III. Configurations of solar cells
The efficiencies of the different configurations of solar cells as well as their evolution are
represented in the figure below. The types of solar cells are classified in generations, in the
chronological order of their development.
Generations of solar cells
The first generation of solar cells comprises p-n junctions in crystalline Si, which have typical
conversion efficiencies of 15−20%. The maximum efficiency of solar cells from crystalline Si
reaches 24%; the conversion efficiency is higher in materials with higher purity. The Si
wafers for solar cells are similar to those in computer chips, but are larger and cheaper. In
addition, Si is nontoxic and abundant, being the second element as availability, after oxygen.
It can be found especially in the form of sand, which is in fact silicon dioxide SiO2; in sand,
Si remains after the oxygen is eliminated at high temperatures. Si solar cells benefit from the
technological advancements in the microelectronic industry. Although the first to appear, the
solar cells based on crystalline Si hold still 73% from the market (in 2009). Their advantages
are the wide absorption spectrum and the high mobilities of charge carriers. The
disadvantages include the high fabrication cost of large-area crystalline Si wafers, the high
recombination rates, which indicate the ease with which photogenerated electrons and holes
recombine, and the fact that a significant part of the energy of high-energy photons (with
ultraviolet and blue wavelengths) is lost as heat.
The necessity to develop other types of solar cells was determined mainly by the high
cost of high-purity Si. In Si-based solar cells, 70% of the production cost refers to materials.
The second generation of solar cells includes thin-film solar cells (based on Si, CdTe,
CuInGaSe2 (CIGS), polymers, etc.) and the Grätzel cells based on the sensitized absorption of
nanocrystalline TiO2. A comparison between the efficiencies and costs of second-generation
solar cells and those of crystalline Si (first generation) is given in the table above. As can be
seen from this table, not only the costs but also the efficiency of second-generation solar cells
are lower than for solar cells based on crystalline Si. In the table above LCOE is the acronym
for levelized cost of energy, a parameter that allows the comparison of total costs of
electricity produced by different methods. The advantage of thin-film solar cells is that they
can be fabricated on all substrates, including flexible substrates and textiles (see the figures
below).
In 2009, the market coverage of thin-film solar cells (from CdTe, CIS or CIGS (copper
indium gallium selenide), DSC (dye-sensitized), organic/polymer materials, Si) was of 7%.
The evolution of their production capacity is represented in the figures below. The production
capacity of CdTe and CIGS solar cells is limited by the materials used for the photovoltaic
cells, while that of dye-sensitized solar cells is limited by the materials used for electrodes.
The total power generated through the photovoltaic effect in different states in the
European Union is given in the table below, followed by two images of the solar energy
production facilities in Lucainena de las Torres, in the region of Andalusia, Spain, which
produces an average power of 23.2 MW (above), and, respectively, in Brandis, in the region
of Saxony, Germany, which produces 40 MW based on CdTe thin films (below). The largest
solar energy production facility at present (December 2008) is of 60 MW, and is located in
Spain, in Olmedilla de Alarcon, in the region of Castilia.
Material-related costs prevail also for thin-film and organic solar cells. Therefore,
solar cells with 2−3 times higher efficiencies are needed. These cells, based on new concepts
and currently studied only in laboratories, form the third generation of solar cells. Among
these we mention the hot-carrier cells (which collect the non-thermalized charge carriers
before interacting with phonons and have a conversion efficiency of η = 85.4%, the maximum
theoretical efficiency being 86.8%), the photovoltaic elements in which several (more than
one) charge carrier pairs are generated per absorbed photon (with an internal quantum
efficiency > 1 due to impact ionizations and conversion efficiency of solar energy as high as
85.9%), and solar cells with multiple energy bands (of tandem or intermediate band types,
which split the solar spectrum and convert it in electricity using several cells with different
energy bands, the conversion efficiency attaining 86.8%).
An alternative at using several layers that convert the solar energy is to alter the solar
spectrum before it reaches the cell. This can be accomplished in two ways: 1) downconversion, in which a photon of high energy (in the ultraviolet spectral region) is converted
in at least two photons with lower energies that can be absorbed by the cell, and 2) upconversion, in which two (or more) photons with lower energy generate a single photon with
higher energy that can be absorbed.
The classification of solar cells in generations is not unique. If based on a criterion
related to the separation of charge carriers instead of the efficiency/cost parameters, the first
generation is still formed from crystalline Si solar cells, the second generation encompasses
thin-film cells, and the third generation includes solar cells that do not require p-n junctions
(or p-i-n structures and heterostructures), i.e. that do not require a built-in electric field, to
separate the charge carriers. The third generation is formed from solar cells based on
nanocrystals, polymers and dye-sensitized cells in which the separation of charge carriers is
achieved by diffusion. According to this criterion, one can identify a fourth generation, which
includes hybrid solar cells, fabricated from inorganic crystals in a polymer matrix. In the
following we study briefly all above-mentioned types of solar cells with the exception of
those based on p-n junctions, which do not require additional considerations.
