Chapter 2

Introduction to Solar Cell Operation

This chapter discusses the very fundamental working principle of a solar cell. It
is intended to briefly motivate the relevance of carrier recombination and carrier
transport to solar cell operation without going into technological detail.

2.1 Doped Semiconductors

2.1.1 The Band Gap

An adequate description of electrons in a crystal (represented by a periodic assembly

of atoms) would require the solution of a many-electron problem with a Hamiltonian
involving both the interaction between electrons and atomic nuclei as well as the
interaction among electrons. However, in order to qualitatively motivate some fun-
damental properties of metals and semiconductors, it may suffice to focus on weakly
bound electrons (Bloch electrons) in an effective one-electron potential representing
the interaction with a matrix of atomic nuclei.1 For all Bravais lattice vectors R of
the crystal lattice, the potential U(r) must obey

U(r + R) = U(r). (2.1)

Bloch’s theorem states that for such a periodic effective one-electron potential U(r),
the eigenstates ψk , which are solutions of the Schrödinger equation
 
2
− ∇ 2 + U(r) ψk (r) = εk ψk (r) (2.2)
2m

1 The main lines of reasoning in this section draw on full derivations given in [1].

J. Giesecke, Quantitative Recombination and Transport Properties in Silicon 7

from Dynamic Luminescence, Springer Theses, DOI: 10.1007/978-3-319-06157-3_2,
© Springer International Publishing Switzerland 2014
8 2 Introduction to Solar Cell Operation

to eigenvalues εk , can be chosen as the product of a plane wave and a function uk (r)
with the periodicity of the Bravais lattice

ψk (r) = eik·r uk (r). (2.3)

With reciprocal lattice vectors K, the periodicity of the Bravais lattice allows expres-
sion of the eigenstates ψk and the potential U(r) in terms of Fourier components ck−K
and UK as

ψk (r) = ck−K ei(k−K)·r (2.4)
K

U(r) = UK eiK·r . (2.5)
K

In the limit of vanishing UK , insertion of this expression for ψk into Eq. 2.2 reveals
a parabolic electron band structure with degenerate eigenvalues at Bragg planes
(Brillouin zone boundaries). For a weak but finite UK , perturbation theory yields a
splitting of otherwise degenerate energy levels at a magnitude of 2 · |UK | [1]. This
finding implies a vanishing density of electronic states at certain energy levels.
According to the Pauli exclusion principle, the ground state of a gas of weakly
bound Bloch electrons (valence electrons) in a crystalline solid implies occupation
of all one-electron levels below a so-called Fermi energy εF , which is distinguished
by requiring the total number of one-electron levels with energies less than εF to be
equal to the number of Bloch electrons in a crystalline solid [1]. The Pauli exclusion
principle is also reflected in the Fermi-Dirac distribution representing the occupation
probability

1
f (εk ) = εk −εF (2.6)
e kB T +1

of free (or quasi-free) one-electron levels. If—for reasons outlined above—there is

a considerable difference between the highest occupied level and the lowest unoccu-
pied level in the ground state of the electron gas (at temperature T = 0), this energy
difference is referred to as a band gap. Adjacent bands are denoted as valence band
(below the band gap) and conduction band (above the band gap).

2.1.2 Classification of Semiconductors

Both electronic and optical properties of crystalline solids are essentially affected by
the existence and width of a band gap and by the characteristics of the band structure
near the band gap. Insulating solids are distinguished by a band gap much greater than
kB T . Intrinsic semiconductors feature a band gap in the range of kB T , whereas metals
2.1 Doped Semiconductors 9

feature a vanishing band gap. The above perception of the term semiconductor [1] is
somewhat inadequate with regard to most semiconductors of technological relevance
(e.g. for silicon, the band gap is more than 40 times greater than room temperature
kB T ). Therefore, here the term semiconductor shall denote crystalline solids with a
band gap in the energy range of visible light.

2.1.3 Interaction with Light—Indirect Semiconductors

Electrons in solids may interact with photons. Such optical transitions enable exci-
tation of electrons from their ground state to a previously unoccupied state. The
interaction between light and matter must obey conservation of energy, momentum,
and angular momentum. For indirect semiconductors such as silicon and germanium,
the locations of conduction band minimum and valence band maximum in k-space
differ such that the interaction between electrons and photons must involve another
excitation (e.g. a phonon) in order to satisfy conservation of momentum. The charac-
teristic of an indirect semiconductor largely affects the optical properties of silicon,
as the prerequisite of excitation or anihilation of phonons renders both absorption of
light as well as radiative recombination very unlikely compared to a direct semicon-
ductor of comparable band gap. This fact is essential when relating recombination
lifetime as inferred from measurements of radiative recombination (luminescence)
to other recombination causes such as scattering of electrons at impurities and crystal
defects.

