SCIENCE CHINA

Physics, Mechanics & Astronomy

•Invited Review• January 2023 Vol. 66 No. 1: 217301
Special Topic: Recent Developments and Challenges in Solar Cells https://doi.org/10.1007/s11433-022-2001-4

Chalcopyrite solar cells

—state-of-the-art and options for improvement
* *
S. Siebentritt , and T. P. Weiss

Laboratory for Photovoltaics, Department of Physics and Materials Science, University of Luxembourg, Belvaux 4422, Luxembourg

Received June 3, 2022; accepted September 16, 2022; published online December 7, 2022

Chalcopyrite solar cells will have to play an important role to mitigate the climate crisis, because of their particularly low carbon
emissions. Doping in these semiconductors is due to native defects and intentional alkali impurities. The recent progress in
efficiency has been made possible by post-deposition treatments with heavy alkalis. Tail states and band gap distribution are the
main limitations for the open circuit voltage in state-of-the-art chalcopyrite solar cells. Further efficiency limitations are due to
the increased diode factor because of metastable defect transitions. Alloying with Ag opens new possibilities of band-edge
engineering, as well as seems to improve the diode factor. In state-of-the-art cells the back contact is passivated by a Ga gradient;
considerable research has been done to passivate the back contact by structured or continuous dielectric layers. A leap forward in
efficiency can be expected from tandem cells. Chalcopyrite solar cells show promising potential as bottom cells as well as top
cells.
doping, tail states, alkali treatment, efficiency

PACS number(s): 71.20.Nr, 71.55.Ht, 81.05.Hd, 84.60.Jt

Citation: S. Siebentritt, and T. P. Weiss, Chalcopyrite solar cells—state-of-the-art and options for improvement, Sci. China-Phys. Mech. Astron. 66, 217301
(2023), https://doi.org/10.1007/s11433-022-2001-4

1 Introduction: Why we need thin film solar [2] (IRENA International Renewable Energy Agency. World
cells Energy Transitions Outlook 2022: 1.5°C Pathway-Executive
summary, https://www.irena.org/publications/2022/Mar/World-
The recent IPCC (Intergovernmental Panel on Climate Energy-Transitions-Outlook-2022), necessitating a huge in-
Change) report lists solar energy as most effective and cost- crease in the annual PV installations [1]. To achieve this goal
efficient tool to mitigate the climate crisis (IPCC. Climate without increasing green house gas (GHG) emissions it is
Change 2022-Mitigation of Climate Change-Summary for necessary to reduce the GHG emissions during the produc-
Policymakers, https://report.ipcc.ch/ar6wg3/pdf/IPCC_AR6 tion of PV modules. A 2016 report by the UNEP (United
_WGIII_SummaryForPolicymakers.pdf), although the tech- Nations’ Environment Programme) compared the green
nical potential of PV (photovoltaics) has long been under- house gas emissions by different energy sources [3] using the
estimated [1]. Various studies on a climate neutral energy then available data and technology projections. Already this
system of the future come to the conclusion that photo- report confirms that solar PV electricity has one to two orders
voltaics will have to be one of the major electricity sources of magnitude lower GHG emissions than any fossil based
power station. Furthermore, the study predicts for the 2030
technologies GHG emissions for Si solar cells of slightly
*Corresponding authors (S. Siebentritt, email: susanne.siebentritt@uni.lu; T. P. Weiss,
email: thomas.weiss@uni.lu) above 30 g CO2e/kW h. It can be stated, that also in this

© The Author(s) 2022. This article is published with open access at link.springer.com. phys.scichina.com link.springer.com
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-2

respect, the potential of PV is underestimated. In fact, today’s band gap gradient, based on a Ga profile [14], which helps to
state-of-the-art Si solar modules emit only 13 to 30 g CO2e/kW h reduce interface recombination [15,16] and the discovery
[4]. A similar study on state-of-the-art thin film solar cells is that the presence of sodium improves the doping level of
missing. However, a 2018 study on current Cu chalcopyrite chalcopyrites [17,18]. Two recent reviews call for a para-
solar modules produced on stainless steel, using different digm change and propose to remove the Ga gradient in
scenarios found in the best case GHG emissions of 5 g CO2e/ chalcopyrite solar cells and to implement selective contacts
kW h [5]. Also the mentioned UNEP study [3] found in instead [19,20]. Both aspects are discussed in further detail
direct comparison to Si solar cells that thin film solar cells below in sects. 4 and 7.
emit a factor 3 to 5 less GHG. Furthermore, a recent long The technological state-of-the-art can be found in refs.
term study found that the performance ratio of Cu(In,Ga)Se2 [21,22], and a short review on the recent developments in ref.
modules, calibrated to 25°C, was the best among all in- [23]. A discussion of issues concerning the efficiency of
vestigated modules, which included several state-of-the-art commercial modules has been presented in ref. [24]. A
Si modules, as well as CdTe modules [6]. Therefore, it is particular advantage of chalcopyrite solar cells is that they
highly desirable to install considerably more PV systems to can be prepared on flexible substrates and made into bend-
mitigate the climate crisis and in particular to install more able and lightweight modules [25,26].
thin film PV systems. Thin film solar cells based on chal-
copyrite absorbers are commercially available. The tech-
nology is particularly suited for building integration [7], as 3 Native defects and metastabilities
well as for utility systems. Chalcopyrite solar modules have
proven stable in the field over decades (for detailed in- The structure of chalcopyrite solar cells is based on a p/n
vestigations see e.g., refs. [8,9]), as well as in accelerated hetero junction between the p-type chalcopyrite and an n-
aging test [7]. Recently the efficiency of this technology has type TCO (transparent conductive oxide). The advantages of
been increased to 23.4% in the laboratory [10] and 19.8% on hetero junctions for solar cells based on highly absorbing
the module level (Avancis. https://www.avancis.de/en/ma- materials have been discussed in ref. [27]. The p-type doping
gazine/pr-efficiency). Several companies in Europe, the US of chalcopyrite absorbers is due to native defects and due to
and Asia produce chalcopyrite modules, either as glass-glass extrinsic doping by alkali impurities, which are discussed in
devices or as flexible products on steel or plastic foil. Pro- sect. 5. Native defects have been studied by photo-
duction is still relatively small at 1.5 GW per year (Fraun- luminescence, admittance spectroscopy and by ab initio
hofer ISE. Photovoltaics Report, https://www.ise.fraunhofer. calculations. A recent review can be found in ref. [28]. We
de/en/publications/studies/photovoltaics-report.html), but its summarise the main results in Figure 1 and Table 1. The p-
potential for GHG reduction should be a good argument to type doping in CuInSe2 and CuGaSe2 is dominantly caused
increase production. In this article we give an overview on by two acceptors, that have been identified as the Cu vacancy
recent scientific and technological developments of chalco- and the Cu-on-III antisite. Under certain conditions a third
pyrite PV solar cells and modules. acceptor is present. Theses acceptors are compensated by
shallow and deep donors, notably the Cu interstitial and the
III-on-Cu antisite. The latter is shallow in CuInSe2 and deep
2 A short overview of recent reviews in CuGaSe2. It is likely that this defect plays a role in the
higher non-radiative loss in the open-circuit voltage, that is
“Chalcopyrite” describes the tetragonal crystal structure of I- observed for high Ga Cu(In,Ga)Se2 [28]. It has been found
III-VI2 compounds. Cu(In,Ga)Se2 is an alloy of CuInSe2 and that Cu-rich chalcopyrite is higher doped than Cu-poor ma-
CuGaSe2 with a bandgap varying with the Ga content be- terial [29-31]. Furthermore, it has been observed by photo-
tween 1.0 and 1.65 eV. Also alloys with S, replacing partly or luminescence that the luminescence due to A2 (Cu-on-III)
completely Se, and with Ag, replacing partly Cu are used as increases with increasing Cu-content, whereas the lumines-
solar cell absorbers. These compounds are discussed in sect. cence due to A1 (Cu vacancy decreases) [32,33]—as one
6 and 8. Collectively, these materials are often labelled would expect. Thus, it appears, that Cu-rich material is
“CIGS” or “CIS”. A number of excellent reviews exist on mostly doped by the Cu-on-III antisite acceptor, however, Cu
material properties and solar cell technology. Fundamental vacancy acceptors are still present in Cu-rich material. The
structural and electronic properties of these materials are lower doping in Cu-poor material is due to the fact, that the
summarised in an early book by Shay and Wernick [11]. A Cu vacancy acceptor is compensated by III-on-Cu antisite
recent review on crystal and band structure by Wada can be donor. In fact, it has been found by neutron scattering in
found in ref. [12]. The historical development of the tech- CuInSe2 that the concentration of Cu vacancies does not
nology is available in refs. [8,13]. Important aspects of the change much in the Cu-poor composition range, whereas the
historical efficiency improvement were the introduction of a concentration of In-on-Cu antisite increases with decreasing
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-3

more deep defects are observed.