Solar cells from amorphous and poly- or microcrystalline Si
Depending on the fabrication procedures and parameters, Si can be obtained in three forms:
crystalline, polycrystalline and amorphous (see the figure below). The efficiency of Si solar
cells is related to the quality of the material: η = 20% for monocrystalline Si solar cells
(which is an expensive material), η = 10% for amorphous Si cells (relatively cheap), and η =
33% for graded solar cells, in which the crystallinity of the material varies continuously from
amorphous to crystalline (this material is, however, very expensive).
The amorphous Si, consisting of randomly positioned Si atoms, has an energy gap of
E g = 1.7−1.9 eV and a direct band structure (in crystalline Si, E g = 1.12 eV and the band
structure is indirect, i.e. the conduction band minimum and the valence band maximum are
positioned at different k values in the reciprocal plane – see the first part of the course).
The advantages of solar cells from amorphous Si are that they can have large areas and
can be n or p doped, as in the case of crystalline Si. In addition, the temperature coefficient of
amorphous Si is smaller than that of crystalline Si, which implies that the solar cells from the
amorphous material are not overheated when illuminated, so that the decrease of the
conversion efficiency due to thermal effects is less important.
Moreover, amorphous Si has a higher absorption coefficient than crystalline Si for
blue light (the region of high energies in the figure below), which is present in diffuse
illumination. As a consequence, the conversion efficiency at weak/diffuse illumination is
better than for crystalline Si. Note that, unlike in a crystal, in an amorphous material the total
momentum k is not conserved at photon absorption (the energy is still conserved); the concept
of (quasi)momentum has no sense in amorphous materials, being only defined in crystalline
solids.
The main disadvantage of amorphous Si is that it is unstable. More precisely, in
contact with sunlight the density of defects increases so that the lifetime of the excess charge
carriers (those contributing to the photocurrent) decreases. Therefore, the performances of
solar cells from amorphous Si depend on the encapsulation and the type of the illuminated
electrode, i.e. the type of the transparent conducting oxide (TCO) used. The conversion
efficiency can decrease with up to 15% in several months if the encapsulation is not proper.
Because the carrier mobilities in amorphous Si is small (< 20 cm2/Vs), it is not
possible to efficiently collect the carriers through diffusion. Therefore, the solar cells based on
this material are p-i-n structures. In addition, the doped p layer is on top and the doped n layer
is at the bottom (see the figure above) because the generation of charge carriers occurs near
the exposed surface and the hole mobility is smaller than the electron mobility. The p layer
must be thin to avoid recombination. The width of the i layer (see the discussion in the second
part of the course) depends on the product between mobility and lifetime, μτ . If this layer is
too thick, there are losses at charge collection since the holes generated far away from the
surface recombine. On the contrary, if the i layer is too thin, the absorption is not enough. The
optimum thickness of this layer in amorphous Si is 300 nm. The typical efficiency of solar
cell consisting of a p-i-n structure in amorphous Si is of 4−6%. In many cases the amorphous
Si is hydrogenated, the resulting semiconductor material having a conductivity that can be
varied through doping.
The polycrystalline Si consists of grains/crystallites of Si with dimensions between 1
μm and 1 mm. The polycrystalline Si has no amorphous phase, has an energy gap similar to
that of the crystalline material, E g = 1.1 eV, and the mobility of charge carriers is with orders
of magnitude higher than in amorphous Si. For this reason, the solar cells from polycrystalline
Si have a p-n junction configuration. Because the diffusion length is, however, low (of several
μm), the solar cells must be thin (between 1.5−3 μm), which requires solutions for an efficient
trapping of light. The polycrystalline Si is harder and cheaper than crystalline Si, but the solar
cells fabricated from this material have lower conversion efficiencies, with typical values
10−14%, because the interfaces between grains hinder the electron transport and thus decrease
the output power.
The microcrystalline Si is a mixed phase, which contains a crystalline fraction and an
amorphous fraction. The diameter of crystallites varies between several nm to tens of nm,
these crystallites appearing as conglomerates with dimensions of 1 μm. The energy gap E g
depends on the amorphous fraction. If the crystalline phase is over 70%, E g = 1.1 eV. The
absorption of this material is higher than in crystalline Si because the light is scattered at the
surface of the crystallites and because photons with energy smaller than E g can be absorbed
due to defects. The degradation of microcrystalline Si as result of sunlight exposure is
between that of amorphous Si and crystalline Si, the latter being actually stable. In particular,
there is no sunlight degradation if the high-energy photons are filtered. The solar cells from
microcrystalline Si have a p-i-n configuration, the optimum width of the i layer being of 1−2
μm (the absorption coefficient is lower than in amorphous Si, and the μτ product is larger).
Structured TCO electrodes are used to increase the efficiency. These are obtained either as a
result of the natural morphology induced in the deposition process or in a subsequent etching
process. The figure above presents a comparison of the spectral region of absorbed photons in
solar cells from amorphous and microcrystalline Si, as well as the obtained photocurrent.
