1-3-3. Energy Band Alignment in Chalcogenide Thin Film Solar Cells From Photoelectron Spectros

Journal of Physics: Condensed Matter
TOPICAL REVIEW Related content

- Synthesis and characterization of
Energy band alignment in chalcogenide thin film (Cd,Zn)S buffer layer for Cu2ZnSnSe4
solar cells
solar cells from photoelectron spectroscopy Khaled Ben Messaoud, Marie Buffière,
Guy Brammertz et al.
- Efficiency limitations of thermally

To cite this article: Andreas Klein 2015 J. Phys.: Condens. Matter 27 134201 evaporated thin-film SnS solar cells
A Schneikart, H-J Schimper, A Klein et al.
- Elongated nanostructures for radial

junction solar cells
Yinghuan Kuang, Marcel Di Vece, Jatindra
View the article online for updates and enhancements. K Rath et al.
Recent citations
- Electronic defects in Cu(In,Ga)Se2 :
Towards a comprehensive model
Conrad Spindler et al
- Rectifying Effect of the Sr 3 Al 2 O 6 /Ga 2

O 3 Heterojunction
Zeng Liu et al
- Band Alignments, Band Gap, Core Levels,

and Valence Band States in Cu3BiS3 for
Photovoltaics
Thomas J. Whittles et al
This content was downloaded from IP address 131.183.180.59 on 10/01/2020 at 21:12

Journal of Physics: Condensed Matter
J. Phys.: Condens. Matter 27 (2015) 134201 (24pp) doi:10.1088/0953-8984/27/13/134201
Topical Review
Energy band alignment in chalcogenide

thin film solar cells from photoelectron
spectroscopy
Andreas Klein
Department of Materials and Geo Sciences, Technische Universität Darmstadt, Surface Science Division,
Jovanka-Bontschits-Straße 2, D-64287 Darmstadt, Germany
E-mail: [email protected]
Received 21 July 2014, revised 30 September 2014

Accepted for publication 15 October 2014
Published 13 March 2015
Abstract
Energy band alignment plays an important role in thin film solar cells. This article presents an
overview of the energy band alignment in chalcogenide thin film solar cells with a particular focus
on the commercially available material systems CdTe and Cu(In,Ga)Se2 . Experimental results from
two decades of photoelectron spectroscopy experiments are compared with density functional theory
calculations taken from literature. It is found that the experimentally determined energy band
alignment is in good agreement with theoretical predictions for many interfaces. These alignments,
in particular the theoretically predicted alignments, can therefore be considered as the intrinsic or
natural alignments for a given material combination. The good agreement between experiment and
theory enables a detailed discussion of the interfacial composition of Cu(In,Ga)Se2 /CdS interfaces in
terms of the contribution of ordered vacancy compounds to the alignment of the energy bands. It is
furthermore shown that the most important interfaces in chalcogenide thin film solar cells, those
between Cu(In,Ga)Se2 and CdS and between CdS and CdTe are quite insensitive to the processing of
the layers.
There are plenty of examples where a significant deviation between experimentally-determined
band alignment and theoretical predictions are evident. In such cases a variation of band alignment
of sometimes more than 1 eV depending on interface preparation can be obtained. This variation can
lead to a significant deterioration of device properties. It is suggested that these modifications are
related to the presence of high defect concentrations in the materials forming the contact. The
particular defect chemistry of chalcogenide semiconductors, which is related to the ionicity of the
chemical bond in these materials and which can be beneficial for material and device properties, can
therefore cause significant device limitations, as e.g. in the case of the CuInS2 thin film solar cells or
for new chalcogenide absorber materials.
Keywords: solar cells, band alignment, photoemission

(Some figures may appear in colour only in the online journal)
1. Introduction chalcogenide semiconductor solar cells (Cu(In,Ga)(S,Se)2 and

CdTe), have considerable production volumes. The increasing
In the last decade, solar cells have experienced a huge commercialization of the latter has led to considerable
increase in manufacturing capacities, which has been enabled advances in cell efficiencies [2]. Both Cu(In,Ga)(S,Se)2 and
by significant advances in manufacturing technologies [1]. CdTe solar cells can now be prepared with efficiencies of
Today, the market is dominated by crystalline Si solar 21.7% [3] and 21.0% [4], respectively. Both are higher than
cells but thin film technologies, in particular thin film that of polycrystalline Si.
0953-8984/15/134201+24$33.00 1 © 2015 IOP Publishing Ltd Printed in the UK

