Vous êtes sur la page 1sur 17

Appl. Phys. A 69, 131147 (1999) / Digital Object Identifier (DOI) 10.

1007/s003399900061

Applied Physics A
Materials
Science & Processing
Springer-Verlag 1999

Electronic properties of Cu(In,Ga)Se2 heterojunction solar cellsrecent


achievements, current understanding, and future challenges
U. Rau, H.W. Schock
Institut fr Physikalische Elektronik (IPE), Universitt Stuttgart, Pfaffenwaldring 47, D-70569, Stuttgart, Germany
(E-mail: uwe.rau@ipe.uni-stuttgart.de)
Received: 12 March 1999/Accepted: 28 March 1999/Published online: 24 June 1999

Abstract. The recent achievements of high-efficiency


Cu(In,Ga)Se2 heterojunction solar cells are reviewed with
a special focus on the understanding of the electronic transport properties of the devices. We discuss the basic limitations
of the device performance, the present understanding of electronic device analysis, as well as the role of intrinsic defects
and of the interfaces for the performance of the solar cells.
PACS: 72.20.Jv; 73.40.Lq; 84.60.Jt
The success of micro-electronics essentially is based on the
perfection of single crystals as an indispensable starting
point [1]. Defects, for example, dopants, are then deliberately introduced in order to control the electronic material
properties and to define the functionality of the device, often
after a number of processing steps of two hundred or more.
Concepts of micro-electronics, for instance the passivation
of semiconductor surfaces, but also the selective doping of
the emitter, also lead to remarkable progress in photovoltaics,
a prominent example being the 24.4% efficient silicon solar
cell of the University of New South Wales [2]. The starting
point is still the perfect crystal of a float-zone silicon wafer.
The picture changes drastically when we turn to thin-film photovoltaics. For this technology, success is not counted in units
of bits/m2 as in the case of micro-electronics but rather in
m2 /$ [3]. It is clear that for such a macro-electronic technology, to be realized in areas of km2 , the challenges become
quite different as compared to micro-electronics. All leading
thin-film photovoltaic technologies [4] have to learn to live
with the imperfect. This postulate holds for amorphous Si
(a-Si) [5] and thin-film crystalline Si (c-Si) [6], as well as for
the compound semiconductors CdTe [7] and Cu(In,Ga)Se2 .
Living with the imperfect is especially exciting in the case
of Cu(In,Ga)Se2 . Beyond the thin-film technologies mentioned above, Cu(In,Ga)Se2 solar cells are those with the
highest efficiencies on the laboratory scale as well as on
the level of large-area modules. The technology has proven
their potential for high efficiencies (> 17% realized by different groups [810]), has achieved a certain degree of maturity, and commercial production is on its way at several

places [1113]. Furthermore, Cu(In,Ga)Se2 thin-film modules show excellent outdoor stability [14] and radiation hardness [15, 16]. At a first sight, this achievement is somewhat
surprising in view of the complexity of the system and the
degree of understanding we actually have on the structural
and electronic properties of Cu(In,Ga)Se2 . The knowledge
on this material is still extremely low compared to what is
known from a-Si or c-Si. Hence, most of the achievements of
Cu(In,Ga)Se2 -based solar cells accumulated during the years
are thus initiated rather by intuition than by knowledge-based
technological design. Briefly: it worked first, and was explained later.
This review represents an attempt to equilibrate the balance between achievements and understanding of
Cu(In,Ga)Se2 -based heterojunction devices. We focus on
three points: (i) The description of the electronic transport
properties of these heterostructures and their dependence
on the different processing steps during device preparation.
(ii) The role of intrinsic defects in Cu(In,Ga)Se2 and their
implications for the device performance. (iii) The basic
options to influence and to design the material properties
of Cu(In,Ga)Se2 . For other important issues we refer the
reader to the literature. More about structural properties of
Cu(In,Ga)Se2 can be found, for example, in [1721]; interface properties of Cu(In,Ga)Se2 and related compounds
were recently reviewed by Scheer [22]; an overview on recent developments can be gained also from [2325]; for the
questions on up-scaling and module technologies see, for example, [26], and for economic aspects [3].
The review is organized as follows. In Sect. 1 we briefly
introduce the material and device properties of high-efficiency
Cu(In,Ga)Se2 heterojunction devices. Here, we restrict ourselves to those types of solar cells yielding the highest efficiencies; variations either in the deposition method and
material composition or in the device structure itself are only
briefly mentioned where it is necessary for completeness or to
assist the understanding of the high-efficiency devices. Section 2 deals with the electrical transport properties of the
heterostructure. Here, we start with an examination of the
limitations to the solar cell output parameters in terms of simple textbook knowledge before we summarize the present

132

state of the understanding of electrical transport attained by


new experimental and theoretical approaches for electrical
analysis. The role of defects in Cu(In,Ga)Se2 is highlighted
in Sect. 3. Section 4 gives a short summary of the ingredients
that are currently used to improve and to design the electronic
properties of Cu(In,Ga)Se2 .
1 Material and device properties
1.1 Recent achievements
The best Cu(In,Ga)Se2 solar cells are realized as heterojunctions containing a 50-nm-thick CdS buffer layer and a ZnO
window layer as shown schematically in Fig. 1 together with
the corresponding band diagram. Table 1 summarizes the
photovoltaic output parameters of the to-date highest efficiency Cu(In,Ga)Se2 solar cell [8] and compares them with
those of the best Si solar cell [2] and of the best data from
the two most common other chalcopyrite-based materials,
namely CuGaSe2 [27] and CuInS2 [28] as well as with that of
the best CuInSe2 solar cell [29]. The main body of this review
will be concerned with explaining these data and with understanding how they have been achieved and how they can be
further improved. We first mention the four important technological steps which, during recent years, led to a considerable
improvement of the efficiencies. These steps are, in our opinion, the key elements of the present Cu(In,Ga)Se2 technology.
(i) The film quality has been substantially improved by utilizing the crystallization mechanism induced by the pres-

Table 1. Absorber band-gap energy E g , efficiency , open-circuit voltage


Voc , short-circuit current density jsc , fill factor FF, and area A of the best
Cu(In,Ga)Se2 , CuInSe2 , CuGaSe2 , CuInS2 , and the record Si solar cell
Material

Eg
eV

Voc
mV

jsc
mA/cm2

FF
%

A
cm2

Ref.

Cu(In,Ga)Se2
CuInSe2
CuGaSe2
CuInS2
Si

1.11a
1.04
1.68
1.57
1.15

17.7
15.4
8.3
11.1
24.4

647
515
861
728
696

34.0
41.2
14.2
21.24
42.0

77.2
72.6
67.9
70.9
83.6

0.414
0.38
0.471
0.48
4.00

[8]b
[30]c
[27]b
[28]b
[2]b

a) estimated from the quantum efficiency spectrum in [8], b) confirmed total


area values, c) effective area values (not confirmed)

ence of Cu y Se (y < 2) [9]. This process is further supported by a substrate temperature close to the softening
point of the glass substrate [30].
(ii) The glass substrate was changed from Na-free glass to
Na-containing soda-lime glass [9, 30]. This change has
led to an enormous improvement of efficiency and reliability of the solar cells as well as to a larger process
tolerance. In contrast to first assumptions that this effect
is due to better match of thermal expansion coefficients,
the beneficial impact of Na diffusing from the substrate
through the Mo back contact on the growth of the absorber layer, its structural and electrical properties was
soon recognized. We will discuss the outstanding role of
Na in Sect. 4.1 in some detail.
(iii) Initially the absorbers consisted of pure CuInSe2 . The
partial replacement of In with Ga [31] is a further noticeable change which increases the band gap of the absorber
from 1.04 eV to 1.11.2 eV for the high-efficiency devices (cf. Table 1).
(iv) The counterelectrode for the CuInSe2 absorber of the
earlier cells was a 2-m-thick CdS layer obtained by
physical vapor deposition (PVD). This has been replaced
by a combination of a 50-nm-thin CdS buffer layer from
chemical bath deposition [3234] and a highly conductive ZnO window layer.
The consequences of the issues (i)(iv) on the electronic
properties and the performance of Cu(In,Ga)Se2 solar cells
will be discussed in detail below.

1.2 Material properties and absorber preparation

Fig. 1. a The different layers of a ZnO/CdS/CuIn(Ga)Se2 heterojunction


solar cell. b Band diagram of the heterojunction with the conduction and
valence band energies E C and E V , an applied bias voltage V . The quanp
tities b and bn denote the barriers for holes and electrons as described
in the text, E Fn is the energy distance between the Fermi level and the
ab and
conduction band energy at the CdS/CuIn(Ga)Se2 heterointerface, E V
E Cab are the valence band and conduction band discontinuities at the buffer
absorber interface

The preparation of Cu(In,Ga)Se2 -based solar cells starts with


the deposition of the absorber material on a Mo-coated glass
substrate. The properties of the Mo film and the choice of
the glass substrate are of primary importance for the final
device quality (cf. Sect. 4.1). The absorber material yielding
the highest efficiencies is Cu(In,Ga)Se2 with a Ga/(Ga+In)
ratio of around 20% prepared by co-evaporation from elemental sources. Material deposition and film formation are
performed predominantly during the same processing step.
A maximum temperature of around 550 C is required for
a certain time during this process, preferably towards the end
of growth. There exist several variations of the co-evaporation
process with different sub-steps or stages for the deposition
and growth as will be shown below. The composition of the

133

material with regard to the metals is controlled by the evaporation rates whereas Se is always evaporated in excess. Reasonable efficiencies are also obtained from absorber prepared
by the selenization of metal precursors in the presence of either H2 Se [3537] or elemental Se [38], and by rapid thermal
processing of stacked elemental layers [39]. On the laboratory scale the efficiencies of these preparation routes are
smaller by about 3% (absolute) when compared to the record
values. However, on the module level the differences are
much smaller and modules prepared by the selenization route
are, at the moment, the closest to commercialization [11]. The
advantage of co-evaporation is the flexibility of the method
with regard to variations in the material composition and
preparation conditions during processing.
Photovoltaic-grade Cu(In,Ga)Se2 films have slightly Inrich compositions. The allowed stochiometry deviations are
astonishingly large, thus yielding a wide process-window
with respect to composition. Devices with efficiencies above
14% are obtained from absorbers with a (In + Ga)/(In +
Ga + Cu) ratio between 52% and 64% if the sample contains
Na [40]. This issue has to be discussed in the context of the
phase diagram below. Cu-rich Cu(In,Ga)Se2 displays the segregation of a secondary Cu y Se (y < 2) phase preferably at
the film surface. Even after removal of the secondary phase
from the surface by etching the absorber in KCN, the use
of this material for photovoltaic applications is limited, most
probably due to the high doping density of 1018 cm3 in the
bulk and by the defects at the surface. However, the importance of the Cu-rich composition is given by its role during
film growth. Cu-rich films have grain sizes in excess of 1 m
whereas In-rich films have much smaller grains. A model for
the film growth under Cu-rich compositions comprises the
role of Cu y Se as a flux agent during the growth process of
co-evaporated films [29, 41]. This model for the growth of
Cu(In,Ga)Se2 in the presence of a quasi-liquid surface film of
Cu y Se is highlighted in Fig. 2. For Cu(In,Ga)Se2 prepared by
RTP the role of Cu y Se is similar [39]. Advanced preparation
sequences, therefore, always include a Cu-rich stage during
the growth process and end up with an In-rich overall composition in order to combine the large grains of the Cu-rich
stage with the otherwise more favorable electronic properties
of the In-rich composition. The first example for this kind
of procedure is the so-called Boeing or bi-layer process [42]
which starts with the deposition of Cu-rich Cu(In,Ga)Se2 and

Fig. 2. Model for the growth of polycrystalline Cu(In,Ga)Se2 in the presence of a quasi-liquid surface film of Cu y Se on top of the Cu(In,Ga)Se2
grains (according to [41])

ends with an excess In rate. Another possibility is the inverted


process where first (In,Ga)2 Se3 (likewise In,Ga, and Se from
elemental sources to form that compound) is deposited at
lower temperatures (typically around 300 C). Then, Cu and
Se are evaporated at an elevated temperature until an overall
composition close to stochiometry is reached [43]. This process leads to a smoother film morphology when compared to
the bi-layer process. The most successful version of the inverted process is the so-called three-stage process [44] which
puts the deposition of In, Ga, and Se at the end of an inverted process as described above to ensure the overall In-rich
composition of the film even if the material had been Cu-rich
during the second stage. The three-stage process leads to the
currently best solar cells [8]. This process allows us also to introduce Ga gradients into the absorber and thus enables us to
design a graded band-gap structure [45].
1.3 Device structure and band diagram
Figure 1a displays the layer sequence for the actual highefficiency ZnO/CdS/Cu(In,Ga)Se2 heterojunction solar cells.
Mo-coated glass is used as the substrate for the deposition
of the Cu(In,Ga)Se2 absorber layer of a typical thickness
of 1.52 m with the various methods discussed above.
The CdS buffer layer obtained by chemical bath deposition
(CBD), usually has a thickness of 50 nm and is followed by
the ZnO window layer of a typical thickness of 0.52 m.
The window layer is deposited by radio-frequency (rf) sputtering from a ZnO target or reactive dc sputtering from
a Zn target or by metal-organic chemical vapor deposition
(MOCVD). The conductivity of the window layer is of the
order of 2 103 1 cm1 , usually achieved by doping with
Al. If sputtering is used for deposition a 50100-nm-thick
layer of undoped ZnO (the so-called i-ZnO layer) is deposited
prior to that with Al doping. Often a post-deposition airannealing step at 200 C for one or two minutes is used for
efficiency optimization. For the early cells, this annealing step
was strictly mandatory (see, for example, [46, 47]), but also
for the present high-efficiency devices this step is reported to
increase the efficiency significantly [8]. Device completion is
achieved by an evaporated Al-grid in the case of laboratory
cells and by three patterning steps for monolithic interconnection and lamination in the case of modules.
The realization of the sequence shown in Fig. 1a requires
five deposition steps (including two for the ZnO), none of
them exceeding a temperature of 600 C and is herewith
reasonably simple and energy-efficient to provide a costeffective and reliable technology for large-area photovoltaics.
However, if we count the number of interfaces in Fig. 1a (five)
and the number of chemical elements involved in the heterostructure (ten or eleven if we include Na), we see that the
system is also rather complex. In the following we will see
that each component has a specific role for the properties of
the device.
The equilibrium band diagram for the heterostructure in
Fig. 1b displays the conduction and valence band energies
E C/V of the Cu(In,Ga)Se2 absorber, the CdS buffer layer, and
the ZnO window consisting of the intrinsic and the Al-doped,
degenerate layer. The diagram is purposely kept simple in
order to allow for a discussion of the basic electronic properties of the heterostructure in Sect. 2.1. (The remainder of the