Recently, however, very thin solar cells from monocrystalline Si have been obtained
(with a thickness of 100 nm, limited by the depth of the junction) with a small surface (of
several μm) deposited on a flexible substrate (see the figure below, left). An array of such
micro-cells is represented in the figure below, center, while an array of lenses that concentrate
light on this structure is illustrated in the figure below, left. The advantage of these cells is the
optimum balance between optical absorption and the separation/collection efficiency of
charge carriers, as well as a cost reduction (with respect to large-area crystalline Si wafers)
due to the minimal quantity of material needed and the relaxed purity requirements. The
conversion efficiency in this case reaches 6−7%, the solar cells having FF = 0.6, j sc = 20
mA/cm2 and Voc = 0.48 V.
Thin-film solar cells
By definition, a thin film is a material created through random nucleation and growth
processes of molecules that condense individually on a substrate. The structural and physical
properties of thin films depend on the deposition processes and thickness, which can vary
between several nm to tens of μm. the advantages of thin films are
1) There are many deposition techniques for thin films from the same material
2) The structure of the films can vary from amorphous to micro/nanocrystalline,
depending on the substrate and the technique and parameters of deposition
3) Can be deposited on many substrates of different forms and surfaces
4) Doping and alloys/compounds in different proportions can be obtained due to
relaxation of the solubility conditions
5) The grain limits and the surfaces can be passivated with suitably chosen materials
6) Several types of electronic junctions can be fabricated: single junction, in tandem, etc
7) It is possible to fabricate materials with desired and/or graded energy gaps/
compositions/lattice constants/reflection coefficients
8) The surfaces and interfaces can be modified to obtain suitable diffusion barriers at
interfaces/electric fields
The thin-film solar cells are grown on a substrate, unlike those on wafer (from
crystalline Si), in which the solar cell is the substrate itself. The thin-film solar cells have
different configurations: p-n junctions, heterojunctions or p-i-n structures, depending on the
material parameters. When heterostructures between materials with different bandgaps are
used (see the table below), their lattice constants must match in order to avoid the
dislocations/defects that would otherwise appear at the interface. These dislocations/defects
act as recombination centers and decrease the performances of solar cells.
Examples of energy bandgaps and lattice constants of semiconductor materials used in the
fabrication of solar cells are presented in the figure below.
Problem: Using the figure below, chose pairs of materials suitable for the fabrication of solar
cells with heterojunctions, and specify in each case which is the illuminated material (the
window).
The thin films have, generally, higher absorption coefficients than the crystals (due to
light scattering in the polycrystalline materials), so that the thickness of solar cells can be
decreased up to 0.3 μm. But, the conversion efficiency is smaller than in crystalline materials,
such that solar cells with larger areas are needed. Thin films from amorphous Si, CdTe,
CuInSe2 (CIS) and other materials have been successfully fabricated. In particular, CIS has
the highest conversion efficiency of all thin films: 17% (in laboratory conditions), and is not
degraded by sunlight, having also a high absorption coefficient: a 0.5-μm-thick layer absorbs
90% of solar spectrum. The disadvantage is that it is difficult to fabricate and that very toxic
materials, such as hydrogen selenide, are used in its fabrication.
A solar cell with a heterojunction between CdS and the compound Cu(Ga,In)(Se,S)2 is
represented in the figure below, left, while the right side illustrates a solar cell in the
superstrate configuration (in which the light enters through the transparent glass supporting
substrate) with a heterojunction between CdS and CdTe. In the first case the substrate needs
not be transparent, while in the second case the junction is graded, in order to minimize the
dislocations/defects at the interface due to the different lattice constants of CdS and CdTe.
In solar cells with variable composition, the energy bandgap changes also. For
example, in the compound Cu(Ga,In)Se2 (or simply CIGS) with a zinc blende-like crystalline
structure, the energy bandgap varies between 1.02 eV, for CuInSe2 (or CIS), and 1.68 eV, for
CuGaSe2. Similarly, in AlGaAs, the energy bandgap varies between 2.16 eV, for AlAs, and
1.42 eV, for GaAs, its dependence in x in the compound AlxGa1-xAs being linear up to x =
0.45: E g (eV) = 1.42 + 1.25 x , and parabolic for larger x values.