J. Phys.: Condens. Matter 27 (2015) 134201 Topical Review
Cu(In,Ga)(S,Se)2 and CdTe solar cells are made by

deposition on glass substrates by various thin film deposition
techniques. High conversion efficiencies are achieved by
suitable deposition processes and material combinations. In
contrast to conventional semiconductor devices, the required
current rectification and the ohmic contacts of the cells are
not achieved by intentional doping of materials but rather by
combining different materials (see sections 4 and 5). This
is a consequence of the low-formation enthalpies of defects
in chalcogenides [5–10], which is related to the ionicity of
the involved compounds. In particular, the dependence of
defect formation enthalpies on the Fermi level position leads
to self-compensation [11, 12]. This prevents the ambipolar
dopability required for the formation of p-n junctions and
degenerate doping levels, which is the conventional approach
for the formation of ohmic contacts [13].
As a consequence of the device structures being based on
a combination of different materials, thin film chalcogenide
solar cells inherently include interfaces between materials
with different energy gaps, lattice constants and partially even
different crystal stuctures. A set of energy gaps and (cubic)
lattice constants for chalcogenide and pnictide compounds is
displayed in figure 1(a). At an interface, the combination of
materials with different energy gaps leads to discontinuities
(called offsets) in the valence band maximum (EVB ) and
conduction band minimum (ECB ), as illustrated in figure 1(b).
Together with the doping of the semiconductors, the
discontinuities EVB and ECB determine the electrostatic Figure 1. (a) Energy gaps and cubic lattice constants for pnictide
potential distribution in the solar cell. The discontinuities (III-V) and chalcogenide (II-VI and I-III-VI2 ) semiconductors; (b)
definition of quantities at semiconductor interfaces: valence band
furthermore form barriers for charge transport at the interface.
maximum EVB , conduction band minimum ECB , vacuum energy
The energy band alignmnent at the various interfaces in a Evac , ionisation potential IP , electron affinity χ, work function φ,
chalcogenide thin film solar cell, which typically include valence band offset EVB , conduction band offset ECB , Schottky
interfaces between two different chalcogenide semiconductors barrier for holes Bp , Schottky barrier for electrons Bn and
and interfaces of such semiconductors with oxides and metals, interface dipole potentials δ.
are therefore of primary importance for the function of the
cell [14]. be found in [15]. The fundamental physics of energy
In this article, the author’s viewpoint on energy band band alignment at semiconductor interfaces had been studied
alignment at interfaces in chalcogenide thin film solar cells is in the last 25 years of the 20th century [16–20]. Until
presented. The article is organized as follows: The underlying then, mostly electron affinities were used to estimate the
physics of energy band alignment intrinsic to a combination difference in conduction band minimum energies, the so-
of materials and a potential origin for a modification of energy called Anderson model, which corresponds to an alignment
band alignment by defects is briefly described in section 2, of the vacuum levels at the interface [21]. According to the
followed by a short introduction to the measurement of energy Anderson model, the energy band alignment should depend
band alignment at semiconductor interfaces in section 3. linearly on the electron affinity of the semiconductor. For
The interfaces in CdTe and Cu(In,Ga)(S,Se)2 thin film solar semiconductor/metal interfaces, the vacuum level alignment,
cells, which form the major part of this contribution, are which is known as the Mott–Schottky model, should lead to a
addressed in sections 4 and 5, respectively. The energy band
linear dependence of barrier height on metal work function.
alignment of new chalcogenide absorber materials, which have
It had been pointed out by Heine in 1965 [22], that the
been recently investigated more intensively to overcome the
coupling of wave functions at an interface can induce electronic
material’s limitation issues, are briefly addressed in section 6.
states in the forbidden energy gap near the interface. These
This paper is concluded with a summary given in section 7.
induced gap states, which are nicely illustrated for a Si/Al
interface using density functional theory (DFT) calculations
2. Energy band alignment at semiconductor by Louie et al [23], have been identified as the origin of
interfaces the Fermi level pinning at interfaces of covalently bonded
semiconductors. The expression Fermi level pinning refers to
2.1. Intrinsic band alignment
the observation that, for many semiconductors barrier heights
The main contributions to energy band alignment are only are almost independent of the metal [24]. As has been worked
briefly reviewed here. A more extensive discussion can out subsequently, the alignment of the energy bands is then
2
Figure 2. Intrinsic energy band alignment of chalcogenide semiconductors discussed in this paper. Semiconductors where insufficient
systematics are available yet are Cu2 O, Cu2 S, Cu2 S, SnS and Bi2 S3 . Alignments for these materials may therefore be affected by pinning
effects as discussed in the text. The influence of cation states in the valence bands is discussed in the text.
described by the alignment of the charge neutrality levels of the for X = S, Se, Te. Such behavior is expressed by the common-
induced gap states [17, 20, 25–27], rather than by an alignment anion-rule [42]. The common-anion-rule is also observed for
of the vacuum energy. Fermi level pinning is particularly a range of oxides [43]. A noticeable exception of this rule
pronounced for covalently bonded materials, in particular for is the alignment between anatase and rutile TiO2 , where the
Si, Ge and GaAs, but less important for more ionically bonded ∼0.7 eV higher valence band maximum of rutile is caused by
materials [28]. Although charge neutrality levels are used to the stronger overlap of the non-bonding O 2pz lone pair orbitals
describe band alignment of ionic semiconductors like oxides compared to the anatase modification [44].
[20, 29–32], dedicated experiments reveal that the pinning is Considerable deviation from the common-anion-rule is
not very pronounced for these materials [28, 33, 34]. This is observed if cation orbitals, like the Cu 3d or Sn 5s, contribute
particularly the case for SiO2 , where the metal work function to the valence band density of states [43, 45–48]. The higher
can be directly used to tune the flat band potential of metal- valence band maximum of CuGaSe2 compared to its binary
oxide-semiconductor diodes in field effect transistors [13]. analogue ZnSe, e.g. is known as p−d repulsion [46]. The con-
The energy band alignment of chalcogenide semi- tribution of Cu 3d states to the valence band maximum can be
conductors has been studied extensively in literature, as shown by energy-dependent photoemission experiments [49].
outlined in detail in this paper. A summary of energy band
alignments obtained from these measurements is given in 2.2. The role of defects on energy band alignment
figure 2, where the valence and conduction band energies are
referenced with respect to the valence band maximum of ZnS As already mentioned, significant deviation from the intrinsic
following Zhang et al [35]. The alignments shown in figure 2 energy band alignment has been observed for a number of
are considered as the intrinsic energy band alignment, i.e. interfaces. Experimental evidence for such modifications
the alignment at structurally-ideal interfaces. The alignment will be discussed in detail for interfaces between transparent
scheme shown in figure 2 does not contain any dependence conducting oxides (TCO) and CdS in section 4.2, for
on interface orientation and/or preparation. The former is interfaces between Cu(In,Ga)Se2 and Zn(O,S) buffer layers
consistent with theoretical expectations [36–38] and with in section 5.3, for In2 S3 /ZnO interfaces in section 5.4, for the
experimental observations [39–41]. But there are a couple CuInS2 /CdS interface in section 5.5, for CdS/ZnO interfaces in
of examples where energy band alignment seems to depend section 5.6 and for interfaces of SnS in section 6, respectively.
on the specific preparation of the interface. Some of them are A modification of energy band alignment with deposition
explicitly described in this paper. In such cases, a modification conditions has furthermore been observed for interfaces
of band alignment from the intrinsic value occurs. A possible between Sn-doped In2 O3 (ITO) and Cu2 O [50] and between
origin for such modifications will be outlined in section 2.2. ITO and Al2 O3 [51].
The energy band alignment shown in figure 2 can be In all cases mentioned in the previous paragraph, the
separated into two different classes. For chalcogenides and modification of the band alignment can be attributed to the
oxides where the valence bands are formed mainly by the limitation of band bending in one or both materials forming
chalcogen p-orbitals, the valence band maximum energy is the contact. A limitation of band bending corresponds to
very similar for materials with the same anion. For example, a restricted movement of the Fermi energy in the energy
the valence band maxima of ZnX is comparable to that of CdX gap of a material. How a restriction of band bending leads
3
Figure 3. (a) Modification of energy band alignment through the absence of band bending [12]. The intrinsic, i.e. the unmodified alignment
is shown on the very left. The valence and conduction band discontinuities are preserved at the interfaces independent on doping if the
bands are allowed to bend. In the absence of band bending, the alignment depends on the Fermi level position in the materials; (b) effect of a
high concentration of deep acceptor traps in a p-type semiconductor on the electrostatic potential at an interface. The shallow dopants
responsible for p-type conductivity are not shown. When the trap level crosses the Fermi energy, they become charged and can develop a
charge density which effectively modifies the band energies at the interface by E.
to a modification of energy band alignment is illustrated in exhibit a binding energy with respect to the valence band
figure 3(a) [12]. Such a restriction can be caused by charged maximum ECL VB
= EB (CL) − EVB , which is characteristic for
crystallographic defects if the evolution of a band bending the given material and can be determined for the substrate and
causes a change of the charge state of the defects [51]. the contact material at the beginning and the end of an interface
In the case of high defect concentrations (1020 cm−3 ), this experiment, respectively.
will lead to a very narrow space charge region, which will The procedure of an interface experiment using PES is
effectively form an interface dipole and thereby modify the illustrated in figure 4. The interface experiment starts with the
band alignment. The mechanism is illustrated in figure 3(b). preparation of a substrate surface and continues by stepwise
deposition of the contact material until the substrate emission
lines are completely attenuated. Using Al Kα radiation as
3. Experimental approach excitation for PES (hν = 1486.6 eV), this corresponds to a
thickness of the contact material of ∼10 nm. To follow the
The procedure to determine energy band alignment using evolution of the chemical and electronic properties, the contact
photoelectron spectroscopy (PES) with x-ray or ultraviolet material has to be deposited in multiple steps, starting with
excitation (XPS, UPS) has been outlined by Waldrop and films of sub-nanometer thickness.
co-workers [52]. It has been extensively used to study The valence band discontinuity is derived from the binding
semiconductor interface formation [17, 18]. Binding energies energy difference of substrate and film core levels ECL ,
in PES are measured with respect to the Fermi energy of which can be measured if the thickness of the deposited film
the spectrometer system and can be easily calibrated by is less than 3–5 times the inelastic mean free path of the
measuring the Fermi edge emission of a metallic sample. For photoelectrons. The latter is a function of the kinetic energy
semiconductor samples, the onset of emission at low binding of the photoelectrons and varies in the range of 0.5–3 nm [53].
energy corresponds to the valence band maximum at a binding In the idealized case, ECL does not depend on the thickness
energy of EF − EVB . The core levels of the semiconductor of the deposited film. However, non-constant ECL may
4
Figure 4. Determination of energy band alignment between two semiconductors by photoelectron spectroscopy following the Kraut
method [52]. Experimentally determined quantities in a photoemission experiment are indicated by asterisks.
occur in the case of strong band bending or if the electronic obtained with a substrate configuration [59]. Between
structure of the deposited film is not yet fully developed, the TCO contact and CdTe layer, a CdS buffer layer is
e.g. if the crystallographic structure of the film depends on typically inserted. After deposition of the CdTe layer the
film thickness [54–56]. For a reliable determination of band so-called activation treatment is performed, in which the
alignment, a series of depositions with stepwise increasing partially completed cell structure is annealed in a chlorine-
film thickness is therefore recommended. If several core contained atmosphere. For example, the sample is dipped into
level lines are available, it is important to check if the ECL a CdCl2 /methanol solution and afterwards heated to ∼400◦ C
of different substrate and film core levels reveal the same for ∼20 min [58, 60–62]. The cell is completed by back contact
values. Valence band discontinuities can further be determined formation, which is one of the crucial issues in thin film CdTe
from photoelectron spectroscopy by valence band difference solar cells [58, 63–66].
spectra [16, 57], although this procedure is less reliable and The performance of the CdTe solar cell is largely
less frequently applied. insensitive to the deposition techniques of the various
layers. CdS and CdTe can be grown by a variety of
techniques like thermal evaporation from the compounds,
4. The CdTe solar cell
close-spaced sublimation (CSS), vapor transport deposition,
4.1. The cell structure
electrodeposition, magnetron sputtering, chemical vapor
deposition, spray pyrolysis and screen printing [58]. CdS can
The CdTe solar cell (figure 5) is typically grown in a superstrate furthermore be deposited by chemical bath deposition [62, 69],
configuration, i.e. starting with the transparent front contact which is, however, not fast enough to grow the CdTe layer with
electrode [58]. Recently, high efficiencies have also been the required thickness of 2 µm within a reasonable time.
5
Figure 5. Schematic layout of a CdTe thin film solar cell, typical electron microscope cross-section and schematic energy band
alignment [67, 68].
The first interface, which is important in the CdTe thin film with those obtained using deposition experiments and indicate
solar cell, is the TCO/CdS front contact, which is related to no significant modification by post-deposition treatments [76].
the buffer layer/TCO contact in the Cu(In,Ga)(S,Se)2 thin film A peculiarity of evaporated CdS films is that the Fermi level
solar cell grown in the reverse sequence. A key interface for the is found in a restricted range of 1.8 eV EF − EVB
cell is the one between the CdS buffer layer and CdTe. It will 2.2 eV. This restriction is likely caused by the particular defect
be shown in section 4.3 that the energy band alignment at this concentrations in such films, albeit that the known defect
interface is quite robustly characterized by a small conduction calculations [77, 78] do not clearly identify which defects
band offset. It is rather the Fermi level position at the interface, are responsible for this restriction. The Fermi levels in the
which seems to be important for the cell efficiency. As already TCO materials exhibit an upper (doping) limit due to self-
mentioned, the back contact is a crucial part of the CdTe solar compensation by formation of intrinsic defects in ZnO (VZn )
cell having attracted significant attention in literature. The and In2 O3 (Oi ) [11, 29, 79, 80] or by ambipolar dopants in SnO2
topic will be dealt with in section 4.5. (interstitial fluorine) [80].
The valence band offsets at the ZnO/CdS (see also
4.2. The TCO/CdS front contact section 5.6) and at the SnO2 /CdS interfaces are very similar
with EVB = 1.4–1.5 eV. This indicates a comparable energy
A broad variety of TCO materials including fluorine-doped of the valence band maximum in both oxides. This can
SnO2 (FTO), Sn-doped In2 O3 (ITO) and CdSnO4 have been be expected from the common-anion-rule [42], which is
investigated as transparent front contacts in CdTe thin film solar closely fulfilled by oxides where only O 2p states contribute
cells with comparable results [70–72]. Very often, a double to the valence band density of states [15, 43]. One therefore
layer structure with a highly-doped conductive electrode would expect a similar valence band offset at the In2 O3 /CdS
and a nominally-undoped secondary layer (mostly SnO2 ) is interface, in particular as EVB = 0.0 eV at the SnO2 /In2 O2
used. Possible roles of the secondary layer, which increase interface [43]. In contrast, the valence band offset determined
efficiency and/or reproducibility, are to prevent diffusion of experimentally is only EVB = 0.8 eV (see figure 6).
atoms from the underlying highly-doped substrate [73], or the As illustrated in the graph on the far right in figure 6
insertion of an additional series resistance to reduce weak diode this deviation can be explained by the Fermi level limitations
effects [74]. The doping of the layer may, however, influence mentioned above. Assuming a valence band offset of
the energy band alignment as demonstrated in the following. EVB = 1.5 eV as expected from transitivity and the common-
A series of valence band offsets obtained from anion-rule alignment, the Fermi level cannot be placed at
photoemission experiments is illustrated in figure 6 [75]. The a value consistent with both limitations for In2 O3 and CdS.
experiments were performed with magnetron sputtered TCO This explanation is consistent with the observation of a higher
films and evaporated CdS films. Energy band alignments valence band offset at the ITO/CdS interface, as higher Fermi
obtained from sputter depth profiling are largely consistent levels are possible in ITO compared to In2 O3 .
6
Figure 6. Energy band alignment at TCO/CdS interfaces as determined from photoelectron spectroscopy using stepwise deposition
experiments and sputter depth profiles. All values are given in electronvolts. The first four graphs show experimental results for ZnO/CdS,
SnO2 /CdS, ITO/CdS and In2 O3 /CdS, whereas the last illustrates why it is not possible to find a Fermi level position consistent with the
possible Fermi level range in In2 O3 and the 1.8–2.2 eV range found for CdS. Consequently, a local dipole occurs at the interface, shifting the
energy bands of In2 O3 upwards as shown in the experimental In2 O3 /CdS band diagram [75].
The modification of energy band alignment by a restricted Fritsche et al performed further studies using laboratory-
range of Fermi levels is not important for the SnO2 /CdS and the based experimental setups with monochromatic Al Kα
ZnO/CdS interfaces. This is related to the larger energy gaps radiation and in-situ sample deposition [89]. As in the real
of ZnO (Eg = 3.3 eV [81]) and SnO2 (Eg = 3.5 eV [82]) solar cell, the prepared films were polycrystalline in nature.
compared to In2 O3 (Eg = 2.8 eV [83]) in conjunction The interface formation has been studied both during stepwise
with similar carrier concentrations in all three TCOs. The deposition of CdTe onto CdS and during subsequent sputter
higher possible values for EF − EVB have been demonstrated depth profiling of the same sample. Both procedures provide
experimentally [12, 75, 84]. not only very similar valence band offsets of EVB = 0.94 ±
In the case of highly-doped FTO layers, the valence band 0.05 eV, but similar Fermi level positions at the interface. The
offset with CdS has been determined to be EVB = 1.7 eV valence band offset has later been slightly revised based on a
[85], which is larger than the value measured for the lower larger number of studied interfaces to EVB = 1.0 eV [90].
doping of SnO2 . In this case, the upper limit of the Fermi level The agreement of the interface properties obtained from
in CdS causes the modification of the band alignment. As the sputter depth profiling is a unique feature of this interface,
modification of the band alignment is always associated with a which is likely related to the absence of preferential sputtering
high defect concentration at the interfaces, it should be avoided effects as a consequence of the similar atomic masses of
in a solar cell, to reduce interface recombination. Cd and Te. A particular advantage of this feature is the
possibility for studying the effect of post-deposition treatments
4.3. The CdS/CdTe interface on the interface properties. The most important treatment is
the activation. The influence of activation on energy band
The first study of energy band alignment at the CdTe/CdS alignment at CdS/CdTe interfaces has been studied in-situ
interface was performed by Niles and Höchst using using a CdCl2 filled effusion cell where the sample has been
synchrotron radiation photoelectron spectroscopy back in 1990 used as top cover [91, 92]. The efficiency of the treatment and
[86]. The interface was prepared by cleaving a CdTe single optimization of the treatment parameters have been validated
crystal along the (1 1 0) cleavage plane and in-situ deposition by cell efficiency measurements [92]. Sputter depth profiling
of CdS by thermal evaporation. The experiments revealed a of as-deposited samples, samples annealed in 5 · 10−5 mbar O2