Another effect related to native defects is metastable be-
haviour. Metastabilities have been observed in Cu(In,Ga)Se2
solar cells early on: light soaking describes the effect that the
open circuit voltage increases with illumination time [40].
There exists a variety of metastable effects, some of them are
clearly related to the buffer or the buffer/absorber interface
[41-44]. Here, we concentrate on metastable effects related
to the Cu(In,Ga)Se2 absorber. Metastable effects are gen-
erally accompanied by an increase in the doping level [45],
also visible as persistent photoconductivity [46]. First ex-
planations already attributed these effects to amphoteric
Figure 1 (Color online) Overview of native defects in Cu(In,Ga)Se2.
defects [47]. From a summary of the then available experi-
Physico-chemical structures are summarised in Table 1. From ref. [28] mental data Rau et al. [48] proposed as early as 2001 that the
(reproduced under CCBY licence). metastable effects are related to Se vacancies. Ab initio
calculations showed in fact that the Se vacancy does show
metastable behaviour in various semiconductors due to the
Table 1 Defect structures. According to ref. [28] (III: In or Ga) formation of cation dimers [49]. Breaking the bonding of
Defect label Defect structure these dimers requires considerable energy, which explains
A1 Cu vacancy the metastable stabilisation of the charge state that forms the
A2 Cu-on-III antisite dimers. In Cu(In,Ga)Se2, the Se vacancy forms a defect
A3 tentatively: In vacancy complex with the Cu vacancy. This defect shows metastable
A3x unknown transitions between donor type and acceptor type [50], which
D1 Cu interstitial, in CuInSe2 also In-on-Cu antisite can explain the metastable increase of the doping level. It
D2 tentatively Se-Cu double vacancy was shown recently that Na and K play a role in the me-
DD2 Ga-on-Cu antisite tastable increase of the doping level [51,52]. The In-on-Cu
DD1 tentatively: lower charge state of Ga on Cu antisite antisite has also been described as a metastable defect [53]
has been proposed tentatively as second charge state which transitions from a shallow donor (the antisite) to a
DS
of Cu-on-III antisite deep donor (a DX defect, where the In atom becomes in-
terstitial) thereby trapping electrons. It should be noted,
however, that more recent calculations found that the tran-
Cu content [34]. This behaviour implies an increasing degree sition to the DX state is not likely under normal experimental
of compensation and thus lower net doping with decreasing conditions [54].
Cu content. Indeed, Cu-poor CuInSe2 without Na and grown In particular the metastable increase of the doping level has
under low Se conditions, is n-type [35,36]. In the contrary, been found to anneal out at temperature above about 200 K
CuGaSe2 is always p-type, independent of the composition. [52,55]. Thus, it could be expected that this metastable ef-
It was found that above around 17% Ga/Ga+In atomic ratio fects have little influence on solar cell efficiency. However,
is needed in Cu(InGa)Se2 to make the Cu-poor material p- even at room temperature but under excitation there will be a
type without Na doping [37]. steady-state quasi-equilibrium between the acceptor state
While the shallow defects are responsible for the doping and the donor state of the metastable defect. The balance
level and the degree of compensation, deep defects like the between the two states will depend on the amount of ex-
DS and the DD defects are clearly unwanted, because they citation, i.e. illumination level or applied voltage. With
act as recombination centres and decrease the quasi-Fermi higher excitation more metastable defects are in the acceptor
level splitting and the open circuit voltage [38]. It is inter- state, thereby increasing the effective doping level. This
esting to note that in all chalcopyrites, including the sulfide behaviour of the metastable defects leads to a shift of the
chalcopyrites, we observe the same deep level luminescence. majority quasi-Fermi level, even in low injection conditions.
The only difference is: with higher band gaps we observe Thus, metastable defects will increase the diode factor above
more deep levels [39]. The luminescence related to the DS 1 even for recombination in the quasi-neutral region and
defect is found in all Cu(In,Ga)(S,Se)2 chalcopyrites, the under low excitation conditions [56]. This diode factor in-
DD2 luminescence is observed in CuInS2 and all chalco- crease can be most easily studied by the excitation depen-
pyrites with a higher bandgap, DD1 luminesce in Cu(In,Ga)S2 dence of the photoluminescence (PL) flux [57,58]. This
from a band gap of 1.6 eV upwards, as well as in CuGaSe2. “optical diode factor” measurement is essentially the same as
In CuGaS2, which has the highest band gap of 2.5 eV, three a jV (current-voltage) measurement in forward direction: in a
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-4

jV measurement the voltage is applied and the current mea- called the Urbach energy [63-66]. Essentially, higher Urbach
sured, whereas in the PL measurement the generation flux, energies mean there are more tail states. The temperature
i.e. the current, is fixed and the measured PL flux is de- dependence of the Urbach energy can be attributed to the
termined by the quasi-Fermi level splitting, i.e., the voltage vibrational disorder caused by phonons, while static disorder
[58]. The increase in the diode factor by metastable defects adds an additional constant contribution [64,67]. The static
has been confirmed experimentally and by simulations [56]. disorder is partly due to structural deviations from the ideal
The effect of the metastable defects is summarised in crystal like grain boundaries, or due to variations in the
Figure 2. We would like to note that without metastable composition [68]. Grain boundaries contribute to tail states,
defects, the diode factor for dominant recombination in the as can be seen from the higher Urbach energy found in
quasi-neutral zone is 1 because only the electron quasi-Fermi polycrystalline films than in single crystalline films [69],
level shifts, whereas it is 2 for recombination in the space however the difference is rather small: Urbach energies vary
charge region, because here both Fermi levels shift with between 10 and 17 meV between Cu-rich and Cu-poor
increasing excitation (by light or by voltage) [59]. In a PL CuInSe2 and the Urbach energies of single crystalline films
experiment no contacts are applied to the absorber, i.e. no are only 1 or 2 meV lower than those of polycrystalline ones.
space charge region exists and recombination takes place in It is also likely that bond length variations in the chalcopyrite
the neutral region. Thus, without metastable defects a diode crystal contribute to the tail states [70]. Furthermore, elec-
factor of 1 would be expected in low excitation (Figure 2 left) trostatic fluctuations due to doping compensation will play a
[56]. role in the formation of tail states. These fluctuations are
A higher diode factor leads to a lower fill factor of the solar greatly reduced at room temperature, but do not disappear
cell and thus to lower efficiency [59]. Although the increased completely [71]. The electrostatic band bending around
doping level after light soaking anneals out at room tem- charged grain boundaries will also contribute to tail states
perature, the steady-state situation under continuous excita- [72,73].
tion between the acceptor and the donor state of the The Urbach energy is determined from the exponential
metastable defects shifts with excitation and thus shifts the behaviour of the absorption coefficient below the band gap.
majority quasi-Fermi level, increasing the diode factor and For an accurate determination of the Urbach energy the
thereby decreasing the fill factor [60]. Thus, metastable de- measurement of very low absorption coefficients is neces-
fects do decrease the efficiency of solar cells. We recently sary [74,75]. Since in general it is easier to detect single
found that alloying with Ag decreases metastable effects and emitted photons than to detect missing photons, photo-
increases the fill factor [60] (see also sect. 6). luminescence (PL) is particularly suited to measure ultra-low
absorption coefficients and thus allows a reliable determi-
nation of the Urbach energy [76,77].
4 Tail states It is been long known that tail states contribute to a ra-
diative loss in open circuit voltage [78]. In that work, tail
Besides deep defects, tail states also lead to losses in the open states were described as a Gaussian variation of the band gap,
circuit voltage [61]. They lead to radiative and non-radiative which is not exactly the same as an Urbach tail, but quali-
losses [62]. In an ideal semiconductor the density of states in tatively the arguments also hold for Urbach tails: the main
the bandgap is zero. However, in real semiconductors, band reason for the radiative VOC loss is the fact that the carriers
tails occur, which are characterised by a density of states, can thermalise to lower energy states in the case of tails. An
decaying exponentially into the gap. The decay constant is expansion of this model is presented in ref. [79], where the
open circuit voltage losses are discussed based on detailed
balance, using an arbitrary shape of the quantum efficiency
spectrum that can include any shape of tail states. Since
alkali postdeposition treatments (see sect. 5) allow to ma-
nipulate the Urbach energy [73], the influence of these states
on the VOC loss can be investigated experimentally [62].
Figure 3 shows a number of different Cu(In,Ga)Se2 solar
cells, together with various cells from the literature ([62] and
references therein). It is evident that every additional meV in
Figure 2 (Color online) p-type semiconductor under low excitation:
without metastable defects (left) and with metastable defects (right). the Urbach energy increases the VOC loss by 20 mV. Several
Without metastable defects only the electron quasi-Fermi level (EFC) shifts, experimental parameters influence the Urbach energy:
whereas in the presence of metastable defects an additional shift of the hole higher process temperature leads to lower Urbach energy
Fermi level (EFV) occurs. This shift depends in turn on the electron con-
centration and increases the diode factor. EF0 depicts the equilibrium Fermi (compare the samples labeled “H” with those labelled “L”);
level. For more details see ref. [56]. also treatment with heavy alkalis, like Rb decreases the
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-5

5 Alkali addition: doping and interface mod-

ification

Na is the first alkali metal, which was identified to increase

the efficiency of CIGS solar cells. The beneficial effect of Na
was discovered, where Na was introduced unintentionally
into the CIGS growth process by using soda-lime-glass
substrates [13]. Generally, average concentrations of ap-
proximately 0.1 at.% are found to be optimal [84], while too
much Na deteriorates the performance [85].
The main benefit of the addition of Na is an increase in net
doping density of the CIGS absorber layer. While the net p-
type doping of the alkali-free Cu-poor absorber is only on the
Figure 3 (Color online) VOC loss with respect to the Shockley-Queisser 14 −3
order of 10 cm , it increases by up to 2 orders of magni-
VOC according to the bandgap of the respective cells, as a function of the
Urbach energy of the tail states. Every increase in the Urbach energy of tude upon the addition of Na [86,87]. Importantly, the in-
1 meV leads to an additional VOC loss of 20 mV. We compare a range of crease in the doping describes well the increase of the open-
different Cu(In,Ga)Se2 solar cells with various other cells from the litera- circuit voltage [86-88], which is the main reason for the
ture. Shown is also the theoretical radiative and the non-radiative loss due
to tail states, assuming everything else remains unchanged. From ref. [62]
improved performance. Often, an increase in the FF is re-
(reproduced under CCBY licence). ported as well, which is however expected due to an in-
creased VOC. Additional improvements of the FF might be
caused by a reduced series resistance due to the increased net
Urbach energy (compare samples labelled “1” with those p-type doping or a reduced diode factor. Unfortunately, these
labelled “2”). Based on the Urbach energy alone it is possible contributions are generally not disentangled.
to calculate the loss in VOC due to radiative recombination The microscopic origin for the increased p-type doping is
(thermalisation into tails states) and due to non-radiative still under debate. Kronik et al. [89] suggested that Na cat-
recombination (Shockley-Read-Hall recombination through alyses the passivation of donor-like Se-vacancies by O. This
tail states), also shown in Figure 3. The actual loss increases model is supported by the correlation of Na and O in the near
stronger than the combined effect of those two losses. The surface region of the CIGS absorber as well as the formation
reason is most likely that additional effects in doping oc- of In–O bonds [84,88]. In addition, Abou-Ras [90] found
cur: lower doping would reduce VOC and increase elec- segregation of Na, K, and O at random grain boundaries and
trostatic potential fluctuations simultaneously. This speculated that OSe might act as an acceptor.
explanation is supported by the observation that alkali Another model assumes NaIn and NaGa acceptor-like point
addition increases the doping level and reduces the Urbach defects, which is based on the presence of Na–Se bonds by
energy (see sect. 5). XPS, while metallic Na bonds (Na–Na, Na–Cu, Na–In and
It should be mentioned that the Urbach energies measured Na–Ga) could be excluded [84]. Contreras et al. [91] propose
in Figure 3 are from tail states deep in the gap. A distribution that Na inhibits the formation of InCu and GaCu antisite donor-
of bandgaps has similar effects [78,79], but would be ob- like defects and thus an increase in the net p-type doping due
served near the effective band gap [80,81]. State-of-the-art to reduced compensation.
chalcopyrite solar cells certainly suffer from a distribution of More recently, Yuan et al. [92] proposed a new mechanism
band gaps, visible in the broadening of the onset of absorp- explaining the increased net p-type doping caused by Na. In
tance [82] or the quantum efficiency [79], which is due to the their calculations, in contrast to previous approaches, the
intentionally graded band gap. The effects of tail states and thermodynamic limiting condition to the elemental chemical
band gap distribution are linked and difficult to disentangle potential of Na is taken into account. As a result, NaCu is the
experimentally. defect with the lowest formation energy. While NaCu is a
No deep defects were found in the absorbers shown in neutral defect, the increase in the p-type conductivity is ex-
Figure 3 [83]. For a given Urbach energy of the absorber, the plained by out-diffusion of Na (from Cu-sites) when cooling
VOC loss can be higher than the dashed line in Figure 3, if down the CIGS absorber. The driving force is a lower so-
additional deep defects or recombination channels at the lubility of Na in the CIGS crystal at room temperature
interfaces exist. However, the loss cannot be lower than the compared to elevated temperatures (growth temperature or
one dictated by the Urbach energy. It can, thus, be sum- PDT temperature). As a result, an increased density of VCu
marized that the open circuit voltage of state-of-the-art form in the CIGS bulk, which are shallow acceptors (see
chalcopyrite solar cells is limited by the effects of tail states Figure 1 and Table 1).
or band gap distribution. Apart from an increase of the net-doping, Na influences
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-6