The solar cells based on CdTe, which is a polycrystalline semiconductor compound
with a high absorption coefficient (a 1 μm thick layer can absorb 90% of the solar spectrum),
are quite cheap, have a conversion efficiency of 7%, but their performances are unstable. In
addition, Cd is toxic. The solar cells with CdTe/CdS heterojunctions are cheaper than those
from crystalline Si, but have a lower efficiency. Such (polycrystalline) thin film solar cells are
usually deposited on a glass substrate, with a typical thickness of 1 mm, but can also be
grown on a flexible substrate with a thickness of 5−10 μm. In the last case the solar cell is 5
μm thick. The CdTe/CdS solar cell is stable at irradiation with protons and electrons, being
suitable for cosmic applications. CdTe has an optimum E g , of 1.5 eV, and a high absorption
coefficient, being used as absorbing layer with a typical thickness of 3−5 μm, the thickness of
the CdS layer, which plays the role of window, being of 0.5−1 μm. On glass substrate, the
CdTe/CdS solar cells have conversion efficiencies higher than 16%, but on flexible polymer
substrates their efficiency drops to 7.3%, whereas on flexible metallic sheets the efficiency is
even lower, of 3.5−8%, since it is difficult to form ohmic contacts. The influence of the
transparent electrode can be seen from the following example: the CdTe/CdS solar cells
grown on polyimide with an ITO contact have 11% efficiency ( Voc = 842 mV, FF = 70.9%,
j sc = 18.5 mA/cm2), the efficiency decreasing up to 8.6% for a top contact from ZnO:Al.
The solar cells from III-V semiconductor compounds (for example GaAs, InGaP) on
Ge substrates have high efficiencies (higher than 20%) and are used as power sources on
satellites. Their widespread use is, however, hampered by the higher costs compared to Si
solar cells, which have comparable efficiencies (higher than 15%). In the second part of the
course we have seen that GaAs is the material with an ideal energy gap (theoretically, with a
maximum conversion efficiency). GaAs has a crystalline structure similar to Si, but a higher
absorption coefficient. The same solar intensity is absorbed in a few μm in GaAs, compared
to 200−300 μm in crystalline Si. GaAs solar cells have higher conversion efficiencies than
those from Si, of 25−30%, and are less sensitive to heat (being suited in cells with
concentrated light) and irradiation (being used in space). Because of their high costs, these
solar cells are used only with concentrators, where the required surface is smaller. The
performances of thin-film solar cells from inorganic materials and of the modules fabricated
from them are summarized in the two tables below (the first and second, respectively).
Solar cells from organic materials
The organic materials differ from inorganic materials from the point of view of solar cell
applications through two characteristics: 1) the energy spectrum of electrons is discrete (does
not have allowed and forbidden bands), and 2) the electrons and holes generated after photon
absorption are not longer free but linked through electrostatic (Coulomb) interactions. These
bound electron-hole pairs are called excitons.
Molecular orbitals
Unlike crystals, which are periodic arrangements of very many atoms, an organic molecule
consists from a relatively small number of atoms that form a chemical bond by sharing a pair
of electrons that are weakly attracted by the nucleus (that are situated on external orbits). In an
isolated atom the electron has discrete states, called atomic orbitals, which are characterized
by a set of quantum numbers. In a chemical bond the atomic orbitals in adjacent atoms
interact and the electrons are strongly localized near the molecule. This situation is opposite
to that in crystalline materials, where conduction electrons move freely inside the crystal.
The formation of a chemical bond through the hybridization of s and p atomic orbitals
and the creation of a highly directional sp molecular orbital is presented in the figure above.
When one s orbital and two p orbitals hybridize, an sp2 molecular orbital is formed, the
resulting molecule having strong planar σ bonding with the atoms placed at 120° (see the
figure below).
The third p orbital is perpendicular on the plane of the sp2 orbitals, and the p orbitals
belonging to adjacent atoms form a π bond, in which the electrons are less localized compared
to the σ electrons (see the figure above).
When three p atomic orbitals hybridize with s orbitals, the result is an sp3 molecular
orbital, in which the electrons are strongly localized in σ bonds and the atoms are placed in
the corners of a tetrahedron (see the figures above).
In a molecule, the bonding between atoms is complex (see the figures above for the
CH4 molecule (left) and water (right)).
As in crystalline materials, irrespective of the molecule type, the electrons occupy the
discrete energy levels according to the Pauli principle, which forbids two electrons with the
same set of quantum numbers to occupy the same energy level. Therefore, there is always an
occupied molecular orbital with the highest energy, called HOMO (highest occupied
molecular orbital), and an empty/unoccupied molecular orbital with the lowest energy, called
LUMO (lowest unoccupied molecular orbital), similar to the valence and conduction bands,
respectively, in semiconductors, except that the energy spectrum is discrete.
Excitons
Because the electron bond is stronger in molecules than in crystals, the probability to generate
interacting electrons and holes as a result of photon absorption is high. The generation of such
an electron-hole pair linked through Coulomb electrostatic interaction in a crystal is presented
in the figure below, left and center. This interacting pair is called exciton.
α
E = hω
Eg
The exciton is a neutral quasi-particle, which does not contribute to electrical
conduction unless it dissociates (it separates in an electron and a hole that move freely in the
crystal) and which describes the elementary excitations of the system of electrons in crystals
in the presence of Coulomb attraction.