134

paper will then be used to discuss all possible modifications


of Fig. 1b.) Given the band gap energies E g of the three semiconductor materials involved, the band diagram is defined
essentially by the conduction band discontinuities E Cab and
E Cbw at the absorber/buffer and the buffer/window interface. The importance of these quantities is obvious. In the absence of interface states at the buffer/absorber interface these
p
two quantities would define the barriers b and bn indicated
p
in Fig. 1b [48]. The barrier b determines the transport of
holes to the buffer/absorber interface; it would be equal to the
Schottky barrier height if the buffer/window heterojunction
part would be replaced by a metal. In Fig. 1b the definition
of bn = E Fn + E Cab with E Fn as the energy difference
between the (electron) Fermi level and the absorber conduction band at the interface is unusual but practical in view of
the electrons which have to surmount this barrier when leaving the absorber if the device is under working conditions. We
p
will discuss the impact of b on interface recombination in
n
Sect. 2.1 and that of b on the fill factor in Sect. 2.3.
Band discontinuities in terms of valence band offsets
E Vab between semiconductor a and b are usually determined
by photoelectron spectroscopy (for a discussion with respect
to Cu-chalkopyrite surfaces and interfaces see [22]). The
valence band offset between a (011)-oriented Cu(In,Ga)Se2
single crystal and CdS deposited by PVD at room temperature was found by Lher et al. [49]. E Vab = 0.8 (0.2) eV
(and, therefore E Cab = E gCdS E gCIS + E Vab 0.55 eV, with
the band gaps E gCdS 2.4 eV and E gCIS 1.05 eV of CdS and
CuInSe2, respectively). Nelson et al. reported a value of
0.8 eV [50] for a non-oriented sample and a CdS deposition temperature of 200 C. Several authors investigated
the band discontinuity between polycrystalline Cu(In,Ga)Se2
films and CdS and values between 0.6 and 1.3 with a clear
center of mass around 0.9 eV corresponding to conduction
band offset of 0.45 [51]. Wei and Zunger [52] calculated
a theoretical value of E Vab = 1.03 eV which would lead
to E Cab 0.3 eV. The band alignment of polycrystalline
CuInSe2 and Cu(In,Ga)Se2 alloys was examined by Schmid
et al. [53, 54] with the result that the valence band offset is almost independent of the Ga content [55]. In turn, the increase
of the absorber band gap leads to a change of E Cab from
positive to negative values. The band offset between the CdS
buffer and the ZnO window layer was determined by Ruckh
et al. to be 0.4 eV [56].

cut-off of the spectral response at the band gap of 2.4 eV only


a total of 35.5 mA cm2 would be available for the shortcircuit current. However, measurements of the external quantum efficiency (EQE) of typical ZnO/CdS/Cu(In,Ga)Se2
heterostructures unveil that the EQE typically drops only by
about a factor of 0.8 in the wavelength range between the
band gap of CdS and that of the ZnO window layer. About
70%80% of the photons in the wavelength range between
440 nm and 510 nm contribute to jsc because the thin buffer
layer does not absorb all photons and still about 50% of
the electron/hole pairs created in the buffer layer contribute
to the photocurrent due to the relatively low recombination
probability for holes at the buffer/absorber interface [59].
For a wavelength > 510 nm the maximum EQE is between
90%95% for high-efficiency cells. Losses are due to reflection at the front surface and to free-carrier absorption in the
highly doped ZnO window layer.
The open circuit voltage Voc of Cu(In,Ga)Se2 solar cells
is below the theoretical limit defined by radiative recombination. Therefore, it is worth comparing the observed Voc
with the limitations imposed by non-radiative recombination as recombination in the bulk of the absorber material
(space-charge region (SCR) or neutral zone), at internal interfaces (grain boundaries), or at the buffer/absorber interface. The different recombination paths in the volume of the
Cu(In,Ga)Se2 absorber are indicated in the band diagram in
Fig. 3 where we have considered recombination in the neutral bulk (A), recombination in the space-charge region (B),
and recombination at the buffer/absorber interface (C). The
dashed lines indicate that the latter two mechanisms may
be enhanced by tunneling in the presence of a high built-in
electrical field (as will be discussed for process B in more
detail below). The basic equations for these recombination
processes (A-B) can be found in text books, such as [60, 61].
For the limitation by recombination in the neutral zone of the
absorber we find


E g kT
qDn NC NV
Voc =

ln
,
(1)
q
q
jsc NA L D

2 Electronic transport
2.1 Basic limitations
The system Cu(In,Ga)(S,Se)2 covers a wide range of band
gaps from 1.04 eV for CuInSe2 to 2.4 eV for CuGaS2 , thus,
including also the theoretical optimum of E g 1.4 eV for
single cells with a theoretical potential for the conversion efficiency of around 30% under one sun [57]. Nevertheless, the
Cu(In,Ga)Se2 solar cell in Table 1 is far from this theoretical
limit and also far from the best Si solar cell with a similar
band gap.
The current density available from the standard
100-mW cm2 solar spectrum (AM1.5) for a band gap of
1.11 eV (that of the best Cu(In,Ga)Se2 solar cell [8]) is about
43 mA cm2 [58]. If the CdS buffer layer would cause a sharp

Fig. 3. Detail of the band diagram of the ZnO/CdS/CuIn(Ga)Se2 heterostructure indicating the three relevant recombination paths recombination
in the neutral bulk (A), in the space-charge region (B), and at the interface
between the Cu(In,Ga)Se2 absorber and the CdS buffer layer (C). The dotted arrows stand for the tunneling enhancement of the recombination paths
B and C

135

where kT/q is the thermal voltage, Dn the diffusion constant for electrons, NC/V the effective density of states in
the conductance/valence band [62], NA the acceptor density,
and L D the diffusion length of the electrons. If this diffusion
length becomes comparable to the thickness d of the quasineutral zone of the absorber the recombination velocity Sb at
the back contact has to be taken into account and L D in (1) has
to be replaced by
L eff = L D

cosh(l 1 ) + sb sinh(l 1 )
,
sb cosh(l 1 ) + sinh(l 1 )

(2)

with sb = Sb L D /Dn and l = L D /d. Using L D = (Dn n )1/2 ,


where n is the lifetime for electrons, the Voc limitation due to
recombination in the SCR of the absorber may be written in
a form comparable to (1):



E g 2kT
kTDn NC NV

ln
.
Voc =
q
q
jsc Fm L 2D

(3)

Here Fm = (2qNA Vbm /s )1/2 is the electrical field at the position of maximum recombination. The quantity Fm depends
on the doping density NA and the band bending Vbm ; s is the
dielectric constant of the absorber. The dependence of (1) as
well as of (3) on the doping density NA is equal in that sense
that an increase of NA by one order of magnitude yields an
increase of Voc of Voc = kT ln(10)/q 60 mV. However,
improving the open-circuit voltage by increasing the doping
density finds its limitation by the increased Auger recombination in the neutral volume and by the enhancement of
tunneling in the SCR (see [63] and below).
We note further, that (1) and (3) may be written in the
same form also for polycrystalline materials if one uses the

Fig. 4. Functional dependence of the open-circuit voltage Voc on the diffusion length L D for electrons in the Cu(In,Ga)Se2 absorber according to
(1) L D > 0.7 m for and according to (4) for L D < 0.7 m. The assumed
doping density is NA = 1016 cm3 . The dashed lines reflect the influence
of the back-surface recombination velocity Sb for values of Sb = 102 cm/s
(upper curve) and of Sb = 105 cm/s (lower curve) according to (2). The
top axis shows the grain-boundary recombination velocity Sg leading for
poly
a grain size of g = 1 m by (3) to the effective diffusion length L eff of the
p
bottom axis. The right axis denotes the allowed barrier height b for a given
interface recombination velocity for holes of Sp = 104 cm/s according to (5)

effective diffusion length for polycrystalline materials:


h
i1/2
2
poly
L eff = L mono
+
2S
/
g)
,
(D
g
n
eff

(4)

where Sg is the recombination velocity at the grain boundaries


of grains with size g [64]. Figure 4 shows the dependence
of the open-circuit voltage Voc on the diffusion length L D in
Cu(In,Ga)Se2 with a band gap energy of E g = 1.11 eV according to (1) and (3). We see that open-circuit voltages of
650 meV are just at the threshold between the limitation by
recombination in the SCR and recombination in the neutral
volume. The top axis expresses the dependence of Voc on the
recombination velocity Sg for a grain size of g = 1 m and
an infinite bulk diffusion length L D . Thus, the values of Sg
in Fig. 4 represent the limitation by grain-boundary recombination, for example, a maximum recombination velocity
of Sg = 3 103 cm/s is allowed for an open-circuit voltage
of Voc = 650 mV. We conclude, in turn, that grain-boundary
recombination must be low in the present high-efficiency devices. For the influence of grain boundaries see also [65].
The dotted lines in Fig. 4 represent the limitations due
to back-surface recombination velocities of Sb = 102 cm/s
and 105 cm/s for a 1.5-m-thick neutral region. Open-circuit
voltages in the range of 650 mV are already slightly influenced by back-surface recombination, further improvements
by increasing material quality or by thinner cells can be only
achieved if Sb is reasonably low.
In Fig. 4, recombination at the CdS/Cu(In,Ga)Se2 interface is also considered. For this recombination path, the open
circuit limitation reads


p
qSp NV

kT
Voc = b
ln
,
(5)
q
q
jsc
where Sp is the interface recombination velocity for holes and
p
the barrier b as indicated in Fig. 1b. As (5) establishes a linp
ear relationship between b and Voc we have used the right
axis in Fig. 4 to demonstrate that for an assumed interface recombination velocity Sp = 104 cm3 already a relatively high
p
barrier b is needed to maintain Voc = 650 mV. A change of
Sp by one order of magnitude would shift the right axis in
p
Fig. 4 by b = kT ln(10) 60 meV.
Electronic loss in the early cells was essentially explained
by recombination at the buffer/absorber interface [66]. These
cells consisted of a thick CdS window layer deposited by
PVD on the CuInSe2 absorber. The doping density in the
CdS and in the CuInSe2 absorber were approximately equal.
Therefore, the band bending in both heterojunction partners
was also equal. In terms of (5) and the band diagram the openp
circuit voltage was determined by a relatively low value of b
and a high value of Sp . The use of a highly doped ZnO and
of only a several 10-nm-thin CdS buffer layer increased the
band bending in the absorber and decreased the recombination velocity for holes at the heterointerface such that in the
actual state-of-the-art high-efficiency devices it is commonly
believed that recombination in the bulk of the absorber is the
dominant electronic loss mechanism. We note that considerations [67, 68] similar to those sketched above lead to the conclusion that efficiencies close to or even above 20% are possible for ZnO/CdS/Cu(In,Ga)Se2 heterostructure solar cells.
The required open-circuit voltage is then about 700 meV. Figure 4 indicates that in this case the diffusion length becomes

136
p

larger than the cell thickness and the necessary barrier b


is 1.05 eV.
2.2 Recombination loss analysis

The modeling and the analysis of the complicated heterostructure shown in Figs. 1b and 3 often requires numerical
analysis by means of device simulation programs. The programs ADEPT from Purdue University [69] and SCAPS from
the Ghent University [70] are especially suited and frequently
used for the simulation of the present heterostructures. However, the analysis of a larger amount of real devices is better
performed by means of analytical models. As we have seen
above understanding of the open-circuit voltage losses requires the understanding of the diode law underlying the
electron transport of the heterojunction. For the three models
mentioned in Sect. 2.1 and those discussed in the following one can write a general expression for the dark-current
density






qV
E a
qV
j = j0 exp
= j00 exp
exp
(6)
AkT
AkT
AkT
at forward voltage V > 3kT/q. Hence, for the open-circuit
voltage,
 