the growth of the CIGS absorber layers resulting in increased was due to reduced surface recombination. This reduced
grain size [91] and preferred 〈112〉 texture [93]. Also the surface recombination was partly attributed to the observed
GGI grading, important to achieve high-efficiency absorbers, widening of the surface band gap [111] which can be caused
is influenced by the presence of Na [94,95], which however by a KInSe2 or an RbInSe2 layer at the surface, which have
is different for polycrystalline and epitaxial films: it appears wide band gaps. However, a wider surface band gap is
that Na supports the intragrain interdiffusion of In and Ga, generally observed in Cu-poor Cu(In,Ga)Se2 [112,113]. A
whereas it hampers the intergrain diffusion. The latter effect reduced surface recombination was also concluded based on
was used, for example, by Zahedi-Azad et al. [96] who used the temperature and generation dependent VOC measure-
substrates with a diffusion barrier to reduce the Na con- ments [100,114], according to the model of ref. [115].
centration during the growth of wide bandgap (high Ga) However, the model proposed in ref. [115] depends on as-
absorbers to smoothen the Ga grading and increase the ef- sumptions that are not correct: e.g., the hole barrier at the
ficiency. front interface (i.e., distance of the Fermi level to the valence
Apart from Na, heavier alkalis are introduced into the band) is interpreted as activation energy of the reverse sa-
growth process of CIGS absorber layers and are important to turation current in case of interface recombination. However,
reach state-of-the-art quality. Already in 1997, Contreras et this is not the case as the hole barrier itself is temperature
al. [91] used thin precursor layers of MF (M = Na, K, Cs) dependent [116,117]. Furthermore, standard Cu-poor devices
prior to the CIGS growth to investigate the influence of some without alkali PDT are not limited by interface recombina-
other (heavier) alkalis. However, no additional beneficial tion [27,69]. It is therefore highly unlikely that the observed
effect was observed using heavier alkalis. Also, the highest VOC improvement should be due to interface recombination.
increase in doping resulted from using a NaF precursor. Only Recently, based on the effort of a large consortium, it was
in 2013, Chirilă et al. [97] introduced potassium as a KF concluded that the heavy alkali treatment has a strong ben-
post-deposition treatment (PDT), i.e., after the growth of the eficial effect on the absorber bulk properties [38,73]. First,
CIGS absorber layer. A record efficiency of 20.4% was the addition of heavy alkalis reduced disorder in the CIGS
achieved, followed by a series of record efficiencies by other bulk material, as evidenced by reduced Urbach tails [62,73],
institutes and companies [98]. One of the benefits of the KF which are shown to correlate well with the VOC losses
PDT is that it enabled the reduction of the CdS buffer layer [61,62]. Second, reduced variations of potential fluctuations
thickness and hence reduced parasitic absorption losses. at grain boundaries are found [118]. In particular, for RbF
Moreover, an increased VOC was observed, which in this case (for which the process was optimized), only an upward band
exceeded the expected increase due to the doping density, bending was found, i.e., electrons (minorities) are repelled
thus suggesting another beneficial effect [86,99,100]. Che- from the GBs. Thus, GB recombination is reduced. The main
mically, the KF resulted in a Cu and Ga depletion and a K findings are summarized in Figure 4.
enrichment at the front surface [97]. It was then suggested by This finding is in agreement with the results from Krause
several studies that a K-In-Se layer forms at the front surface et al. [119], who recently reported that grain boundary re-
[101-103]. Lepetit [104] also suggested that this surface combination contributes significantly to VOC losses in state-
layer reacts during the CBD CdS deposition to form a of-the-art CIGS absorbers. Alkalis tend to segregate at GBs
CdIn2S4 layer. In any case, an increased diffusion of Cd into [120], in particular the heavy alkalis are almost exclusively
the Cu depleted surface was proposed as a reason for the found at the grain boundaries [73]. This trend can be ex-
improved VOC [86]. The model is based on CdCu donor de- plained by the thermodynamics of the segregation of Alkali-
fects, which help to increase the surface inversion and In-Se compounds versus the formation of mixed (Cu-Alka-
thereby prevent interface recombination [27]. It is likely that li)-In-Se compounds. Heavy alkalis (K, Rb, Cs) segregate at
these defects form in any case, but they form faster in the Cu- growth conditions, whereas Na and Li form mixed com-
depleted layer, which is obtained after the PDT, allowing for pounds [110]. Although it was shown that alkali accumula-
a shorter CdS deposition time and thinner buffer layer. It was
also shown that a PDT by RbF acts very similar to a KF PDT
[105], and a Rb-In-Se phase, identified as a RbInSe2 struc-
ture, was directly observed by TEM analysis [106]. The
formation of the RbInSe2 phase was found to depend on the
Cu content of the absorber and on the availability of Cu
vacancies to be occupied by Rb [107]. An overtreatment was
found to hamper the fill factor [108,109], which is attributed
to the formation of a barrier because of the wide bandgap of
the RbInSe2 layer [110]. Figure 4 (Color online) Summary of effects due to alkali doping and
Several authors have suggested that the VOC improvement heavy-alkali post-deposition treatment.
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-7

tion does not reduce the recombination velocity at grain miscibility gap mentioned above [129]. This degradation did
boundaries [121], it is likely that alkalis change the charge not happen for samples with AAC = 0.5 and the same GGIs,
state of defects at grain boundaries and thus the band bending which is surprising as samples with AAC = 0.5 are calculated
around them, which will influence tail states and the effective to have the highest free energy of mixing [129] and should
grain boundary recombination [73]. On the other hand, KF therefore show a faster degradation/decomposition.
postdeposition treatment was also found to increase the Another interesting prediction from the calculation of the
quasi-Fermi level splitting in single crystalline films [122]. total free energy of an ACIGS absorber is the anti-correlation
And similar reductions of the Urbach tails were observed in of the Ag/(Ag+Cu) profile to the (fixed, because group-III
single crystalline films upon alkali postdeposition treatment diffusion is much slower) Ga/(Ga+In) profile [129], which is
as in polycrystalline films [123]. Both effects indicate that generally observed experimentally as well [126,129,131,
the improvement goes beyond grain boundary effects. 132].
Early studies reported a very similar 3-stage deposition
process for ACIGS [126], as it was applied for high-effi-
6 Ag compounds ciency Cu(In,Ga)Se2 [14]. In particular, a Ga notch forms,
i.e., the Ga concentration is graded and increases towards the
The addition of Ag to form (Ag,Cu)(In,Ga)Se2 (ACIGS) front and back surfaces [126,132,133]. However, Essig et al.
absorbers enables another handle to modify the electronic [133] performed interruption process, where the ACIGS
structure of the CIGS absorber as well as the growth process. absorbers grown on glass/Mo substrates are taken out during
Avon et al. [124] showed that the Ag(In,Ga)Se2 and Cu(In, the 3rd stage, where the composition is stoichiometric and/or
Ga)Se2 compounds are not entirely miscible over the whole (Ag,Cu)-poor again. It is found that Ag-rich islands form for
Ag-Cu compositional range, based on samples grown in (Ag+Cu)/(Ga+In) atomic ratios greater >0.9. It is speculated
sealed quartz ampules at temperatures between 600°C and that these islands are a result of a faster reaction of Cu with
800°C. However, growth of thin-films by co-evaporation the (In,Ga)2Se3 layer deposited in the 1st stage [133,134].
showed complete miscibility, i.e., the entire compositional Also Valdes et al. [135] observes a Ag-rich surface for (Ag,
space of (Ag,Cu)(In,Ga)Se2 results in single-phase material Cu)InSe2 absorbers. For compositions with (Ag+Cu)/(Ga
(i.e., possible secondary phases are below detection limit of +In) < 0.9, local variations in the Ag/(Ag+Cu) and Ga/(Ga
X-ray diffractometry (XRD)) [125-128]. Sopiha et al. [129] +In) ratio are still observed [133], probably caused due to
studied the phase diagram theoretically based on density similar reasons as the Ga and Ag anti-correlation mentioned
functional theory (DFT) calculations and experimentally by above, which are expected to be detrimental to device per-
glow discharge optical emission spectroscopy (GDOES) and formance [129]. Another consequence of the Ag-rich islands
by transmission electron microscopy (TEM). The authors is that the surface shows an increase of the Ag content, i.e.,
showed that spinodal and binodal decomposition of the the anti-correlation of the Ga/(Ga+In) and Ag/(Ag+Cu) is
ACIGS absorber can occur for most Ga/(Ga+In) and Ag/(Ag not maintained in that region. This is in stark contrast to the
+Cu) atomic ratios at room temperature. At 350°C (typical observations of Edoff et al. [136], who showed a strong
PDT temperatures), the decomposition may only occur for group I depletion with thicknesses ranging from 50 to
Ga/(Ga+In) ratios > 0.4, conditions, which however are 300 nm at the front surface. At the same time, a K enrich-
found towards the back contact in graded (A)CIGS absor- ment is observed within this front surface layer. Noteworthy,
bers. However, it seems that the kinetics are rather slow at this group I depletion and K enrichment was only observed
room temperature, which explains the single-phase thin-film when using a high strainpoint glass having a high con-
materials mentioned above [125-128]. It is noted that some centration of K and a low concentration of Na. No such
decomposition at the front surface is observed under certain surface layer was found when using SLG. Further studies are
process conditions (overall high Se flux and high substrate needed to clarify the role of alkalis on the group-I depleted
temperature in the first stage of a 3-stage co-evaporation surfaces in ACIGS absorbers, as it might help to reduce
process) for films with Ag/(Ag+Cu) = 0.5 and Ga/(Ga+In) = inhomogeneities of Ag/(Ag+Cu) and Ga/(Ga+In).
0.85 [126], which is the least stable region reported in ref. An important property of the Ag alloying is that it lowers
[129]. Keller et al. [130] find a significant formation of order the melting point [11] of the chalcopyrite compound. Con-
vacancy compound (OVC) phases as the front and back sequently, enlarged grains are generally observed for absor-
contact for high Ag (Ag/(Ag+Cu) = 0.8) and high Ga (Ga/ ber layers with increasing Ag content or when compared to
(Ga+In) = 0.72-0.85) absorbers. Also, degradation of finished Ag free absorbers grown under the same process conditions
solar cells after 8 month storage in air for samples with Ag/ [132,133]. For instance, Essig et al. [133] found grain sizes
(Ag+Cu) = 0.8 and Ga/(Ga+In) = 0.72-0.85 is observed most >10 μm for ACIGS absorbers with Ag/(Ag+Cu) = 0.2 grown
likely linked to formation of ordered vacancy compounds at a maximum temperature of 650°C, while crevices form at
(OVC) [130]. These features might be associated with the the same time, which can lead to shunt paths. Also, the
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-8