2.1.4 Doping

In a semiconductor with nv valence electrons per atom, adding atoms with nv ± 1

valence electrons induces shallow electronic states in the band gap. Such states are
much confined (∼kB T ) to an energy range close to the conduction band edge for
nv + 1 (donors, n-type doping, e.g. phosphorous doping of silicon) whereas they
are much confined to an energy range close to the valence band edge for nv − 1
(acceptors, p-type doping, e.g. boron doping of silicon).
Adding of donors (free electrons) or acceptors (free vacancies or holes) to a
semiconductor brings about a significant change of electrical conductivity, which
can be explained by a high degree of room temperature ionization of the electronic
states induced via doping—or in other words by a high free carrier concentration.
Acceptor levels allow thermal excitation of electrons from the adjacent valence band,
whereas donor electrons are easily thermally excited into the adjacent conduction
band.
10 2 Introduction to Solar Cell Operation

With a density of electronic states per energy and volume in a solid D(εk ), the
dark densities of electrons and holes are
∞
ne0 = dεk f (εk )D(εk ) (2.7)
εC
εV
 
nh0 = dεk 1 − f (εk ) D(εk ). (2.8)
−∞

Here, εC and εV denote edges of the conduction and valence band, respectively. For
a symmetric shape of D(εk ) with respect to the band gap (intrinsic semiconductor),
this implies a Fermi energy close to the center of the band gap, leading to an intrinsic
free carrier concentration in silicon of ni ≈ 1010 cm−3 at room temperature [2–5].
However, with a substantial amount of electronic states near either conduction band
or valence band, the symmetry of D(εk ) is destroyed, leading to a shift of Fermi
energy toward the respective band edge in order to satisfy Eqs. 2.7 and 2.8. For
T = 0, the Fermi energy must lie between the respective band edge and the adjacent
energy level of the donors or acceptors—leading to a vanishing concentration of free
carriers. Yet, at an increasing temperature the Fermi level advances toward the center
of the bandgap, such that at room temperature free carrier concentration (ionized
dopant concentration) nearly corresponds to dopant concentration. At typical dopant
concentrations of 1015 –1016 cm−3 this brings about an increase of free carrier density
and hence of electrical conductivity by 5–6 orders of magnitude.
The density of free carriers due to ionized dopant atoms is denoted as majority
carrier density, whereas the density of the complementary charge carrier species is
denoted as minority carrier density. The shift of Fermi energy in a doped semiconduc-
tor away from the minority carrier band leads to an inverse proportionality between
dark minority carrier density and dopant concentration.

2.2 Quasi-Fermi Energy—Electrochemical Potential

While the band population via free electrons (and—equivalently—holes) may be

described via Fermi-Dirac statistics according to Eq. 2.7, this description is compli-
cated in the presence of additional generation of excess carrier density n beyond
dark equilibrium. The concept of so-called Quasi Fermi levels aims at an adequate
description of such a state of electrochemical non-equilibrium.
2.2 Quasi-Fermi Energy—Electrochemical Potential 11

2.2.1 Quasi-Fermi Levels



Due to a square root shape of the density of states D(εk ) ∝ ± εk − εC/V , free
carrier densities in the dark may be accurately expressed as2
εF −εC
ne0 = NC e kB T , (2.9)
εV −εF
nh0 = NV e kB T , (2.10)

with NC and NV denoting effective densities of states of the conduction and valence
band
 3
2πme∗ kB T 2
NC = 2 , (2.11)
h2
  23
2πmh∗ kB T
NV = 2 , (2.12)
h2

and with effective masses of electrons and holes [1] me/h ∗ and the Planck constant h.

In the case of excess carrier generation beyond thermal equilibrium, the pop-
ulation of valence and conduction band through excess carriers is no longer ade-
quately described via Fermi-Dirac statistics. Whereas in the dark state, the product
of minority and majority carrier densities must correspond to intrinsic carrier den-
sity ne0 nh0 = ni2 , now both carrier species feature enhanced densities, requiring a shift
of Fermi energy toward both band edges. This dilemma is resolved by the concept
of quasi-Fermi energies, which allow the description of a state of electrochemical
non-equilibrium of a semiconductor under illumination via Fermi-Dirac statistics.
Quasi-Fermi energies εFC and εFV of electrons and holes replace Fermi energy εF in
the Fermi-Dirac distribution (cf. Eq. 2.6), respectively. Electron and hole densities
can thus be expressed as3
εFC −εC
ne = ne0 + n = NC e kB T , (2.13)
εV −εFV
nh = nh0 + n = NV e kB T . (2.14)

The separation of quasi-Fermi energies of electrons and holes relates to the implied
voltage U of a semiconductor under illumination εFC − εFV = qU. The following
derivation reveals the relation between quasi-Fermi levels and the so-called electro-
chemical potential of carriers.