In the absence of Coulomb interaction, the independent evolution of the electron
(subscript n in the equations below) and the hole (subscript p) in the exciton would be
described by Schrödinger equations, as in crystals (the reference energy is the maximum of
the valence band):
⎛
h2 2 ⎞
⎜⎜ E g −
∇ n ⎟⎟ϕ n (rn ) = E nϕ n (rn ) ,
2
m
n
⎝
⎠
h2
−
∇ 2pϕ p ( r p ) = E pϕ p ( r p ) ,
2m p
and the exciton wavefunction, ϕ (rn , r p ) = ϕ n (rn )ϕ p (r p ) , would satisfy the equation
⎞
⎛
h2 2
h2
⎜ Eg −
∇
−
∇ 2p ⎟⎟ϕ (rn , r p ) = Eϕ (rn , r p ) .
n
⎜
2m n
2m p
⎠
⎝
In the presence of Coulomb interaction, ϕ (rn , r p ) ≠ ϕ n (rn )ϕ p (r p ) , and satisfies the equation
2
2
⎛
e2
⎜ E g − h ∇ n2 − h ∇ 2p −
⎜
2m n
2m p
4πε | rn − r p
⎝
⎞
⎟ϕ (rn , rv ) = Eϕ (rn , r p )
| ⎟⎠
where ε is the dielectric constant of the material. The equation above can be separated in a
part that describes the evolution of the center of mass, characterized by the coordinate
R=
mn rn + m p r p
M
with M = mn + m p , and a part describing the relative motion of the electron and the hole,
characterized by the coordinate
r = rn − r p .
In the new coordinates, the evolution equation for the exciton wavefunction
ϕ ( r , R) = g ( R) f (r ) becomes
⎛
h2 2
h2 2
e2 ⎞
⎜⎜ E g −
⎟ϕ (r , R) = Eϕ (r , R)
∇R −
∇r −
2M
2mr
4πε r ⎟⎠
⎝
where mr = mn m p /(mn + m p ) is the relative mass of the electron-hole pair, g (R) evolves
according to
−
h2 2
∇ g ( R) = E1 g ( R) ,
2M R
which describes the evolution of a free particle of mass M, for which E1 = (h 2 K 2 ) / 2 M with
K = k n + k p the total wavevector of the exciton, and f (r ) satisfies the equation
⎛
h2 2
e2 ⎞
⎜⎜ E g −
⎟ f (r ) = E2 f (r ) ,
∇r −
2mr
4πε r ⎟⎠
⎝
which expresses the evolution of a particle with mass mr and energy E2 − E g around a fixed
point to which it is attracted by the Coulomb force.
As for the hydrogen atom, if E2 − E g < 0 , the exciton energy spectrum consists of
discrete levels inside the energy gap (see the figure above, center), given by
E2 − E g = −
mr e 4
1
Eex
=
−
32π 2ε 2 h 2 n 2
n2
with n integer, and an exciton radius, similar to the Bohr radius of the hydrogen atom a B , can
be defined as
aex =
4πε h 2
m
= aB 0 ε r ,
4
mr
e mr
where m0 is the free electron mass. In GaAs, aex = 120 Å, and in Ge aex = 80 Å. The
electron-hole binding energy Eex is given by the difference between the energy necessary to
create a free electron-hole pair and that necessary to create an exciton; the value of this
parameter strongly depends on the crystal. In semiconductors with large ε, the exciton radius
is large and Eex is small, of the order of 10–2 eV, so that the excitons dissociate easily (for
instance, by absorbing phonons) even at moderate temperatures, resulting in a free electron
and a free hole. For example, Eex = 27 meV for CdS, 15 meV for CdSe, 5.1 meV in InP, and
4.9 meV in GaAs. Room temperature excitons can only be observed in nanometer
semiconductor structures where, due to the confinement of electron wavefunctions in
conduction and valence bands, the exciton binding energy is much higher than in bulk
semiconductor. In addition, large exciton binding energies (small exciton radius) are
encountered in molecular crystals in which Eex ≅ 1 eV (in particular, the dielectric constant ε
is high in these materials). The simplest method to create excitons is absorption of
electromagnetic radiation, the dependence of the absorption coefficient on the photon energy
E looking, in this case, as in the figure above, right. The dotted line illustrates the spectral
dependence of the absorption coefficient for the direct, band-to-band absorption mechanism
(see the first part of the course). Note that, since the excitonic levels are inside the energy gap,
the absorption coefficient is significant even for E < E g .
Organic solar cells
The photovoltaic cells in organic materials have, in particular, small mobilities and high
absorption coefficients (with up to three orders of magnitude higher than in inorganic
semiconductors, due to larger dipole moments), but the absorption is significant only in a
narrow bandwidth situated usually at high energies. In noncrystalline organic materials the
concepts of valence and conduction bands are replaced, respectively, by the HOMO and
LUMO discrete levels. In general, in photovoltaic elements based on organic materials the
built-in field originates in the different workfunctions of electrodes. The organic photovoltaic
cells have relatively low fabrication costs and the position of their HOMO and LUMO energy
levels can be tuned in a controllable way. The disadvantage is that most organic materials
degrade if exposed at ultraviolet radiations.