growth temperature can be reduced, without a sacrifice in band with increasing Ag concentration, the interface with
device performance when adding Ag to the CIGS growth CdS (or another buffer layer) can be changed from a “cliff”
process [137]. to a “spike”-like band alignment [131]. While Ag alloying
Apart from the thermodynamic properties, which influence allows to engineer the interface quality and thus reduce front
the growth of ACIGS absorbers, opto-electronic properties surface recombination, recent studies suggest that absorbers
are strongly influenced by Ag-alloying. In particular, the with high Ga contents (GGI > 0.5) still suffer from a lower
bandgap is increased upon Ag alloying, which was experi- bulk quality, as for instance evidenced in ref. [144] by a small
mentally determined by transmission and reflection mea- diffusion length of minority carriers (electrons). Thus, the
surements [127,138], however, the two studies find reason for a VOC saturation when increasing the bandgap of
somewhat different bandgap values for high Ag and high Ga the absorber by increasing the Ga concentration seems not be
values. These discrepancies are probably caused due to the solved when alloying additionally Ag. However, further
2
non-linear behaviour in the Tauc plots ((αE) versus E) and defect studies are needed to draw more substantiated con-
thus the derived bandgap values depend on the choice of the clusions.
fitting region. Bandgap values are also reported based on Recently, Ag is added/alloyed to low Ga CIGS (i.e., low
DFT calculations, where it was shown additionally that the bandgap around 1.0 to approximately 1.25 eV) as well due to
increase in the bandgap is caused by a downshift of the the properties mentioned above. Edoff et al. [136] demon-
valence- and conduction-band [131]. This phenomenon is strated that the mean voltage loss with respect to the bandgap
unlike Ga alloying to CuInSe2, where mainly the conduction is decreased for ACIGS devices (449 mV) compared with
band shifts upward [139]. A comparison of various band CIGS devices (460 mV), which is mentioned to be sig-
alignments in the chalcopyrite system is shown in Figure 5 nificant, where the bandgap for the ACIGS devices is ap-
[131,139]. The implications for this modification of the band proximately 50 meV higher (around 1.22 eV). This
alignment with ACIGS absorbers are discussed below. characteristic might be beneficial for modules, due to a lower
The early work of Ag alloying to CIGS focused on high current density, which then requires a thinner TCO. Finally, a
bandgap, i.e., high Ga, absorber layers. The first high- champion device with a 20.9% power conversion efficiency,
bandgap (Eg = 1.70 eV) Ag(In,Ga)Se2 solar cell was reported a VOC of 814 mV with a bandgap of 1.22 eV is achieved (with
by Nakada et al. [140] in 2005 with an efficiency of 9.3% and Ag/(Ag+Cu) = 0.2 and Ga/(Ga+In) = 0.4) [136]. Essig et al.
a VOC of 0.949 V. Shafarman et al. [128] showed that Ag [133] achieved a 20.5% efficient device with Ag/(Ag+Cu) =
allows to increase the bandgap, without losses in efficiency, 0.05 and Ga/(Ga+In) = 0.3. Yang et al. [137] used a 15 nm
i.e., beyond the threshold of Eg= 1.25 eV, which is associated Ag precursor layer prior to a low temperature CIGS growth
with larger VOC losses [141]. Erslev et al. [142] find that the process [137]. Significantly higher efficiencies than without
addition of Ag results in Urbach energies (determined in both Ag are obtained when lowering the substrate temperature. In
contributions by transient photocapacitance spectroscopy) particular, efficiencies of 19.6% and 18.5% are demonstrated
between 10 and 15 meV, which are significantly reduced for substrate temperatures of 353°C and 303°C [137]. Also,
compared to CIGS absorbers with Urbach energies between similar as for high Ga containing absorbers [142], lower
18 and 26 meV at that time [143]. The authors attribute the Urbach energies are measured for these samples with the Ag
reduced Urbach energies due to a reduction of structural precursor layer [137].
defects due to the lower melting point upon Ag alloying. The Another important electronic modification is the doping
decrease of the Urbach energy with the addition of Ag is an density, which often is reported to be lower upon Ag alloy-
important finding, due to the correlation of the VOC loss, as ing. In 2009, Erslev et al. [142] reported a very low carrier
14 −3
discussed in sect. 4. Keller et al. [131] used Ag alloying in concentration on the order of 10 cm for Ag-alloyed ab-
order to modify the conduction band offset at the front in- sorbers, even though they are grown on SLG at a growth
terface. In particular, due to the downshift of the conduction temperature of 550°C, in agreement with an improved carrier
collection due to enlarged space charge region widths [125].
However, a later study showed a smaller difference of carrier
concentrations between rather high efficient CIGS and
ACIGS devices, where ACIGS devices have free carrier
15 −3
densities around 2 × 10 cm as measured by Drive Level
Capacitance Profiling (DLCP) [145]. The authors speculated
that the increased carrier densities results from a decrease in
contamination during the growth process. Valdes et al. [146]
found a reduced carrier concentration for ACIS compared to
Figure 5 (Color online) Band alignments between different chalcopyrites
CIS (i.e., absorbers without Ga). For low temperature grown
and the most used buffer CdS. Data taken from refs. [131,139]. CIGS (substrate temperatures between 350°C and 410°C)
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-9

and the addition of Ag via a precursor layer, Yang et al. [137] the VOC shows the opposite trend, which indicates an im-
observed also a lower apparent doping density for ACIGS provement in the diode factor. It is noted however, that their
than for CIGS. On the other hand, Kim et al. [147] found an CuInSe2 and (Ag,Cu)InSe2 devices show an opposite trend,
increase in carrier density when using Ag precursor layers higher VOC with Ag, but lower FF. Diode factors greater than
for low temperature (450°C) (A)CIGS absorbers. Keller et 1 are often associated with contributing SCR recombination
al. [144] demonstrated that the doping density depends on as diode factors close to 2 are expected for this recombina-
the stoichiometry of the absorber, i.e., the group-I over tion channel [59]. However, the diode factor can also be
group-III ratio for rather large bandgap ACIGS absorbers measured optically on the bare absorber, i.e., without SCR,
(Ag/(Ag+Cu) = 0.53, Ga/(Ga+In) = 0.66). Interestingly, the by intensity dependent photoluminescence spectroscopy
doping density decreases the closer the absorber shifts to- [57,58,148], often labeled optical diode factor (see also sect.
wards the 1:1:2 stoichiometry point [144], which is in con- 3). It is found that for CIGS absorbers the optical diode factor
trast to the observations for Ag-free CIGS absorbers [69]. If is already greater than 1 for these measurements, where no
these results also hold for lower bandgap ACIGS devices, SCR is present, and with commonly reported values around
variations in the stoichiometry could explain the different 1.3 [56,58]. Weiss et al. [56] showed that metastable defects,
results. such as the VSe-VCu, can explain an increase in the diode
The importance of the doping density is that it has a direct factor. The metastable defects change from a donor-type to
impact on the VOC. An improvement/deterioration of the VOC an acceptor-type defect upon injection of minority carriers
by 60 mV can be expected for an increase/decrease of the (electrons). Thus, the diode current is altered during a vol-
doping density by 1 order of magnitude [59]. Valdes et al. tage sweep, which results in increased diode factors. Inter-
[146] investigated low bandgap (1.0 eV) (Ag,Cu)InSe2 based estingly, upon alloying CIGS with Ag, the smallest optical
solar cells with and without Ag alloy. In this study, the ACIS diode factors (around 1.1) as well as electrical diode factors
solar cells had a reduced VOC of approximately 30 mV, which (around 1.14 obtained from IV characteristics) are obtained
could be attributed to the decrease of the doping density. In [60]. In addition, it is demonstrated that this Ag-alloyed
particular, for low Ag contents (Ag/(Ag+Cu) < 0.25), no samples shows much weaker metastability as detected by
increase in the bandgap is expected for (Ag,Cu)InSe2 ab- capacitance based methods [60].
sorbers [131], which facilities in this case the analysis. In conclusion, it seems promising that Ag alloying may
However, for samples including significant amounts of Ga contribute to the success of the CIGS thin-film photovoltaic
and Ag, a detailed detanglement of the different contribu- technology. It reduces the Urbach energy and allows to
tions to VOC due to bandgap and doping density is missing. fabricate higher bandgaps high efficiency CIGS devices. In
Nevertheless, Edoff et al. [136] stated that a comparison of addition, due to the lowered melting point, the growth pro-
various ACIGS and CIGS samples with the comparable cess may be modified such that a faster deposition or lowered
bandgap indicated a small performance advantage for temperature can be implemented without a sacrifice in effi-
ACIGS. Unfortunately, measurements of the doping density ciency and thus might be particularly important for industry.
were missing to judge the quality of the ACIGS material However, there are still open questions, such as the impact on
better. the doping density, including metastable defects, and its in-
Interestingly, some studies show large fill factors for fluence on the VOC and FF.
ACIGS devices [60,128,133]. It is stressed that FF values are
not always comparable between various samples with sig-
nificant different VOC values, because of the dependence of 7 Back contact passivation
FF on VOC [60]. A more suitable parameter for comparison is
the diode factor of the solar cell, which is the main parameter In any solar cell the (necessary) metal back contact leads to
governing the FF, assuming that the solar cell does not suffer high recombination. State-of-the-art PERC Si solar cells are,
2
from too large series (> 0.4 Ω cm ) or too low shunt therefore, equipped with a dielectric layer between the ab-
2
(< 1 kΩ cm ) resistances [60]. Unfortunately, this parameter sorber and the back contact with small holes to let the current
is not always determined for the published IV characteristics. pass [149]. The standard way for backside passivation in
Shafarman et al. [128] reported a diode factor of 1.1 for the chalcopyrite solar cells is a Ga gradient, which results in a
device with the highest FF, which the authors attributed to gradient of the conduction band edge that prevents the
improved carrier collection. Also Edoff et al. [136] reported electrons from reaching the highly recombinative back
that ACIGS devices generally have a lower diode factor contact [14,15]. A sketch of both approaches is shown in
compared with CIGS devices, whereas the diode factors are Figure 6. We have recently shown quantitatively that a Ga
still rather large around 1.4 for their record device. Valdes et gradient reduces the backside recombination as effectively as
al. [135] reported a higher FF for low bandgap (1.04 eV) a dielectric layer [150]: both increase the quasi-Fermi level
ACIGS compared to CIGS (also Eg = 1.04 eV), even though splitting by almost 50 meV for a mediocre absorber with a
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-10