valence band offset of EVB = 0.65 eV. Later on, Fritsche and activated samples all reveal identical valence band offsets
et al studied interfaces using the same growth sequence for of EVB = 1.0 eV [90]. While annealing in oxygen seems
differently-oriented CdTe single crystals [87]. They suggested to have no effect on the interface properties, the activation
a valence band offset of ∼1 eV for all investigated orientations. treatment leads to a broadening of the chemical composition
Unfortunately, neither the study of Niles and Höchst nor that profile at the interface, indicating interdiffusion. Interdiffusion
of Fritsche et al included any sulfur core levels due to the low at the CdS/CdTe interface has already been suggested by
photon energies at the used synchrotron beamlines. McCandless et al as being one major consequence of the
The orientation dependence of the energy band alignment activation treatment [60, 93, 94]. The photoemission study
at the CdS/CdTe interface has subsequently been studied of Fritsche et al further revealed that the Fermi level at the
by Siepchen et al using high energy synchrotron radiation, interface is shifted by ∼0.2 eV towards the conduction band
which included Cd and S core levels [40]. CdS single minimum after activation. This could be explained either by
crystals with (0 0 0 1) and (1 0 1̄ 0) orientation and CdTe single passivation of interface defects or by an enhanced donor doping
crystals with (1 1 1) and (1 1 0) orientation have been used of the CdS film by chlorine. The observations are summarized
as substrates. Also different substrate temperatures were in figure 7.
used during growth. Despite a complex nucleation behavior Except for the first study of the band alignment by Niles
involving several intermediate species [40, 88], the valence and Höchst [86], all photoemission experiments consistently
band offsets for all studied interfaces vary only in the range of show a valence band offset at the CdS/CdTe interface of
1.0–1.05 eV [40]. EVB = 1.0±0.05 eV. This corresponds to a conduction band
7
Figure 7. Energy band diagrams of the CdS/CdTe interface before and after activation treatment as obtained using photoelectron
spectroscopy and sputter depth profiling [90].
offset of about ECB ≈ −0.1 eV (the conduction band of CdS in CdS, while the ECB of ZnS is 1.14 eV higher than that
is at a lower energy than that of CdTe), which is ideally suited of CdS (see figure 2). Substituting Cd by Zn will therefore
for high conversion efficiencies. Such a conduction band offset mainly raise the conduction band minimum and leaves the
is consistent with that derived by Nishi et al using internal valence band maximum mostly at the same energy. This
photoemission [95]. The band alignment is in very good is a consequence of the same anion in CdS and ZnS. The
agreement with supercell calculations using density functional conduction band offset will therefore change from the small
theory [6]. So far there is no evidence that the alignment cliff of ECB = −0.1 eV at the CdS/CdTe interface to a
depends on interfacial orientation and growth sequence despite spike ECB > 0 for (Cd,Zn)S/CdTe. As evident from
significant differences in nucleation behavior [40, 88]. The Cu(In,Ga)(S,Se)2 solar cells, a small spike (ECB 0.3 eV)
insensitivity of band alignment on interface orientation is does still allow for high efficiencies [14, 98]. This provides
in very good agreement with the bond polarity model of a range for increasing the buffer layer band gap of at least
Lambrecht and co-workers for interfaces between heterovalent 0.4 eV, in good agreement with results from CdTe solar cells
semiconductors [36–38]. Moreover, the activation treatment using (Cd,Zn)S buffer layers [99, 100]. The amount of Zn
does not modify the band alignment. which has to be added to obtain such an increase can hardly
be predicted due to strong optical bowing effects on the band
4.4. Other buffer layer materials gaps of mixed chalcogenide semiconductors [6, 101–103].
Systematic energy band alignment studies at (Cd,Zn)S/CdTe
Electron hole pairs generated in the CdS layer do not interfaces are also not available. While (Cd,Zn)S buffer layers
contribute to the photocurrent, which is evident from the are a reasonable approach for improving the photocurrent of
characteristic reduction of quantum efficiency for wavelengths CdTe solar cells, the only moderate increase of buffer layer
550 nm [58]. It is therefore desirable to either reduce the energy gap consistent with low conduction band offsets can
CdS film thickness or to use other buffer layer materials hardly explain the high quantum efficiency at low wavelengths
with a larger energy gap. A similar issue arises for the of the best cells [97].
Cu(In,Ga)(S,Se)2 solar cell system (see section 5). The A buffer layer material, which has been studied
probability of pinholes increases, however, for lower CdS film extensively for Cu(In,Ga)(S,Se)2 solar cells is Zn(O,S) (see
thickness. This can lead to direct contacts between the TCO section 5.3). According to the calculated energy band
front electrode and the CdTe, which leads to weak diodes and alignments by Zhang et al, the conduction band of ZnO is
reduced efficiency [68]. Wu et al have overcome this problem 0.4 eV lower than that of CdTe [35] (see figure 2). This offers
by using a CdSnO4 /ZnSnO4 front electrode. The CdS layer is a potential for raising the conduction band minimum by up
largely consumed by additional intermixing with the ZnSnO4 to ∼0.7 eV for Zn(O,S) solid solutions. However, mixing
layer [96]. up to ∼50% ZnS to ZnO leaves ECB almost unchanged but
Latest record cells reported by General Electric and First raises EVB to the value of pure ZnS [102]. Only further
Solar exhibit significantly-enhanced photocurrent densities substitution of oxygen by sulfur raises the conduction band
and external quantum efficiencies >90% for wavelengths down minimum. As EVB is at the same energy as that of ZnS for such
to ∼350 nm [97]. Such properties are likely obtained using compositions, the maximum Zn(O,S) band gap compatible
buffer layers with higher energy gaps. An increase in the with ECB 0.3 eV should be of the same magnitude as
energy gap might be achieved by alloying CdS with ZnS that of (Zn,Cd)S.
(Eg = 3.6 eV [81].) Following the calculations of Wei Another buffer layer material successfully applied in CdTe
et al, the EVB of ZnS is only 0.18 eV lower than the EVB solar cells is Cd(O,S), which can be prepared by reactive
8
Figure 8. Cd(O,S) films deposited on glass substrates by reactive magnetron sputtering with a varying amount of oxygen in the sputter gas:
(top) photographic images; (middle) transmission spectra; (bottom) XPS core level spectra.
magnetron sputtering from CdS targets [71, 104, 105]. By which led to transparent films. Chemical phase analysis using
addition of only a few percent of oxygen to the sputter gas, XPS indicates that the as-deposited films are approximately
the films become almost completely transparent as evident composed of 60% CdSO4 , 20% CdO, 12% CdS and 8%
from the photographs and the transmission spectra depicted in CdSO3 [85]. The composition of the Cd(O,S) layers has been
figure 8 (see [71, 104–106]). In contrast to Zn(O,S) [103, 107], assessed after a temperature treatment, which is comparable to
as-deposited Cd(O,S) films grown by reactive sputtering in that during CSS deposition of the CdTe layer as well as after
Ar/O2 atmosphere from CdS targets do not form a solid heat treatment and after activation with a CdTe cover layer.
solution of CdS and CdO but rather a mixture of CdS, CdO, Basically, the CdS and CdO fractions increase after treatment,
CdSO3 and CdSO4 . This is evident from a combined analysis while that of CdSO3 disappears and that of CdSO4 is slightly
of core level spectra shown in figure 8. The existence of a phase reduced but still dominates the film composition [85]. For
mixture is an advantage compared to a CdS-CdO solid solution O2 /Ar ratios >5%, pure CdSO4 films were obtained, resulting
as the direct band gap of CdO is not larger than that of CdS in insulating properties of the respective films and poor solar
[81]. The solid solution would thus provide no optical benefit. cell performance with such buffer layers.
Moreover, CdO has a very small indirect energy gap [81], The standard one-dimensional description of the energy
which should lead to considerable distortion of the electrical band diagrams of thin film solar cells is inadequate for
properties of the cell. the understanding of the interface properties in the case of
In our laboratory, the best CdTe solar cells were prepared multiphase buffer layers such as Cd(O,S). The situation of
using Cd(O,S) films grown with 4–5% O2 in the sputter gas, Cd(O,S) is particularly intriguing as the layer mainly consists
9
Figure 9. Suggested composition and contact formation of Cd(O,S) films prepared by reactive magnetron sputtering from a CdS target [85].
Electrical conductivity is maintained via conductive paths including CdS and CdO phases embedded in an insulating CdSO4 matrix.
Extraction of electrons from the CdTe layer is achieved only via the conductive CdS and CdO phases, while CdSO4 /CdTe interfaces lead to a
good surface passivation of the CdTe layer.
of CdSO4 , an electrically-insulating material. The electrical interface of B ≈ 0.55 eV [116], which still causes non-ideal
contact between Cd(O,S) and CdTe may therefore be similar cell behavior as evident from the characteristic rollover of cells
to that present in Si solar cells with point contacts. In that case, prepared with such contacts [64, 108].
CdSO4 may serve as a more or less ideal surface passivation for Another disadvantage of the Te contact is the absence of a
CdTe while electrical contact to CdTe and transport through barrier in the conduction band, which is generally favorable
the buffer layer is established by the CdS and CdO fractions as it can repel electrons [68, 128]. For such a barrier an
of Cd(O,S). The situation is sketched in figure 9. intermediate layer with a larger energy than that of CdTe is
required. ZnTe with an energy gap of ∼2.3 eV [81] is an ideal
4.5. The back contact candidate for such a contact. The valence band offset at the
CdTe/ZnTe interface is ∼0 eV as determined independently
The back contact of the CdTe solar cell has been intensively by Rioux et al [117] and by Späth et al [118, 119]. In
investigated, as good contact properties are difficult to achieve. addition, ZnTe can be doped highly p-type using Cu, N or
Very often a back contact barrier exists, which causes low fill
Sb [119, 129–131]. The CdTe/ZnTe:N system is the only
factors and the so-called rollover, which is a flattening of the
one in which the Fermi level at the CdTe/ZnTe interface has
current-voltage curve for U > Voc [14, 58, 63, 64, 108]. Due
been found close to the valence band maximum [119]. A
to the large ionisation potential of CdTe of ∼5.7 eV [109], it is
disadvantage of the ZnTe layer limiting its applicability is the
typically suggested to use high work function metals in order
incompatibility of the temperature treatments required for the
to obtain a good ohmic contact to a p-type semiconductor.
activation of the cell and for the doping of ZnTe. Highly-doped
However, deposition of metals onto CdTe typically leads to an
ZnTe layers can only be prepared at temperatures ∼400◦ C
interfacial reaction, which causes the Fermi energy in CdTe
[119, 131], which leads to an over-activation of the cells as the
at the interface to rise to EF − EVB ≈ 1 eV [110–113]. This
activation treatment is typically performed before back contact
is a very unfavourable situation for the cell. Growth of CdTe
formation.
on other chemically inert high work function materials such as
VSe2 and TiSe2 results in a similar Fermi level position [114]. The valence band discontinuity at the CdTe/Cu2 Te
The high Fermi level position has been associated with defects interface has been determined to be EVB = 0.8 eV [126].
induced by the reaction at the interface [110, 111]. Interstitial The higher valence band maximum of Cu2 Te is explained by
Cd is a likely candidate for the pinning of the surface/interface the p −d repulsion effect, i.e. the hybridization of Cu 3d states
Fermi level in the upper half of the band gap [7]. This defect with Te 5p states. Despite being a highly p-doped material,
is likely as the reactions result in the bonding of Te to the Cu2 Te itself is therefore not suitable for making good ohmic
deposited metal and the release of Cd [111–113]. contacts to CdTe.
In order to avoid the back contact barrier formation by the The best efficiencies of CdTe solar cells are obtained
interface reaction, various intermediate layers have been tried. with Cu containing back contacts [59, 62], which is ascribed
Among them are various tellurides such as Te [115, 116], to diffusion of Cu into the CdTe. Such contacts have been
ZnTe [117–119], Sb2 Te3 [120–123] and Cu2 Te [124–126]. realized using doped graphite paste [70, 132, 133], ZnTe [130],
Tellurium is a small gap semiconductor (Eg = 0.33 eV [81]), HgTe [62] and Cu2 Te [124]. It is widely accepted that Cu
which can easily be evaporated. A Te layer is formed on top of diffusion in CdTe leads to p-type doping. Perrenoud et al
the CdTe layer by the etching of activated samples [116, 127]. have recently shown that most of the Cu introduced into CdTe
The thickness of the Te layer, which forms by etching with is found at grain boundaries and that the concentration of
a nitric-phosphorous acid (NP) etch, is about 20 nm. The Cu in the CdTe grains of <1015 cm−3 is only ∼0.1% of the
etching procedure removes the oxide on the CdTe surface and total Cu concentration in the films [134]. While the low
the thickness of the Te layer is sufficient to avoid the reaction Cu concentration in the bulk can explain the typical acceptor
of the deposited metals with CdTe and thereby the formation of concentrations and space charge layer widths found in CdTe
interstitial Cd. However, there is still a barrier at the CdTe/Te thin film solar cells [65, 68, 122, 135], such a low doping level
10
Figure 10. Schematic layout of a Cu(In,Ga)(S,Se)2 (CIGS) thin film solar cell (top left) and a corresponding cross-section from an electron
microscope (right). The schematic energy band diagram (bottom left) of the cell shows the energetic situation in the front contact region
without a Cu-poor surface layer.
does not explain the beneficial effect of Cu on the back contact β-phase [141]. The β-phase can be understood by an ordered
formation as it is orders of magnitude too low for forming an arrangement of neutral (2VCu InCu )x defect pairs [142]. This is
ohmic tunneling contact [136]. why such CuIn3 Se5 (135) or CuIn5 Se8 (158) phases are often
The results obtained by Perrenoud et al led to the spec- called ordered vacancy compounds (OVC). After deposition,
ulation that the ohmic contact using Cu is not homogeneous the surface of the Cu(In,Ga)Se2 absorbers exhibit an OVC
but similar to the suggested point contact for Cd(O,S). Elec- composition and, in contrast to the p-type doping of the bulk
trical contact is probably formed at the grain boundaries only, of the films, an n-type surface [143, 144]. The thickness
where most of the Cu is found in CdTe. This would explain the of the Cu-poor surface layer is, however, still under debate
high barriers observed by XPS for almost any contactmaterial [145–147].
and processing, even if electrical measurements indicate that The front contact of the cell is made up of a combination
transport is not limited by high barriers. The discrepancy in of various different layers. A degenerately-doped transparent
barrier heights between XPS and electrical measurements orig- conductive oxide, mostly doped ZnO, serves as the current
inates from the fact that the current is dominated by areas with collector. Various buffer layers are inserted between the
low barriers [137], potentially at grain boundaries, while XPS Cu(In,Ga)(S,Se)2 and the current collector. The most
measures the total area and the signal is therefore dominated widely used is CdS, which is deposited by a chemical bath
by the grain surfaces [138]. and reproducibly gives high efficiencies [148, 149]. Due
to its small band gap of only 2.4 eV, which limits the
photocurrent, and the toxicity of Cd, intensive research has
5. The Cu(In,Ga)(S,Se)2 solar cell been devoted to the development of alternative buffer layer
materials [148, 149]. Today Zn(O,S) is widely used as buffer
5.1. The cell structure layer (see [150] and references therein) with comparable
The typical structure of a solar cell with Cu(In,Ga)(S,Se)2 efficiencies [2, 151].
as light absorber is shown in figure 10 [14]. The standard
substrate for making cells and modules is soda-lime glass. 5.2. The Cu(In,Ga)Se2 /CdS interface
Diffusion of Na from the substrate during cell deposition is The energy band alignment at the Cu(In,Ga)(S,Se)2 /CdS
beneficial for cell efficiencies. Very recently, efficiencies interfaces for varying absorber compositions has been
above 20% have been achieved with flexible polyimide calculated by the group of A. Zunger using density-
substrates using a KF post-deposition treatment [139]. The functional theory [5, 35, 101, 152]. Within the Cu(In,Ga)Se2
standard back contact material is Mo. During deposition or composition range, no strong dependence of the valence band
formation of the Cu(In,Ga)(S,Se)2 layer, a MoSe2 or MoS2 offset EVB on the Ga content is expected, because the
layer is formed at the surface of the Mo film (see [140] and introduction of Ga into CuInSe2 mainly shifts the position
references therein), resulting in a rather ideal ohmic contact. of the conduction band to higher values [35]. Experimental
The Cu(In,Ga)(S,Se)2 absorber layer is intrinsically observations are consistent with this expectation [153].
p-doped. This can be ascribed to the presence of Cu-vacancies, Compared to chalcogenide semiconductors without Cu (or
which are shallow acceptors in these materials [5, 9]. For Ag), the valence band maximum of Cu(In,Ga)Se2 compounds
Cu(In,Ga)Se2 the deposition is generally made under Cu-poor is approximately 1 eV higher in energy [154], as the
conditions, resulting in an overall Cu-deficiency of a few hybridization of the chalcogen p-orbitals with the metal
% compared to the stoichiometric chalcopyrite composition d-orbitals leads to an upward shift of the valence band
and a mixture of the chalcopyrite α-phase and a Cu-poor maximum, known as p-d repulsion [46].
11
Figure 11. Evolution of the Cu/Ga intensity ratio (top left) and of the valence band maximum position (top right) of a Cu-poor and a
stoichiometric Cu(In,Ga)Se2 film in the course of CdS deposition as obtained from XPS measurements (after [111]). During the CdS
deposition, the compositions and valence band maxima of the Cu-poor (135) and the stoichiometric (112) surface approach each other. This
is related to the mobility of Cu as illustrated in the lower part of the figure. The dashed lines in the top left graph indicate the expected
dependence on film thickness due to the different surface sensitivity of the Cu 2p and the Ga 2p photoemission core levels.
Experimental determinations of the valence band offset has been attributed to the fact that the Cu content near the
at the interface between CuInSe2 and CdS obtain values of surface may vary in the course of film deposition. A variation
EVB = 0.8–1.06 eV [143, 144, 150, 153, 155–161], which is of the Cu content at the surface has also been observed during
mostly lower than the calculated offset of 1.07 eV [101, 152]. deposition of CdS and Na onto vacuum-cleaved CuInSe2 single
Due to the lower amount of Cu and the reduced p-d repulsion, crystals [163, 164].
the valence band maximum of the OVC is lower than that of The change of Cu content is possible due to the high
the chalcopyrite (112)-phase. Correspondingly, the valence mobility of Cu in Cu(In,Ga)Se2 compounds [9, 10, 141, 165].
band offset between the CuIn5 Se8 (158) ordered vacancy It has therefore been suggested that the resulting Cu content at
compound and a CdS buffer layer is calculated to be 0.42 eV the Cu(In,Ga)Se2 /CdS interface is independent on the starting
lower than that between Cu(In,Ga)Se2 and CdS [101], and that condition. CdS deposition then results in a Cu content in
between CuIn3 Se5 (135) and CdS is 0.34 eV smaller than that between the stoichiometric Cu(In,Ga)Se2 and the Cu-poor
between Cu(In,Ga)Se2 and CdS [5]. The lower offset found Cu(In,Ga)5 Se8 composition, which explains the valence band
experimentally for CuInSe2 /CdS interfaces compared to the offset [111]. It is mentioned that the band gap is affected
theoretical calculations may thus be related to the presence of by the Cu-content and will therefore depend on the final
an OVC between the chalcopyrite and CdS. composition [155].
Following the suggestion that the interface composition
Schulmeyer et al have explicitly studied the energy
is more or less independent of the starting composition, it is
band alignment at the interfaces of CdS with Cu-poor and
interesting to have a more detailed look at the dependence of
stoichiometric Cu(In,Ga)Se2 [111]. The Cu-poor surface was
the energy band alignment at Cu(In,Ga)Se2 /CdS interfaces on
prepared by de-capping of a Cu(In,Ga)Se2 layer, which was Ga content. From DFT calculations no strong dependence
capped by a Se layer directly after deposition in order to avoid is expected [35]. Valence band offsets measured by
oxidation of the substrate surface [144]. The stoichiometric photoemission interface experiments using single crystalline
surface was prepared by etching a Cu(In,Ga)Se2 layer grown and polycrystalline substrates for 4 different In/Ga ratios are
under Cu-rich conditions using KCN in order to remove the displayed in figure 12. There is a clear trend towards a
Cu2 Se layer segregated at the surface [162]. The Cu-content higher EVB with increasing Ga content, suggesting that
and the Fermi level position were distinctly different before the CuInSe2 /CdS interface has a tendency for a Cu-poor
CdS deposition but approached each other in the course of CdS surface layer while the CuGaSe2 /CdS interface tends to be
deposition as illustrated in figure 11 [111]. This observation stoichiometric at the interface.
12
interpretation of the band alignment data given above. The