2 This approximation is deemed accurate for |εF − εC/V |  3kB T .

3 Here, ne = nh = n is assumed.
12 2 Introduction to Solar Cell Operation

2.2.2 The Electrochemical Potential

The change of internal energy dE of a statistical ensemble such as a gas of electrons

or holes corresponds to the balance of changes of all relevant energy forms4

dE = TdS T − pdV + μc dN + φdQ. (2.15)

Here, changes of thermal energy, compressional energy, chemical energy, and elec-
trical energy are considered. The system is distinguished by the intensive variables
temperature T , pressure p, chemical potential μc , and electrical potential φ. Further,
the extensive variables giving rise to a change of internal energy are the entropy S T ,
the volume V , the particle number N, and the charge Q. The energy which a solar
cell delivers to a load is given by the sum of the changes of free energy dFe/h of
electrons and holes, respectively.
 
dF = dE − d TS T = S T dT − pdV + μc dN + φdQ (2.16)

With constant semiconductor volume and temperature during solar cell operation,
this expression simplifies accordingly such that

dF = (ηe + ηh ) dN, (2.17)

with the electrochemical potential

ηe/h = μc,e/h ∓ qφ, (2.18)

and with the positive elementary charge q. With a free energy increment according
to Eq. 2.17, the electrochemical potential represents the essential potential of excess
carriers in semiconductors. Its gradient gives rise to net currents of excess carriers.
Consider the mean energy per particle of a gas of Ne/h free electrons/holes at
temperature T

T +p V +η
TSe/h
3 e/h e/h
εe/h
= ±εC/V + kB T = . (2.19)
2 Ne/h

Making use of the modified Sackur-Tetrode equation expressing the entropy of an

ideal gas [6]
 
5 NC/V
T
Se/h = kB Ne/h + ln (2.20)
2 ne/h

and exploiting the equation of state of an ideal gas

4 This derivation proceeds along the lines of a derivation given in [6].

2.2 Quasi-Fermi Energy—Electrochemical Potential 13

pe/h V = Ne/h kB T , (2.21)

one obtains
ηe −εC
ne = NC e kB T , (2.22)
ηh +εV
nh = NV e kB T . (2.23)

Comparison to Eqs. 2.13 and 2.14 reveals

ηe = εFC (2.24)
ηh = −εFV . (2.25)

Therefore, the implied voltage U of a semiconductor under illumination can be

expressed via the sum of electrochemical potentials

qU = εFC − εFV = ηe + ηh = η . (2.26)

With a chemical potential of the form

ne/h
μc,e/h = kB T ln (2.27)
ae/h

with constant densities ae/h ,5 according to Eq. 2.18 the electrochemical equilibrium
condition (vanishing separation of quasi-Fermi energies η = 0) yields

ae ah = ne0 nh0 = ni2 . (2.28)

Consequently, the separation of quasi-Fermi energies η can be expressed in terms

of the product of minority and majority carrier density as
ne nh
η = kB T ln . (2.29)
ni2

2.3 The pn-Junction

The essential task of a photovoltaic device is to separate electrical charge released

upon absorption of light—preferrably at the highest possible conversion efficiency
from photon energy to electrical energy (as the product of voltage and electrical
charge). This essential task requires efficient mechanisms of both light absorption

5 ae/h enable a dimensionless logarithmic argument and incorporate a constant offset of μc,e/h .
14 2 Introduction to Solar Cell Operation

and charge separation. In semiconductor photovoltaic devices, this task is com-

monly accomplished by a pn-junction within an absorber. This is the interface region
between n- and p-doped semiconductor volumes. The light absorbing semiconductor
bulk supplies electrical charge, while the pn-junction provides a means to separate it.
The charge current released by an illuminated pn-junction device with a separation
of quasi-Fermi energies qU = η gives rise to the delivery of work to an external
load. In the following, this charge current is motivated—substantiating the relevance
of transport properties such as carrier mobilities, diffusion coefficients, and dopant
concentrations.