difficult to scale up to industrial scales. Another approach to

a passivated back contact is to use ultrathin dielectric layers,
which can be tunnelled through or are conductive enough,
not to block the current. A 3 nm thick TiO2 layer has been
shown to effectively reduce backside recombination and to
lead to a considerable increase of VOC and jSC for a 1 μm thick
Cu(In,Ga)Se2 absorber without backside gradient [165]. We
believe that more work is needed on continuous dielectrics or
wide gap semiconductors that can serve as selective hole
contact at the back side and reduce recombination at the back
Figure 6 (Color online) Band diagram schematics of back contact pas- contact.
sivation. (a) Ga gradient that reduces electron concentration at (highly
recombinative) metal back contact; (b) dielectric layer that is either thin
enough to allow tunnelling of holes or is structured to open pathways for
the current or a wide gap semiconductor with sufficient hole conductivity. 8 Tandem cells and wide bandgap chalcopyrites

bulk lifetime of 40 ns. In a good absorber with a lifetime of While single junction chalcopyrite solar cells can still be
200 ns [151] the improvement increases to 80 meV [150]. improved, e.g., with a new device design [19,20], much
These experiments and simulations were performed with higher efficiencies can be expected with tandem solar cells
rather thick absorbers of 3 μm thickness. In contrast to some [166,167]. For the schematics of the loss mechanisms in
studies in the literature we find that even at this thickness single and tandem solar cells see Figure 7. Low bandgap
back contact recombination does have a significant influ- Cu(InGa)Se2 has been combined with perovskite solar cells,
ence. to reach efficiencies above 26% [168-171]. Monolithic 2-
However, it is desirable to decrease absorber thickness, to terminal, as well as stacked 4-terminal architectures have
reduce production cost and in particular reduce the con- been used. One challenge for the monolithic tandem is the
sumption of costly and rare indium. Although it was recently rather rough surface of the Cu(In,Ga)Se2 bottom solar cells.
shown that accessible In resources are 35 times higher than Using optimised Cu(In,Ga)Se2 processing and solution based
previously assumed [152]. First experiments to reduce ab- deposition of the perovskite solar cell, this problem has been
sorber thickness were already performed 20 years ago [153]. overcome [169]. Interestingly, a comparatively simple self-
Various light management techniques have been used to assembled monolayer on top of the TCO contact of the
improve the absorption in thin Cu(In,Ga)Se2 layers, includ- Cu(In,Ga)Se2 cell acts as a tunnel junction between the two
ing plasmonic structures and improved back reflectors, see devices [169] A certified efficiency of 24.2% has been
e.g., refs. [154-159]. With decreased absorber thickness back achieved [169]. Since no procedure exists yet to certify
contact recombination becomes ever more important. Thus stacked tandem cells, only in-house measurements are
the concept of the PERC solar cell has been transferred to
Cu(In,Ga)Se2 solar cells: dielectric layer passivation at the
back contact with point contact openings to let the current
pass [160]. The challenge compared to Si solar cells is the
shorter diffusion length in chalcopyrite solar cells, which
requires a much denser grid of contact holes [160]. Various
dielectrics have been successfully used as passivating layers
at the back contact: Al2O3 [160], SiO2 [161], TiO2 [162].
Patterning methods range from arbitrarily distributed nano-
particles [160] through nano-imprint lithography [162] and
laser interference lithography [161] to e-beam lithography
[163]. A review can be found in ref. [164]. In general an
improvement of the open-circuit voltage is observed with a
passivated back contact; the difference increases with de-
creasing thickness [164]. This is expected since the re-
combination at the back contact is reduced. An increase in
the short circuit current is also observed in many cases [164]
which can be attributed to optical effects [161]. Figure 7 (Color online) Main losses in single junction solar cells: in-
sufficient absorption and thermalisation. With a lower band gap bottom cell
While the approach of a structured dielectric layer to more photons can be absorbed and the higher band gap top cell reduces the
passivate the back contact has been successful, it might be thermalisation losses in both cells.
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-11

available. The highest reported efficiency is 26.2%, based on Table 2 Best reported efficiencies of wide gap chalcopyrite solar cells
a highly efficient Cu(In,Ga)(S,Se)2 solar cell in combination Absorber material Band gap (eV) Efficiency (%) Reference
with an improved transparent perovskite solar cell [171]. (Ag,Cu)(In,Ga)Se2 1.45 15.1 [131]
Recently, a triple junction with 28% efficiency was obtained, Cu(In,Ga)Se2 1.50 14.2 [176]
combining a InGaP/AlGaAs tandem with a Cu(InGa)Se2 Cu(In,Ga)S2 1.50 15.5 [177]
bottom cell [172]. Considerable progress has been made Cu(In,Ga)S2 1.60 15.2 [178]
recently improving low band gap Cu(In,Ga)Se2 solar cells to CuGaSe2 1.65 11.9 [179]
19.2% efficiency [173]. Cu(In,Ga)S2 1.65 14.2 [180]
Another approach is to use widegap chalcopyrite in com-
bination with a Si bottom cell or a low bandgap Cu(In,Ga)Se2
cell [174]. This approach requires obviously highly efficient, (Zn,Mg)O, can avoid this problem and lead to better fill
transparent wide gap chalcopyrite solar cells. So far, most factors [183]. As discussed in sect. 6, the situation is different
wide gap chalcopyrites solar cells have been optimised on a for the Ag compounds: here both band edges shift down-
Mo back contact, thus, they can only be considered a first wards and CdS is still a useful buffer [131].
step towards a top cell in a tandem device. However, some
promising improvements in the efficiency have been re-
ported recently. In general, wide gap semiconductors can be 9 Outlook
expected to have a higher VOC loss because of increased non-
radiative recombination due to an increasing number of State-of-the-art chalcopyrite solar cells are limited by bulk
different deep defects. More deep defect levels are in fact recombination due to tail states. Interface recombination
found in widegap Cu(In,Ga)(S,Se)2 (see sect. 3) [39]. does not play a role, because the front surface is passivated
The targeted bandgap for top cells is around 1.5 to 1.7 eV. by the buffer layer [186] and the back surface is passivated
In fact, it was recently shown that in combination with a by the Ga gradient. Efficient back side passivation has also
realistic bifacial Si solar cell a top cell band gap of 1.5 eV been demonstrated by structured or continuous layers of
can be sufficient [175]. Several options exist to achieve wide various dielectrics.
gap chalcopyrites: high Ga Cu(In,Ga)Se2 or pure CuGaSe2, The recent progress in efficiency has been made possible
alloying with S or pure sulfide Cu(In,Ga)S2, or alloying with by alkali postdeposition treatments. The main effect of this
Ag (Ag,Cu)(In,Ga)Se2. For the latter see sect. 6. An over- treatment is a reduction of the non-radiative recombination
view of current best published efficiencies of wide gap in the bulk due to the reduction of tail states. Tail states are
chalcopyrites is given in Table 2 [131,176-180]. As becomes the main source of VOC loss in state-of-the-art chalcopyrite
obvious from the table, all wide gap chalcopyrites, besides solar cells.
CuGaSe2, reach efficiencies around 15%. This is about the Several recent reviews have called for a paradigm change
critical efficiency for a top cell in a tandem solar cell [19,20] in the development of chalcopyrite cells: replacing
[181,182]. the Ga gradient by a selective contact at the back. Dielectric
Obviously further improvements are necessary, so it is layers can also passivate the back contact. An absorber
worth having a look at current limitations. As mentioned without band gap grading would show a sharper absorption
above: in general wide gap semiconductors are more prone to edge and thus less radiative and non-radiative VOC loss due to
non-radiative recombination because they host more deep tail states. However, without the Ga gradient new ways must
defect levels [39]. Another issue is the interface between the be found to passivate the back contact. Much more research
absorber and the buffer. With increasing Ga or S the con- is needed on selective and passivated back contacts. The
duction band shifts upwards [139]. Therefore, above a cer- electrostatic contributions to tail states (doping compensa-
tain Ga or S content the absorber conduction band edge will tion and charged grain boundaries) can be reduced by using
be above the buffer conduction band edge, forming a cliff, absorbers with higher doping level. In general higher Cu
which increases interface recombination [59], in particular content leads to higher net doping [39,69], thus it appears
when using the standard CdS buffer. In general, alternative desirable to grow the chalcopyrite absorbers as close as
buffer layers are necessary to avoid this interface re- possible to stoichiometry. It is necessary to maintain a Cu-
combination. Examples are Zn(S,O) [178], (Zn,Mg)O poor composition, to avoid the problems of the Cu-rich
[177,183], (Zn,Sn)O [184]. However, the alternative buffer surface [187,188]. The higher doping level will also be
layers lead to a large cliff towards the standard i-ZnO layer, beneficial for the open circuit voltage.
which limits the forward current and thus the fill factor [183]. The efficiency of state-of-the-art chalcopyrite cells is fur-
A high resistive i-layer is generally necessary to avoid the ther limited by the rather high diode factor which leads to a
influence of inhomogeneities and local diodes with a low VOC low fill factor. Metastable defect transitions increase the
[185]. An i-layer with a higher conduction band edge, like diode factor. This implies another call for more research: to
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-12