low temperature (flatband) valence band maximum binding
energy of the CuIn0.7 Ga0.3 Se2 sample is not modified by CdS
deposition, which is consistent with an OVC at this interface.
Currently, it is neither clear why the CuInSe2 /CdS
interface tends to be Cu-poor while the CuGaSe2 /CdS interface
prefers to be stoichiometric, nor if this observation is unique
for thermally-evaporated CdS films or present for CdS films
deposited by other techniques like chemical bath deposition.
Concerning the former question, a possible explanation
could be related to different chemical reactivities of the
chalcopyrites with CdS. Considering chemical reactivities
requires knowledge of the formation enthalpies of the
chalcopyrites and of the ordered vacancy compounds from
comparable sources in dependence on Ga composition. Such
data are not available.
5.3. Other II-VI buffer layers

Figure 12. Valence band offsets at Cu(In,Ga)Se2 /CdS interfaces in
dependence on the Ga content as obtained from photoelectron Schulmeyer et al have studied the band alignment between
spectroscopy employing vacuum-cleaved CuInSe2 and CuGaSe2 Cu(In,Ga)Se2 and all binary II-VI compounds [167, 171]. All
single crystals [160, 166], a de-capped epitaxial CuInSe2 film [144]
and de-capped polycrystalline Cu(In,Ga)Se2 thin films [153, 167]. valence band offsets are consistent with values expected from
The solid lines correspond to expected valence band offsets for DFT calculations by Zhang et al [35] and illustrated in figure 2.
Cu(In,Ga)Se2 /CdS interfaces as anticipated from calculated offsets Together with other band alignments summarized in this
for CuInSe2 /CdS [152], CuIn3 Se5 /CdS [5] and work, there remains little doubt that the intrinsic energy band
CuInSe2 /CuGaSe2 [101], neglecting any bowing effects and
alignment between various Cu(In,Ga)(S,Se)2 absorbers and
assuming that the offset between CuGaSe2 and CuGa3 Se5 is
identical to that between CuInSe2 and CuIn3 Se5 . The dashed line II-VI based buffer layer materials follow the general picture
indicates the trend of experimental offsets. given by Zhang et al [35]. This picture is further confirmed
The interpretation of the band alignment data in figure 12 by the alignment reported for the CuGaSe2 /ZnSe interface
in terms of different Cu content of the Cu(In,Ga)Se2 layer by Bauknecht et al (EVB = 0.6 eV) [154]. Pettenkofer
near the interface is supported by temperature-dependent and co-workers studied in more detail the dependence of the
measurements of clean de-capped Cu(In,Ga)Se2 surfaces with valence band offset at the CuInSe2 /ZnSe interface and reported
and without a thin CdS layer. Corresponding measurements comparable offsets of EVB = 0.7 eV for epitaxially grown
are shown in figure 13. In case there is any bending of energy (0 0 1) and (1 1 2) oriented interfaces [39], which are also in
bands at the surface of the measured absorber, the energy bands line with the other studies.
will progressively become flatter during the photoemission Among the alternative buffer layer materials applied in
measurement when the sample is cooled [168]. This is a Cu(In,Ga)(S,Se)2 solar cells Zn(O,S), (Zn,Mg)O and In2 S3
consequence of the photovoltage, which is induced by the have been more intensively investigated [148, 149]. Zn(O,S)
electron-hole pairs generated by the photon source during is a mixture of ZnO (Eg = 3.2 eV [81]) and ZnS (Eg = 3.6 eV
measurements [169] and the temperature dependence of the [81]), which are immiscible at elevated temperature but can
photovoltage. The position of the valence band maximum at be prepared, e.g. by magnetron sputtering with crystalline
low substrate temperature is therefore a measure of the Fermi (wurtzite) structure at lower substrate temperature [103, 107].
level position under flatband condition. The energy gap of ZnO1−x Sx exhibits a very strong bowing
Clean sample surfaces were prepared using the de-capping and a minimum energy gap of ∼2.6 eV at x ≈ 0.45 [103].
procedure resulting in Cu-poor surface compositions due to Experimental valence band offsets for Cu(In,Ga)Se2 /-
the Cu-deficient growth conditions [153]. The valence band ZnO have been reported by Platzer-Björkman et al (EVB =
maximum binding energies of the uncovered samples at low 2.26 eV) [172] and by Klein et al (EVB = 2.15 eV) [55].
temperature are ∼0.5 eV below the Fermi energy. This is too The values are close to each other and to the calculation
low for the typical p-type nature of these films but can be for stoichiometric CuInSe2 (EVB = 2.23 eV) [35].
understood by the presence of an OVC phase at the surface Considerable variation of experimental valence band offsets
with its lower valence band maximum energy (see bottom left of EVB = 0.6–1.2 eV is reported for Cu(In,Ga)Se2 /ZnS
graph in figure 13). interfaces [143, 167, 172–174]. Except for the smallest value
The flatband situation changes after deposition of a thin of 0.6 eV reported by Schmid et al [143], the variation
CdS layer onto the CuGaSe2 sample. Now the valence band is, however, within the range calculated for CuInSe2 /ZnS
maximum is found 0.26 eV below the Fermi energy. Such a (1.23 eV) and CuIn5 Se8 /ZnS (0.81 eV) [35, 101]. It may
binding energy is no more consistent with the presence of an therefore be explained by different Cu contents near the
OVC (see bottom right graph in figure 13). Apparently, the interface, which could be related to different preparation
OVC disappeared with CdS deposition, in agreement with the techniques of ZnS as thermal evaporation [143, 167, 173],
13
Figure 13. (top) Temperature dependence of the valence band maximum position at the surfaces of polycrystalline de-capped
CuIn0.7 Ga0.3 Se2 and CuGaSe2 before (filled symbols) and after (open symbols) deposition of a thin CdS layer [170]. (bottom) Sketch of the
energetic situations at low temperatures for the uncovered and the CdS covered surfaces. Flatband situation is assumed at low temperature.
magnetron sputtering [174] and atomic layer layer deposition the CuInS2 /CdS interfaces (see section 5.5) and at CdS/TCO
[172] have been used in these experiments. interfaces (see section 5.6) [150].
The valence band offsets at the Cu(In,Ga)Se2 /Zn(O,S)
interfaces depend strongly on the O/S ratio due to the 5.4. In2 S3 buffer layers
pronounced dependence of the Zn(O,S) band edge energies
on the composition, which has been calculated by Persson In2 S3 buffer layers can be prepared by a variety of techniques
et al [102] and confirmed experimentally by atomic layer and have resulted in solar cells with efficiencies >15%
deposited Zn(O,S) layers by Platzer-Björkman et al [172] and [149, 177, 178]. The optically-determined energy gap of In2 S3
for magnetron sputtered Zn(O,S) layers by Kieven et al [174]. varies between 1.9–2.8 eV [179–182]. The differences in band
gap are probably related to different phases and/or In/S ratios.
In contrast to chemical bath deposited (CBD) CdS
Thin films prepared by evaporation of In2 S3 powder show
layers, which are close to CdS stoichiometry, CBD
a strong dependence of In/S ratio on substrate temperature
deposited ‘ZnS’ layers are typically a ZnO1−x Sx mixture
during deposition, while post-deposition heat treatments of
with x = 0.5–0.7 [150, 173, 175]. The significantly different
films deposited onto ZnO do not change the composition (see
oxygen incorporation in CBD-CdS and CBD-‘ZnS’ has been
figure 15). The changes in composition are accompanied
attributed by Hubert et al to different solubilities [176]. The
by reproducible changes of the Fermi level position with
considerable amount of ZnO and even Zn(OH)2 in CBD-‘ZnS’ respect to the valence band maximum of thermally-evaporated
films during chemical bath deposition is related to the small In2 S3 films. It has been demonstrated that these Fermi level
solubility differences of these compounds. In contrast, CdS positions are almost unchanged during interface formation
and CdO have strongly different solubilities, which explain with ZnO [55, 111]. In dependence on the doping of the ZnO
the lower oxide and hydroxide incorporation in CBD-CdS layers, the valence band offset can therefore vary by >1 eV.
compared to CBD-‘ZnS’. The pinning effects described in section 2.2 are therefore
The EVB of chemical bath deposited Zn(O,S) layers particularly pronounced for thermally-evaporated In2 S3 films.
with Cu(In,Ga)Se2 substrates only approximately follow the While the chemistry of the interface with Cu(In,Ga)(S,Se)2
calculated dependence on O/S ratio and deviate by up to absorbers is rather complicated and significant reactions have
∼0.5 eV from theoretical calculations and the offsets obtained been reported by a number of groups [178, 184–187], measure-
with ALD and sputter deposited Zn(O,S) as illustrated in ments of the band line-up at Cu(In,Ga)(S,Se)2 /In2 S3 interfaces
figure 14 [150]. Such a process-dependent valence band are scarce. Schulmeyer et al have studied the alignment be-
offset is not observed for the Cu(In,Ga)Se2 /CdS interface (see tween thermally-evaporated In2 S3 onto de-capped CuGaSe2 at
section 5.2) and not for the CdS/CdTe interface (section 4) and room temperature. For this combination the substrate and film
is therefore considered to be caused by extrinsic contributions. have different cations, thus the difficulty of analyzing spectra
It has been suggested that the deviation, which cannot be of the overlapping Se 3p/S 2p and Se 3s/S 2s core levels can
explained by a variation in Cu content, is related to defect- thereby be avoided. The cation core levels exhibit parallel
induced Fermi level pinning (see section 2.2), which has also shifts with increasing film thickness and the valence band off-
been suggested to explain the process-dependent offsets at set has been derived as EVB = 0.8 eV [179]. Higher values
14
Figure 14. Valence band offsets between Cu(In,Ga)Se2 and

ZnO1−x Sx in dependence on sulfur content x [150]. Closed and
open symbols for CBD- and MS-Zn(O,S) correspond to values
obtained from stepwise deposition and sputter depth profiles,
respectively. Calculated valence band offsets for pure ZnO and ZnS
from Zhang et al [35] are included. Calculated EVB of ZnO1−x Sx in
dependence on x (dashed lines) are taken from Persson et al [102].
The curve is offset in order to provide an estimate for the
dependence of the Cu(In,Ga)Se2 /Zn(O,S) valence band offset on
buffer layer composition. The thin solid line represents the
dependence of the Zn(O,S) valence band maximum energy given by
Persson et al [102] adapted to the theoretical valence band offsets
given by Zhang et al for CuInSe2 /ZnS and CuInSe2 /ZnO [35]. The
dashed line corresponds to the same trend but for the ordered defect
compound CuIn5 Se8 [35, 101]. Conduction band energies are Figure 15. (top) Composition and Fermi level position with respect
obtained by adding energy gaps which are calculated using to the valence band maximum of evaporated In2 S3 films as derived
equation (2) in [150] with a bowing parameter b = 3 eV. from in-situ XPS measurements [183]. (bottom) Energy level
diagrams of In2 S3 films prepared by evaporation either at room
of EVB = 1.2–1.4 eV have been obtained by Sterner et al us- temperature or at 250◦ C and Al-doped ZnO films prepared by
ing ALD deposition of In2 S3 onto a Cu(In,Ga)Se2 [188]. The magnetron sputtering either in pure Ar or in an Ar/O2
larger offset obtained by Sterner et al may suffer from the diffi- mixture [55, 111]. Due to the absence of strong band bending at the
culty of using overlapping anion core levels, which may be the interface, the energy band alignment for any In2 S3 /ZnO
combination is obtained approximately by alignment of the Fermi
reason for a non-parallel shift of the substrate’s Se 3d and the energies. The resulting valence band offsets of EVB = 1.6, 2.3 and
film’s S 2p core level (see table II in [188]). The differences 2.8 eV are all observed experimentally.
may, however, be caused by a stronger interface reaction due
to the higher substrate temperature. Scheer [193], the band alignment at the CuInS2 /CdS interface
may be one of the reasons why CuInS2 based solar cells
show much lower efficiencies than the selenides. The cliff,
5.5. The CuInS2 /CdS interface
in particular, may be responsible for too low a photovoltage of
As the S 3p orbital has a lower energy compared to the Se 4p such cells [193].
orbital [189], the valence band of CuInS2 is 0.28 eV lower The different valence band offsets might be explained
in energy than that of CuInSe2 [101]. Correspondingly, the by the limitation of Fermi level variation caused by a high
valence band offset with CdS is expected to be smaller by defect concentration. It is to be expected that the type and
the same amount. This has been verified experimentally concentration of such defects depend on the preparation of
for a vacuum-cleaved CuInS2 single crystal and evaporated a material. In fact, the different valence band offsets have
CdS with a EVB of ∼0.6 eV [57]. There are, however, a been measured using either a high temperature (>1000◦ C)
couple of studies where different preparation techniques have melt grown CuInS2 single crystals [57] or low-temperature
been used for CuInS2 and CdS, in which considerably larger (<600◦ C) processed thin films [190–192]. Moreover, thin
valence band offsets of EVB = 1.2–1.5 eV and consequently films for CuInS2 solar cells are typically grown using Cu excess
unfavorable cliffs in the conduction band are observed (see resulting in segregation of Cux S phases, which need to be
figure 16) [190–192]. There is no comparable dependence removed by chemical etching before cell preparation [193].
of the valence band offset on the deposition techniques for Due to the overall good agreement between experi-
the Cu(In,Ga)Se2 /CdS interface, where the variation of band mental and calculated valence band offsets for the
alignment can be mostly explained by the expected variation Cu(In,Ga)Se2 /buffer layer interfaces (see sections 5.2 and 5.3),
on Ga and Cu content (see section 5.2). As pointed out by it is expected that the interface between single crystal CuInS2
15
of EVB = 1.2 eV is derived, which agrees with the calculated