2.3.1 Dark pn-Junction

Electron and hole charge current density in semiconductors can be expressed in terms
of the gradient of the electrochemical potential. With the electrical conductivity

σe/h = qne/h μe/h (2.30)

and with μe/h denoting electron and hole carrier mobilities, electron or hole charge
current density is
σe/h
jQ,e/h = ± ∇ηe/h . (2.31)
q

Insertion of the electrochemical potential of Eq. 2.18 with a chemical potential given
by Eq. 2.27 yields charge currents
 
σe/h kB T
jQ,e/h = ± ∇ne/h ± qE , (2.32)
q ne/h

with a local electrical field E = −∇φ. In the dark, i.e. in electrochemical equilibrium,
net current must vanish everywhere in the semiconductor, which is equivalent to the
condition
kB T
∇ne/h = ±qE. (2.33)
ne/h

Thus, in dark equilibrium, the gradient of electron and hole densities must be bal-
anced by a local electrical field. Integrating Eq. 2.32 across the pn-junction from the
n-contact to the p-contact (indicated via superscript here) yields
n
p kB T ne/h
φne/h − φe/h =± ln p . (2.34)
q ne/h
2.3 The pn-Junction 15

p
The quantity φne/h − φe/h —which is also referred to as a dark diffusion voltage
[6]—accounts for the compensation of electrical and chemical potential differences
across the pn-junction such that in electrochemical equilibrium the voltage between
the terminals of a (dark) pn-junction device is zero.
The large difference between free electron and hole densities across the pn-
junction leads to diffusion of carriers from the region where they represent majority
carriers to the region where they represent minority carriers. This diffusion process
brings about a buildup of local charge within a so-called space charge region.

2.3.2 Illuminated pn-Junction

Illumination of a pn-junction with photon energies above the band gap leads to band-
to-band photon absorption and generation of excess charge carrier density n. This
process increases the number of free electrons and holes by the same amount. Dark
majority and minority carrier concentrations commonly differ by a factor of

0
 0
2
nmaj nmaj
0
= ≫ 1. (2.35)
nmin ni

Therefore, the relative increase of minority carrier concentration upon illumination

is much greater than the relative increase of majority carrier concentration, and the
space charge and the diffusion voltage of the pn-junction are substantially reduced
as a consequence of excess carrier generation—as can already be concluded from
Eq. 2.34.
   
 n p   p 
φe/h − φe/h  > φne/h − φe/h  (2.36)
n=0 n>0

In the metal contacts of a pn-junction, there is no band gap and therefore excess
carrier density must smoothly approach zero toward the terminals at the side faces of
a pn-junction. This again implies that the difference of chemical potentials between
the terminals of a pn-junction must not change upon excess carrier generation
   
 n p   n p 
ln ne/h − ln ne/h  = ln ne/h − ln ne/h  . (2.37)
n=0 n>0

Thus, the voltage delivered at the metal electrodes of an illuminated pn-junction

corresponds to the amount by which the diffusion voltage of the pn-junction
(cf. Eq. 2.34) is reduced through a finite excess carrier density. This state of elec-
trochemical non-equilibrium is equivalent to a finite separation of quasi-Fermi ener-
gies 0 = η = qU. Figure 2.1 depicts the space charge, the band edges, and the
quasi-Fermi levels of electrons and holes in a pn-junction device both, in (dark)
electrochemical equilibrium, and under illumination or carrier injection.
16 2 Introduction to Solar Cell Operation

Fig. 2.1 Depiction of local charge density ρQ , valence and conduction band edges, and the quasi-
Fermi energies (± electrochemical potentials) of electrons and holes across a pn-junction. The state
of electrochemical equilibrium without illumination (left) is compared to the non-equilibrium state
under illumination—featuring a finite separation of quasi-Fermi energies η = qU corresponding
to the implied voltage U. εnV denotes the energy of the valence band edge at the n-terminal

2.3.3 Charge Current of a pn-Junction

As yet, the implied voltage of a pn-junction device under illumination has been
motivated. However, what has not been motivated so far is the nature of the charge
current out of the pn-junction device, which gives rise to the delivery of work in an
external load. With electrical conductivity as specified in Eq. 2.30 and along the lines
of Eq. 2.31, the total charge current density in a pn-junction device corresponds to
the sum of charge current densities of electrons and holes

1
jQ = ±σe/h ∇ηe/h . (2.38)
q
e/h

The charge current released at the terminals of a pn-junction device is a majority

carrier drift current. As such, the solar cell current is induced by an electrical field—
albeit not explicitly by the field of the space charge region, which does not deliver
work on carriers [6]. In order to motivate the driftlike characteristic of a solar cell
current, the nature of drift and dielectric relaxation has to be reflected upon. As one
may conclude from Eqs. 2.17 and 2.18, minimization of free energy in a solar cell is
equivalent to minimization of the sum of chemical potentials (via diffusion/random
motion) and of electrical potentials (via field-driven drift/dielectric relaxation) of
charge carriers, respectively.
2.3 The pn-Junction 17

2.3.3.1 Dielectric Relaxation

Considering the dissipation of a local charge density ρQ , the combination of the

conservation of charge6

∂ρQ
+ ∇jQ = 0 (2.39)
∂t
with the charge current density

jQ = σE (2.40)

and with the differential representation of Gauss’s law

0 ∇E = ρQ (2.41)

yields the ordinary equation

 
d σ
+ ρQ = 0. (2.42)
dt

Its solution for an initial value ρQ (t = 0) = ρQ,0 and with a dielectric relaxation
time τdiel =