The organic solar cells usually consist from a donor and an acceptor material (see the
figure below, left). A typical example is the polymer(donor)/fullerene C60(acceptor) or the
CuPc(donor)/PV(acceptor) solar cell, as in the figure below, right. Many undoped conjugated
polymers become electron donors after photoexcitation.
As a result of light absorption, an electron is excited from HOMO in LUMO and excitons are
created, the strong Coulomb interaction between electrons and holes in a pair being
characterized by Eex = 0.1−1.4 eV. (The high exciton binding energy in organic materials/
polymers is caused a low dielectric constant; the relative permittivity ε takes values between 2
and 4.) Excitons can be created in both donor and acceptor materials. To generate a
photocurrent, the exciton must dissociate in free carriers that would be transported to
electrodes before recombination. After creation, the excitons diffuse at the donor/acceptor
interface, such that the hole remains in the donor material (with a small electronic affinity)
and the electron remains in the acceptor (with high electron affinity). At this interface there is
a voltage drop that can (or not) separate the electron and the hole (see the sequence of
processes illustrated in the figure below). In most cases, after (a first) exciton dissociation, a
still Coulomb attracted charge pair (geminate pair) forms at the donor/acceptor interface,
which finally dissociates when the charge carriers reach the contacts. At contacts, the excitons
dissociate in the local electric fields that originate by the different workfunctions of
electrodes. The charge separation is, thus, a two-step process.
It must be emphasized that the separation of electric charges due to metallic electrodes
with asymmetric/different workfunctions is possible even in a single layer (not a
heterostructure) of polymer, as in the figure below. At direct polarization, the holes in the
metal with higher workfunction and the electrons in the metal with smaller workfunction are
injected in the thin film of a bulk organic semiconductor. Because of the asymmetric
workfunction, the currents at direct polarization are orders of magnitude higher than at inverse
polarization. Asymmetric electrodes are, for example, ITO and Al. In this case, however, the
charge separation is not efficient since the potential different between electrodes is not high
enough. This simple configuration can be improved by using Schottky junctions.
On the other hand, the diodes with conjugated polymer/C60 heterojunctions are
analogous to p-n junctions even if the electrodes have a similar workfunction. The reason is
that one polarization (electron injection in the semiconducting polymer or hole injection in
C60) is not favorable from an energetic point of view. This polarization induces very low
current densities, unlike the high current densities obtained at the other, energetically
favorable polarization (see the figure below).
Although the quantum efficiency of photocarrier separation is almost 1 for a
donor/acceptor pair, the conversion efficiency in a two-layer molecular heterojunction is
limited since efficient charge separation occurs only near the donor/acceptor interface, inside
the diffusion length of excitons. This diffusion length is of only 3−10 nm. As a consequence
the conversion efficiency is limited by the number of photons absorbed in this narrow region
in which charge separation occurs, so that materials with high absorption coefficients are
needed. If the interface between the donor and acceptor materials is planar, as in the figure
below, left, the active layer (in which excitons are generated) must be thin so that all excitons
reach the interface before recombination. If excitons form further away from the
heterojunction, they have a smaller probability to be collected.
To enhance significantly the number of absorbed photons, and thus the conversion
efficiency, bulk heterojunctions, as in the figure above, center, or ordered heterojunctions, as
in the figure above, right, are used. In bulk heterojunctions the two materials must intermix on
distances smaller than the exciton diffusion length, and the charge transport in both phases
must be assured such that both electrons and holes reach the contacts. Unlike in a planar
interface, in which the surface of effective interaction between the donor and acceptor
components equals the geometric interface, in the bulk heterojunction this surface becomes
the whole volume occupied by the composite material.
Ordered heterostructures are especially encountered in hybrid solar cells (consisting of
an organic and an inorganic material) and make use of TiO2, CdSe, CdS, fullerene (C60) or
ZnO nanowires or nanoparticles, for example, or even carbon nanotubes as type-n
semiconductors, i.e. electron acceptors/materials with higher electronic affinity. These
nanowires or nanoparticles have high absorption coefficients (higher than in bulk materials),
which can be controlled (from the point of view of the spectral dependence) through the
dimensions of the wires/particles. Hybrid solar cells combine the advantages of low cost and
wide range of electrical and optical properties of organic materials with the higher mobility of
charge carriers in inorganic materials, which minimizes the recombination losses. In hybrid
solar cells the charge transfer takes place between the inorganic semiconductor with a high
electronic affinity and the organic/polymer molecules with a small ionization potential. The
charge-tranfer rate increases if the organic substance is chemically bonded to the inorganic
nanocrystallites, which have high densities of electronic states. In the figures below one can
see hybrid solar cells in the superstrate configuration with planar (top, right), mixed
planar/bulk (top, left), bulk (bottom, left) and ordered (bottom, right) heterojunctions. In bulk
heterojunctions, the TiO2 nanocrystals are incorporated in polymer such that a heterogeneous
composite with a large-surface interface is formed. In the figure below P3HT stands for poly3-hexylthiophene and PEDOT:PS is the abbreviation for poly(3,4-ethylenedioxythiophene)
poly(styrenesulfonate). The conversion efficiency of organic and hybrid solar cells is
relatively low (of 5−6%).