offset by Zhang et al [35]. The experiment was performed by
evaporating CdS onto an ex-situ prepared sputtered ZnO film.
An identical valence band offset has later been reported by Rao
et al using magnetron sputtering of nominally-undoped ZnO
onto evaporated CdS layers [54].
In an extension of the latter experiment, Säuberlich has
investigated the band alignment between CdS and Al-doped
ZnO (AZO) films [55, 194]. The ZnO films were grown
by magnetron sputtering from ceramic ZnO targets doped
with 2 wt% Al either using pure Ar (AZO:Ar) as sputter
gas or an Ar/O2 mixture (AZO:O2 ). The AZO:Ar films are
degenerately-doped and show a high Fermi level position well
above the conduction band minimum while the AZO:O2 films
Figure 16. Energy band alignment at CuInS2 /CdS interfaces as exhibit a high resistivity and a Fermi inside the band gap (see
derived (left) using a vacuum-cleaved CuInS2 single crystal and [84]). A valence band offset of EVB = 1.2 eV is observed
thermally-evaporated CdS [57] and (right) using polycrystalline for AZO:O2 films (see figure 17), reproducing the previous
CuInS2 thin films and chemical bath deposited CdS [190–192]. An experiments [54, 195] and the calculations [35]. In contrast, a
interface dipole, which might explain the different band line-up
obtained for the thin films is schematically indicated in the significantly higher valence band offset of EVB = 1.6 eV is
diagram (right). found at the CdS/AZO:Ar interface. This is reproduced by two
independent experiments as shown in figure 17 together with
and evaporated CdS [57], for which EVB corresponds to the CdS/ZnO:O2 interface experiment. The different valence
the theoretical value [101, 152], shows the intrinsic band band offsets can be explained by the limited range of Fermi
alignment while the band line-up at the interface between thin level positions in CdS and in the AZO films (see section 4.1).
film CuInS2 and CBD-CdS is modified by defects. To explain For the reverse deposition sequence, where CdS films
a valence band offset of EVB 1.2 eV, it is necessary to were deposited by evaporation onto AZO:Ar and AZO:O2
assume, e.g. a limitation of the Fermi energy in CuInS2 of substrates, a valence band offset of EVB = 1.4 eV is
EF − EVB 0.3 eV and in CdS of EF − EVB 1.5 eV. obtained for both substrates. This offset is in between the two
A limitation of the Fermi level variation in CdS also leads different values observed for sputter deposition onto CdS. This
to a modification of the band line-up at CdS/TCO interfaces is possible as both substrates develop a band bending during
(see section 5.6). A defect, which may be responsible for a CdS deposition. The Fermi level in the substrate can therefore
limitation of the Fermi level in CdS to values EF − EVB be adjusted to match the restricted Fermi level position in
1.5 eV, is the sulfur vacancy. It shows a charge transition level the deposited CdS film. For deposition of AZO onto CdS,
(2+/0) at approximately this energy [77, 78]. For Fermi levels the Fermi level position in the AZO films reaches it bulk
above this energy, the sulfur vacancies are electrically neutral. value already after a film thickness of ∼2 nm according to the
Lowering the Fermi level below this energy causes them to be evolution of binding energies with film thickness (figure 17).
charged. The dipoles, which give rise to the change of band There is thus no band bending in the AZO interface, in contrast
alignment, are indicated on the right in figure 16. to the situation observed during deposition of CdS onto AZO.
The modification of band line-up requires a corresponding The absence of band bending in AZO when deposited onto
limitation of the Fermi level in the CuInS2 thin film. Whether CdS has been related to the nucleation of the sputter deposited
or not a limitation of Fermi level variation is present in CuInS2 films [55, 194], which likely starts growing with an amorphous
thin films remains uncertain based on the available literature layer containing a high defect concentration [196–198]. This
and would require further investigation. If present, the origin amorphous layer may explain why the binding energies of the
of the possible limitation would likely be related to the Cu-rich Zn 2p and O 1s core levels show a different evolution with
growth conditions. film thickness at low ZnO coverage (compare figure 17; see
also [54]).
5.6. The CdS/ZnO interface Rao et al studied the influence of partial substitution of
Zn by Mg in (Zn,Mg)O on the valence band offset [54].
Highly-doped ZnO is almost exclusively used as transparent In agreement with optical measurements of ZnO/(Zn,Mg)O
front contact in Cu(In,Ga)(S,Se)2 solar cells in combination quantum well structures [199] and with the common-
with a nominally-undoped ZnO or (Zn,Mg)O secondary oxide anion-rule, the larger energy gap of (Zn,Mg)O compared to
layer. It has been suggested that the secondary oxide layer ZnO [107] mainly results in higher ECB . EVB remains
introduces an additional series resistance, which can reduce approximately independent on Mg content, at least for not too
effects of sample inhomogeneity [74]. But there may be high an Mg content.
an influence of the doping on the energy band alignment, as
suggested by systematic studies of the CdS/ZnO interface by
5.7. The Cu(In,Ga)Se2 /ZnO interface
Säuberlich et al [55, 194].
The first measurement of energy band alignment at the The interface formation between Cu(In,Ga)Se2 and ZnO
ZnO/CdS interface is reported in [195]. A valence band offset has been studied by Platzer-Björkman et al using atomic
16
Figure 17. Evolution of valence band maxima with film thickness as obtained from core level binding energies in the course of ZnO
deposition onto CdS (left) and in the course of CdS deposition onto ZnO (right) [55]. The ZnO films are prepared by magnetron sputtering
from a ceramic ZnO:Al target in pure Ar (red and green lines) and in an Ar/O2 mixture (blue lines), respectively. The two lines for each
experiment are obtained from Zn 2p and O 1s and from Cd 3d and S 2p core levels, respectively.
layer deposition of ZnO [172, 200] and by Säuberlich the binary Cu chalcogenides, valence band offsets have
using magnetron sputtering of ZnO [55, 194] revealing been measured at the Cu2 S/CdS (EVB = 1.2 eV [205]),
EVB = 2.26 eV and 2.15 eV, respectively. Together with the ZnO/Cu2 S (EVB = −2.1 eV [206]), the Cu2 S/Cu2 O
a valence band offset at Cu(In,Ga)Se2 /CdS (EVB = 0.9 eV, (EVB = 0.2 eV [206]), TiO2 /Cu2 S (EVB = 2.9 eV [207])
section 5.2) and at CdS/ZnO (EVB = 1.2 eV, section 5.6), and the CdTe/Cu2 Te (EVB = 0.8 eV [126]) interface.
transitivity of valence band offsets is fulfilled in the Moreover, initiated by the interest in using Cu2 O as a solar
Cu(In,Ga)Se2 /CdS/ZnO system. The insertion of the CdS cell absorber [208–210] and as a p-type thin film transistor
buffer does therefore not lead to a modification of the band material [211], the energy band alignment at interfaces
alignment between the absorber and the ZnO window layer. between Cu2 O and ITO [50] and between Cu2 O and ZnO
This may contribute to the reproducibly high efficiencies with have been studied [210, 212, 213]. The valence band offset at
CdS buffer layers but does not explain why solar cells without the Cu2 O/ITO interface varies between EVB = 2.1–2.6 eV,
CdS buffer layers show typically much lower efficiencies. depending on the Fermi level position in ITO [50]. For
Due to the high Fermi level position in ZnO, direct the Cu2 O/ZnO interface, the reported valence band offsets
deposition of ZnO onto Cu(In,Ga)Se2 should raise the Fermi vary between EVB = 1.3 eV [212] and 2.9 eV [213]. Our
level towards the conduction band. This may lead to a measurements show a variation between EVB = 1.5–2.7 eV
significant reduction in the Cu content at the Cu(In,Ga)Se2 depending on the preparation of the Cu2 O and the ZnO
surface as reported for Na and CdS deposition [163, 164, 166]. layer [214].
As already discussed for many interfaces in this work, the
Such a Cu depletion is driven by the defect formation energy
valence band offset at such interfaces can significantly depend
of Cu vacancies, which becomes negative for high Fermi level
on interface preparation, which is very likely related to the
positions [8, 9] and which is enabled by the low activation
abundance of defects in ionic compounds. This is evident
energies for Cu diffusion in these compounds [10].
from studies of another novel absorber material: SnS. The
It is mentioned that transitivity is not generally fulfilled
interfaces of evaporated SnS compounds with a variety of
in the Cu(In,Ga)Se2 /In2 S3 /ZnO system. This is particularly materials have been studied by Schneikart et al [183, 215].
caused by the strong dependence of band alignment on depo- The Fermi level position at the surfaces and interfaces of
sition conditions at the In2 S3 /ZnO interface as demonstrated SnS is quite reproducibly measured at EF − EVB ≈ 0.5 eV,
in section 5.4. The particular importance of the In2 S3 /ZnO which indicates that it is difficult to move the Fermi level away
interface has been inferred from solar cell preparation with from this position. This is very likely the origin of the low
CdS/In(OHx ,Sy ) double buffer layer structures [201, 202]. photovoltages, which are the major problem of SnS solar cells
[183]. Nevertheless, the contribution of Sn 5s orbitals to the
6. New absorber materials valence band states in SnS are the origin of a significantly
higher EVB of this material [216].
The scarcity of In and Te and the toxicity of Cd drive
the search for alternative absorber materials for thin film 7. Summary
solar cells. Intensively-studied materials are the kesterite
compounds (Cu2 ZnSn(S,Se)4 ), which ideally are mixtures This article summarizes the author’s view on energy band
of Cu2 (S,Se), Zn(S,Se) and Sn(S,Se)2 [203, 204]. Among alignment in chalcogenide thin film solar cells, which is
17
derived from extensive photoemission studies and comparison CuInS2 /CdS and CdS/ZnO interfaces, respectively. The
with DFT calculations. There is plenty of evidence that the CdS/CdTe and the Cu(In,Ga)Se2 /CdS interface, which are the
general picture of energy band alignment depicted in figure 2, technologically most important interfaces in chalcogenide thin
which is based on theoretical calculations, but for many film solar cells, do not seem to be affected by such modification
interfaces confirmed by experiment, is valid. This alignment but rather behave intrinsically. For Cu(In,Ga)Se2 this may
may be considered as the intrinsic alignment of energy bands at particularly be related to the Cu-poor growth conditions of
interfaces in thin film chalcogenide solar cells. By intrinsic, we Cu(In,Ga)Se2 , as the interface between Cu-rich CuInS2 and
mean that this alignment is naturally established at interfaces CdS is subject to a modification and is not suitable for high
when no defects are present. conversion efficiencies.
So far, there is no evidence that the energy band alignment The energy band alignment at the CdS/CdTe and the
at interfaces of chalcogenides and oxides depends on interface Cu(In,Ga)Se2 /CdS interfaces is found to be quite insensitive
orientation and/or on lattice mismatch. This is expected for on interface preparation. This seems to disagree with the huge
interfaces between homovalent materials. Dislocations, which work invested to optimize the device structures, in particular
cannot be avoided at interfaces of materials with high lattice the interface properties. However, energy band alignment is
mismatch are not expected to modify the band alignment. The not the only interface property which affects device efficiency.
reason is that an electric dipole layer is required to modify Other aspects, mainly interface recombination, is as relevant
band alignment. Dislocations can be charged, but they occur as band alignment. The observed interdiffusion at many of
ususally only on one side of the interface. While the charge can these interfaces, which according to the results presented in
change the Fermi level position at an interface, the alignment this paper does not affect band alignment, can have a major
should not be affected. effect on interface states. While interface states can not change
The energy band alignment between chalcogenides and the band alignment they are crucial for interface recombination
by inserting recombination centers and modifying the Fermi
oxides can be estimated using the common-anion-rule if no
level position at the interface.
cation states contribute to the valence band states. This
There is evidence that the contacts in highly-efficient thin
results in similar valence band maximum energies of the
film CdTe solar cells are not homogeneous but rather form
oxides, the sulfides, the selenides and the tellurides with
point contacts. This is on the one hand demonstrated by
EVB (MO) < EVB (MS) < EVB (MSe) < EVB (MTe). In
the high conversion efficiencies obtained with Cd(O,S) buffer
principle, this relation holds if cation states contribute to
layers, which is a multiphase material mostly consisting of
the valence bands. This is particularly relevant for Cu (or
insulating CdSO4 . On the other hand, ohmic contacts are
Ag) containing compounds like the chalcopyrites or the Cu- typically obtained by Cu diffusion into CdTe from various
chalcogenides. The admixture of cation d-states and of contact materials. High barriers are found at these interfaces by
cation s-states, like e.g. in Sn(II), Pb(II), Sb(III) or Bi(III) photoemission, which are not consistent with high conversion
compounds, contributes to the valence bands and leads to efficiencies. The doping level achieved by Cu diffusion is
significantly higher EVB . furthermore not sufficient to facilitate a tunneling contact as
Bulk defects, in particular vacancies and interstitials, can most of the Cu is present at grain boundaries.
have low-formation enthalpies in chalcogenides and are then In summary, photoemission experiments can contribute
present in high concentrations. The Fermi energy dependence detailed knowledge on the interface formation and energy
of defect-formation enthalpies can cause self-compensation. band alignment in chalcogenide thin film solar cells. The
Intentional doping of the chalcogenides is therefore difficult to phenomena occurring at these interfaces are much more
control and mostly not part of chalcogenide thin film solar cell complex than those dominating the well-understood interfaces
technologies. The change of the Fermi energy at an interface of elementary and III-V compound semiconductors. The
can even induce defect formation. This is particularly relevant additional complexity is related to the low defect-formation
for the chalcopyrites, where interface composition seems to be enthalpies in ionic compounds, the presence of additional
correlated with Fermi level position. The observation that the stable phases and the low-temperature deposition techniques.
CuInSe2 /CdS interface prefers to have a Cu-poor composition The latter are required for cost-effective production of thin film
while the CuGaSe2 /CdS interface prefers to be stoichiometric, solar cells. Growth of materials using such techniques results
requires a more comprehensive understanding of interface in far-from-equilibrium conditions. While this makes the
chemistry of chalcogenides, particularly if intermediate phase understanding of the materials and their interfaces an extensive
like the ordered vacancy compounds can be present. The issue, which is far from complete, the success of chalcogenide
complex phase formation at such interfaces and its relation thin film solar cells demonstrates that it is possible to find
to solar cell efficiency is not yet sufficiently explored. As optimum conditions with suitable deposition techniques and
the Fermi level position may affect phase stability, it can be parameters. The fundamental research on interface properties
expected that electrochemically active species like Na play a may contribute to further enhancing device properties, but may
role in this context. direct future research on alternative absorber materials.
The presence of high defect concentrations has been
suggested to be the origin of process-dependent energy Acknowledgments
band alignments. Such observations are quite frequent for
chalcogenides and oxides. Examples elaborated in detail in The viewpoint on energy band alignment at chalcogenide
this paper are TCO/CdS, Cu(In,Ga)Se2 /Zn(O,S), In2 S3 /ZnO, semiconductor interfaces presented in this article has been
18
developed during more than 20 years of research, which [18] Franciosi A and Van de Walle C G 1996 Heterojunction band
was initiated by Wolfram Jaegermann at the Hahn-Meitner- offset engineering Surf. Sci. Rep. 25 1
[19] Peressi M, Binggeli N and Baldereschi A 1998 Band
Institute in Berlin (now Helmholtz-Zentrum für Materialien engineering at interfaces: theory and numerical
und Energie) and continued at the Technische Universität experiments J. Phys. D 31 1273–99
Darmstadt. During this period many researchers, mainly [20] Mönch W 2003 Electronic Properties of Semiconductor
PhD students, have contributed to the development of the Interfaces (Heidelberg: Springer)
measurement techniques and the understanding. Important [21] Anderson R L 1962 Experiments on Ge-GaAs
heterojunctions Solid-State Electron. 5 341
contributions are from W Jaegermann, C Pettenkofer, H [22] Heine V 1965 Theory of surface states Phys. Rev. 138 A1689
Scheer, R Hunger, T Löher, T Schulmeyer, F Säuberlich, A [23] Louie S G, Chelikowsky J R and Cohen M L 1977 Ionicity
Thißen, J Fritsche, D Kraft, B Späth, T Adler, A Schneikart, S and the theory of Schottky barriers Phys. Rev. B
Siol and A Fuchs. Financial support was mainly received from 15 2154–62