σ0 is
t
−τ
ρQ (t) = ρQ,0 e diel . (2.43)

With electrical conductivity as of Eq. 2.30, τdiel must be much shorter for majority
carriers than for minority carriers. Therefore, local charge imbalances in doped (and
nondegenerate) semiconductors dissipate via dielectric relaxation of majority carri-
ers. With the dielectric constant of silicon
≈ 12, the vacuum permittivity [4]
0 =
8.854 × 10−14 A s V−1 cm−1 , a carrier mobility of roughly μ ≈ 103 cm2 V−1 s−1 ,
and the elementary charge q = 1.602 × 10−19 A s, τdiel may be roughly approxi-
mated as

7 × 103
τdiel ≈ cm−3 s, (2.44)
nmaj

which is below the range of picoseconds for typical dopant concentrations (e.g.
τdiel ≈ 5 × 10−13 s for 1  cm p-type material). Within this timescale, the character-
istic travelling distance of drift-induced transport can be conservatively estimated by
solar cell thickness (d ≈ 2 × 10−2 √ cm). Comparison with carrier diffusion distances
within this timescale of typically Dτdiel  5 × 10−6 cm d (diffusion coeffi-
cient D) allow the conclusion that drift and dielectric relaxation of local excess charge

6 Derivation proceeds along the lines of [6].

18 2 Introduction to Solar Cell Operation

proceeds instantaneously on relevant timescales of carrier diffusion through the solar

cell. This justifies the assumption of local quasi charge neutrality n = ne ≈ nh
outside the space charge region.

2.3.3.2 Diffusion Currents

0 on both sides of a pn-junction such

Let us assume constant dopant concentrations ne/h
that ∇ne/h = ∇n holds locally, and let the chemical potential drop due to changes
of electron and hole concentration across the pn-junction be entirely compensated
by an inverse electrical potential drop in the space charge region. Let us further
assume a vanishing electrical potential gradient beyond the space charge region—
i.e. vanishing majority carrier drift. In this hypothetical scenario, any gradient of
the electrochemical potential would have to be due to a gradient of excess carrier
density, and any electrical current would have to be caused by diffusion. With the
assumptions stated above, and accounting for electrical conductivity according to
Eq. 2.30, Eq. 2.31 would then become

jQ,e/h = ±kB T μe/h ∇n. (2.45)

Consequently, (with ∇ne ≈ ∇nh ) the sum of charge current densities of electrons
and holes would have to locally compensate (except for the minor difference between
their mobilities). Thus, diffusion currents from the pn-junction to the device terminals
do not give rise to a charge release at the terminals of a pn-junction device. Diffusion
currents can be detrimental as recombination currents associated with recombination
in the bulk and interfaces of an absorber, and at the same time they are vital for solar
cell operation as they ensure minority carrier transport to the pn-junction. However,
diffusion currents alone cannot explain solar cell operation. Rather, it is a majority
carrier drift induced by an electrical potential gradient, which gives rise to charge
release at the terminals of a pn-junction device.

2.3.3.3 Selectivity of the pn-Junction

The capability of the pn-junction to actually separate the charge of excess carriers
is denoted here as selectivity. Two essential features account for the selectivity of a
pn-junction: the electrical field of the space charge region, and the large difference
between majority and minority carrier conductivities.
The electrical field of the space charge region enables substantial carrier transport
against a very pronounced chemical potential gradient. Such net carrier transport is
essentially rendered energetically favorable due to elastic backscattering of majority
carriers in the space charge region’s field. Optical generation of excess carriers then
causes a considerable pressure of highly conductive excess majority carriers (and
of their corresponding charge) at the device electrodes, respectively. In spite of the
relatively high chemical potential of excess majority carriers, this pressure cannot
2.3 The pn-Junction 19

relax via diffusion across the pn-junction, as this is prevented by the electrical field.
Conversely, when connecting the electrodes of an illuminated solar cell via a load, the
excess majority carrier charge pressure may relax through this external circuit. Thus,
the resulting majority carrier drift current is eventually driven by a faint electrical
potential gradient due to the connection of the device electrodes via a load. It appears
noteworthy that, although the electrical field of the space charge region does not
deliver work on charge carriers, it is nonetheless an electrical field which drives—or
rather pulls—majority carrier currents out of the electrodes of a solar cell: this is
the electrical field established between the electrodes or terminals of a solar cell by
connecting them through a load—sustained and fed by the steady-state generation
of excess carriers.

