not explain degradation of the dye (D) in darkness, in the presence of air and water, observed in the present work. This mechanism could, however, still explain the degradation in darkness, if long-lived electrons accumulated in the TiO2 from before the dark treatment exist and are responsible for the degradation. If degradation effects from such long-lived electrons are negligible, one can suspect that the dye also degrades from the ground state (D).. 28.

References

Related documents

The practice of using GLD is still rather novel and previous studies have focused on the chemical, physical and mineralogical characterization of GLD

Nc Effective density of conduction band electrons ε Molar extinction coefficient η Solar cell efficiency ηcc Charge collection efficiency ηreg Regeneration efficiency λ Inelastic

Architecture of the Dyes The architecture of the dye affects the electron transfer kinetics in the solar cell, as well as the interaction of the dye-sensitized working electrode

Table 3. This section asks general.. questions regarding the students view on EM as a way for them to find potential employers. The first question regarded what the students

Until  now  genetic  operations  haven’t  been  mentioned  and  without  them  there  will  be  no  change  of  the  individuals  but  only  a  convergence  of 

Design of organic dyes and cobalt polypyridine redox mediators for high-efficiency dye-sensitized solar cells.. High-efficiency dye-sensitized solar cells with molecular

Photo-induced absorption spectroscopy (PIA) will be described in detail in Chapter 5.1.3. In this section, the influence of the Stark effect on photo-induced absorption spectra will

Instead of measuring single particles, we measured particle populations with simultaneous measurements of mass concentration and number concentration, as other authors have done