E a AkT
j00
Voc =

ln
.
(7)
q
q
jsc
The values for the diode ideality factors A, the activation
energy E a , and the weakly temperature-dependent reference
current density j00 for the thermal activation of the saturation
current density j0 can be obtained from (1), (3), (5) for either
case. The practical analysis of devices is performed by measuring currentvoltage characteristics under different illumination intensities rather than the dark characteristics [71]. By
a semilogarithmic plot of the short-circuit current densities
vs. the open-circuit voltages the saturation current density and
the ideality factor are obtained at a given temperature. This
method avoids complications from series resistance effects
which may heavily influence the dark characteristics as will
be discussed in Sect. 2.4. Furthermore, if the diode ideality
factor A is only weakly dependent or independent of temperature a plot of Voc vs. temperature allows one to extrapolate
the activation energy E a . It is found that for high-efficiency
Cu(In,Ga)Se2 and CuIn(Se,S)2 devices the activation energy
is close to the band gap energy of the respective material.
This indicates that Voc is dominated by recombination in the
SCR according to (2) rather than interface recombination
p
(with E a = b according to (3)) [71]. However the ideality factor of Cu(In,Ga)Se2 -based heterojunctions is not 2 but
generally around 1.5 at room temperature. Walter and coworkers [71, 72] proposed a model based on an exponential
distribution of trap states to explain the temperature dependence of the diode ideality factor. The expression derived
in [71, 72] may be written as


1
1
T
=
1+ ,
(8)
A 2
T
where kT is the characteristic energy of the distribution.
Thus, a plot of the inverse ideality vs. temperature should

yield a straight line with an axis intercept at 1/A = 1/2 at


T = 0 K [73]. This evaluation method works remarkably well
in a temperature range between 350 and 200 K [7175] and
therefore explains the relevant recombination losses at room
temperature. However, below 200 K and/or for devices with
a higher doping concentration the ideality factor may become
larger than two, a situation which is not covered by the theory of [71, 72]. A model that accounts also for the tunneling
enhancement of SCR recombination is proposed in [75]. The
current density is still given by (5) and the ideality reads


2
E 00
1
1
T
=
1+
,
(9)
A 2
T
3(kT )2
with the well-known characteristic tunneling energy E 00 [76].
This expression is suitable to describe the temperature dependence of the diode ideality factor of Cu(In,Ga)Se2 -based
devices in a temperature range between 350 K and 100 K [75]
as well as that of the to-date most efficient CuGaSe2 -based
solar cells [77]. An example is given in Fig. 5 where we display the temperature dependence of the inverse diode ideality
factor 1/A of a typical ZnO/CdS/Cu(In,Ga)Se2 heterojunction device. In the temperature range above 250 K the data
follow the straight dashed line given by (8). Thus, tunneling is not significant around room temperature but becomes
more and more important at lower temperatures. The theory
of tunneling-enhanced recombination [75] explains the behavior of the diode ideality factor in the hole temperature
range (full line in Fig. 5 corresponding to (9)).
The most common characterization method of
Cu(In,Ga)Se2 -based heterojunction solar cells in excess of
currentvoltage analysis is the measurement of the quantum
efficiency. The external quantum efficiency (EQE) at a given
wavelength is defined as the number of electron/hole pairs
contributing to the photocurrent divided by the number of
photons incident on the solar cell. A quantitative evaluation
of the EQE can be used to determine the diffusion length L D
if the data are properly corrected for reflection losses and absorption losses in the window material and if the absorption
data of the absorber material is known for the wavelength
regime where the absorption length is in the order of L D [78].
This analysis has been performed in the past by several

Fig. 5. Temperature dependence of the inverse diode ideality factor 1/ A of


a ZnO/CdS/Cu(In,Ga)Se2 heterojunction solar cell (symbols). The dashed
line describes recombination without tunneling (8). The full line corresponds to (9) and explains the tunneling contribution to the recombination
current (indicated by the arrows). For more details see [75]

137

authors for different types of devices [67, 79, 80]. A more


elaborate analysis is possible by using the bi-facial spectral
response of Cu(In,Ga)Se2 solar cells prepared on a semitransparent Mo back contact [81]. However, this method requires
especially prepared test devices. Information on the recombination probability of charge carriers photogenerated in the
buffer layer can be obtained from the blue wavelength regime
of the EQE spectra. Engelhardt et al. found a recombination
probability of about 0.4 for carriers photogenerated in ZnSe
as well as in CdS buffers [59].
An alternative possibility to determine the diffusion
length in solar cells is provided by electron-beam-induced
current (EBIC) measurements. Two approaches are possible:
the planar EBIC, where the electron beam is scanned perpendicular to the device surface, and junction EBIC where
the device is cleaved and the beam is scanned normal to the
cross section [46, 82]. It is important to note that the values
for L D extracted from the EQE measurements [67, 79, 80] as
well as from EBIC [46, 82] are in the order of 0.51.5 m,
and thus, in general somewhat larger than what would lead
to the observed values of the open-circuit voltage (cf. Fig. 4).
We assume that this contradiction is due to the difference of
the material properties at the point of maximum recombination some 50100 nm from the absorber/buffer interface
when compared to the properties deeper in the bulk where the
diffusion length is measured.
Direct access to the minority carrier lifetime n of photogenerated electrons is provided by the investigation of
the photoluminescence decay in Cu(In,Ga)Se2 absorbers
after pulsed photo-excitation at room temperature. Ohnesorge et al. [83] found lifetimes n in a range between 5 and
15 ns and a strong correlation between n measured in bare
Cu(In,Ga)Se2 thin films prepared by RTP and the open circuit
voltage Voc of solar cells processed from respective identical
absorbers. Huang et al. [84] used a (steady-state) dual-beam
optical-modulation technique to determine the lifetime n .
These authors investigated absorbers covered with different
types of buffer layers and find a strong dependence of n
(extending from 1 to 15 ns) on the buffer process.
2.3 Back-contact issues
Recombination at the back contact can become a critical issue
(c.f. Sect. 2.1). Back-contact recombination becomes dominant as soon as the diffusion length becomes equal to or
larger than the thickness of the absorber. Furthermore, if the
Mo/Cu(In,Ga)Se2 contact were a Schottky contact we would
face substantial problems with resistive losses due to the
Schottky barrier. Jaegermann et al. [85] reported a band bending of 0.8 eV for non-reacted Mo on vacuum-cleaved CuInSe2
crystals. This value is not compatible with an ohmic back
contact and would rather indicate a rectifying junction with
minority carrier collection properties in the opposite polarity with respect to the main heterojunction. For Cu(In,Ga)Se2
thin films deposited on Mo it was believed that anomalies in
the low-temperature current voltages (the so-called doublediode behavior) could be due to the blocking character of the
back contact [86]. However, this assumption has been falsified recently, by three point-probe measurements across the
back contact where no voltage drop has been found across
the Mo/Cu(In,Ga)Se2 contact although a double-diode behavior was found in the overall device characteristic [87].

Moreover, analysis of current collection by EBIC indicates


a good collection efficiency for electrons which are generated
close to the back contact [30, 88]. An explanation for these
findings is provided by the fact that during absorber deposition a MoSe2 film forms at the Mo surface [89, 90]. MoSe2 is
a layered semiconductor with p-type conduction, a band gap
of 1.3 eV and weak van der Waals bonding along the c axis.
If the layer were oriented parallel to the contact plane, the
MoSe2 would inhibit adhesion of the absorber as well as electronic transport. Fortunately, due to goodwill of nature, the
c axis is found in parallel and the van der Waals planes thus
perpendicularly to the interface [89]. Due to the larger band
gap of the MoSe2 compared to that of standard Cu(In,Ga)Se2
films this semiconductor layer provides a back-surface field
for photogenerated electrons and at the same time provides
a low resistive contact for the holes.
2.4 Barrier effects at the buffer and the window layer
At first sight the band offset E Cab 0.3 eV between the absorber and the buffer layer as discussed in Sect. 1.1 seems
to represent a substantial barrier hindering the collection of
photogenerated electrons. The buffer/absorber interface as
shown in Fig. 3 can be roughly looked at as a Schottky barrier
with a height bn = E Cab + E Fn and the photocurrent as the
reverse current over this barrier. A simple calculation shows
that the barrier does not provide a large series resistance to
the cell as long as bn < 0.5 eV. The conductance of a Schottky contact with an effective Richardson constant A =
100 A cm2 K2 is G = qA T/k exp[bn /(kT )] = 1.4 S/cm2
for bn = 0.5 eV and at a temperature T = 300 K. A series
resistance of 1 cm2 is acceptable for a device with a reasonable open-circuit voltage. It has been shown by numerical calculations that a conduction band offset E Cab below
0.4 eV does not affect the device performance [91, 92]. However, at lower temperatures or for devices where the type
inversion at the Cu(In,Ga)Se2 is less pronounced, i. e., the
quantity E Fn being larger than 200 meV, the barrier bn becomes important as shown by numerical simulations [93, 94].
The above-mentioned double-diode behavior is nowadays
uniquely ascribed to the influence of the front electrode.
2.5 Metastabilities
The long time relaxation (i.e., for hours and days) of the
open-circuit voltage of Cu(In,Ga)Se2 -based solar cells during
illumination is a commonly observed phenomenon [95, 96].
Fortunately, it turns out that in most cases the open-circuit
voltage increases with illumination time, a situation which
is more favorable than that encountered in a-Si [97]. Consequently, light soaking treatments are systematically used to
re-establish the cell efficiency after thermal treatments [98,99].
A first model for the open-circuit voltage relaxation of
Cu(In,Ga)Se2 solar cells was proposed by Ruberto and Rothwarf [95] in 1986. This model relies on the reduction of
interface recombination at the CdS/Cu(In,Ga)Se2 interface
by additional charges introduced into the CdS buffer layer
either by illumination under open-circuit conditions or by application of forward bias in the dark. The model is based on
the assumption that interface recombination is the dominant

138

recombination mechanism in the solar cells. The increase of


positive charges in the buffer layer is assumed to increase the
p
barrier b and, thus, reduce interface recombination. However, as we have stated above, the open-circuit voltage of the
recent high-efficiency devices are limited by recombination in
the bulk (i.e., in the SCR) rather than at the interface. Since
these devices also show light-soaking effects another mechanism, possibly additional to that proposed in [95] must be at
work.
Important experimental evidence for metastable electrical behavior in the bulk of Cu(In,Ga)Se2 is provided by the
work of Igalson and co-workers [100, 101]. By means of
deep-level transient spectroscopy (DLTS) and photoluminescence these authors have observed a characteristic metastability in the defect distribution of CuInSe2-based thin-film
solar cells: minority carrier injection caused a transformation
of a shallow minority carrier (electron) trap into a majority carrier trap at 260 meV above the valence band. Igalson
et al. attributed this behavior to different charge states D ,
D0 , D+ of the same defect. A similar metastable behavior
was also observed in admittance spectra of different types
of Cu(In,Ga)(Se,S)2 -based solar cells [102]. An important
observation is that of persistent photoconductivity (PPC) in
Cu(In,Ga)Se2 thin films [103] and in single crystals [104].
Relating the persistent trapping of electrons on the one hand
with PPC and on the other hand with a persistent increase of
the charge density in the SCR of the heterojunction as determined by capacitance measurements [73, 105, 106] led to
the model that the gradual decrease of the electrical field in
the SCR leads to a decrease of space-charge recombination
and, finally, to the increase of the open-circuit voltage during
illumination [106]. The band diagrams in Fig. 6 schematically compare the model of Ruberto and Rothwarf [95] with
the more recent considerations of the consequence of PPC in
the bulk of Cu(In,Ga)Se2 [73, 105, 106]. In Fig. 6 solid and
dotted lines represent the band diagram before and after illumination, respectively. An increase of positive charge in the
p
buffer layer increases the barrier b , thus decreasing interface
recombination (cf. (5)). An increase of negative charge in
the buffer layer will decrease the width of the space-charge
region wp in the absorber, and thus increase the electrical field Fm at the point of maximum recombination (cf.
(3)). The active region of bulk recombination will therefore decrease. Thus, light-soaking experiments can be also
used to determine the dominant electronic loss mechanism
in Cu(In,Ga)Se2 -based devices since SCR recombination is
the only loss mechanism that decreases upon increase of
the space-charge density. Interface recombination as well
as all tunneling-related recombination processes would be
enhanced [77, 106].
As PPC seems to be the fundamental mechanism leading to metastabilities in the Cu(In,Ga)Se2 absorber material
it would be interesting to know more about the mechanism
of PPC. In principle, two distinct models are available to
explain PPC which is a phenomenon occurring in a wide
variety of semiconductor materials [107]. The recombination of photogenerated electronhole pairs is either hindered
by macroscopic potential barriers or by a lattice relaxation
caused by carrier capture in a deep trap providing a large time
constant for the re-emission of the carrier. Potential fluctuations on different length scales in Cu(In,Ga)Se2 have been
used by Dirnstorfer et al. [108] to explain photolumines-