2.3.3.4 Drift or Diffusion

As outlined above, solar cell operation is enabled by the interplay of drift and diffusion
currents. While the kinetics of transport of minority carriers to the pn-junction is
governed by their random motion with a diffusion coefficient D, the kinetics of the
transport of majority carriers to the terminals of a photovoltaic device is governed
by drift or dielectric relaxation. Both carrier mobilities μe/h and concentrations ne/h
are crucial determinants of this transport process.

2.4 Solar Cell Efficiency—Loss Mechanisms

The efficiency η of photovoltaic energy conversion is defined as the ratio

Pel
η= (2.46)
Pph

between the electrical power Pel = Umpp Impp delivered to a load by a photovoltaic
device and the light power Pph to which the latter is exposed. Here, Umpp and Impp
denote the steady-state (maximum-power-point) operation voltage and current deliv-
ered by a photovoltaic device.

2.4.1 Idealized Reversible Conversion Process

Photovoltaic energy conversion efficiency is inherently costrained to η < 1, as

follows from thermodynamic considerations. An idealized treatment by Landsberg
[7, 8] considers the internal energy, the entropy, and the radiation pressure of isotropic
blackbody radiation of temperature Tp (pump) surrounding a photovoltaic converter
at a temperature Tc in thermal equilibrium with a thermal sink of temperature Ts = Tc .
20 2 Introduction to Solar Cell Operation

The resulting Landsberg efficiency is

 4
4 Ts 1 Ts
ηL = 1 − + . (2.47)
3 Tp 3 Tp

It represents a reversible process of energy conversion (i.e. no generation of entropy),

and therefore provides an upper limit of photovoltaic energy conversion efficiency.
Landsberg efficiency ηL differs from the efficiency ηC of an idealized Carnot engine
(acting between the gas of blackbody radiation at temperature Tp and a converter at
temperature Tc = Ts )

Ts
ηC = 1 − (2.48)
Tp

in that it not only accounts for the entropy of radiation, but also for radiation of the
converter. Minor inaccuracies of the Landsberg efficiency limit stem from the fact
that it assumes perfect blackbody radiation of both the sun (pump) and the converter.
Assuming a converter Temperature of Tc = Ts = 300 K and an effective temperature
of solar black body radiation of Tp ≈ 5,800 K [9], Carnot efficiency is ηC = 0.948,
whereas Landsberg efficiency is ηL = 0.931.
Real solar energy conversion features efficiencies below the Landsberg limit.
Landsberg efficiency represents the upper efficiency limit of an energy conversion
process with respect to the second law of thermodynamics, as any energy conversion
process of practical relevance is irreversible (i.e. involves generation of entropy) and
accompanied by power losses.

2.4.2 Single Junction-Detailed Balance Limit

In 1961, Shockley and Queisser [10] identified a limit of the solar energy conversion
efficiency via semiconductors, which essentially accounts for irreversibility, and
which incorporates all loss mechanisms to be outlined in the following. The principle
of detailed balance, which requires radiative recombination of a light absorbing
semiconductor, sets an upper limit to the maximally achievable electrical power
Umpp Impp of a photovoltaic device exposed to sunlight.
Shockley and Queisser first proposed an idealized calculus, assuming the solar
(6,000 K [10]) black body spectrum to be absorbed and to entirely contribute to
solar cell current for photon energies above the semiconductor bandgap εG , whereas
photon energies below εG would not be absorbed at all. This calculus yielded an
ultimate efficiency limit of η = 0.44 peaking at a band gap of εG = 1.1 eV,
and suggesting silicon (with a band gap slightly above the latter value) as an opti-
mal single junction photovoltaic device. Accounting for radiative recombination of
2.4 Solar Cell Efficiency—Loss Mechanisms 21

absorbed light without any concentration of sunlight, the resulting detailed balance
limit encountered at εG = 1.1 eV is η = 0.30.7