find ways to suppress the metastable defects. Alloying with 11 J. L. Shay, and J. H. Wernick, Ternary Chalcopyrite Semiconductors:
Growth, Electronic Properties, and Application, In: International
Ag appears beneficial in that context. Series of Monographs in the Science of the Solid State, Vol. 7, edited
A massive increase in efficiency can only be expected by B. R. Pamplin (Pergamon Press, Oxford, 1975).
from tandem cells. Chalcopyrite cell shows convincing po- 12 T. Wada, Jpn. J. Appl. Phys. 60, 080101 (2021).
tential as bottom cells as well as for top cells. It should be 13 D. Abou-Ras, S. Wagner, B. J. Stanbery, H. W. Schock, R. Scheer, L.
Stolt, S. Siebentritt, D. Lincot, C. Eberspacher, K. Kushiya, and A.
noted, though, that a recent economic study found, that N. Tiwari, Thin Solid Films 633, 2 (2017).
tandem cells are certainly the preferred choice where the area 14 A. M. Gabor, J. R. Tuttle, D. S. Albin, M. A. Contreras, R. Noufi,
is limited, e.g., on roofs, however, for utility scale project it and A. M. Hermann, Appl. Phys. Lett. 65, 198 (1994).
might economically more attractive to use a highly efficient 15 T. Dullweber, O. Lundberg, J. Malmström, M. Bodegård, L. Stolt, U.
Rau, H. W. Schock, and J. H. Werner, Thin Solid Films 387, 11
and low cost single junction cell [189,190]. Thus, it is cer- (2001).
tainly worthwhile to further improve the efficiency of single 16 A. Chirilă, S. Buecheler, F. Pianezzi, P. Bloesch, C. Gretener, A. R.
junction chalcopyrite solar cells. Uhl, C. Fella, L. Kranz, J. Perrenoud, S. Seyrling, R. Verma, S.
Nishiwaki, Y. E. Romanyuk, G. Bilger, and A. N. Tiwari, Nat. Mater.
10, 857 (2011).
We thank Marika Edoff, Roland Scheer and Wolfram Witte for many useful
17 A. Rockett, Thin Solid Films 480-481, 2 (2005).
discussions and for their support in collecting some of the data, in particular
18 D. Rudmann, D. Brémaud, H. Zogg, and A. N. Tiwari, J. Appl. Phys.
Table 2.
97, 084903 (2005).
19 M. Ochoa, S. Buecheler, A. N. Tiwari, and R. Carron, Energy En-
Open Access This article is licensed under a Creative Commons Attri- viron. Sci. 13, 2047 (2020).
bution 4.0 International License, which permits use, sharing, adaptation, 20 B. J. Stanbery, D. Abou-Ras, A. Yamada, and L. Mansfield, J. Phys.
distribution and reproduction in any medium or format, as long as you give D-Appl. Phys. 55, 173001 (2021).
appropriate credit to the original author(s) and the source, provide a link to 21 M. Powalla, S. Paetel, E. Ahlswede, R. Wuerz, C. D. Wessendorf,
the Creative Commons licence, and indicate if changes were made. The and T. Magorian Friedlmeier, Appl. Phys. Rev. 5, 041602 (2018).
images or other third party material in this article are included in the article’s 22 T. Feurer, P. Reinhard, E. Avancini, B. Bissig, J. Löckinger, P. Fuchs,
Creative Commons licence, unless indicated otherwise in a credit line to the R. Carron, T. P. Weiss, J. Perrenoud, S. Stutterheim, S. Buecheler,
material. If material is not included in the article’s Creative Commons li- and A. N. Tiwari, Prog. Photovolt.-Res. Appl. 25, 645 (2017).
cence and your intended use is not permitted by statutory regulation or 23 S. Siebentritt, Curr. Opin. Green Sustain. Chem. 4, 1 (2017).
24 V. Bermudez, and A. Perez-Rodriguez, Nat. Energy 3, 466 (2018).
exceeds the permitted use, you will need to obtain permission directly from
25 R. Carron, S. Nishiwaki, T. Feurer, R. Hertwig, E. Avancini, J.
the copyright holder. To view a copy of this licence, visit http://creative-
Löckinger, S. Yang, S. Buecheler, and A. N. Tiwari, Adv. Energy
commons.org/licenses/by/4.0/.
Mater. 9, 1900408 (2019).
26 J. Ramanujam, D. M. Bishop, T. K. Todorov, O. Gunawan, J. Rath,
1 F. Creutzig, P. Agoston, J. C. Goldschmidt, G. Luderer, G. Nemet, R. Nekovei, E. Artegiani, and A. Romeo, Prog. Mater. Sci. 110,
and R. C. Pietzcker, Nat. Energy 2, 17140 (2017). 100619 (2020).
2 M. Ram, D. Bogadnov, A. Aghahosseini, A. S. Oyewo, A. Gulagi, 27 R. Klenk, Thin Solid Films 387, 135 (2001).
M. Child, H.-J. Fell, and C. Breyer, Global Energy System based on 28 C. Spindler, F. Babbe, M. H. Wolter, F. Ehré, K. Santhosh, P. Hilgert,
100% Renewable Energy-Power Sector (Lappeenranta University of F. Werner, and S. Siebentritt, Phys. Rev. Mater. 3, 090302 (2019).
Technology and Energy Watch Group, 2017). 29 A. Gerhard, W. Harneit, S. Brehme, A. Bauknecht, U. Fiedeler, M. C.
3 E. G. Hertwich, J. Aloisi de Larderel, A. Arvesen, P. Bayer, J. Ber- Lux-Steiner, and S. Siebentritt, Thin Solid Films 387, 67 (2001).
gesen, E. Bouman, T. Gibon, G. Heath, C. Peña, P. Purohit, A. Ra- 30 Y. Aida, V. Depredurand, J. K. Larsen, H. Arai, D. Tanaka, M.
mirez, and S. Suh, eds., Green Energy Choices: The Benefits, Risks Kurihara, and S. Siebentritt, Prog. Photovolt.-Res. Appl. 23, 754
and Trade-Offs of Low-Carbon Technologies for Electricity Pro- (2015).
duction, Report of the International Resource Panel, United Nations’ 31 A. Jasenek, U. Rau, V. Nadenau, and H. W. Schock, J. Appl. Phys.
Environment Programme, 2016. 87, 594 (2000).
4 A. Müller, L. Friedrich, C. Reichel, S. Herceg, M. Mittag, and D. H. 32 A. Bauknecht, S. Siebentritt, J. Albert, and M. C. Lux-Steiner, J.
Neuhaus, Sol. Energy Mater. Sol. Cells 230, 111277 (2021). Appl. Phys. 89, 4391 (2001).
5 S. Amarakoon, C. Vallet, M. A. Curran, P. Haldar, D. Metacarpa, D. 33 S. Siebentritt, N. Rega, A. Zajogin, and M. C. Lux-Steiner, Phys.
Fobare, and J. Bell, Int. J. Life Cycle Assess 23, 851 (2018). Stat. Sol. (c) 1, 2304 (2004).
6 B. Gao, Y. Shao, W. Liu, H. Xiang, Y. Yu, and Z. Liu, Sol. Energy 34 C. Stephan, S. Schorr, M. Tovar, and H. W. Schock, Appl. Phys. Lett.
Mater. Sol. Cells 236, 111490 (2022). 98, 091906 (2011).
7 J. Palm, L. Tautenhahn, J. Weick, R. Kalio, J. Kullmann, A. Heiland, 35 S. Siebentritt, Thin Solid Films 403-404, 1 (2002).
S. Grïnsteid, N. Schmidt, P. Borowski, and F. Karg, in BIPV mod- 36 F. Werner, D. Colombara, M. Melchiorre, N. Valle, B. El Adib, C.
ules: Critical requirements and customization in manufacturing: Spindler, and S. Siebentritt, J. Appl. Phys. 119, 173103 (2016).
Proceedings of the in 7th IEEE World Conference on Photovoltaic 37 O. Ramírez, E. M. Lanzoni, R. G. Poeira, T. P. Weiss, R. Leturcq, A.
Energy Conversion (WCPEC)/A Joint Conference of 45th IEEE Redinger, and S. Siebentritt, APL Mater. 10, 061108 (2022).
PVSC/28th PVSEC/34th EU PVSEC, Waikoloa, 2018. 38 S. Siebentritt, T. P. Weiss, M. Sood, M. H. Wolter, A. Lomuscio, and
8 M. Powalla, S. Paetel, D. Hariskos, R. Wuerz, F. Kessler, P. Lechner, O. Ramirez, J. Phys. Mater. 4, 042010 (2021).
W. Wischmann, and T. M. Friedlmeier, Engineering 3, 445 (2017). 39 S. Siebentritt, A. Lomuscio, D. Adeleye, M. Sood, and A. Dwivedi,
9 B. H. King, J. S. Stein, D. Riley, C. B. Jones, and C. D. Robinson, in Phys. Rapid Res. Ltrs 16, 2200126 (2022).
Degradation assessment of fielded CIGS photovoltaic arrays: Pro- 40 M. N. Ruberto, and A. Rothwarf, J. Appl. Phys. 61, 4662 (1987).
ceedings of the in 2017 IEEE 44th Photovoltaic Specialist Con- 41 N. Naghavi, S. Temgoua, T. Hildebrandt, J. F. Guillemoles, and D.
ference (PVSC), Washington, 2017. Lincot, Prog. Photovolt.-Res. Appl. 23, 1820 (2015).
10 M. Nakamura, K. Yamaguchi, Y. Kimoto, Y. Yasaki, T. Kato, and H. 42 J. Serhan, Z. Djebbour, W. Favre, A. Migan-Dubois, A. Darga, D.
Sugimoto, IEEE J. Photovol. 9, 1863 (2019). Mencaraglia, N. Naghavi, G. Renou, J. F. Guillemoles, and D. Lin-
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-13