Dye-sensitized solar cells
The dye-sensitized solar cell (or DSSC) is a photoelectrochemical cell based on hybrid
(organic-inorganic) technology. The photoelectrochemical solar cells consist from a
semiconductor photoanode and a metallic cathode immersed in electrolyte, and charge
separation occurs not due to the semiconductor only, but due to the semiconductor and the
electrolyte. These cells have efficiencies of about 10% and good performances in diffuse light
conditions (cloudy sky, or indoors). These solar cells mimic the photosynthesis process, and
were first fabricated by covering TiO2 semiconductor crystals with a chlorophyll layer. Their
very small efficiency, of only 0.01%, was considerably enhanced by using TiO2 nanoparticles,
with a diameter of 20 nm (see the figure below, right), covered by a very thin layer of dye/
pigment. This last configuration of dye-sensitized solar cell is also called Gräetzel cell.
The use of nanoparticles increases the effective absorption surface of light of up to
1000 times, and the presence of dyes allows the absorption of photons with lower energies
(larger wavelengths – see the figure below, left) than in TiO2 nanocrystals. This is the reason
that these cells are called dye-sensitized: they are sensitive to photons with energy lower than
the (wide) energy gap of the nanocrystalline semiconductor material.
Observation: The nanocrystalline TiO2 differs from the crystalline TiO2 through its small
dimensions, which confine the electron motion along all three directions, x, y and z. In
particular, for a nanocrystalline particle with dimensions Lx , L y and L z , the energetic
spectrum of electrons inside the nanoparticle no longer consists of allowed and forbidden
bands, as in crystals, but is discrete (see the figure below), the energy levels being given by
ρ0D
z
y
x
E111
h 2 ⎛ pπ ⎞
h 2 ⎛⎜ qπ
⎜⎜
⎟⎟ +
E (k x , k y , k z ) = Ec +
2m ⎝ L x ⎠
2m ⎜⎝ L y
2
E112
2
E113
2
⎞
⎛
⎟ + h ⎜⎜ rπ
⎟
2m ⎝ L z
⎠
E
2
⎞
⎟⎟ = E pqr
⎠
with p, q, r integer numbers. Such a discrete spectrum is similar to that in atoms or molecules,
so that the nanoparticles can be considered as artificial atoms. A nanoparticle is called
quantum dot if its discrete energy levels can be observed, i.e. if the difference between two
discrete levels is higher than the thermal vibration energy k B T . This condition implies
h 2π 2 / 2mL2x , h 2π 2 / 2mL2y , h 2π 2 / 2mL2z > k B T , or Lx , L y , Lz < h 2π 2 / 2mk B T .
The sensitive layer binds to the surface of TiO2 nanocrystals through anchor groups:
carboxyl, phosphonate, or hidroxamate. TiO2 is initially undoped/isolator, and becomes
conductor through the photogenerated electron. Common dyes include Ru bipyridine, other
compounds of rutenium and osmium, porphyrin, cyanides, and phtalocyanine. The dye has
general structure ML2(X)2, where L = 2,2’-bipyridyl-4-4’-dicarboxylic acid, M (metal) = Ru
or Os, X = halides, cyanides, thiocyanates, or even water. The excitation of the Ru complex
via photon absorption occurs according to the following mechanism: the dye HOMO is close
to the Ru atom (metal), whereas LUMO is close to the bipyridyle rings. At excitation, the
electron passes from HOMO in LUMO. The LUMO level, which extends to the “anchor”
group COOH is spatially close to the surface of TiO2, such that there is spatial overlap
between the wavefunction of the LUMO level of the dye and the conduction band in TiO2. As
a consequence, the electron is rapidly transferred at the dye/TiO2 surface.
R
The working principle of the dye-sensitized solar cells is illustrated in the figure
above. The solar cell consists of a 10 μm thick, optically transparent and mesoporous layer,
which contains TiO2 nanoparticles, and which is deposited on the optically transparent and
conducting ITO (indium tin oxide) glass. The TiO2 layer is in chemical contact with a
photosensitive dye, an electrolyte solution containing a redox mediator and another nonilluminated electrode, called counterelectrode, which can even be metallic. An example of
electrolyte is HClO4 with redox systems (I3-/I- groups and Li+ as counterion).
As we have already discussed, the excitation of dye as a result of illumination is
followed by the injection of resulting electrons in the conduction band of the semiconductor
(TiO2), from where they reach the anode (conductive ITO glass or plastic). The regeneration
of electrons in dye occurs through their donation by the redox electrolyte, which is in contact
with the dye. This process is mediated by an organic solvent that contains the couple
iodide/tri-iodide. The tri-iodide is reduced at the metallic electrode (counterelectrode), and the
electron migration from anode to cathode (counterelectrode) closes the circuit. More
precisely, the dye molecule is regenerated by the redox system, which in turn is regenerated at
the metallic electrode by the electrons that pass through the external circuit/the resistance R.