2.4.3 Classification of Power Losses

One way to classify the power loss mechanisms of a solar cell is to distinguish between
thermal, optical, and electrical power losses. The categories overlap to some extent.
Yet, this classification proves to be convenient for the specification of the subject of
this work.
1. Thermal losses denote dissipation of photon energy in excess of a solar cell band
gap upon photon absorption. Once excited, electrons and holes quickly relax to
the lowest unoccupied states in their respective bands via generation of phonons.
In an indirect semiconductor, this relaxation takes place on a timescale much less
than excess carrier recombination. Thermal losses imply a major loss mechanism
in single-junction devices such as silicon solar cells. They can be reduced via
band gap engineering, involving an appropriate design and sequence of materials
with different band gaps (e.g. compound semiconductors or quantum wells) tuned
to the solar spectrum.
2. Optical losses denote deviations from total light absorption by a photovoltaic
absorber. They incorporate losses due to reflection, transmission, and parasitic (i.e.
non-band-to-band) absorption. High-efficiency solar cells provide miscellaneous
concepts to minimize optical power losses, ranging from antireflective coating
and light trapping techniques to approaches of up- and/or downconversion of
remanent solar irradiation.
3. Electrical losses denote losses due to recombination or the imperfection of trans-
port of excess carriers. Recombinative losses involve radiative recombination,
recombination due to the interaction among excess carriers (Auger recombina-
tion), and recombination at crystal defects and impurities causing electronic states
in the band gap (Shockley-Read-Hall (SRH) recombination [11, 12]). In indirect
semiconductors, recombinative losses are mostly associated with phonon genera-
tion. Thus, they imply thermal losses as well. Transport related loss mechanisms
involve resistive losses (finite series resistance of pn-junction device) as well as
shunt losses (finite parallel resistance of pn-junction devices accompanied by an
ohmic reflow of excess charge).
There is a substantial gap between the above stated detailed balance limit of
photovoltaic energy conversion efficiency via single junction devices of η = 0.30 and
the highest efficiencies reached on laboratory type silicon solar cells of η = 0.250 ±
0.005 [13–15]. This gap can be attributed to optical and electrical causes. As for

7 Note that the maximum of the detailed balance limit (without concentration) as a function of band

gap energy εG is shifted to ∼1.3 eV as opposed to the maximum at ∼1.1 eV encountered for the
ultimate efficiency limit.
22 2 Introduction to Solar Cell Operation

optical causes, Shockley and Queisser assumed a Heaviside-like quantum efficiency.

In reality, capture of light near the band gap energy εG has proven to be among the
most demanding challenges on the path toward the Shockley-Queisser-Limit [13].
As for electrical causes, recombination at metal contacts probably represents the
most dominant loss mechanism [16, 17].

2.5 Carrier Recombination and Transport

The above-mentioned category of electrical power losses is closely related to both

recombination and transport properties of a solar cell. While concepts to adequately
and quantitatively describe recombination rates and essential carrier transport prop-
erties are detailed in the following chapter, this section is intended to qualitatively
sketch the relevance of carrier recombination and transport to solar cell operation.
Upon optical generation of excess electrons and holes in a semiconductor solar
cell, excess carriers pile up in the absorber—bringing about a separation of quasi-
Fermi levels η of electrons and holes equivalent to an implied device voltage
η
U ≈ q , which enables the delivery of work dW = η dN to an external load by
releasing dN carriers of charge q. The steady-state excess carrier density of a solar
cell in operation results from a balance between excess carrier generation on the one
hand and both excess carrier recombination and charge extraction on the other hand.

2.5.1 Recombination

To some extent, excess charge carrier recombination is inherent to the operation of

a semiconductor solar cell. This particularly applies to radiative recombination and
Auger recombination. However, SRH recombination through electronic states in the
band gap is by no means inherent to solar cell operation. Rather, it largely depends
on material quality and on the formation or passivation of electronic states in the
band gap during the solar cell production process. Therefore, the prevention and the
efficient passivation of such electronic states is the essential task of all solar cell
technologists concerned about excess charge carrier recombination.
SRH recombination occurs both at interfaces as well as in the semiconductor
bulk. Whereas bulk recombination may be limiting for multicrystalline and possi-
bly upgraded metallurgical grade silicon wafers [18] as well as for p-type materials
featuring substantial recombination related to impurity-acceptor complexes [19, 20],
the dominant recombination rate of most high efficiency solar cell concepts repre-
sents emitter or interface recombination, and particularly recombination at contacts
between a semiconductor and metal electrodes [16, 17].
2.5 Carrier Recombination and Transport 23

The optimization of the solar cell production process with regard to excess charge
carrier recombination calls for accurate and precise measurement techniques of
recombination properties, and for an accurate localization of recombination rates
in bulk or interfaces.

2.5.2 Transport

Equation 2.38 indicates how electrical charge is delivered to the metal contacts of a
solar cell. As can be seen already from this very simple and schematic representa-
tion, transport properties such as carrier mobilities, dopant concentrations, and the
minority carrier diffusion coefficient largely affect solar cell operation. Therefore,
accurate knowledge of these properties is vital for the further improvement of solar
cell technologies.
Particularly carrier mobilities in crystalline silicon are often assumed to be accu-
rately known, and as such they are occasionally used in measurements and device
simulations as if they featured no uncertainty. Yet, considerable deviations of mobil-
ity data and models have been reported in literature (even for monocrystalline silicon)
[21–23], and in spite of speculations about potential measurement artifacts, the cause
to these deviations has not yet been resolved. With novel techniques to measure both
majority and minority carrier mobility, the present work provides new experimen-
tal evidence in an unresolved controversy on the one hand. On the other hand, the
techniques to be presented open up new pathways to access carrier mobilities in
multicrystalline and particularly in upgraded metallurgical grade silicon.