Fig. 6. a Light-generated excess positive charges are persistently captured in


p
the buffer layer and lead to an increase of the barrier b . The full (dashed)
lines correspond to the band diagram before (after) illumination. b Lightgenerated excess negative charges persistently trapped in the Cu(In,Ga)Se2
absorber layer lead to a decrease of the width wp of the p-side part of the
space-charge region

cence results on In-rich Cu(In,Ga)Se2 . On the other hand,


the results of Igalson et al. [101] support a lattice relaxation
model. Nadazdy et al. came to a similar conclusion after investigation of the voltage bias dependence of DLTS spectra
of Cu(In,Ga)Se2 [109]. Furthermore, there are some hints
that the migration of Cu (see below) exhibits interference
with light-induced metastabilities [102, 109, 110]. Thus, possibly a Cu-related defect may be responsible for the metastability in Cu(In,Ga)Se2 . Despite of the metastability in the
bulk of the Cu(In,Ga)Se2 absorber material, long-term relaxations in ZnO/CdS/Cu(In,Ga)Se2 heterojunction devices
may be also caused by carrier capture in the buffer material. This effect mainly affects the fill factor of the devices
and its relaxation is evoked by illumination with light that
is absorbed in the buffer layer. For a comparison of this socalled blue-illumination effect with that caused by light absorbed in the Cu(In,Ga)Se2 (red-illumination effect) see, for
example, [102].
3 Bulk and surface properties of Cu(In,Ga)Se2 films
3.1 Phase diagram
Compared to all other materials used for thin-film photovoltaics Cu(In,Ga)Se2 has by far the most complicated phase
diagram. The phase diagram of CuInSe2 has recently been
re-investigated by Haalboom et al. [111] with a special focus on the temperature and composition range relevant for

139

that were actually achieved in recent years by the use of Nacontaining substrates as well as by the use of Cu(In,Ga)Se2
alloys.
3.2 Defect physics of Cu( In, Ga) Se2
The role of defects in the ternary compound CuInSe2 , and
even more in Cu(In,Ga)Se2 , is of special importance because of the large number of possible intrinsic defects and,
of course, because of the role of deep recombination centers
for the performance of the solar cell devices [113]. The 12
possible intrinsic defects in CuInSe2 are listed in Table 2. According to Zhang, Wei, Zunger, and Katayama-Yoshida [114]
the challenge of defect physics in Cu(In,Ga)Se2 is to explain
three unusual effects in this material: (i) the ability to dope
Cu(In,Ga)Se2 with native defects, (ii) the structural tolerance
to large off-stochiometries, and (iii) the electrical benign nature of the structural defects. It is obvious that the answer
to the questions (i)(iii) significantly contributes to the explanation of the photovoltaic performance of this material. It
is known that the doping of CuInSe2 is achieved by intrinsic defects. Samples with p-type conductivity are grown if
the material is Cu-poor and is annealed under high Se vapour
pressure, whereas Cu-rich material with Se deficiency tends
to be n-type [115, 116]. Thus, the Se vacancy VSe is held for
the dominant donor in n-type material (and also for the compensating donor in p-type material) and the Cu vacancy VCu
for the dominant acceptor in Cu-poor p-type material.
By calculating the metal-related defects in CuInSe2 and
CuGaSe2 Zhang et al. [114] found that the defect formation
energies for some intrinsic defects are so low that they can
be heavily influenced by the chemical potential of the components (i.e., by the composition of the material) as well as by
the electrochemical potential of electrons. For VCu in Cu-poor
and stochiometric material even a negative formation energy
is calculated. This would imply the spontaneous formation
of large amounts of this defect under equilibrium conditions.
Low (but positive) formation energies are also found for the
Cu on In anti-site CuIn in Cu-rich material (a shallow acceptor which could be responsible for the p-type conductivity of
Cu-rich, non-Se deficient CuInSe2). The dependence of the
defect formation energies on the electron Fermi-level could
explain the strong tendency of CuInSe2 for self-compensation
and the difficulties in achieving extrinsic doping. The results
of Zhang et al. [114] provide a good theoretical base on defect
formation energies and defect transition energies that exhibits
good agreement with experimentally obtained data.

Fig. 7. Detail of the phase diagram of CuInSe2 along the quasi-binary tie
line Cu2 Se-In2 Se3 according to [110, 111]

the preparation of thin films. Figure 7 shows the four different phases that have been found to be relevant in this range:
the -phase (CuInSe2), the -phase (CuIn3 Se5 ), the -phase
(the high-temperature sphalerite phase), and Cu2y Se. The
useful range of absorber compositions, in terms of the ratio
x = In/(In + Cu), of CuInSe2-based cells is limited by the
segregation of Cu2y Se on the copper-rich side and by a transition from p- to n-type conductivity with increasing x. The
phase diagram [111] suggests that the latter effect arises from
the occurrence of the secondary -phase which shows weak
n-type conductivity [112]. The existence range of the -phase
in pure CuInSe2 on the quasi-binary tie line Cu2 Se-In2Se3 extends from a Cu content of 24% to 24.5% (see Fig. 7). Thus,
the existence range of single-phase CuInSe2 is astonishingly
small and does even not involve the proper stochiometric
composition with 25% Cu. The Cu content of absorbers for
thin-film solar cells varies typically between 22 and 24 at. %
Cu. At the growth temperature this lies within the singlephase region of the -phase. However, at room temperature
one enters the two-phase + region in the equilibrium
phase diagram of Haalboom [111]. Hence, one would expect a tendency for phase separation in photovoltaic-grade
CuInSe2 after deposition. Fortunately, it turns out that partial
replacement of In with Ga as well as the use of Na-containing
substrates considerably widens the process window in terms
of (In+Ga)/(In+Ga+Cu) ratios [110]. Thus, the phase diagram already gives a hint to the substantial improvements
Table 2. Electronic transition energies (difference to the valence/conduction band energy
for acceptor/donor states) and the energies of
formation E for the 12 intrinsic defects in
CuInSe2 . Transition energies in italics are derived from [122] and the formation energies in
brackets from [119]. All bold numbers stem
from [114]. Here the defect-formation energy
is that of the neutral defect in stochiometric
material

Defects (Transition energies and formation energies in eV)


Transition
(/0)

VCu
0.03
0.03

(/2)
(2 /3)
(0/+)

VIn
0.17
0.04
0.41
0.67

Vse
0.04

Cui

Ini

Sei
0.07

0.2
0.11b 0.08
0.60
2.9

a) covalent, b) ionic

3.04
2.8

2.6

2.88
4.4

9.1

CuIn
0.29
0.05
0.58

0.25
0.04
0.34

0.07

(+/2+)
E/eV

InCu

22.4

3.34
1.4

SeCu CuSe SeIn


0.13

0.08

0.06
1.54
1.5

7.5

InSe

0.09

7.5

5.5

5.0

140

A further important result in [114] are the formation energies of defect complexes as (2VCu ,InCu), (CuIn ,InCu ), and
(2Cui ,CuIn ). These are even lower than those of the respective isolated defects. Interestingly, the (2VCu ,InCu ) does not
exhibit an electronic transition within the forbidden gap, in
contrast to the isolated InCu anti-site which is a deep recombination center. As the (2VCu ,InCu ) complex is most likely
to occur in In-rich material it can accommodate for a large
amount of excess In (or likewise deficient Cu) and, at the
same time, maintain the electrical performance of the material. Furthermore, ordered arrays of this complex can be
thought of as the building stones of a series of Cu-In-Se compounds such as CuIn3 Se5 , CuIn5Se8 [117].
The work of Zhang and co-workers [114, 117] on formation energies of defects is based on computations from
first principles. However, as the algorithm used in [114, 117]
was not able to reproduce the proper band gap energy of
CuInSe2, assumptions had to be made to derive the ionization energies of the defects. Also the work does not cover
Se-related defects which however are considered to be important [21, 113, 118]. Earlier work of Neumann [119] and
other groups [120] used the simpler macroscopic cavity
model [119]. Table 2 summarizes the ionization energies and
the defect formation energies of the 12 intrinsic defects in
CuInSe2. The energies (bold values in Table 2) for VCu , VIn ,
Cui , CuIn , InCu , are obtained from a first-principles calculation [114] whereas the formation energies in italics (Table 2)
and for the other defects are calculated from the macroscopic
cavity model [119]. The ionization energies used in Table 2
are either taken from Zhang et al. [114] or from the data
compiled in [121]. Note that the data given in [119, 121] for
the cation defects differ significantly from those computed
in [114].
There are numerous experimental results on defects in
Cu(In,Ga)Se2 obtained from photoluminescence [121124],
Hall measurements [125], DLTS [126128], admittance
spectroscopy [129133], and positron annihilation [134, 135].
As these experiments are conducted on single crystals, epitaxial films, or polycrystalline thin films of different stochiometry, a comparison and a definite picture of the defects is
difficult to achieve. There are different and, unfortunately,
contradictory attempts for an assignment of the observed transitions to the respective defect levels [114, 120, 122].
Here we concentrate on the defects detected in photovoltaic-grade (and thus In-rich) polycrystalline films. In-rich material, in general, is highly compensated with a net acceptor
concentration in the order of 1016 cm3 . The shallow acceptor level VCu (around 30 meV above the valence band [121])
is assumed to be the main dopant in this material. As compensating donors the Se vacancy VSe and the double-donor
InCu are considered. The most prominent defect is an acceptor
level about 270 meV above the valence band which is reported
by several groups [122, 129, 130] and is also detected in single crystals [122, 126]. The transition exhibits a broadened
energy distribution with a tail in the defect density towards
larger energies [130]. This defect is detected not only in Inrich, but to equal amounts also in Cu-rich polycrystalline
materials [136]. Therefore, an assignment of this defect to the
CuIn anti-site is in agreement with the theoretical calculations
of Zhang and co-workers [114] as well as with the proposition of several experimenters [102, 109]. The importance of
this transition is given by the fact that its concentration is

related to the open-circuit voltage of the device [137] and


that the defect seems to be involved in the defect metastability [101, 109] as we have discussed above.
Another transition due to the capture and emission of electrons is frequently detected in high-efficiency Cu(In,Ga)Se2
solar cells. Since the activation energy of this transition varies
between 50 meV and 250 meV dependent on air-annealing
prior to the measurement this transition is attributed to a continuum of donor states at the surface of the Cu(In,Ga)Se2 due
to surface Se vacancies [48, 133]. The activation energy is
then the energetic distance E Fn between the Fermi energy
and the energy of the conduction band at the film surface.
The next puzzle in the defect physics is that of Cu migration. The mobility of Cu+ ions in electrical fields has been
established by the observation of field-generated transistor
structures [138] and by radioactive tracer experiments [139].
The change of doping profiles (obtained from capacitance
voltage measurements) after application of reverse bias at an
elevated temperature of 60 C [102, 110] and at 220 C [109]
is also ascribed to Cu drift caused by the excess electrical
field. This Cu drift is reversible if the device is kept at zero
bias and an elevated temperature for a longer time. Cu migration also takes place during air-annealing of the completed
heterostructure devices at a temperature of 200 C. Here an
additional driving force for the Cu migration is the release of
Cu atoms from its chemical bonds close to the surface due to
oxygenation [48, 140].
Guillemoles and Cahen have shown that the situation of
Cu migration in Cu(In,Ga)Se2 is quite different from that
in Cu2 S/CdS solar cells [141] for two reasons: (i) because
the chemical potential of Cu in Cu(In,Ga)Se2 is far below
that of Cu in Cu2 S such that the system is not able to plate
elemental Cu as in the case of Cu2 S; and (ii) because the
built-in field in the Cu(In,Ga)Se2 absorber opposes the Cu migration into CdS, i.e. Cu is driven towards the bulk of the
Cu(In,Ga)Se2 absorber. This is also observed in most experiments [48, 109, 136]. Interestingly, in some experiments it is
found that the light-induced, short-range (electronic) metastability (PPC) in Cu(In,Ga)Se2 exhibits some interference with
the voltage-bias-induced, long-range (ionic) changes (Cu migration) [102, 109, 110]. These findings indicate that Curelated defects are involved into the short-range processes.
3.3 The free Cu( In, Ga) Se2 surface
The free surfaces of as-grown Cu(In,Ga)Se2 films exhibit
two prominent features: (i) the energy distance of the valence band energy E V to the Fermi level E F is about 1.1 eV
for pure CuInSe2 films [53]. This energy is larger than the
band-gap energy E g . This finding was taken as an indication for a widening of band gap at the surface of the film
in [53]. For Cu(In1x ,Gax )Se2 it was found E F E V = 0.8 eV
(almost independently of the Ga content if x > 0) [142].
(ii) The surface composition of Cu-poor CuInSe2 as well as
that of Cu(In,Ga)Se2 films corresponds to a surface composition of (Ga+In)/(Ga+In+Cu) of approximately 0.75 for
a range of bulk compositions of 0.5 < (Ga + In)/(Ga + In +
Cu) < 0.75. Both observations have led to the assumption
that a phase segregation of Cu(In, Ga)3 Se5 occurs at the surface of the films. The segregation of this -phase would be
compatible with the phase diagram (see above) and, as this