cot, Thin Solid Films 519, 7606 (2011). 77 G. Rey, C. Spindler, F. Babbe, W. Rachad, S. Siebentritt, M. Nuys, R.
43 I. Repins, S. Glynn, K. Bowers, B. Stevens, C. L. Perkins, and L. Carius, S. Li, and C. Platzer-Björkman, Phys. Rev. Appl. 9, 064008
Mansfield, Sol. Energy Mater. Sol. Cells 215, 110597 (2020). (2018).
44 I. Repins, S. Glynn, T. J. Silverman, R. Garris, K. Bowers, B. Ste- 78 U. Rau, and J. H. Werner, Appl. Phys. Lett. 84, 3735 (2004).
vens, and L. Mansfield, Prog. Photovolt.-Res. Appl. 27, 749 (2019). 79 U. Rau, B. Blank, T. C. Müller, and T. Kirchartz, Phys. Rev. Appl. 7,
45 S. Siebentritt, M. Igalson, C. Persson, and S. Lany, Prog. Photovolt- 044016 (2017).
Res. Appl. 18, 390 (2010). 80 G. Rey, G. Larramona, S. Bourdais, C. Choné, B. Delatouche, A.
46 U. Rau, M. Schmitt, J. Parisi, W. Riedl, and F. Karg, Appl. Phys. Jacob, G. Dennler, and S. Siebentritt, Sol. Energy Mater. Sol. Cells
Lett. 73, 223 (1998). 179, 142 (2018).
47 M. Igalson, and H. W. Schock, J. Appl. Phys. 80, 5765 (1996). 81 E. M. Spaans, J. de Wild, T. J. Savenije, and B. Vermang, J. Appl.
48 U. Rau, K. Weinert, Q. Nguyen, M. Mamor, G. Hanna, A. Jasenek, Phys. 130, 123103 (2021).
and H. W. Schock, MRS Proc. 668, H9.1 (2001). 82 S. Siebentritt, U. Rau, S. Gharabeiki, T. P. Weiss, A. Prot, T. Wang,
49 S. Lany, and A. Zunger, Phys. Rev. B 72, 035215 (2005). D. Adeleye, M. Drahem, and A. Singh, Faraday Discuss., doi:
50 S. Lany, and A. Zunger, J. Appl. Phys. 100, 113725 (2006). 10.1039/d2fd00057a.
51 A. Czudek, A. Urbaniak, A. Eslam, R. Wuerz, and M. Igalson, IEEE 83 F. Werner, M. H. Wolter, S. Siebentritt, G. Sozzi, S. Di Napoli, R.
J. Photovol. 10, 1926 (2020). Menozzi, P. Jackson, W. Witte, R. Carron, E. Avancini, T. P. Weiss,
52 A. Czudek, A. Urbaniak, A. Eslam, R. Wuerz, and M. Igalson, Phys. and S. Buecheler, Prog. Photovolt.-Res. Appl. 26, 911 (2018).
Rapid Res. Ltrs 16, 2100459 (2022). 84 D. W. Niles, K. Ramanathan, J. Granata, F. Hasoon, R. Noufi, B. J.
53 S. Lany, and A. Zunger, Phys. Rev. Lett. 100, 016401 (2008). Tielsch, and J. E. Fulghum, Vacuum Sci. Tech. A 15, 3044 (1997).
54 J. Pohl, and K. Albe, Phys. Rev. B 87, 245203 (2013). 85 A. Rockett, J. S. Britt, T. Gillespie, C. Marshall, M. M. Al Jassim, F.
55 M. Igalson, M. Cwil, and M. Edoff, Thin Solid Films 515, 6142 Hasoon, R. Matson, and B. Basol, Thin Solid Films 372, 212 (2000).
(2007). 86 F. Pianezzi, P. Reinhard, A. Chirilă, B. Bissig, S. Nishiwaki, S.
56 T. P. Weiss, F. Ehre, V. Serrano-Escalante, T. Wang, and S. Sieben- Buecheler, and A. N. Tiwari, Phys. Chem. Chem. Phys. 16, 8843
tritt, Sol. RRL 5, 2100063 (2021). (2014).
57 T. Trupke, R. A. Bardos, M. D. Abbott, and J. E. Cotter, Appl. Phys. 87 D. Rudmann, A. F. da Cunha, M. Kaelin, F. Kurdesau, H. Zogg, A.
Lett. 87, 093503 (2005). N. Tiwari, and G. Bilger, Appl. Phys. Lett. 84, 1129 (2004).
58 F. Babbe, L. Choubrac, and S. Siebentritt, Sol. RRL 2, 1800248 88 M. Ruckh, D. Schmid, M. Kaiser, R. Schaffler, T. Walter, and H.
(2018). Schock, Sol. Energy Mater. Sol. Cells 41-42, 335 (1996).
59 R. Scheer, and H. W. Schock, Chalcogenide Photovoltaics: Physics, 89 L. Kronik, D. Cahen, and H. W. Schock, Adv. Mater. 10, 31 (1998).
Technologies, and Thin Film Devices (Weinheim, Wiley-VCH, 90 D. Abou-Ras, S. S. Schmidt, R. Caballero, T. Unold, H. W. Schock,
2011). C. T. Koch, B. Schaffer, M. Schaffer, P. P. Choi, and O. Cojocaru-
60 T. P. Weiss, O. Ramírez, S. Paetel, W.Witte, J. Nishinaga, T. Feurer, Mirédin, Adv. Energy Mater. 2, 992 (2012).
and S. Siebentritt, arXiv: 2205.00826. 91 M. A. Contreras, B. Egaas, P. C. Dippo, J. Webb, J. Granata, K.
61 S. De Wolf, J. Holovsky, S. J. Moon, P. Löper, B. Niesen, M. Le- Ramanathan, S. Asher, A. Swartzlander, and R. Noufi, in On the role
dinsky, F. J. Haug, J. H. Yum, and C. Ballif, J. Phys. Chem. Lett. 5, of Na and modifications to Cu(In,Ga)Se2 absorber materials using
1035 (2014). thin-MF (M=Na, K, Cs) precursor layers: Proceedings of the Con-
62 M. H. Wolter, R. Carron, E. Avancini, B. Bissig, T. P. Weiss, S. ference Record of the Twenty Sixth IEEE Photovoltaic Specialists
Nishiwaki, T. Feurer, S. Buecheler, P. Jackson, W. Witte, and S. Conference, Anaheim, 1997, pp. 359-362.
Siebentritt, Prog. Photovol. 30, 702 (2022). 92 Z. K. Yuan, S. Chen, Y. Xie, J. S. Park, H. Xiang, X. G. Gong, and S.
63 H. Sumi, and Y. Toyozawa, J. Phys. Soc. Jpn. 31, 342 (1971). H. Wei, Adv. Energy Mater. 6, 1601191 (2016).
64 M. V. Kurik, Phys. Stat. Sol. (a) 8, 9 (1971). 93 J. Hedstrom, H. Ohlsen, M. Bodegard, A. Kylner, L. Stolt, D. Har-
65 T. Skettrup, Phys. Rev. B 18, 2622 (1978). iskos, M. Ruckh, and H.-W. Schock, in ZnO/CdS/Cu(In,Ga)Se2 thin
66 S. R. Johnson, and T. Tiedje, J. Appl. Phys. 78, 5609 (1995). film solar cells with improved performance: Proceedings of the in
67 G. D. Cody, T. Tiedje, B. Abeles, B. Brooks, and Y. Goldstein, Phys. Conference Record of the Twenty Third IEEE Photovoltaic Specia-
Rev. Lett. 47, 1480 (1981). lists Conference, Louisville, 1993.
68 C. Ayik, I. Studenyak, M. Kranjec, and M. Kurik, Optics 4, 76 94 D. Colombara, F. Werner, T. Schwarz, I. Cañero Infante, Y. Fleming,
(2014). N. Valle, C. Spindler, E. Vacchieri, G. Rey, M. Guennou, M.
69 S. Siebentritt, L. Gütay, D. Regesch, Y. Aida, and V. Deprédurand, Bouttemy, A. G. Manjón, I. Peral Alonso, M. Melchiorre, B. El Adib,
Sol. Energy Mater. Sol. Cells 119, 18 (2013). B. Gault, D. Raabe, P. J. Dale, and S. Siebentritt, Nat. Commun. 9,
70 C. S. Schnohr, S. Eckner, P. Schöppe, E. Haubold, F. d’Acapito, D. 826 (2018).
Greiner, and C. A. Kaufmann, Acta Mater. 153, 8 (2018). 95 D. Rudmann, D. Brémaud, A. F. da Cunha, G. Bilger, A. Strohm, M.
71 J. K. Larsen, K. Burger, L. Gütay, and S. Siebentritt, in Temperature Kaelin, H. Zogg, and A. N. Tiwari, Thin Solid Films 480-481, 55
dependence of potential fluctuations in chalcopyrites: Proceedings of (2005).
the in 37th IEEE Photovoltaic Specialist Conference, Seattle, 2011, 96 S. Zahedi-Azad, M. Maiberg, and R. Scheer, Prog. Photovolt.-Res.
pp. 396-401. Appl. 28, 1146 (2020).
72 S. Siebentritt, Sol. Energy Mater. Sol. Cells 95, 1471 (2011). 97 A. Chirilă, P. Reinhard, F. Pianezzi, P. Bloesch, A. R. Uhl, C. Fella,
73 S. Siebentritt, E. Avancini, M. Bär, J. Bombsch, E. Bourgeois, S. L. Kranz, D. Keller, C. Gretener, H. Hagendorfer, D. Jaeger, R. Erni,
Buecheler, R. Carron, C. Castro, S. Duguay, R. Félix, E. Handick, D. S. Nishiwaki, S. Buecheler, and A. N. Tiwari, Nat. Mater. 12, 1107
Hariskos, V. Havu, P. Jackson, H. Komsa, T. Kunze, M. Malitckaya, (2013).
R. Menozzi, M. Nesladek, N. Nicoara, M. Puska, M. Raghuwanshi, 98 P. Reinhard, F. Pianezzi, B. Bissig, A. Chirila, P. Blosch, S. Nishi-
P. Pareige, S. Sadewasser, G. Sozzi, A. N. Tiwari, S. Ueda, A. Vi- waki, S. Buecheler, and A. N. Tiwari, IEEE J. Photovol. 5, 656
lalta-Clemente, T. P. Weiss, F. Werner, R. G. Wilks, W. Witte, and M. (2015).
H. Wolter, Adv. Energy Mater. 10, 1903752 (2020). 99 P. Pistor, D. Greiner, C. A. Kaufmann, S. Brunken, M. Gorgoi, A.
74 W. B. Jackson, and N. M. Amer, Phys. Rev. B 25, 5559 (1982). Steigert, W. Calvet, I. Lauermann, R. Klenk, T. Unold, and M. C.
75 M. Vaněček, J. Kočka, J. Stuchlík, and A. Tříska, Solid State Lux-Steiner, Appl. Phys. Lett. 105, 063901 (2014).
Commun. 39, 1199 (1981). 100 T. Kato, A. Handa, T. Yagioka, T. Matsuura, K. Yamamoto, S. Hi-
76 E. Daub, and P. Würfel, Phys. Rev. Lett. 74, 1020 (1995). gashi, J. L. Wu, K. F. Tai, H. Hiroi, T. Yoshiyama, T. Sakai, and H.
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-14