References

1. N.W. Ashcroft, D.N. Mermin, Solid State Physics (Saunders College, Philadelphia, 1976)
2. M.A. Green, Intrinsic concentration, effective densities of states, and effective mass in Silicon.
J. Appl. Phys. 67, 2944–2954 (1990)
3. A.B. Sproul, M.A. Green, Improved Value for the Silicon intrinsic carrier concentration from
275 to 375 K. J. Appl. Phys. 70, 846–854 (1991)
4. A.B. Sproul, M.A. Green, Intrinsic carrier concentration and minority-carrier mobility of Sili-
con from 77 to 300 K. J. Appl. Phys. 73, 1214–1225 (1993)
5. P.P. Altermatt, A. Schenk, F. Geelhaar, G. Heiser, Reassessment of the intrinsic carrier density
in Crystalline Silicon in view of band-gap narrowing. J. Appl. Phys. 93, 1598–1604 (2003)
6. P. Würfel, Physics of Solar Cells: From Basic Principles to Advanced Concepts (Wiley-VCH,
Weinheim, 2009)
7. P.T. Landsberg, G. Tonge, Thermodynamic energy conversion efficiencies. J. Appl. Phys. 51,
R1–R20 (1980)
8. P. Baruch, A. De Vos, P.T. Landsberg, J.E. Parrott, On some thermodynamic aspects of photo-
voltaic Solar Energy conversion. Solar Energy Mater. Solar Cells 36, 201–222 (1995)
9. D.R. Williams, Sun Fact Sheet, NASA (2012). http://nssdc.gsfc.nasa.gov/planetary/factsheet/
sunfact.html
24 2 Introduction to Solar Cell Operation

10. W. Shockley, H.J. Queisser, Detailed balance limit of efficiency of p-n junction solar cells.
J. Appl. Phys. 32, 510–519 (1961)
11. R.N. Hall, Germanium rectifier characteristics. Phys. Rev. 83, 228 (1951)
12. W. Shockley, W.T. Read, Statistics of the recombinations of holes and electrons. Phys. Rev.
87, 835–842 (1952)
13. J. Zhao, A. Wang, M.A. Green, F. Ferrazza, 19.8 % efficient “Honeycomb” textured mul-
ticrystalline and 24.4 % monocrystalline silicon solar cells. Appl. Phys. Lett. 73, 1991–1993
(1998)
14. M.A. Green, The path to 25 % silicon solar sell efficiency: history of silicon cell evolution.
Prog. Photovoltaics: Res. Appl. 17, 183–189 (2009)
15. M.A. Green, K. Emery, Y. Hishikawa, W. Warta, E.D. Dunlop, Solar cell efficiency tables
(Version 41). Prog. Photovoltaics: Res. Appl. 21, 1–11 (2012)
16. J. Benick, B. Hoex, M.C.M. Van de Sanden, W.M.M. Kessels, O. Schultz, S.W. Glunz, High
efficiency n-type Si solar cells on AlO-passivated Boron Emitters. Appl. Phys. Lett. 92, 253504
(2008)
17. P.J. Cousins, D.D. Smith, H.C. Luan, J. Manning, T.D. Dennis, A. Waldhauer, K.E. Wilson,
G. Harley, W.P. Mulligan, Generation 3: improved performance at lower cost, in Proceedings
of the 35th IEEE PVSC (Honolulu, 2010), pp. 275–278
18. B. Michl, M. Rüdiger, J.A. Giesecke, M. Hermle, W. Warta, M.C. Schubert, Efficiency limiting
bulk recombination in multicrystalline silicon solar cells. Solar Energy Mater. Solar Cells 98,
441–447 (2012)
19. D. Macdonald, T. Roth, P.N.K. Deenapanray, K. Bothe, P. Pohl, J. Schmidt, Formation rates of
Iron-Acceptor pairs in crystalline Silicon. J. Appl. Phys. 98, 083509 (2005)
20. K. Bothe, J. Schmidt, Electronically activated Boron-Oxygen-related recombination centers in
crystalline Silicon. J. Appl. Phys. 99, 013701 (2006)
21. D.B.M. Klaassen, A unified mobility model for device simulation—I. Model equations and
concentration dependence. Solid State Electron. 35, 953–959 (1992)
22. D.B.M. Klaassen, A unified mobility model for device simulation—II. Temperature depen-
dence of carrier mobility and lifetime. Solid State Electron. 35, 961–967 (1992)
23. A.B. Sproul, M.A. Green, A.W. Stephens, Accurate determination of minority carrier- and
lattice scattering-mobility in Silicon from photoconductance decay. J. Appl. Phys. 72,
4161–4171 (1992)
http://www.springer.com/978-3-319-06156-6