141

material displays n-type conductivity, could yield the explanation for the type inversion. The findings of [53] led to an
increasing interest in the Cu(In,Ga)3 Se5 compound [143] and
also to the fabrication of solar cells based on Cu(In,Ga)3 Se5
absorbers [144].
Unfortunately, the existence of a separate phase on
top of standard Cu(In,Ga)Se2 thin films has to our knowledge not yet been confirmed by structural methods such as
X-ray diffraction, high-resolution transmission electron microscopy, or electron diffraction. Furthermore, if the surface
phase would exhibit the weak n-type conductivity of bulk
Cu(In,Ga)3 Se5 , this would not be sufficient to achieve the
type inversion, as can be concluded from simple chargeneutrality estimations [110]. Moreover, it has been shown
that exposure of the free thin-film surface to ambient air
immediately reduces the type inversion [48]. From surface
photovoltage measurements performed on samples stored in
air it is routinely found that the surface band-bending is below
50 meV, i.e., the Cu(In,Ga)Se2 surface after a long exposure
to air is nearly in flat-band conditions [145].
Based on these arguments, another picture of the surface of Cu(In,Ga)Se2 thin films and of junction formation
has emerged [48, 110]. As sketched in Fig. 8, the type inversion can be viewed as a result of Fermi-level pinning due
to shallow surface acceptors. The positively charged surface
acceptors in Fig. 8 are expected on theoretical grounds for
the metal-terminated (112) surface of CuInSe2 due to the
dangling bond to the missing Se [146, 147]. Under this assumption, the Cu-poor surface composition is a consequence
of the type inversion (rather than its origin) as the Cu is driven
away from the surface by the built-in electrical field [110].
The reason for the vanishing of the type inversion upon air exposure is given by the passivation of the surface donors due
to the reaction of oxygen with the metal terminated surface as
discussed in Sect. 4.1.
Now we are in the situation that, on the one hand, the
simple picture of a buried p/n junction as a consequence
of a segregated surface phase of the Cu(In,Ga)Se2 absorber
is misleading as it suggests that any change at the absorber
surface (air exposure, buffer deposition) would have only minor impact on the electronic properties of the device. This is
obviously not the case. On the other hand, electrical measurements indicate that the Cu-poor surface region (with thickness
> 0) has in fact different electronic properties compared to

the bulk of the absorber [148]. A possible way out is the assumption of a defect-rich, disordered surface layer [110, 136]
which can even have a larger band gap than that of the absorber material.
3.4 Heterojunction formation and the role of buffer
deposition
As mentioned above, one of the four most important technological steps which led to the actual high-efficiency devices
is the introduction of the chemical bath for the deposition
of the CdS buffer layer. The benefit of the CdS layer is
manifold: (i) CBD deposition of CdS provides complete coverage of the rough polycrystalline absorber surface already
at a film thickness of 10 nm [149]. (ii) The layer provides
a protection against damages and chemical reactions resulting from the subsequent ZnO-deposition process. (iii) The
chemical bath removes the natural oxide from the film surface [150] and, thus, re-establishes positively charged surface
states and, as a consequence, the natural type inversion at
the CdS/Cu(In,Ga)Se2 interface. (iv) The Cd ions, reacting
first with the absorber surface, remove elemental Se possibly
by the formation of CdSe. (v) They also diffuse to a certain
extent into the Cu-poor surface layer of the absorber material [151, 152] where they possibly form CdCu donors, thus
providing additional positive charges enhancing the type inversion of the buffer/absorber interface. (vi) As we have discussed above the open-circuit voltage limitations by interface
recombination require also a low surface-recombination velocity in addition to the type inversion of the absorber surface.
Thus one might conclude that also interface states (except
those shallow surface donors responsible for the type inversion) are also passivated by the chemical bath.
4 Film modification options
The purpose of this section is to discuss the fundamental
options that were used so far to modify and to design the
electronic properties of the Cu(In,Ga)Se2 during and after
preparation. We will show that a number of ingredients such
as Na, Ga, O, S (note that we use the symbols of the chemical
in the following, regardless of ionization states or chemical
bindings of the atoms) can be utilized to optimize the electronic properties of the material. However, our present picture
of the impact of these materials is far from being complete
and the interactions to be considered are in any case manifold.
4.1 Influence of sodium

Fig. 8. Model for the formation of the Cu-poor surface of Cu(In,Ga)Se2


films as a consequence of positively charged surface donors and the migration of interstitial Cu away from the interface (according to [110])

The outstanding role of Na for the growth of the Cu(In,Ga)Se2


films has been realized some years ago [9, 30, 40]. In most
cases Na diffuses from the glass substrate in to the absorber [40, 153, 154]. But there are also approaches where
Na is incorporated by the use of Na-containing precursors such as NaSe [155, 156], Na2 O2 [56], NaF [157], and
Na2 S [158]. Also other alkali precursors were investigated
in [157] where Na-containing precursors yielded the best cell
results. The most obvious effects of Na incorporation are
a better film morphology [153, 154] and a higher conductivity of the films [153, 154]. Furthermore, the incorporation of

142

Na induces changes in the defect distribution of the absorber


films [159, 160].
The explanations for the beneficial impact of Na are manifold and it is most likely that the Na incorporation, in fact,
results in a variety of consequences. During film growth the
Na incorporation leads to the formation of NaSex compounds
which slows down the growth of CuInSe2 and could at the
same time facilitate the Se incorporation into the film [161].
Also, the widening of the existence range of the -(CuInSe2)
phase in the phase diagram as discussed above as well as
the reported larger tolerance to the Cu/(In+Ga) ratio of Nacontaining thin films [56] could be explained in this picture.
Furthermore, also the higher conductivity of Na-containing
films could result from the diminished number of compensating Vse donors. Wolf et al. [162] investigated the influence
of Na incorporation on the formation of CuInSe2 films from
stacked elemental layers by means of thin-film calorimetry.
The addition of Na inhibits the growth of CuInSe2 at temperatures below 380 C. According to [162] the retarded phase
formation is responsible for the better morphology in the case
of Na-containing samples.
Another aspect forwarded by Kronik and co-workers [163]
is Na-promoted oxygenation and passivation of grain boundaries. This picture can account for the observed enhancement of the net film doping by Na incorporation due to the
diminished positive charge at the grain boundaries. It has
been observed in fact that the surfaces of Na-containing films
are more prone to oxygenation when compared to Na-free
films [164]. As this model is closely related to the impact of
oxygen we will discuss this aspect in the next section more
closely.
The above explanations deal with the role of Na during growth. However, the amount of Na in device-quality
Cu(In,Ga)Se2 films is in the order 0.1 at. %, i.e., a concentration of 1020 cm3 [160, 165] and one may ask the question:
Where are these tremendous quantities of Na in the finished
absorber? The electronic effect, i.e., the change of effective doping, resulting from Na incorporation is of the order
1016 cm3 , i. e., four orders of magnitude below that of the
absolute Na content [153, 160]. It was believed early that the
main part of Na is situated at the film surface and at the grain
boundaries [154]. Final evidence for this hypothesis was recently found by Niles and co-workers with the help of highspatial-resolution Auger electron spectroscopy [166]. Heske
et al. investigated the behavior of Na [167, 168] on the surface
of polycrystalline Cu(In,Ga)Se2 films by X-ray photoelectron
spectroscopy (XPS). They observed two different species of
Na: (i) The first one, denoted as reacted, was observed on
the air-exposed sample or after storing the sputter-cleaned
sample for three days in ultrahigh vacuum (UHV). The second one, denoted as metallic, was found on clean samples
either after annealing at 410 K in UHV as well as after deliberate Na deposition from a metallic source [167]. Thus, the
latter species is considered as that being active during crystal growth. In addition Heske et al. [168] found an increase of
band bending by approximately 150 meV induced by the deposition of Na. This finding as well of the occurrence of two
different Na species is consistent with results obtained from
vacuum-cleaved single crystals [169].
Another interpretation of the beneficial effect of Na is
based on the incorporation of Na into the Cu(In,Ga)Se2 lattice [165]. Niles and co-workers identified NaSe bonds by

means of XPS and concluded that the Na is built into the lattice replacing In or Ga. The extrinsic defect NaIn/Ga should
then act as an acceptor and improving the p-type conductivity. The incorporation of Na into the Cu(In,Ga)Se2 lattice is
supported by X-ray diffraction measurements indicating an
increased volume of the unit cell [157]. The authors of [157]
assume the Na on a Cu place preventing the formation of
the deep double-donor InCu which would also have a beneficial effect. Schroeder and Rockett [170] found that Na
driven into epitaxial Cu(In,Ga)Se2 films at a temperature of
550 C decreased the degree of compensation by up to a factor 104 . Schroeder and Rockett attributed their findings to
a Na-enhanced re-organization of the defects, which allows
the defects to built electrically passive clusters [170].
We see from the numerous approaches that, despite the
significance of Na incorporation, the benefit is far from being
explained in terms of simple models. However, we feel that in
view of the amounts of Na (0.1 at. %) necessary for optimum
film preparation arguments considering the effect on the film
growth are slightly in favor with respect to those dealing with
the incorporation of Na into the completed film.
4.2 Influence of oxygen
Air-annealing has been one important process step crucial
for the efficiency especially of the early solar cells based
on CuInSe2 [171, 172]. Also often not mentioned explicitly,
an oxygenation step is still used for most of the present-day
high-efficiency devices [8]. The beneficial effect of oxygen
was explained within the defect chemical model of Cahen
and Noufi [146]. In this model the surface defects at grain
boundaries are positively charged Se vacancies VSe . During air-annealing, these sites are passivated by O atoms. By
the decreased charge at the grain boundary the band bending as well as the recombination probability for photogenerated electrons is reduced [146]. As we have already seen
above the surface donors and their neutralization by oxygen are also important for the free Cu(In,Ga)Se2 surface as
well as for the formation of CdS/Cu(In,Ga)Se2 interface [48].
Electrical analysis of oxidized and unoxidized samples revealed the validity of the CahenNoufi model for the earlier
CdS/CuInSe2 devices [173] as well as for the more recent
ZnO/CdS/Cu(In,Ga)Se2 heterostructures [174]. Also the extension of this model for the catalytic effect of Na [163] could
be verified [174].
The intriguing interplay between surface oxygenation and
the deposition of the CdS buffer layer (cf. Sect. 3.4) is visualized in Fig. 9. In the initial state (Fig. 9a) the film surface as well as the grain boundaries are electrically active
due to the positively charged dangling bonds. These charges
cause a large space-charge region within the grain. Airannealing passivates the dangling bonds at both interfaces.
The bands essentially become flat: space-charge essentially
vanishes (Fig. 9b). Eventually, the chemical bath removes the
passivating oxygen and thus re-establishes the beneficial type
inversion of the film surface (Fig. 9c).
4.3 Influence of gallium
The best Cu(In,Ga)Se2 -based solar cells (efficiency of 17.7%)
have Ga contents between 10% and 30% with respect to

143

a)

-phase is less probable in the case of Ga-containing alloys.


As in the discussion of the impact of Na we notice, that in
addition, or likewise, in competition, to the arguments based
on the defects in the finished absorbers, Ga also has a similar
impact to Na on the growth of the absorber material. Braunger
et al. [161] found by in situ conductivity measurements during the growth of CuInSe2 and CuGaSe2 that the growth of
the latter compound is considerably slower compared to the
former. The slower growth in the presence of Ga then might
have the same beneficial consequences on the film properties
as it has the Na-related slowing down of film growth.
4.4 Influence of sulfur

b)

c)
Fig. 9ac. Illustration of the effect of air-annealing of polycrystalline
Cu(In,Ga)Se2 films and that of subsequent CdS deposition by the chemical
bath. a In the initial state grain boundaries and the film surface are positively charged. The space-charge region within the grains counterbalances
these charges. b Air-annealing passivates both surfaces. c The chemical bath
during CdS deposition re-establishes positively charged donors at the film
surface only

the sum of Ga+In and a band-gap energy from 1.1 to


1.2 eV. However, the benefit of Ga incorporation is limited to
Ga/(Ga+In) ratios of approximately 0.3 [175177]. Within
this range, alloying the absorber with moderate contents of
Ga opens the important option to design a graded band gap
through the absorber [45]. Solar cells with CuGaSe2 - and
Cu(In,Ga)Se2 -absorbers with a high Ga-content suffer from
poor collection efficiency [177] and from too high doping
densities leading to tunneling-enhanced recombination. Consequently, the to-date best CuGaSe2 cells are those with the
lowest net charge density in the SCR [27, 77].
However, for Cu(In,Ga)Se2 in the optimum range Ga
seems to play a similar role to Na. The existence range of the
-phase is substantially widened and, consequently, the tolerance to deviations from the Cu/(Cu+In+Ga) stochiometry is
improved by the addition of Ga [110]. Similar to Na incorporation, Ga increases the free-carrier density. The conductivity
of Ga-containing films is found to be one order of magnitude
larger when compared to thin films without Ga [178, 179].
Wei, Zhang, and Zunger [180] explained these features of Ga
addition in terms of the different defect physics of CuGaSe2
as compared to CuInSe2 . The formation energy of the deep
donor GaCu is higher when compared to its counterpart InCu .
Both defects are thought to be the important compensating
donor in the respective Ga- or In-rich materials. Moreover,
the formation of the complex (2VCu , MCu ) with M = Ga or In
is less likely in CuGaSe2 . Thus, it is concluded in [180] that
the ordering of these defects leading to the segregation of the

Within the pentenary system Cu(In,Ga)(S,Se)2 sulfur is one


possibility to tune the band gap of the semiconductor material. The major drawback of the system CuIn(S,Se)2 is the
low charge-carrier density [181] which forces to grow the
film, in general, Cu-rich and to remove segregated CuS by
a KCN etch. Recent achievements with heterojunction solar
cells based on the pentenary alloy Cu(In,Ga)(S,Se)2 [182]
demonstrate a possible way out to compensate the respective
complementary disadvantages of the high-band-gap systems
CuGaSe2 and CuInS2 .
An important aspect is the impact of small amounts
of sulfur within the absorber material prepared by multistage processes within the last stage of absorber preparation [183185]. Improvements of the open-circuit voltages
in the range of more than 100 mV are reported upon sulfurization when compared to un-sulfurized but otherwise
identical samples [184, 185]. The sulfurization effect seems
to be important, especially when preparing heterostructures
with alternative (i.e., non-CdS) buffer layers [183]. It is interesting to note that the addition of S does not seem to influence
the band-gap energy as determined from the spectral response
of the finished devices [184, 186]. The increase of the opencircuit voltage goes rather along with a dramatic change of
the electrical transport most probably due to a passivation of
recombination centers [160, 186].
5 Conclusions
Working with a multinary compound implies dealing with
multinary interdependencies. However, the challenge for the
design of semiconductor devices is always mastering of the
electronic properties of the bulk and those of the interfaces.
The fortunate situation in Cu(In,Ga)Se2 is due to the benign,
forgiving nature of bulk and interfaces, this means of bulk
and interface defects. As the trail to the top, to efficiencies
close to or above 20% could become more and more narrow a profound knowledge of what is to the left and to the
right becomes more and more vital. As the electronic properties of semiconductors are determined by their defects a much
more detailed understanding of the rich defect physics in
Cu(In,Ga)Se2 is mandatory to enable the tailoring of more
efficient solar cells. For thin films, with the required low number of processing steps, this implies that control over defects
must be achieved during two, or at most three, steps, during
film growth, during heterojunction formation, and possibly
during one post-deposition treatment step.