the German Department of Education and Research (BMBF) [24] Bardeen J 1947 Surface states and rectification at a metal
semiconductor contact Phys. Rev. 71 717–27
under various research projects. [25] Tersoff J 1984 Theory of semiconductor heterojunction: the
role of quantum dipoles Phys. Rev. B 30 4874–7
[26] Tersoff J 1984 Schottky barrier heights and the continuum of
References gap states Phys. Rev. Lett. 52 465–8
[27] Cardona M and Christensen N E 1987 Acoustic deformation
[1] Jäger-Waldau A 2013 PV Status Report Scientific Report potentials and heterostructure band offsets in
European Commission Joint Research Center semiconductors Phys. Rev. B 35 6182–94
http://iet.jrc.ec.europa.eu/remea/pv-status-report-2013 [28] Kurtin S, McGill T C and Mead C A 1969 Fundamental
[2] Green M A, Emery K, Hishikawa Y, Warta W and transition in the electronic nature of solids Phys. Rev. Lett.
Dunlop E D 2014 Solar cell efficiency tables (version 43) 22 1433–6
Prog. Photovolt. Res. Appl. 22 1–9 [29] Robertson J and Clark S J 2011 Limits to doping in oxides
[3] www.zsw-bw.de Phys. Rev. B 83 075205
[4] www.firstsolar.com [30] Robertson J 2013 Band offsets, Schottky barrier heights and
[5] Zhang S B, Wei S-H, Zunger A and Katayama-Yoshida H their effects on electronic devices J. Vac. Sci. Technol. A
1998 Defect physics of the CuInSe2 chalcopyrite 31 050821
semiconductor Phys. Rev. B 57 9642 [31] Mönch W 2005 On the band structure lineup of ZnO
[6] Wei S-H, Zhang S B and Zunger A 2000 First-principles heterostructures Appl. Phys. Lett. 86 162101
calculation of band offsets, optical bowings and defects in [32] Mönch W 2007 On the electric-dipole contribution to the
CdS, CdSe, CdTe and their alloys J. Appl. Phys. valence-band offsets in semiconductor-oxide
87 1304–11 heterostructures Appl. Phys. Lett. 91 042117
[7] Wei S-H and Zhang S B 2002 Chemical trends of defect [33] Chen F, Schafranek R, Li S, Wu W and Klein A 2010 Energy
formation and doping limit in II-VI semiconductors: the band alignment between Pb(Zr,Ti)O3 and high and low
case of CdTe Phys. Rev. B 66 155211 work function conducting oxides—from hole to electron
[8] Zhao Y-J, Persson C, Lany S and Zunger A 2004 Why can injection J. Phys. D: Appl. Phys. 43 295301
CuInSe2 be readily equilibrium-doped n-type but the [34] Schafranek R, Li S, Chen F, Wu W and Klein A 2011
wider-gap CuGaSe2 cannot? Appl. Phys. Lett. 85 5860–2 PbTiO3 /SrTiO3 interface: energy band alignment and Its
[9] Pohl J and Albe K 2010 Thermodynamics and kinetics of the relation to the limits of Fermi level variation Phys. Rev. B
copper vacancy in CuInSe2 , CuGaSe2 , CuInS2 and 84 045317
CuGaS2 from screened-exchange hybrid density [35] Zhang S B, Wei S-H and Zunger A 1998 A phenomenological
functional theory J. Appl. Phys. 108 023509 model for systemization and prediction of doping limits in
[10] Pohl J, Klein A and Albe K 2011 Role of copper interstitials II-VI and I-III-VI2 compounds J. Appl. Phys. 83 3192–6
in CuInSe2 : first-principles calculations Phys. Rev. B [36] Lambrecht W R L and Segall B 1988 Theory of
84 121201 semiconductor heterojunction valence band offsets: from
[11] Walukiewicz W 2001 Intrinsic limitations to the doping of supercell band-structure calculation towards a simple
wide-gap semiconductors Physica B 302–3 123–34 model Phys. Rev. Lett. 61 1764–7
[12] Klein A 2013 Transparent conducting oxides: electronic [37] Lambrecht W R L, Segall B and Anderson O K 1990
structure—property relationship from photoelectron Self-consistent dipole theory of heterojunction band
spectroscopy with in-situ sample preparation J. Am. offsets Phys. Rev. B 41 2813–31
Ceram. Soc. 96 331–45 [38] Lambrecht W R L and Segall B 1990 Interface-bond-polarity
[13] Sze S M and Ng K K 2007 Physics of Semiconductor Devices model for semiconductor heterojunction band offsets Phys.
3rd edn (Hoboken: John Wiley) Rev. B 41 2832–48
[14] Scheer R and Schock H-W 2011 Chalcogenide [39] Hofmann A and Pettenkofer C 2011 Surface orientation
Photovoltaics: Physics, Technologies and Thin Film dependent band alignment for CuInSe2 ZnSeZnO Appl.
Devices (Weinheim: Wiley) Phys. Lett. 98 113503
[15] Klein A 2012 Energy band alignment at interfaces of [40] Siepchen B, Schimper H, Klein A and Jaegermann W 2013
semiconducting oxides: a review of experimental SXPS studies of single crystalline CdTe/CdS interfaces
determination using photoelectron spectroscopy and J. Electron Spectrosc. Relat. Phenom. 190 54–63
comparison with theoretical predictions by the electron [41] Chen F, Schafranek R, Wachau A, Zhukov S, Glaum J,
affinity rule, charge neutrality levels and the common Granzow T, von Seggern H and Klein A 2010 Barrier
anion rule Thin Solid Films 520 3721–8 heights, polarization switching and electrical fatigue in
[16] Capasso F and Margaritondo G (ed) 1987 Heterojunction Pb(Zr,Ti)O3 ceramics with different electrodes J. Appl.
Band Discontinuities (Amsterdam: North-Holland) Phys. 108 104106
[17] Yu E T, McCaldin J O and McGill T C 1992 Band offsets in [42] McCaldin J O, McGill T C and Mead C A 1976 Correlation
semiconductor heterojunctions Solid State Phys. 46 1 for III-V and II-VI Semiconductors of the Au Schottky
19
barrier energy with anion electronegativity Phys. Rev. Lett. [63] Nollet P, Burgelman M and Degrave S 2000 The back contact
36 56–8 influence on characteristics of CdTe/CdS solar cells Thin
[43] Li S et al 2014 Intrinsic energy band alignment of functional Solid Films 361–2 293–7
oxides Phys. Status Solidi 8 571–6 [64] McMahon T J and Fahrenbruch A L 2000 Insights into the
[44] Pfeifer V et al 2013 Energy band alignment between anatase nonideal behavior of CdS/CdTe solar cells Proc. of the
and rutile TiO2 J. Phys. Chem. Lett. 4 4182–7 28th IEEE Photovoltaic Specialists Conference
[45] Li S et al 2013 Influence of orbital contributions to valence (Anchorage, AK, 15–22 September 2000) pp 539–42
band alignment of Bi2 O3 , Fe2 O3 , BiFeO3 and [65] Roussillon Y, Karpov V G, Shvydka D, Drayton J and
Bi0.5 Na0.5 TiO3 Phys. Rev. B 88 045428 Compaan A D 2004 Back contact and reach-through diode
[46] Jaffe J E and Zunger A 1984 Theory of the band-gap effects in thin-film photovoltaics J. Appl. Phys. 96 7283–8
anomaly in ABC2 chalcopyrite semiconductors Phys. [66] Fahrenbruch A L 2008 Exploring back contact technology to
Rev. B 29 1882 increase CdS/CdTe solar cell efficiency Mater. Res. Soc.
[47] Wei S-H and Zunger A 1987 Role of d orbitals in Symp. Proc. 1012 283
valence-band offsets of common-anion semiconductors [67] Fritsche J, Kraft D, Thißen A, Mayer T, Klein A and
Phys. Rev. Lett. 59 144–7 Jaegermann W 2002 Band energy diagram of CdTe thin
[48] Wei S-H and Zunger A 1996 Chemical trends in band offsets film solar cells Thin Solid Films 403–4 252–7
of Zn- and Mn-based II-VI superlattices: d-level pinning [68] Jaegermann W, Klein A and Mayer T 2009 Interface
and offset compression Phys. Rev. B 53 R10457 engineering of inorganic thin film solar cells materials
[49] Löher T, Klein A, Pettenkofer C and Jaegermann W 1997 science challenges for advanced physical concepts Adv.
Partial density of states in the CuInSe2 valence bands Mater. 21 4196–206
J. Appl. Phys. 81 7806–9 [69] Han J, Spanheimer C, Haindl G, Fu G, Krishnakumar V,
[50] Deuermeier J, Gassmann J, Brötz J and Klein A 2011 Schaffner J, Fan C, Zhao K, Klein A and Jaegermann W
Reactive magnetron sputtering of Cu2 O: dependence on 2011 Optimized chemical bath deposited CdS layers for
oxygen pressure and interface formation with ITO J. Appl. the improvement of CdTe solar cells Sol. Energy Mater.
Phys. 109 113704 Sol. Cells 95 816–20
[51] Bayer T J M, Wachau A, Fuchs A, Deuermeier J and Klein A [70] Britt J and Ferekides C 1993 Thin-film CdS/CdTe solar cell
2012 Atomic layer deposition of Al2 O3 onto Sn-doped with 15.8% efficiency Appl. Phys. Lett. 62 2851
In2 O3 : absence of self-limited adsorption during initial [71] Wu X, Yan Y, Dhere R G, Zhang Y, Zhou J, Perkins C and To
growth by oxygen diffusion from the substrate and band B 2004 Nanostructured CdS:O film: preparation,
offset modification by Fermi level pinning in Al2 O4 Chem. properties and application Phys. Status Solidi c 1 1062–6
Mater. 24 4503–10 [72] Ferekides C S, Mamazza R, Balasubramanian U and
[52] Waldrop J R, Grant R W, Kowalczyk S P and Kraut E A 1985 Morel D L 2005 Transparent conductors and buffer layers
Measurement of semiconductor heterojunction band for CdTe solar cells Thin Solid Films 480–1 224–9
discontinuities by x-ray photoemission spectroscopy [73] Krishnakumar V, Klein A and Jaegermann W 2013 Studies
J. Vac. Sci. Technol. A 3 835 on CdTe solar cell front contact properties using x-ray
[53] Tanuma S, Powell C J and Penn D R 1991 Calculations photoelectron spectroscopy Thin Solid Films 545 548–57
of electron mean free path Surf. Interface Anal. [74] Rau U, Grabitz P O and Werner J H 2004 Resistive
17 911–27 limitations to spatially inhomogeneous electronic losses in
[54] Venkata Rao G, Säuberlich F and Klein A 2005 Influence of solar cells Appl. Phys. Lett. 85 6010–2
Mg content on the band alignment at CdS/(Zn,Mg)O [75] Klein A, Körber C, Wachau A, Säuberlich F, Gassenbauer Y,
interfaces Appl. Phys. Lett. 87 032101 Harvey S P, Proffit D E and Mason T O 2010 Transparent
[55] Klein A and Säuberlich F 2008 Surfaces and interfaces of conducting oxides for photovoltaics: manipulation of
sputter-deposited ZnO films Transparent Conductive Zinc Fermi Level, work function and energy band alignment
Oxide: Basics and Applications in Thin Film Solar Cells Materials 3 4892–914
ed K Ellmer et al (Berlin: Springer) pp 125–85 [76] Krishnakumar V, Ramamurthi K, Klein A and Jaegermann W
[56] Li S, Wachau A, Schafranek R, Klein A, Zheng Y and 2009 Band alignment of differently treated TCO/CdS
Jakoby R 2010 Energy level alignment and electrical interface Thin Solid Films 517 2558–61
properties of (Ba,Sr)TiO3 /Al2 O3 interfaces for tunable [77] Nishidate K, Sato T, Matsukura Y, Baba M, Hasegawa M and
capacitors J. Appl. Phys. 108 014113 Sasaki T 2006 Density-functional electronic structure
[57] Klein A, Löher T, Tomm Y, Pettenkofer C and Jaegermann W calculations for native defects and Cu impurities in CdS
1997 Band lineup between CdS and ultra high Phys. Rev. B 74 035210
vacuum-cleaved CuInS2 single crystals Appl. Phys. Lett. [78] Wu J-C, Zheng J, Wu P and Xu R 2011 Study of native
70 1299–301 defects and transition-metal (Mn, Fe, Co and Ni) doping in
[58] McCandless B E and Sites J R 2003 Cadmium Telluride a zinc-blende CdS photocatalyst by DFT and hybrid DFT
Solar Cells (Handbook of Photovoltaic Science and calculations J. Phys. Chem. C 115 5675–82
Engineering) ed A Luque and S Hegedus (Chichester: [79] Lany S and Zunger A 2007 Dopability, intrinsic conductivity
John Wiley) pp 617–62 and nonstoichiometry of transparent conducting oxides
[59] Kranz L et al 2013 Doping of polycrystalline cdte for Phys. Rev. Lett. 98 045501
high-efficiency solar cells on flexible metal foil Nat. [80] Ágoston P, Körber C, Klein A, Puska M J, Nieminen R M
Commun. 4 2306 and Albe K 2010 Limits for n-type doping in In2 O3 and
[60] McCandless B E, Moulton L W and Birkmire R W 1997 SnO2 : a theoretical approach by first-principles
Recrystallization and sulfur diffusion in CdCl2 treated calculations using hybrid-functional methodology J. Appl.
CdTe/CdS thin film solar cells Prog. Photovolt. Res. Appl. Phys. 108 053511
5 249–60 [81] Madelung O (ed) 1996 Semiconductors Basic Data 2nd edn
[61] Paulson P D and Dutta V 2000 Study of in situ CdCl2 (Berlin: Springer)
treatment on CSS deposited CdTe films and CdS/CdTe [82] Schleife A, Varley J B, Fuchs F, Rödl C, Bechstedt F,
solar cells Thin Solid Films 370 299–306 Rinke P, Janotti A and Van de Walle C G 2011 Tin dioxide
[62] Wu X 2004 High-efficiency polycrystalline CdTe thin-film from first principles: quasiparticle electronic states and
solar cells Solar Energy 77 803–14 optical properties Phys. Rev. B 83 035116
20
[83] Walsh A et al 2008 The nature of the bandgap in In2 O3 ZnO1−x Sx enhances p-type nitrogen doping of ZnO-like
revealed by first-principles calculations and x-ray alloys Phys. Rev. Lett. 97 146403
spectroscopy Phys. Rev. Lett. 100 167402 [103] Meyer B K, Polity A, Farangis B, He Y, Hasselkamp D,
[84] Klein A, Körber C, Wachau A, Säuberlich F, Gassenbauer Y, Kramer T and Wang C 2004 Structural properties and
Schafranek R, Harvey S P and Mason T O 2009 Surface bandgap bowing of ZnO1−x Sx thin films deposited by
potentials of magnetron sputtered transparent conducting reactive sputtering Appl. Phys. Lett. 85 4929–31
oxides Thin Solid Films 518 1197–203 [104] Kephart J M, Geisthardt R and Sampath W S 2012 Sputtered,
[85] Fuchs A 2014 Der Frontkontakt der oxygenated CdS window layers for higher current in
CdTe-Dünnschichtsolarzelle: charakterisierung und CdS/CdTe thin film solar cells Proc. of the 38th IEEE
modifizierung von puffer- und fensterschichten und deren Photovoltaic Specialists Conf. (Austin, TX, 3–8 June 2012)
grenzflächen PhD Thesis Technische Universität, pp 854–8
Darmstadt [105] Paudel N R and Yan Y 2013 Fabrication and characterization
[86] Niles D W and Höchst H 1990 Band offsets and interfacial of high-efficiency CdTe-based thin-film solar cells on
properties of cubic CdS grown by molecular-beam epitaxy commercial SnO2 : F-coated soda-lime glass substrates
on CdTe(110) Phys. Rev. B 41 12710 Thin Solid Films 549 30–5
[87] Fritsche J, Thißen A, Klein A and Jaegermann W 2001 [106] Shim Y-G, Sakamoto J, Suzuki A, Khalilova K, Wakita K,
Oriented growth and band lineup at the CdS/CdTe Mamedov N, Bayramov A, Huseynov E and Hasanov I
interface Thin Solid Films 387 158–60 2011 Spectroscopic ellipsometry studies of CdS:O layers
[88] Siepchen B, Klein A and Jaegermann W 2008 Interface for solar cells Japan. J. Appl. Phys. 50 05FC14
formation in CdTe solar cells: nucleation of CdTe on [107] von Wenckstern H et al 2009 Anionic and cationic
CdS(0 0 0 1) and (10–10) Phys. Status Solidi 2 169–71 substitution in ZnO Prog. Solid State Chem. 37 153–72
[89] Fritsche J, Schulmeyer T, Kraft D, Thißen A, Klein A and [108] Niemegeers A and Burgelman M 1997 Effects of the
Jaegermann W 2002 Utilization of sputter depth Au/CdTe back contact on IV and CV characteristics of
profiling for the determination of band alignment at Au/CdTe/CdS/TCO solar cells J. Appl. Phys. 81 2881–6
polycrystalline CdTe/CdS heterointerfaces Appl. Phys. [109] Fritsche J, Kraft D, Thißen A, Mayer T, Klein A and
Lett. 81 2297–9 Jaegermann W 2001 Interface engineering of chalcogenide
[90] Fritsche J, Klein A and Jaegermann W 2005 Thin film solar semiconductors in thin film solar cells: CdTe as an
cells: materials science at interfaces Adv. Eng. Mater. example Mater. Res. Soc. Symp. Proc. 668 H6.6
7 914–20 [110] Dharmadasa I M 1998 Recent developments and progress on
[91] Fritsche J, Schulmeyer T, Thißen A, Klein A and Jaegermann electrical contacts to CdTe, CdS and ZnSe with special
W 2003 Interface modification of CdTe thin film solar cell reference to barrier contacts to CdTe Prog. Cryst. Growth
by CdCl2 activation Thin Solid Films 431/432 267–71 Charact. 36 249–90
[92] Schulmeyer T, Fritsche J, Thißen A, Klein A, Jaegermann W, [111] Klein A, Säuberlich F, Späth B, Schulmeyer T and Kraft D
Campo M and Beier J 2003 Effect of in situ UHV 2007 Non-stoichiometry and electronic properties of
CdCl2 -activation on the electronic properties of CdTe thin interfaces J. Mater. Sci. 42 1890–900
film solar cells Thin Solid Films 431/432 84–9 [112] Friedman D J, Lindau I and Spicer W E 1988
[93] McCandless B E, Youm I and Birkmire R W 1999 Noble-metal-CdTe interface formation Phys. Rev. B
Optimization of vapor post-deposition processing for 37 731
evaporated CdS/CdTe solar cells Prog. Photovolt. Res. [113] Shaw J L, Viturro R E, Brillson L J, Kilday D, Kelly M K and
Appl. 7 21–30 Margaritondo G 1988 interfacial deep-level formation and
[94] McCandless B E, Engelmann M G and Birkmire R W 2001 its effect on band bending at metal/CdTe interfaces J. Vac.
Interdiffusion of CdS/CdTe thin films: modeling x-ray Sci. Technol. A 6 2752–6
diffraction line profiles J. Appl. Phys. 89 988–94 [114] Kraft D, Weiler U, Thißen A, Tomm Y, Klein A and
[95] Nishi K, Ohyama H, Siuzuki T, Mitsuyu T and Tomimasu T Jaegermann W 2003 Alternative back contacts for CdTe
1997 Evaluation of the CdS/CdTe interface using solar cells: a photoemission study of the VSe2 /CdTe and
free-electron laser internal photoemission technique Appl. TiSe2 /CdTe interface formation Thin Solid Films
Phys. Lett. 70 3585–7 431/432 382–6
[96] Wu X, Asher S, Levi D H, King D E, Yan Y, Gessert T A and [115] Niles D W, Xiaonau L, Albin D, Rose D, Gessert T and
Sheldon P 2001 Interdiffusion of CdS and Zn2SnO4 layers Sheldon P 1996 Evaporated Te on CdTe: a vacuum
and its application in CdS/CdTe polycrystalline thin-film compatible approach to making back contact to CdTe solar
solar cell J. Appl. Phys. 89 4564–9 cell devices Prog. Photovolt. Res. Appl. 4 225
[97] Gloeckler M, Sankin I and Zhao Z 2013 CdTe solar cells at [116] Kraft D, Thißen A, Brötz J, Flege S, Mayer T, Klein A,
the threshold to 20% efficiency IEEE J. Photovolt. Jaegermann W and Campo M 2003 Characterization of
3 1389–93 tellurium layers for back contact formation on close to
[98] Gloeckler M and Sites J R 2005 Efficiency limitations for technology treated CdTe surfaces J. Appl. Phys.
wide-band-gap chalcopyrite solar cells Thin Solid Films 94 3589–98
480–1 241–5 [117] Rioux D, Niles D W and Höchst H 1993 ZnTe: a potential
[99] Chu T L, Chu S S, Britt J, Ferekides C and Wu C Q 1991 interlayer to form low resistance back contacts in
Cadmium zinc sulfide films and heterojunctions J. Appl. CdS/CdTe solar cells J. Appl. Phys. 73 8381
Phys. 70 2688–93 [118] Späth B, Fritsche J, Säuberlich F, Klein A and