The open-circuit voltage generated in the dye-sensitized solar cell is determined by the
difference between the Fermi level of the electron in TiO2 and the redox potential of the
electrolyte/mediator.
Observation: Redox designates the reduction-oxidation reaction, i.e. the chemical reaction in
which atoms change their oxidation number/state. An increase in the oxidation state implies a
(virtual or real) decrease of electron number, whereas a decrease of the oxidation state
corresponds to a (virtual or real) increase in the number of electrons. The oxidation state is a
parameter describing the oxidation degree of an atom in a chemical compound. It is the hypothetical charge of an atom if all its bonds with atoms of other elements would be 100% ionic.
Unlike in conventional solar cells with p-n junctions, where the semiconductor has a
double role: of light absorption and transport of charge carriers, in dye-sensitized solar cells
these processes are separated. Light is absorbed by the sensitive element/dye, which is
anchored at the surface of a semiconductor with a large E g , while charge separation occurs at
the liquid/solid interface via injection of the photogenerated electrons from the dye into the
conduction band of the solid material and its subsequent transport to the charge
collector/electrode.
The efficiency of dye-sensitized solar cells is of 11%, but decreases to 5−7% in
photovoltaic modules. Instead of TiO2 (with E g = 3.2 eV), one can use nanoparticles or
nanowires (see the figure above) of other wide bandgap oxides such as ZnO, Nb2O5, or one
can use combinations of two complementary dyes, which absorb light from visible in the near
infrared, such as combinations of de porphyrines and phtalocyanines. Alternatively,
semiconducting quantum dots with variable diameters can be used as sensitive elements. In
this case, their diameter must be smaller than the pores in the nanoporous TiO2 electrodes.
As can be seen from the figure below, left, the efficiency of dye-sensitized solar cells
increases as the temperature increases, unlike in Si solar cells, in which the efficiency
decreases as the temperature increases. A better efficiency than for Si cells is also obtained for
low intensities of the incident radiation (see figure below, right), respectively for diffuse light,
the efficiency of dye-sensitized solar cells decreasing if the light intensity increases above an
optimum value.
Some recent types of dye-sensitized solar cells use p-doped semiconductors or organic
materials with hole conduction instead of the electrolyte, which allows an easier manipulation
of the cells and avoids possible electrolyte leaks.
In this type of dye-sensitized solar cells (see the figure above), the illuminated electrode is a
metallic mesoporous oxide, TiO2, which is in contact with a solid hole conductor. A
monolayer of sensitized dye is deposited at the surface of the nanocrystalline electrode film.
Following the dye excitation, an electron is injected in the conduction band of the
semiconductor oxide electrode. The dye regenerates due to the electron received from the hole
conductor. In the solar cell the charge transport is electronic, but in the liquid electrolyte or
polymer it is ionic. The hole conductor must transfer holes from the dye after the latter has
injected electrons in TiO2, which implies that the upper edge of the valence band in the p-type
semiconductor must be above the fundamental state of the dye (see the figure below). In
addition, the hole conductor interpenetrates the nanocrystalline porous layer and must be
transparent in the visible spectral region or, if absorbing, it must have at least the same
efficiency of electron injection as the dye. Common type p conductors used in dye-sensitized
solar cells are CuI and CuBr. For example, the efficiency of the structure n-TiO2/Ru
bipyrydil/p-CuI is 6%, for FF = 45%. The efficiency drops to 4.5% if the light intensity is
higher than 100 mW/cm2. An example of an organic semiconductor is OMETAD (2,2’,7,7’tetranis(N,N-di-p-methoxyphenyl-amine)9,9’-spi-robifluorene.
E
A polymeric gel as electrolyte (as hole conductor) is a compromise between a liquid
and a solid electrolyte. Such a gel is, for example, a mixture between NaI, ethylene carbonate,
propylene carbonate and polyacrylonitrile, or Epichlomer-16 doped with I2/NaI. The
efficiency of solid dye-sensitized solar cells is about 5−7%.
Very thin solar cells are similar to solid dye-sensitized solar cells, because a very thin
layer of CuInS2, CdTe or CuSCN replaces the dye that covers TiO2. The advantage is that the
incident light is better collected since the resulting heterojunction has a very large (collecting)
interface and, in addition, there are multiple scatterings. The measured efficiency for the
TiO2/CuInS2 structure is of 4%.
An alternative to this solar cell is to use as electron acceptor a heterojunction
sensitized with an n-type inorganic nanocrystalline semiconductor with a large energy gap,
the neutrality of the dye charge being ensured by the holes originating from the
complementary p-type, organic or inorganic semiconductor. The electrolyte can be replaced
by spirobifluorene, for example, which interpenetrates the TiO2 nanocrystals.