144

Three out of the four cornerstones (i)(iv) for the recent achievements mentioned in Sect. 1.1 concern the growth
of the films: the optimized deposition conditions (i) and the
incorporation of Na (ii) and Ga (iii). However, no detailed
model is available to describe conclusively the growth of
Cu(In,Ga)Se2 and, especially, the impact of Na which in our
opinion is the most important under the different ingredients
available to tune the electronic properties of the absorber.
A clearer understanding of the Cu(In,Ga)Se2 growth would
allow us to find optimized conditions in the wide parameter
space available and, thus, reduce the number of recombination centers and compensating donors and to optimize the
number of shallow acceptors.
The deposition of the buffer layer, or more generally
speaking, the formation of the heterojunction is another critical issue. The surface chemistry taking place during heterojunction formation and also during post-deposition treatments
is decisive for the final device performance. Both processes
heavily affect not only the surface defects (i.e., recombination and charge) and therefore the charge distribution in the
device but also immediately the defects in the bulk of the
absorber material. As recombination under open-circuit conditions takes place only some 50100 nm away from the
surface it is well understood that mastering of the (near) surface chemistry is an extremely challenging business in the
future. Concentrated effort and major progress in these tasks
would not only allow us to push the top efficiencies further
towards 20% but is also required to provide a sound knowledge base for the various attempts for commercialization of
Cu(In,Ga)Se2 solar cells.
Note added in proof : during proof-reading of the article
we learned that a new record efficiency for Cu(In,Ga)Se2 of
18.8% was achieved by the group at the National Renewable
Energy Laboratories (Golden, USA) [187].
Acknowledgements. The authors thank all our colleagues in the IPE for various discussions and the fruitful collaboration with all the groups at our
institute. We are especially grateful to J.H. Werner for his continuous support and his interest in this work. Critical reading of the manuscript by
D. Braunger, A. Jasenek, and M. Grauvogl is especially acknowledged. This
work is to a great part supported by the German Ministry of Research,
Technology, and Education (BMBF) under contract 0328059F. Collaborations with D. Cahen, J.-F. Guillemoles, L. Kronik, and D. Lincot are also
acknowledged. U.R. also thanks colleagues at the University Oldenburg and
the various groups of the FORSOL project (supported by the Bayerische
Forschungsstiftung) for collaboration and discussions.

References
1. H.J. Queisser: Annu. Rev. Mater. Sci. 22, 1 (1992); The Conquest of
the Microchip (Harvard Univ. Press, Cambridge, MA 1985)
2. J. Zhao, A. Wang, M.A. Green, A. Ferrazza: Appl. Phys. Lett. 73, 1991
(1998)
3. K. Zweibel: Prog. Photovolt. Res. Appl. 3, 279 (1995)
4. For a recent overview on thin-film photovoltaics, see: W. Fuhs,
R. Klenk: In Proc. 2nd World Conf. on Photovolt. Energy Conv., ed.
by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (E.
C. Joint Res. Centre, Luxembourg 1998) p. 381
5. For a-Si see: B. Recht, H. Wagner: this issue; H. Keppner: this issue
6. J.H. Werner, R.B. Bergmann, R. Brendel: Adv. Solid State Phys. 34,
115 (1994); see also: R.B. Bergmann: this issue
7. D. Bonnet: Intern. J. Solar Energy 12, 1 (1992); M. Burgelman: this
issue; K. Durose: this issue

8. J.R. Tuttle, J.S. Ward, T.A. Berens, M.A. Contreras, K.R. Ramanathan,
A.L. Tennant, J. Keane, E.D. Cole, K. Emery, R. Noufi: Mater. Res.
Soc. Symp. Proc. 426, 143 (1996)
9. J. Hedstrm, H. Ohlsen, M. Bodegard, A. Kylner, L. Stolt, D. Hariskos,
M. Ruckh, H.W. Schock: In Proc. 23rd IEEE Photovolt. Spec. Conf.
(IEEE, New York 1993) p. 364
10. T. Negami, M. Nishitani, N. Kohara, Y. Hashimoto, T. Wada: Mater.
Res. Soc. Symp. Proc. 426, 278 (1996)
11. R. Gay: Press conference at the 2nd World Conf. on Photovolt. Energy
Conv., Vienna 1998
12. F. Karg, H. Aulich, W. Riedl: In Proc. 14th Europ. Photovolt. Solar Energy Conf., ed. by H.A. Ossenbrink, P. Helm, H. Ehmann (Stephens,
Bedford 1977) p. 2012
13. B. Dimmler, E. Gross, D. Hariskos, F. Kessler, E. Lotter, M. Powalla,
J. Springer, U. Stein, G. Voorwinden, M. Gaeng, S. Schleicher: In Proc.
2nd World Conf. on Photovolt. Energy Conv., ed. by J. Schmid,
H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (E. C. Joint Res.
Centre, Luxembourg 1998) p. 419
14. H.S. Ullal, K. Zweibel, B. v. Roedern: In Proc. 26th IEEE Photovolt.
Spec. Conf. (IEEE, New York 1997) p. 301
15. M. Yamaguchi: J. Appl. Phys. 78, 1476 (1995)
16. H.W. Schock, K. Bogus: In Proc. 2nd World Conf. on Photovolt. Energy Conv., ed. by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann,
E.D. Dunlop (E. C. Joint Res. Centre, Luxembourg 1998) p. 3586
17. J.L. Shay, J.H. Wernick: Ternary Chalkopyrite Semiconductors: Growth,
Electronic Properties, and Applications (Pergamon Press, Oxford
1975)
18. T.J. Coutts, L.L. Kazmerski, S. Wagner (Eds.): Copper Indium Diselenide for Photovoltaic Applications (Elsevier, Amsterdam 1986)
19. L.L. Kazmerski, S. Wagner: In Current Topics in Photovoltaics, ed. by
T.J. Coutts, J.D. Meakin (Academic Press, Orlando 1985) pp. 41110
20. A. Rockett, R.W. Birkmire: J. Appl. Phys. 70, R81 (1991)
21. A. Rockett, F. Abou-Elfotouh, D. Albin, M. Bode, J. Ermer, R. Klenk,
T. Lommason, T.W.F. Russell, R.D. Tomlinson, J. Tuttle, L. Stolt,
T. Walter, T.M. Peterson: Thin Solid Films 237, 1 (1994)
22. R. Scheer: Trends in Vacuum Science and Technology 2, 77 (1997)
23. D. Ginley, A. Catalano, H.W. Schock, C. Eberspacher, T.M. Peterson,
T. Wada (Eds.): Thin Films for Photovoltaic and Related Device Applications, Mater. Res. Soc. Symp. Proc. 426 (Mater. Res. Soc.,
Pittsburgh 1996) pp. 143314
24. V. Nadenau, D. Braunger, D. Hariskos, M. Kaiser, Ch. Kble, A. Oberacker, M. Ruckh, U. Rhle, R. Schffler, D. Schmid, T. Walter,
S. Zweigart, H.W. Schock: Prog. Photov. Res. Appl. 3, 363 (1995)
25. H.W. Schock: Adv. Solid State Phys. 34, 147 (1994)
26. B. Dimmler, H.W. Schock: Prog. Photov. Res. Appl. 4, 425 (1996)
27. V. Nadenau, D. Hariskos, H.W. Schock: In Proc. 14th Europ. Photovolt. Solar Energy Conf., ed. by H.A. Ossenbrink, P. Helm, H. Ehmann
(Stephens, Bedford 1977) p. 1250
28. J. Klaer, J. Bruns, R. Henninger, K. Tpper, R. Klenk, K. Ellmer,
D. Brunig: In Proc. 2nd World Conf. on Photovolt. Energy Conv., ed.
by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (E.
C. Joint Res. Centre, Luxembourg 1998) p. 537
29. R. Klenk, T. Walter, H.W. Schock: Jpn. J. Appl. Phys. 32 (suppl.3), 57
(1993)
30. L. Stolt, J. Hedstrm, J. Kessler, M. Ruckh, K.O. Velthaus, H.W. Schock:
Appl. Phys. Lett. 62, 597 (1993)
31. W.E. Devaney, W.S. Chen, J.M. Stewart, R.A. Mickelsen: IEEE Trans.
Electron Devices ED-37, 428 (1990)
32. R.R. Potter, C. Eberspacher, L.B. Fabick: In Proc. 18th IEEE Photovolt. Spec. Conf. (IEEE, New York 1985) p. 1659
33. R.W. Birkmire, B.E. McCandless, W.N. Shafarman, R.D. Varrin: In
Proc. 9th E. C. Photovolt. Solar Energy Conf., ed. by W. Palz
G.T. Wrixon, P. Helm. (Kluwer, Dordrecht 1989) p. 134
34. R.H. Mauch, M. Ruckh, J. Hedstrm, D. Lincot, J. Kessler, R. Klinger,
L. Stolt, J. Vedel, H.W. Schock: In Proc. 10th E. C. Photovolt. Solar Energy Conf, ed. by A. Luque, G. Sala, W. Palz, G. dos Santos,
P. Helm. (Kluwer, Dordrecht 1991) p. 1415; D. Lincot, J. Vedel: ibid.,
p. 931
35. J.J.M. Binsma, H.A. Van der Linden: Thin Solid Films 97, 237 (1982)
36. T.L. Chu, S.S. Chu, S.C. Lin, J. Yue: J. Electrochem. Soc. 131, 2182
(1984)
37. V.K. Kapur, B.M. Basol, E.S. Tseng: Solar Cells 21, 65 (1987);
B.M. Basol, V.K. Kapur, R.C. Kullberg: Solar Cells 27, 299 (1989)
38. J. Kessler, H. Dittrich, F. Grunwald, H.W. Schock: In Proc. 10th
E. C. Photovolt. Solar Energy Conf., ed. by A. Luque, G. Sala,

145

39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.

60.
61.
62.

63.
64.

65.
66.
67.

68.
69.
70.