[100] Clayton A J, Irvine S J C, Jones E W, Kartopu G, Barrioz V Jaegermann W 2005 Studies of sputtered ZnTe films as
and Brooks W S M 2012 MOCVD of interlayer for the CdTe thin film solar cell Thin Solid Films
Cd(1-x)Zn(x)S/CdTe PV cells using an ultra-thin absorber 480–1 204–7
layer Sol. Energy Mater. Sol. Cells 101 68–72 [119] Späth B, Fritsche J, Klein A and Jaegermann W 2007
[101] Wei S-H and Zunger A 1995 Band offsets and optical Nitrogen doping of ZnTe and Its influence on CdTe/ZnTe
bowings of chalcopyrites and Zn-based II-VI alloys interfaces Appl. Phys. Lett. 90 062112
J. Appl. Phys. 78 3846 [120] Bätzner D L, Romeo A, Zogg H, Wendt R and Tiwari A N
[102] Persson C, Platzer-Björkman C, Malmström J, Törndahl T 2001 Development of efficient and stable back contacts on
and Edoff M 2006 Strong valence-band offset bowing of CdTe/CdS solar cells Thin Solid Films 387 151–4
21
[121] Abken A E and Bartelt O J 2002 Sputtered Mo/Sb2 Te3 and Cu(In,Ga)Se2 /Mo and Cu(In,Ga)(S,Se)2 /Mo interfaces
Ni/Sb2 Te3 layers as back contacts for CdTe/CdS solar cells Phys. Rev. B 78 075404
Thin Solid Films 403–4 216 [141] Herberholz R, Rau U, Schock H W, Haalboom T, Gödecke T,
[122] Durose K, Boyle D, Abken A, Ottley C J, Nollet P, Ernst F, Beilharz C, Benz K W and Cahen D 1999 Phase
Degrave S, Burgelman M, Wendt R, Beier J and Bonnet D segregation, Cu migration and junction formation in
2002 Key aspects of CdTe/CdS solar cells Phys. Status Cu(In, Ga)Se2 Eur. Phys. J. Appl. Phys. 6 131–9
Solidi B 229 1055–64 [142] Zhang S B, Wei S-H and Zunger A 1997 Stabilization of
[123] Kraft D, Späth B, Thißen A, Klein A and Jaegermann W ternary compounds via ordered arrays of defect pairs Phys.
2003 Chemical and electronic properties of metal/Sb2 Te3 / Rev. Lett. 78 4059
CdTe contacts for CdTe thin film solar cells studied by [143] Schmid D, Ruckh M, Grunwald F and Schock H W 1993
photoelectron spectroscopy Mater. Res. Soc. Symp. Proc. Chalcopyrite/defect chalcopyrite heterojunctions on the
763 B3.3 basis of CuInSe2 J. Appl. Phys. 73 2902
[124] Uda H, Ikegami S and Sonomura H 1998 Annealing effect of [144] Schulmeyer T, Hunger R, Klein A, Jaegermann W and Niki S
Cu2 Te/Au contact to evaporated CdTe film on photovoltaic 2004 Photoemission study and band alignment of the
properties of CdS/CdTe solar cell Sol. Energy Mater. Sol. CdS/CuInSe2 (0 0 1) heterojunction Appl. Phys. Lett.
Cells 50 141–6 84 3067–9
[125] Yun J H, Kim K H, Lee D Y and Ahn B T 2003 Back contact [145] Persson C and Zunger A 2003 Anomalous grain boundary
formation using Cu2 Te as a Cu-doping source and as an physics in polycrystalline CuInSe2 : the existence of a hole
electrode in CdTe solar cells Sol. Energy Mater. Sol. Cells barrier Phys. Rev. Lett. 91 266401
75 203–10 [146] Mönig H et al 2009 Surface Cu depletion of Cu(In,Ga)Se2
[126] Späth B, Lakus-Wollny K, Fritsche J, Ferekides C S, Klein A films: an investigation by hard x-ray photoelectron
and Jaegermann W 2007 Surface science studies of Cu spectroscopy Acta Mater. 57 3645–51
containing back contacts for CdTe Solar Cells Thin Solid [147] Siebentritt S, Igalson M, Persson C and Lany S 2010 The
Films 515 6172–4 electronic structure of chalcopyrites—bands, point defects
[127] Sarlund J, Ritala M, Leskelä M, Siponmaa E and Zilliacus R and grain boundaries Prog. Photovolt. Res. Appl.
1996 Characterization of etching procedure in 18 390–410
preparation of CdTe solar cells Sol. Energy Mater. Sol. [148] Hariskos D, Spiering S and Powalla M 2005 Buffer layers in
Cells 44 177–90 Cu(In,Ga)Se2 solar cells and modules Thin Solid Films
[128] Würfel P 2004 Physics of Solar Cells (Weinheim: Wiley) 480–1 99–109
[129] Wu L, LiangHuan F, Li W, Zhang J, Li B, Lei Z, Cai W, [149] Naghavi N et al 2010 Buffer layers and transparent
Cai Y and Zheng J 2007 Effect of ZnTe/ZnTe:Cu complex conducting oxides for chalcopyrite Cu(In,Ga)(S,Se)2
back- contact on device characteristics of CdTe solar cells based thin film photovoltaics: present status and current
Sci. China Ser. E 50 199–205 developments Prog. Photovolt. Res. Appl. 18 411–33
[130] Gessert T A, Asher S, Johnston S, Young M, Dippo P and [150] Adler T, Botros M, Witte W, Hariskos D, Menner R,
Corwine C 2007 Analysis of CdS/CdTe devices Powalla M and Klein A 2014 Valence band offsets at
incorporating a ZnTe: Cu/Ti contact Thin Solid Films Cu(In,Ga)Se2 /Zn(O,S) interfaces Phys. Status Solidi A
515 6103–6 211 1972–80
[131] Barati A, Klein A and Jaegermann W 2009 Deposition and [151] www.solar-frontier.com 2014
characterization of highly p-type antimony doped ZnTe [152] Wei S-H and Zunger A 1993 Band offsets at the
thin films Thin Solid Films 517 2149–52 CdS/CuInSe2 heterojunction Appl. Phys. Lett. 63 2549
[132] Ferekides C S, Marinskiy D, Viswanathan V, Tetali B, [153] Schulmeyer T, Kniese R, Hunger R, Jaegermann W,
Palekis V, Selvaraj P and Morel D L 2000 High Powalla M and Klein A 2004 Influence of Cu(In,Ga)Se2
efficiency CSS CdTe solar cells Thin Solid Films band gap on the valence band offset with CdS Thin Solid
361–2 520–6 Films 451–2 420–3
[133] Wu X, Keane J C, DeHart C, Dhere R G, Albin D S, Duda A [154] Bauknecht A, Blieske U, Kampschulte T, Albert J, Sehnert H,
and Gessert T A 2001 16.4% CdS/CdTe polycrystalline Lux-Steiner M C, Klein A and Jaegermann W 1999 Band
thin-film solar cell Proc. of the 17th European offsets at the ZnSe/CuGaSe2 (001) heterointerface Appl.
Photovoltaic Solar Energy Conf. (Munich, 22–26 October, Phys. Lett. 74 1099–101
2001) p PC1.1 [155] Morkel M, Weinhardt L, Lohmüller B, Heske C, Umbach E,
[134] Perrenoud J, Kranz L, Gretener C, Pianezzi F, Nishiwaki S, Riedl W, Zweigart S and Karg F 2001 Flat
Buecheler S and Tiwari A N 2013 A comprehensive conduction-band alignment at the CdS/CuInSe2 thin-film
picture of Cu doping in CdTe solar cells J. Appl. Phys. solar-cell heterojunction Appl. Phys. Lett. 79 4482–4
114 174505 [156] Terada N et al 2005 Characterization of interface nature and
[135] Sites J and Pan J 2008 Strategies to increase CdTe solar-cell band alignment in CBD-CdS/Cu(In,Ga)Se2 bi-layer
voltage Thin Solid Films 515 6099–102 structure by photoemission and inverse photoemission
[136] Rhoderick E H and Williams R H 1988 Metal-Semiconductor spectroscopy Thin Solid Films 480–1 183–7
Contacts (Monographs in Electrical and Electronic [157] Okano Y, Nakada T and Kunioka A 1998 XPS analysis of
Engeneering vol 19) 2nd edn (Oxford: Clarendon) CdS/CuInSe2 heterojunctions Sol. Energy Mater. Sol. Cells
[137] Tung R T 1992 Electron transport at metal-semiconductor 50 105–10
interfaces: general theory Phys. Rev. B 45 13509–23 [158] Niles D W, Contreras M, Ramanathan K and Noufi R 1996
[138] Schlaf R, Klein A, Pettenkofer C and Jaegermann W 1993 Determination of the valence-band offset of CdS/CIS solar
Laterally inhomogeneous surface potential distribution cell devices by target factor analysis Conf. Record of the
and photovoltage at clustered In/WSe2 (0 0 0 1) interfaces 25th IEEE Photovoltaic Specialists Conf. (Washington,
Phys. Rev. B 48 14242–52 DC, 13–17 May 1996) pp 833–6
[139] Chirilǎ A et al 2013 Potassium-induced surface modification [159] Hunger R, Lebedev M V, Sakurai K, Schulmeyer T,
of Cu(In,Ga)Se2 thin films for high-efficiency solar cells Mayer T, Klein A, Niki S and Jaegermann W 2007
Nat. Mater. 12 1107–11 Junction formation of CuInSe2 with CdS: a comparative
[140] Bär M, Weinhardt L, Heske C, Nishiwaki S and study of ‘dry’ and ‘wet’ interfaces Thin Solid Films
Shafarman W N 2008 Chemical structures of the 515 6112–8
22
[160] Löher T, Jaegermann W and Pettenkofer C 1995 Formation Cu(In,Ga)(S,Se)2 solar cells Phys. Status Solidi A
and electronic properties of the CdS/CuInSe2 (0 1 1) 205 2335–9
heterointerface studied by synchrotron-induced [177] Naghavi N, Spiering S, Powalla M, Cavana B and Lincot D
photoemission J. Appl. Phys. 77 731–8 2003 High-efficiency copper indium gallium diselenide
[161] Schmid D, Ruckh M and Schock H W 1996 A (CIGS) solar cells with indium sulfide buffer layers
comprehensive characterization of the interfaces in deposited by atomic layer chemical vapor deposition
Mo/CIS/CdS/ZnO solar cell structures Sol. Energy Mater. (ALCVD) Prog. Photovolt. Res. Appl. 11 437–43
Sol. Cells 41–2 281–94 [178] Abou-Ras D, Kostorz G, Strohm A, Schock H W and
[162] Niki S, Fons P J, Yamada A, Lacroix Y, Shibata H, Tiwari A N 2005 Interfacial layer formations between
Oyanagi H, Nishitani M, Negami T and Wada T 1999 Cu(In,Ga)Se2 and Inx Sy layers J. Appl. Phys. 98 123512
Effects of the surface Cu2−x Se phase on the growth and [179] Schulmeyer T, Klein A, Kniese R and Powalla M 2004 Band
properties of CuInSe2 films Appl. Phys. Lett. 74 1630–2 offset at the In2 S3 /CuGaSe2 heterointerface Appl. Phys.
[163] Klein A, Löher T, Pettenkofer C and Jaegermann W 1996 Lett. 85 961–3
Chemical interaction of Na with cleaved (0 1 1) surfaces of [180] Nomura R, Konishi K and Matsuda H 1991 Single-source
CuInSe2 J. Appl. Phys. 80 5039 organometallic chemical vapour deposition process for
[164] Klein A and Jaegermann W 1999 Fermi-level dependent sulphide thin films: introduction of a new organometallic
defect formation in Cu-chalcopyrite semiconductors Appl. precursor Bun In(SPri )2 and preparation of In2 S3 thin films
Phys. Lett. 74 2283–5 Thin Solid Films 198 339–45
[165] Gartsman K, Chernyak L, Lyahovitskaya V, Cahen D, [181] Barreau N, Bernède J C, Maliki H E, Marsillac S, Castel X
Didik V, Kozlovsky V, Malkovich R, Skoryatina E and and Pinel J 2002 Recent studies on In2 S3 containing
Usacheva V 1997 Direct evidence for diffusion and oxygen thin films Solid State Commun. 122 445–50
electromigration of Cu in CuInSe2 J. Appl. Phys. [182] Spiering S, Hariskos D, Powalla M, Naghavi N and Lincot D
82 4282–5 2003 Cd-free Cu(In,Ga)Se2 thin-film solar modules with
[166] Klein A, Fritsche J, Jaegermann W, Schön J H, Kloc C and In2 S3 buffer layer by ALCVD Thin Solid Films
Bucher E 2000 Fermi level dependent defect formation at 431 359–63
Cu(In,Ga)Se2 interfaces Appl. Surf. Sci. 166 508–12 [183] Schneikart A, Schimper H-J, Klein A and Jaegermann W
[167] Schulmeyer T, Hunger R, Fritsche R, Jäckel B, Jaegermann 2013 Efficiency limitations of thermally evaporated
W, Klein A, Kniese R and Powalla M 2005 Interfaces of thin-film SnS solar cells J. Phys. D: Appl. Phys. 46 305109
chalcogenide solar cells: a study of the composition at the [184] Gall S, Barreau N, Harel S, Bernède J C and Kessler J 2005
Cu(In,Ga)Se2 /CdS contact Thin Solid Films 480–1 110–7 Material analysis of PVD-grown indium sulphide buffer
[168] Jacobi K, Myler U and Althainz P 1990 Determination of layers for Cu(In,Ga)Se2 -based solar cells Thin Solid Films
band bending at the Si(1 1 3) surface from 480–1 138–41
photovoltage-induced core-level shifts Phys. Rev. B [185] Bär M, Allsop N, Lauermann I and Fischer C H 2007
41 10721–6 Deposition of In2 S3 on Cu(In,Ga)(S,Se)2 thin film solar
[169] Alonso M, Cimino R and Horn K 1990 Surface photovoltage cell absorbers by spray ion layer gas reaction: evidence of
effects in photoemission from metal-GaP(110) interfaces: strong interfacial diffusion Appl. Phys. Lett. 90 132118
importance for band bending evaluation Phys. Rev. Lett. [186] Eicke A, Spiering S, Dresel A and Powalla M 2008 Chemical
64 1947 characterisation of evaporated In2 Sx buffer layers in
[170] Schulmeyer T 2005 Mechanismen der Cu(In,Ga)Se2 thin-film solar cells with SNMS and SIMS
Grenzflächenausbildung des Cu(In,Ga)Se2 -Systems PhD Surf. Interface Anal. 40 830–3
Thesis Technische Universität, Darmstadt [187] Bär M et al 2010 Nondestructive depth-resolved
(http://tuprints.ulb.tu-darmstadt.de/000617/) spectroscopic investigation of the heavily intermixed
[171] Schulmeyer T, Hunger R, Jaegermann W, Klein A, Kniese R In2 S2 /Cu(In,Ga)Se2 interface Appl. Phys. Lett. 96 184101
and Powalla M 2003 Interface formation between [188] Sterner J, Malmström J and Stolt L 2005 Study on ALD
polycrstalline Cu(In,Ga)Se2 and II-VI-compounds Proc. of In2 S3 /Cu(In,Ga)Se2 interface formation Prog. Photovolt.
the 3rd World Conf. on Photovoltaic Energy Conversion Res. Appl. 13 179–93
(Osaka, Japan, 18 May 2003) pp 364–7 [189] Harrison W A 1989 Electronic Structure and the Properties
[172] Platzer-Björkman C, Törndahl T, Abou-Ras D, Malmström J, of Solids (New York: Dover)
Kessler J and Stolt L 2006 Zn(O,S) buffer layers by atomic [190] Scheer R, Luck I and Lewerenz H J 1994 Photoemission
layer deposition in Cu(In,Ga)Se2 based thin film solar study on CuInS2 thin films for efficient solar energy
cells: band alignment and sulfur gradient J. Appl. Phys. conversion Proc. of the 12th European Photovoltaic Solar
100 044506 Energy Conf. (Amsterdam, The Nederlands, April 1994)
[173] Nakada T, Hongo M and Hayashi E 2003 Band offset of high pp 1751–4
efficiency CBD-ZnS/CIGS thin film solar cells Thin Solid [191] Hashimoto Y, Takeuchi K and Ito K 1995 Band alignment at
Films 431–2 242–8 CdS/CuInS2 heterojunction Appl. Phys. Lett. 67 980
[174] Kieven D, Grimm A, Lauermann I, Lux-Steiner M C, Palm J, [192] Weinhardt L, Fuchs O, Groß D, Storch G, Umbach E,
Niesen T and Klenk R 2011 Band alignment at sputtered Dhere N G, Kadam A A, Kulkarni S S and Heske C 2005
ZnSx O1−x /Cu(In,Ga)(Se,S)2 heterojunctions Phys. Status Band alignment at the CdS/Cu(In,Ga)S2 interface in
Solidi 6 294–6 thin-film solar cells Appl. Phys. Lett. 86 062109
[175] Nakada T and Mizutani M 2000 Improved efficiency of [193] Scheer R 2012 Open questions after 20 years of CuInS2
Cu(In,Ga)Se2 thin film solar cells with chemically research Prog. Photovolt. Res. Appl. 20 507–11
deposited ZnS buffer layers by air-annealing-formation of [194] Säuberlich F 2006 Oberflächen und Grenzflächen
homojunction by solid phase diffusion Proc. of the 28th polykristalliner kathodenzerstäubter
IEEE Photovoltaic Specialists Conf. (Anchorage, AK, Zinkoxid-Dünnschichten PhD Thesis Technische
15–22 September 2000) pp 529–34 Universität, Darmstadt
[176] Hubert C, Naghavi N, Etcheberry A, Roussel O, Hariskos D, [195] Ruckh M, Schmid D and Schock H W 1994 Photoemission
Powalla M, Kerrec O and Lincot D 2008 A better studies of the ZnO/CdS interface J. Appl. Phys. 76 5945–8
understanding of the growth mechanism of Zn(S,O,OH) [196] Sieber I, Wanderka N, Urban I, Dörfel I, Schierhorn E,
chemical bath deposited buffer layers for high efficiency Fenske F and Fuhs W 1998 Electron microscopic
23
characterization of reactively sputtered ZnO films with [206] Siol S, Sträter H, Brüggemann R, Brötz J, Bauer G H,
different Al-doping levels Thin Solid Films 330 108–13 Klein A and Jaegermann W 2013 PVD of Copper Sulfide
[197] Mirica E, Kowach G, Evans P and Du H 2004 Morphological Cu2 S for PIN-structured solar cells J. Phys. D: Appl. Phys.
evolution of ZnO thin films deposited by reactive 46 495112
sputtering Cryst. Growth Des. 4 147–56 [207] Liu G, Schulmeyer T, Thissen A, Klein A and Jaegermann W
[198] Yoshino Y, Inoue K, Takeuchi M and Ohwada K 1998 Effects 2003 In situ preparation and interface characterization of
of interface micro structure in crystallization of ZnO thin TiO2 /Cu2 S heterointerface Appl. Phys. Lett. 82 2269–71
films prepared by radio frequency sputtering Vacuum [208] Minami T, Nishi Y, Miyata T and Nomoto J-I 2011
51 601–7 High-efficiency oxide solar cells with ZnO/Cu2 O
[199] Ohtomo A, Kawasaki M, Ohkubo I, Koinuma H, heterojunction fabricated on thermally oxidized Cu2 O
Yasuda T and Segawa Y 1999 Structure and optical sheets Appl. Phys. Express 4 062301
properties of ZnO/Mg0.2 Zn0.8O superlattices Appl. Phys. [209] Minami T, Nishi Y and Miyata T 2013 High-efficiency
Lett. 75 980 Cu2 O-based heterojunction solar cells fabricated using a
[200] Platzer-Björkman C, Lu J, Kessler J and Stolt L 2003 Ga2 O3 thin film as N-Type layer Appl. Phys. Express
Interface study of CuInSe2 /ZnO and Cu(In,Ga)Se2 /ZnO 6 044101
devices using ALD ZnO buffer layers Thin Solid Films [210] Meyer B K et al 2012 Binary copper oxide semiconductors:
431–2 321–5 from materials towards devices Phys. Status Solidi B
[201] Nguyen Q, Rau U, Mamor M, Orgassa K, Schock H W and 249 1487–509
Werner J H 2001 Electrical metastabilities in [211] Matsuzaki K, Nomura K, Yanagi H, Kamiya T, Hirano M and
Cu(In,Ga)Se2 -based solar cells with Inx (OH,S)y , CdS and Hosono H 2008 Epitaxial growth of high mobility Cu2 O
combined buffer layers Proc. of the 17th European thin films and application to p-channel thin film transistor
Photovoltaic Solar Energy Conf. (Munich, 2001) pp Appl. Phys. Lett. 93 202107
1107–10 [212] Ichimura M and Song Y 2011 Band alignment at the
[202] Nguyen Q, Orgassa K, Kötschau I, Rau U and Schock H W Cu2 O/ZnO heterojunction Japan. J. Appl. Phys. 50 051002
2003 Influence of heterointerfaces on the performance of [213] Yang M, Li Y, Cao L and Guo Y 2013 Asymmetric interface
Cu(In,Ga)Se2 solar cells with CdS and In(OHx ,Sy ) buffer band alignments of Cu2 O/ZnO and ZnO/Cu2 O
layers Thin Solid Films 431–2 330–4 heterojunctions J. Alloys Compounds 578 143–7
[203] Barkhouse D A R, Gunawan O, Gokmen T, Todorov T K and [214] Siol S 2014 Kupferchalkogenide für photovoltaische
Mitzi D B 2012 Device characteristics of a 10.1% Anwendungen PhD Thesis Technische Universität,
hydrazine-processed Cu2 ZnSn(Se,S)4 solar cell Prog. Darmstadt
Photovolt. Res. Appl. 20 6–11 [215] Schneikart A 2014 Herstellung und Charakterisierung von
[204] Siebentritt S 2013 Why Are kesterite solar cells not 20% SnS Dünnschichtsolarzellen PhD Thesis Technische
efficient? Thin Solid Films 535 1–4 Universität, Darmstadt
[205] Liu G, Schulmeyer T, Brötz J, Klein A and Jaegermann W [216] Burton L A and Walsh A 2013 Band alignment in SnS
2003 Interface properties and band alignment of Cu2 S/CdS thin-film solar cells: possible origin of the low conversion
thin film solar cells Thin Solid Films 431/432 477–82 efficiency Appl. Phys. Lett. 102 132111
24