W. Palz, G. dos Santos, P. Helm (Kluwer, Dordrecht 1991) p.


879; K. Kushia, A. Shimizu, K. Saito, A. Yamada, M. Konagai:
In Proc. 1st World Conf. Photovolt. Energy Conv. (IEEE, New York
1994) p. 87
V. Probst, F. Karg, J. Rimmasch, W. Riedl, W. Stetter, H. Harms,
O. Eibl: Mater. Res. Soc. Proc. 426, 165 (1996)
M. Ruckh, D. Schmid, M. Kaiser, R. Schffler, T. Walter, H.W. Schock:
In Proc. 1st World Conf. Photovolt. Energy Conv. (IEEE, New York
1994) p. 156
R. Klenk, T. Walter, H.W. Schock, D. Cahen: Adv. Mater. 5, 114
(1993)
R.A. Mickelsen, W.S. Chen: Appl. Phys. Lett. 36, 371 (1980); USPatent 4,335,266 (1982)
J. Kessler, D. Schmid, H. Dittrich, H.W. Schock: In Proc. 12th Europ. Photovolt. Solar Energy Conf., ed. by R. Hill, W. Palz, P. Helm
(Stephens, Bedford 1994) p. 648
A.M. Gabor, J.R. Tuttle, D.S. Albin, M.A. Contreras, R. Noufi,
A.M. Hermann: Appl. Phys. Lett. 65, 198 (1994)
A.M. Gabor, J.R. Tuttle, A. Swartzlander, A.L. Tennant, M.A. Contreras, R. Noufi: In Proc. 1st World Conf. Photovolt. Energy Conv.
(IEEE, New York 1994) p. 83
R.J. Matson, R. Noufi, R.K. Ahrenkiel, R.C. Powell, D. Cahen: Solar
Cells 16, 495 (1986)
M. Roy, J.E. Phillips: In Proc. 21st IEEE Photovolt. Spec. Conf.
(IEEE, New York 1990) p. 743
U. Rau, D. Braunger, R. Herberholz, H.W. Schock, J.-F. Guillemoles,
L. Kronik, D. Cahen: J. Appl. Phys. 86, 497 (1999)
T. Lher, W. Jaegermann, C. Pettenkofer: J. Appl. Phys. 77, 731 (1995)
A.J. Nelson, C.R. Schwerdtfeger, S.-H. Wei, A. Zunger, D. Rioux,
R. Patel, H. Hchst: Appl. Phys. Lett. 62, 2557 (1993)
cf. the compilation of data given in [22]
S.-H. Wei, A. Zunger: Appl. Phys. Lett. 63, 2549 (1993)
D. Schmid, M. Ruckh, F. Grunwald, H.W. Schock: J. Appl. Phys. 73,
2902 (1993)
D. Schmid, M. Ruckh, H.W. Schock: Appl. Surf. Sci. 103, 409 (1996)
D. Schmid, D. Braunger: private communication
M. Ruckh, D. Schmid, H.W. Schock: J. Appl. Phys. 76, 5945 (1994)
W. Shockley, H.J. Queisser: J. Appl. Phys. 32, 510 (1961)
cf. Appendix C in: M.A. Green: Silicon Solar Cells, Advanced Principles and Practice (Univ. New South Wales, Sydney 1995)
F. Engelhardt, L. Bornemann, M. Kntges, Th. Meyer, J. Parisi,
E. Pschorr-Schoberer, B. Hahn, W. Gebhardt, W. Riedl, F. Karg,
U. Rau: In Proc. 2nd World Conf. on Photovolt. Energy Conv., ed.
by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (E.
C. Joint Res. Centre, Luxembourg 1998) p. 1153; Prog. Photov. Res.
Appl. 1999 (in press)
R.H. Bube: Photoelectronic Properties of Semiconductors (University
Press, Cambridge 1992)
H.J. Hovel: In Semiconductors and Semimetals, Vol. 11, ed. by
Willardson and Beer (Academic Press, New York 1975)
For our calculations we have used the values NC = 6.7 1017 cm3
and NV = 1.5 1019 cm3 resulting from the (density of states) effective masses m e = 0.09m 0 and m p = 0.71m 0 for electrons and holes (m 0
being the free electron mass) according to: H. Neumann: Solar Cells
16, 317 (1986)
M.A. Green: Prog. Photovolt. Res. Appl. 4, 375 (1996)
R. Brendel, U. Rau: In Polycrystalline Semiconductors V - Bulk Materials: Thin Films, and Devices, ed. by J.H. Werner, H.P. Strunk,
H.W. Schock, (Scitech Publ., Uettikon am See, Switzerland 1999)
p. 81
J.R. Sites, J.E. Granata, J.F. Hiltner: Solar Energy Mater. Solar Cells
55, 43 (1998)
M. Eron, A. Rothwarf: Appl. Phys. Lett. 44, 131 (1984), J. Appl. Phys.
57, 2275 (1985)
K.W. Mitchell, H.I. Liu: In Proc. 20th IEEE Photovolt. Spec. Conf.
(IEEE, New York, 1988) p. 1461; K.W. Mitchell: In Proc. 9th E. C.
Photovolt. Solar Energy Conf., ed. by W. Palz G.T. Wrixon, P. Helm
(Kluwer, Dordrecht 1989) p. 292
R. Menner, H.W. Schock: In Proc. 11th E. C. Photovolt. Solar Energy
Conf., ed. by L. Guimaraes, W. Palz, C. de Reyff, H. Kiess, P. Helm
(Harwood, Chur, Switzerland 1993) p. 834
J.L. Gray: In Proc. 25th IEEE Photovolt. Spec. Conf. (IEEE, New York
1996) p. 905
A. Niemeegers, M. Burgelman: In Proc. 25th IEEE Photovolt. Spec.
Conf. (IEEE, New York 1996) p. 901

71. T. Walter, R. Menner, Ch. Kble, H.W. Schock: In Proc. 12th Europ.
Photovolt. Solar Energy Conf., ed. by R. Hill, W. Palz, P. Helm
(Stephens, Bedford 1994) p. 1755
72. T. Walter, R. Herberholz, H.W. Schock: Solid State Phen. 51-52, 309
(1996)
73. F. Engelhardt, M. Schmidt, Th. Meyer, O. Seifert, J. Parisi, U. Rau:
Phys. Lett. A 245, 489 (1998)
74. J.E. Phillips, W.N. Shafarman, E. Shan: In Proc. 1st World Conf. Photovolt. Energy Conv. (IEEE, New York 1994) p. 303
75. U. Rau: Appl. Phys. Lett. 74, 111 (1999)
76. F.A. Padovani, R. Stratton: Solid-State Electron. 9, 695 (1966), see
also E.H. Rhoderick, R.H. Williams: Metal-Semiconductor Contacts,
2nd ed. (Clarendon Press, Oxford 1988) p. 96
77. V. Nadenau, U. Rau, A. Jasenek, H.W. Schock: unpublished
78. N.D. Arora, S.G. Chamberlain, D.J. Roulston: Appl. Phys. Lett. 37,
325 (1980)
79. R. Klenk, H.W. Schock: In Proc. 12th Europ. Photovolt. Solar Energy
Conf., ed. by R. Hill, W. Palz, P. Helm (Stephens, Bedford 1994) p.
1588
80. J. Parisi, D. Hilburger, M. Schmitt, U. Rau: Solar Energy Mater. Solar
Cells 50, 79 (1998)
81. J.E. Phillips: In Proc. 21st IEEE Photovolt. Spec. Conf. (IEEE, New
York 1990) p. 782
82. G. Jger-Waldau, D. Schmid, A. Jger-Waldau: J. Phys. IV 1, C6-132
(1991)
83. B. Ohnesorge, R. Weigand, G. Bacher, A. Forchel, W. Riedl, F. Karg:
Appl. Phys. Lett. 73, 1224 (1998)
84. C.H. Huang, S.S. Li, B.J. Stanberry, C.H. Chang, T.J. Anderson: In
Proc. 26th IEEE Photovolt. Spec. Conf. (IEEE, New York 1997)
p. 407
85. W. Jaegermann, T. Lher, C. Pettenkofer: Crystal Res. Techn. 31, 273
(1996)
86. W. Miller, L.C. Olsen: IEEE Trans. Electron. Devices ED-31, 654
(1984)
87. W.N. Shafarman, J.E. Phillips: In Proc. 25th IEEE Photovolt. Spec.
Conf. (IEEE, New York 1996) p. 841
88. P.E. Russell, O. Jamjoum, R.K. Ahrenkiel, L.L. Kazmerski, R.A. Mickelsen, W.S. Chen: Appl. Phys. Lett. 40, 995 (1982)
89. T. Wada, N. Kohara, T. Negami, M. Nishitani: Jpn. J. Appl. Phys. 35,
1253 (1996)
90. R. Takei, H. Tanino, S. Chichibu, H. Nakanishi: J. Appl. Phys. 79,
2793, (1996)
91. G.B. Turner, R.J. Schwartz, J.L. Gray: In Proc. 20th IEEE Photovolt.
Spec. Conf. (IEEE, New York 1988) p. 1457
92. A. Niemegeers, M. Burgelman, A. De Vos: Appl. Phys. Lett. 67, 843
(1995)
93. M. Topic, F. Smole, J. Furlan: Solar Energy Mater. Solar Cells 49, 311
(1997)
94. I.L. Eisgruber, J.E. Granata, J.R. Sites, J. Hou, J. Kessler: Solar Energy
Mater. Solar Cells 53, 367 (1998)
95. M.N. Ruberto, A. Rothwarf: J. Appl. Phys. 61, 4662 (1987)
96. R.A. Sasala, J.R. Sites: In Proc. 23rd IEEE Photovolt. Spec. Conf.
(IEEE, New York 1993), p. 543
97. D.L. Staebler, C.R. Wronski: Appl. Phys. Lett. 31, 292 (1977)
98. K. Kushia, M. Tachiyuki, T. Kase, I. Sugiyama, Y. Nagoya, D. Okumura, M. Sato, O. Yamase, H. Takeshita: Solar Energy Mater. Solar
Cells 49, 277 (1997)
99. F. Karg, H. Calwer, J. Rimmasch, V. Probst, W. Riedl, W. Stetter,
H. Vogt, M. Lampert: Inst. Phys. Conf. Ser. 152E, 909 (1998)
100. M. Igalson, R. Bacewicz: In Proc. 12th Europ. Photovolt. Energy
Conf., ed. by R. Hill, W. Palz, P. Helm (Stephens, Bedford 1994) p.
1584; Cryst. Res. Technol. 31, 445 (1996)
101. M. Igalson, H.W. Schock: J. Appl. Phys. 80, 5765 (1996)
102. U. Rau, A. Jasenek, R. Herberholz, H.W. Schock, J.F. Guillemoles,
D. Lincot, L. Kronik: In Proc. 2nd World Conf. on Photovolt. Energy Conv., ed. by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann,
E.D. Dunlop (E. C. Joint Res. Centre, Luxembourg 1998) p. 428
103. U. Rau, M. Schmitt, J. Parisi, W. Riedl, F. Karg: Appl. Phys. Lett. 73,
223 (1998)
104. O. Seifert, F. Engelhardt, Th. Meyer, M.T. Hirsch, J. Parisi, C. Beilharz, M. Schmitt, U. Rau: Inst. Phys. Conf. Ser. 152E, 253 (1998)
105. Th. Meyer, F. Engelhardt, J. Parisi, M. Schmidt, U. Rau: In Proc. 2nd
World Conf. on Photovolt. Energy Conv., ed. by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop, (E. C. Joint Res. Centre,
Luxembourg 1998) p. 1157

146
106. Th. Meyer, M. Schmidt, F. Engelhardt, J. Parisi, U. Rau: unpublished
107. D. Redfield, R.H. Bube: Photoinduced Defects in Semiconductors
(Cambridge Univ. Press, Cambridge 1996)
108. I. Dirnstorfer, Mt. Wagner, D.M. Hoffmann, M.D. Lampert, F. Karg,
B.K. Meyer: Phys. Status Solidi A 168, 163 (1998)
109. V. Nadazdy, M. Yakushev, E.H. Djebbar, A.E. Hill, R.D. Tomlinson:
J. Appl. Phys. 84, 4322 (1998)
110. R. Herberholz, H.W. Schock, U. Rau, J.H. Werner, T. Haalboom,
T. Gdecke, F. Ernst, C. Beilharz, K.W. Benz, D. Cahen: In Proc. 26th
IEEE Photovolt. Spec. Conf. (IEEE, Piscataway 1998) p. 323; R. Herberholz, U. Rau, H.W. Schock, T. Haalboom, T. Gdecke, F. Ernst,
C. Beilharz, K.W. Benz, D. Cahen: Eur. Phys. J. AP 6, 131 (1999)
111. T. Haalboom, T. Gdecke, F. Ernst, M. Rhle, R. Herberholz,
H.W. Schock, C. Beilharz, K.W. Benz: Inst. Phys. Conf. Ser. 152E,
249 (1998)
112. M.A. Contreras, H. Wiesner, D. Niles, K. Ramanathan, R. Matson,
J. Tuttle, J. Keane, R. Noufi: In Proc. of the 25th IEEE Photovolt. Spec.
Conf. (IEEE, New York 1996) p. 809
113. For an instructive introduction into the defect physics of CuInSe2 ,
see: D. Cahen: In Proc. 7th Intern. Conf. Tern. Multin. Comp. (Mater.
Res. Soc., Pittsburgh 1987) p. 433; for recent discussions on the
topic, see: M. Burgelman, F. Engelhardt, J.-F. Guillemoles, R. Herberholz, M. Igalson, R. Klenk, M. Lampert, T. Meyer, V. Nadenau,
A. Niemegeers, J. Parisi, U. Rau, H.W. Schock, M. Schmitt, O. Seifert,
T. Walter, S. Zott: Prog. Photovolt. Res. Appl. 5, 121 (1997)
114. S.B. Zhang, S.H. Wei, A. Zunger, H. Katayama-Yoshida: Phys. Rev. B
57, 9642 (1998)
115. P. Migliorato, J.L. Shay, H.M. Kasper, S. Wagner: J. Appl. Phys. 43,
2469 (1975)
116. R. Noufi, R. Axton, C. Herrington, S.K. Deb: Appl. Phys. Lett. 45, 668
(1984)
117. S.B. Zhang, S.H. Wei, A. Zunger: Phys. Rev. Lett. 78, 4059 (1997)
118. C. Rincon, C. Bellabarba, J. Gonzalez, G. Sanchez Perez: Solar Cells
16, 335 (1986)
119. H. Neumann: Cryst. Res. Techn. 18, 901 (1983)
120. C. Rincon, S.M. Wasim: In Proc. 7th Intern. Conf. Tern. Multin.
Comp. (Mater. Res. Soc., Pittsburgh 1987) p. 443
121. F.A. Abou-Elfotouh, H. Moutinho, A. Bakry, T.J. Coutts, L.L. Kazmerski: Solar Cells 30, 151 (1991)
122. F.A. Abou-Elfotouh, L.L. Kazmerski, H.R. Moutinho, J.M. Wissel,
R.G. Dhere, A.J. Nelson, A.M. Bakry: J. Vac. Sci. Technol. A 9, 554
(1991)
123. S. Zott, K. Leo, M. Ruckh, H.W. Schock: Appl. Phys. Lett. 68, 1144
(1996); J. Appl. Phys. 82, 356 (1997)
124. Mt. Wagner, I. Dirnstorfer, D.M. Hofmann, M.D. Lampert, F. Karg,
B.K. Meyer: Phys. Status Solidi A 167, 131 (1998)
125. S.M. Wasim: Solar Cells 16, 289 (1986)
126. A.L. Li, I. Shih: J. Electron. Mater. 22, 195 (1993)
127. N. Christoforou, J.D. Leslie, S. Damaskinos: Solar Cells 26, 197 (1989)
128. W.W. Lam, L.S. Yip, J.E. Greenspan, I. Shih: Solar Energy Mater. Solar Cells 50, 57 (1998)
129. M. Schmitt, U. Rau, J. Parisi: In Proc. 13th Europ. Photovolt. Solar Energy Conf., ed. by W. Freiesleben, W. Palz, H.A. Ossenbrink,
P. Helm (Stephens, Bedford 1995) p. 1969
130. R. Herberholz, T. Walter, C. Mller, T. Friedlmeier, H.W. Schock,
M.C. Lux-Steiner, V. Alberts: Appl. Phys. Lett. 69, 2888 (1996)
131. T. Walter, R. Herberholz, C. Mller, H.W. Schock: J. Appl. Phys. 80,
4411 (1996)
132. N.D. Singh, R. Scheer, K. Kliefoth, W. Fssel, W. Fuhs: In Proc. 26th
IEEE Photovolt. Spec. Conf. (IEEE, New York 1997) p. 467
133. R. Herberholz, M. Igalson, H.W. Schock: J. Appl. Phys. 83, 318
(1998)
134. A.J. Nelson, A.M. Gabor, M.A. Contreras, J.R. Tuttle, R. Noufi,
P.E. Sobol, P. Asoka-Kumar, K.G. Lynn: J. Appl. Phys. 78, 269 (1995)
135. A. Polity, R. Krause-Rehberg, T.E.M. Staub, M.J. Puska, J. Klais,
H.J. Mller, B.K. Meyer: J. Appl. Phys. 83, 71 (1998)
136. R. Herberholz: Inst. Phys. Conf. Ser. 152E, 733 (1998)
137. R. Herberholz, D. Braunger, H.W. Schock, T. Haalboom, F. Ernst: In
Proc. 14th Europ. Photovolt. Solar Energy Conf., ed. by H.A. Ossenbrink, P. Helm, H. Ehmann (Stephens, Bedford 1977) p. 1246
138. L. Chernyak, K. Gartsman, D. Cahen, O.M. Staffsud: J. Phys. Chem.
Sol. 56, 1165 (1995)
139. K. Gartsman, L. Chernyak, V. Lyahovitskaya, D. Cahen, V. Didik,
V. Kozlovsky, R. Malkovich, E. Skoryatina, V. Usacheva: J. Appl.
Phys. 82, 4282 (1997)

140. L. Kronik, D. Cahen, U. Rau, R. Herberholz, H.W. Schock, J.-F. Guillemoles: In Proc. 2nd World Conf. on Photovolt. Energy Conv., ed. by
J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (E. C.
Joint Res. Centre, Luxembourg 1998) p. 453
141. J.-F. Guillemoles, D. Cahen: Cryst. Res. Technol. 31, 147 (1996); Ionics 2, 143 (1996)
142. D. Schmid, M. Ruckh, H.W. Schock: Solar Energy Mater. Solar Cells
41/42, 281 (1996)
143. M.A. Contreras, H. Wiesner, R. Matson, J. Tuttle, K. Ramanathan,
R. Noufi: Mater. Res. Soc. Proc. 426, 243 (1996)
144. N. Kohara, T. Negami, M. Nishitani, Y. Hashimoto, T. Wada: Appl.
Phys. Lett. 71, 835 (1997)
145. L. Kronik, L. Burstein, M. Leibovitch, Y. Shapira, D. Gal, E. Moons,
J. Beier, G. Hodes, D. Cahen, D. Hariskos, R. Klenk, H.W. Schock:
Appl. Phys. Lett. 67, 1405 (1995)
146. D. Cahen, R. Noufi: Appl. Phys. Lett. 54, 558 (1989), Solar Cells 30,
53 (1991)
147. J.A. Rodriguez, L. Quiroga, A. Camacho, R. Baquero: Braz J. Phys.
26, 274 (1996)
148. A. Niemeegers, M. Burgelman, R. Herberholz, U. Rau, D. Hariskos,
H.W. Schock: Prog. Photovolt. Res. Appl. 6, 407 (1998)
149. T.M. Friedlmeier, D. Braunger, D. Hariskos, M. Kaiser, H.N. Wanka,
H.W. Schock: In Proc. 25th IEEE Photovolt. Spec. Conf. (IEEE, New
York 1996) p. 845
150. J. Kessler, K.O. Velthaus, M. Ruckh, R. Laichinger, H.W. Schock,
D. Lincot, R. Ortega, J. Vedel: In Proc. 6th Intern. Photovolt. Science
Eng. Conf., ed. by B.K. Das, S.N. Singh (Oxford Publ., New Dehli,
India 1992) p. 1005
151. K. Ramanathan, H. Wiesner, S. Asher, D. Niles, R.N. Bhattacharya,
J. Keane, M.A. Contreras, R. Noufi: In Proc. 2nd World Conf. on Photovolt. Energy Conv., ed. by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (E. C. Joint Res. Centre, Luxembourg 1998) p. 477
152. T. Wada, S. Hayashi, Y. Hashimoto, S Nishiwaki, T. Sato, M. Nishitina:
In Proc. 2nd World Conf. on Photovolt. Energy Conv., ed. by
J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (E. C.
Joint Res. Centre, Luxembourg 1998) p. 403
153. B.M. Basol, V.K. Kapur, C.R. Leidholm, A. Minnick, A. Halani: In
Proc. 1st World Conf. on Photovolt. Energy Conv. (IEEE, New York
1994) p. 148; D.F. Dawson-Elli, C.B. Moore, R.R. Gay, C.L. Jensen:
ibid., p. 152
154. M. Bodegard, L. Stolt, J. Hedstrom: In Proc. 12th Europ. Solar Energy
Conf., ed. by R. Hill, W. Palz, P. Helm (Stephens, Bedford 1994) p. 1743
155. J. Holz, F. Karg, H. v. Phillipsborn (Eds.): In Proc. 12th Europ. Solar Energy Conf., ed. by R. Hill, W. Palz, P. Helm (Stephens, Bedford
1994) p. 1592
156. T. Nakada, D. Iga, H. Ohbo, A. Kunioka: Jpn. J. Appl. Phys. 36, 732
(1997)
157. M.A. Contreras, B. Egaas, P. Dippo, J. Webb, J. Granata, K. Ramanathan, S. Asher, A. Swartzlander, R. Noufi: In Proc. 26th IEEE
Photov. Spec. Conf. (IEEE, New York 1997) p. 359
158. T. Nakada, T. Mise, T. Kume, A. Kunioka: In Proc. 2nd World Conf.
on Photovolt. Energy Conv., ed. by J. Schmid, H.A. Ossenbrink,
P. Helm, H. Ehmann, E.D. Dunlop (E. C. Joint Res. Centre, Luxembourg 1998) p. 413
159. B.M. Keyes, F. Hasoon, P. Dippo, A. Balcioglu, F. Aboulfotuh: In Proc.
26th IEEE Photovolt. Spec. Conf. (IEEE, New York 1997) p. 845
160. U. Rau, M. Schmitt, F. Engelhardt, O. Seifert, J. Parisi, W. Riedl,
J. Rimmasch, F. Karg: Solid State Commun. 107, 59 (1998)
161. D. Braunger, S. Zweigart, H.W. Schock: In Proc. 2nd World Conf. on
Photovolt. Energy Conv., ed. by J. Schmid, H.A. Ossenbrink, P. Helm,
H. Ehmann, E.D. Dunlop, (E. C. Joint Res. Centre, Luxembourg 1998)
p. 1113
162. D. Wolf, G. Mller, W. Stetter, F. Karg: In Proc. 2nd World Conf. on
Photovolt. Energy Conv., ed. by J. Schmid, H.A. Ossenbrink, P. Helm,
H. Ehmann, E.D. Dunlop, (E. C. Joint Res. Centre, Luxembourg 1998)
p. 2426
163. L. Kronik, D. Cahen, H.W. Schock: Adv. Mater. 10, 31 (1998)
164. M. Ruckh, D. Schmid, M. Kaiser, R. Schffler, H.W. Schock: Solar Energy Mater. Solar Cells 41/42, 335 (1996)
165. D.W. Niles, K. Ramanathan, F. Haason, R. Noufi, B.J. Tielsch, J. E
Fulghum: J. Vac. Sci. Technol. A 15, 3044 (1997)
166. D.W. Niles, M. Al-Jassim, K. Ramanathan: J. Vac. Sci. Technol. A 17,
291 (1999)
167. C. Heske, R. Fink, E. Umbach, W. Riedl, F. Karg: Appl. Phys. Lett. 68,
3431 (1996)

147
168. C. Heske, Z. Chen, R. Fink, E. Umbach, W. Riedl, F. Karg: J. Appl.
Phys. 82, 2411 (1997)
169. A. Klein, T. Lher, C. Pettenkofer, W. Jaegermann: J. Appl. Phys. 80,
5039 (1996)
170. D.J. Schroeder, A.A. Rockett: J. Appl. Phys. 82, 4982 (1997)
171. R. Noufi, R.J. Matson, R.C. Powell, C. Herrington: Solar Cells 16, 479
(1986)
172. R.J. Matson, R. Noufi, R.K. Ahrenkiel, R.C. Powell, D. Cahen: Solar
Cells 16, 495 (1986)
173. R.A. Sasala, J.R. Sites: Solar Cells 30, 101 (1991)
174. U. Rau, D. Braunger, H.W. Schock: In Polycrystalline Semiconductors
V - Bulk Materials: Thin Films, and Devices, ed. by J.H. Werner,
H.P. Strunk, H.W. Schock (Scitech Publ., Uettikon am See, Switzerland 1999) p. 409
175. W.N. Shafarman, R. Klenk, B.E. McCandleness: In Proc. 25th IEEE
Photovolt. Spec. Conf. (IEEE, New York 1996), p. 763; J. Appl. Phys.
79, 7324 (1996)
176. R. Herberholz, V. Nadenau, U. Rhle, C. Kble, H.W. Schock, B. Dimmler: Solar Energy Mater. Solar Cells 49, 227 (1997)
177. R.J. Matson, M.A. Contreras, J.R. Tuttle, A.B. Swartzlander, P.A. Parilla, R. Noufi: Mater. Res. Soc. Symp. Proc. 426, 183 (1996)
178. R. Schffler, D. Hariskos, M. Kaiser, M. Ruckh, U. Rhle, H.W. Schock:
Cryst. Res. Technol. 31, 543 (1996)

179. D.J. Schroeder, J.L. Hernandez, G.D. Berry, A.A. Rockett: Inst. Phys.
Conf. Ser. 152E, 749 (1998)
180. S.H. Wei, S.B. Zhang, A. Zunger: Appl. Phys. Lett. 72, 3199 (1998)
181. R. Scheer, M. Alt, I. Luck, R. Schieck, H.J. Lewerenz: Mater. Res.
Soc. Symp. Proc. 426, 309 (1996)
182. T.M. Friedlmeier, H.W. Schock: In Proc. 2nd World Conf. on Photovolt.
Energy Conv., ed. by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann,
E.D. Dunlop (E. C. Joint Res. Centre, Luxembourg 1998) p. 1117
183. K. Kushiya, S. Kuriyagawa, T. Kase, M. Tachiyuki, I. Sugiyama,
Y. Satoh, M. Satoh, H. Takeshita: In Proc. 25th IEEE Photov. Spec.
Conf. (IEEE, New York 1997) p. 989
184. T. Nakada, H. Ohbo, T. Watanabe, H. Nakazawa, M. Matsui, A. Kunioka: Solar Energy Mater. Solar Cells 49, 285 (1997)
185. V. Probst, J. Rimmasch, W. Stetter, H. Harms, W. Riedl, J. Holz,
F. Karg: In Proc. 13th Europ. Photov. Solar Energy Conf., ed. by
W. Freiesleben, W. Palz, H.A. Ossenbrink, P. Helm (Stephens, Bedford
1995) p. 2123
186. U. Rau, M. Schmitt, D. Hilburger, F. Engelhardt, O. Seifert, J. Parisi,
W. Riedl, J. Rimmasch, F. Karg: In Proc. 25th IEEE Photovolt. Spec.
Conf. (IEEE, New York 1996) p. 1005
187. M.A. Contreras, B. Egaas, K. Ramanathan, J. Hiltner, A. Swartzlander,
F. Haason, R. Noufi: Prog. Photor. Res. Appl. 7 (1999) in press

Vous aimerez peut-être aussi