1-3-3. Energy Band Alignment in Chalcogenide Thin Film Solar Cells From Photoelectron Spectros

Uploaded by

Copyright:

Available Formats

1-3-3. Energy Band Alignment in Chalcogenide Thin Film Solar Cells From Photoelectron Spectros

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1-3-3. Energy Band Alignment in Chalcogenide Thin Film Solar Cells From Photoelectron Spectros

Uploaded by

Copyright:

Available Formats

Journal of Physics: Condensed Matter

TOPICAL REVIEW Related content

- Efficiency limitations of thermally

- Elongated nanostructures for radial

- Rectifying Effect of the Sr 3 Al 2 O 6 /Ga 2

- Band Alignments, Band Gap, Core Levels,

This content was downloaded from IP address 131.183.180.59 on 10/01/2020 at 21:12

Energy band alignment in chalcogenide

Received 21 July 2014, revised 30 September 2014

Keywords: solar cells, band alignment, photoemission

1. Introduction chalcogenide semiconductor solar cells (Cu(In,Ga)(S,Se)2 and

0953-8984/15/134201+24$33.00 1 © 2015 IOP Publishing Ltd Printed in the UK

Cu(In,Ga)(S,Se)2 and CdTe solar cells are made by

interpretation of the band alignment data given above. The

5.3. Other II-VI buffer layers

Figure 14. Valence band offsets between Cu(In,Ga)Se2 and

of EVB = 1.2 eV is derived, which agrees with the calculated

You might also like