Sugimoto, IEEE J. Photovol. 7, 1773 (2017). Proceedings of the in 2009 34th IEEE Photovoltaic Specialists
101 E. Handick, P. Reinhard, R. G. Wilks, F. Pianezzi, T. Kunze, D. Conference (PVSC), Philadelphia, 2009.
Kreikemeyer-Lorenzo, L. Weinhardt, M. Blum, W. Yang, M. Gorgoi, 126 G. M. Hanket, J. H. Boyle, W. N. Shafarman, and G. Teeter, in Wide-
E. Ikenaga, D. Gerlach, S. Ueda, Y. Yamashita, T. Chikyow, C. bandgap (AgCu)(InGa)Se2 absorber layers deposited by three-stage
Heske, S. Buecheler, A. N. Tiwari, and M. Bär, ACS Appl. Mater. co-evaporation: Proceedings of the in 2010 35th IEEE Photovoltaic
Interfaces 9, 3581 (2017). Specialists Conference, Honolulu, 2010.
102 T. Lepetit, S. Harel, L. Arzel, G. Ouvrard, and N. Barreau, IEEE J. 127 J. H. Boyle, B. E. McCandless, W. N. Shafarman, and R. W. Birk-
Photovol. 6, 1316 (2016). mire, J. Appl. Phys. 115, 223504 (2014).
103 P. Reinhard, B. Bissig, F. Pianezzi, H. Hagendorfer, G. Sozzi, R. 128 W. Shafarman, C. Thompson, J. Boyle, G. Hanket, P. Erslev, and J.
Menozzi, C. Gretener, S. Nishiwaki, S. Buecheler, and A. N. Tiwari, D. Cohen, in Device characterization of (AgCu)(InGa)Se2 solar
Nano Lett. 15, 3334 (2015). cells: Proceedings of the in 2010 35th IEEE Photovoltaic Specialists
104 T. Lepetit, Influence of KF Post Deposition Treatment on the Poly- Conference, Honolulu, 2010.
crystalline Cu(In,Ga)Se2/CdS Heterojunction Formation For Pho- 129 K. V. Sopiha, J. K. Larsen, O. Donzel-Gargand, F. Khavari, J. Keller,
tovoltaic Application, Dissertation for the Doctoral Degree M. Edoff, C. Platzer-Björkman, C. Persson, and J. J. S. Scragg, J.
(University of Nantes, Nantes, 2015). Mater. Chem. A 8, 8740 (2020).
105 E. Avancini, R. Carron, T. P. Weiss, C. Andres, M. Bürki, C. 130 J. Keller, L. Stolt, K. V. Sopiha, J. K. Larsen, L. Riekehr, and M.
Schreiner, R. Figi, Y. E. Romanyuk, S. Buecheler, and A. N. Tiwari, Edoff, Sol. RRL 4, 2000508 (2020).
Chem. Mater. 29, 9695 (2017). 131 J. Keller, K. V. Sopiha, O. Stolt, L. Stolt, C. Persson, J. J. S. Scragg,
106 N. Taguchi, S. Tanaka, and S. Ishizuka, Appl. Phys. Lett. 113, T. Törndahl, and M. Edoff, Prog. Photovolt. Res. Appl. 28, 237
113903 (2018). (2020).
107 T. Kodalle, T. Bertram, R. Schlatmann, and C. A. Kaufmann, IEEE J. 132 L. Chen, J. W. Lee, and W. N. Shafarman, IEEE J. Photovol. 4, 447
Photovol. 9, 1839 (2019). (2014).
108 T. Kodalle, M. D. Heinemann, D. Greiner, H. A. Yetkin, M. Klupsch, 133 S. Essig, S. Paetel, T. M. Friedlmeier, and M. Powalla, J. Phys.
C. Li, P. A. van Aken, I. Lauermann, R. Schlatmann, and C. A. Mater. 4, 024003 (2021).
Kaufmann, Sol. RRL 2, 1800156 (2018). 134 L. Chen, S. Soltanmohammad, J. W. Lee, B. E. McCandless, and W.
109 T. P. Weiss, S. Nishiwaki, B. Bissig, R. Carron, E. Avancini, J. N. Shafarman, Sol. Energy Mater. Sol. Cells 166, 18 (2017).
Löckinger, S. Buecheler, and A. N. Tiwari, Adv. Mater. Interfaces 5, 135 N. Valdes, J. W. Lee, and W. Shafarman, Sol. Energy Mater. Sol.
1701007 (2018). Cells 195, 155 (2019).
110 M. Malitckaya, H. P. Komsa, V. Havu, and M. J. Puska, J. Phys. 136 M. Edoff, T. Jarmar, N. S. Nilsson, E. Wallin, D. Hogstrom, O. Stolt,
Chem. C 121, 15516 (2017). O. Lundberg, W. Shafarman, and L. Stolt, IEEE J. Photovol. 7, 1789
111 E. Handick, P. Reinhard, J. H. Alsmeier, L. Köhler, F. Pianezzi, S. (2017).
Krause, M. Gorgoi, E. Ikenaga, N. Koch, R. G. Wilks, S. Buecheler, 137 S. C. Yang, J. Sastre, M. Krause, X. Sun, R. Hertwig, M. Ochoa, A.
A. N. Tiwari, and M. Bär, ACS Appl. Mater. Interfaces 7, 27414 N. Tiwari, and R. Carron, Sol. RRL 5, 2100108 (2021).
(2015). 138 J. Boyle, G. Hanket, and W. Shafarman, in Optical and quantum
112 M. Morkel, L. Weinhardt, B. Lohmüller, C. Heske, E. Umbach, W. efficiency analysis of (Ag,Cu)(In,Ga)Se2 absorber layers: Proceed-
Riedl, S. Zweigart, and F. Karg, Appl. Phys. Lett. 79, 4482 (2001). ings of the in 2009 34th IEEE Photovoltaic Specialists Conference
113 M. Bär, S. Nishiwaki, L. Weinhardt, S. Pookpanratana, O. Fuchs, M. (PVSC), Philadelphia, 2009.
Blum, W. Yang, J. D. Denlinger, W. N. Shafarman, and C. Heske, 139 A. Klein, J. Phys.-Condens. Matter 27, 134201 (2015).
Appl. Phys. Lett. 93, 244103 (2008). 140 T. Nakada, K. Yamada, R. Arai, H. Ishizaki, and N. Yamada, MRS
114 S. Ishizuka, H. Shibata, J. Nishinaga, Y. Kamikawa, and P. J. Fons, Proc. 865, F11.1 (2005).
Appl. Phys. Lett. 113, 063901 (2018). 141 M. A. Contreras, L. M. Mansfield, B. Egaas, J. Li, M. Romero, R.
115 J. V. Li, S. Grover, M. A. Contreras, K. Ramanathan, D. Kuciauskas, Noufi, E. Rudiger-Voigt, and W. Mannstadt, Prog. Photovolt.-Res.
and R. Noufi, Sol. Energy Mater. Sol. Cells 124, 143 (2014). Appl. 20, 843 (2012).
116 R. Scheer, J. Appl. Phys. 105, 104505 (2009). 142 P. Erslev, G. M. Hanket, W. N. Shafarman, and D. J. Cohen, MRS
117 M. Sood, A. Urbaniak, C. Kameni Boumenou, T. P. Weiss, H. Proc. 1165, 1165 (2009).
Elanzeery, F. Babbe, F. Werner, M. Melchiorre, and S. Siebentritt, 143 J. T. Heath, J. D. Cohen, W. N. Shafarman, D. X. Liao, and A. A.
Prog. Photovol. 30, 263 (2022). Rockett, Appl. Phys. Lett. 80, 4540 (2002).
118 N. Nicoara, R. Manaligod, P. Jackson, D. Hariskos, W. Witte, G. 144 J. Keller, P. Pearson, N. Shariati Nilsson, O. Stolt, L. Stolt, and M.
Sozzi, R. Menozzi, and S. Sadewasser, Nat. Commun. 10, 3980 Edoff, Sol. RRL 5, 2100403 (2021).
(2019). 145 C. P. Thompson, L. Chen, W. N. Shafarman, J. Lee, S. Fields, and R.
119 M. Krause, A. Nikolaeva, M. Maiberg, P. Jackson, D. Hariskos, W. W. Birkmire, in Bandgap gradients in (Ag,Cu)(In,Ga)Se2 thin film
Witte, J. A. Márquez, S. Levcenko, T. Unold, R. Scheer, and D. solar cells deposited by three-stage co-evaporation: Proceedings of
Abou-Ras, Nat. Commun. 11, 4189 (2020). the in 2015 IEEE 42nd Photovoltaic Specialist Conference (PVSC),
120 A. Vilalta-Clemente, M. Raghuwanshi, S. Duguay, C. Castro, E. New Orleans, 2015.
Cadel, P. Pareige, P. Jackson, R. Wuerz, D. Hariskos, and W. Witte, 146 N. H. Valdes, J. W. Lee, and W. N. Shafarman, IEEE J. Photovol. 9,
Appl. Phys. Lett. 112, 103105 (2018). 906 (2019).
121 D. Abou-Ras, A. Nikolaeva, S. Caicedo Dávila, M. Krause, H. Gu- 147 G. Kim, W. M. Kim, J. K. Park, D. Kim, H. Yu, and J. H. Jeong, ACS
threy, M. Al-Jassim, M. Morawski, and R. Scheer, Sol. RRL 3, Appl. Mater. Interfaces 11, 31923 (2019).
1900095 (2019). 148 P. Caprioglio, C. M. Wolff, O. J. Sandberg, A. Armin, B. Rech, S.
122 O. Ramírez, M. Bertrand, A. Debot, D. Siopa, N. Valle, J. Schmauch, Albrecht, D. Neher, and M. Stolterfoht, Adv. Energy Mater. 10,
M. Melchiorre, and S. Siebentritt, Sol. RRL 5, 2000727 (2021). 2000502 (2020).
123 O. Ramírez, J. Nishinaga, F. Dingwell, T. Wang, A. Prot, M. H. 149 M. Hermle, F. Feldmann, M. Bivour, J. C. Goldschmidt, and S. W.
Wolter, V. Ranjan, and S. Siebentritt, arXiv: 2212.01603. Glunz, Appl. Phys. Rev. 7, 021305 (2020).
124 J. E. Avon, K. Yoodee, and J. C. Woolley, J. Appl. Phys. 55, 524 150 T. Wang, F. Ehre, T. P. Weiss, B. Veith-Wolf, V. Titova, N. Valle, M.
(1984). Melchiorre, O. Ramirez, J. Schmidt, and S. Siebentritt, Adv. Energy
125 G. M. Hanket, J. H. Boyle, and W. N. Shafarman, in Characteriza- Mat. 2022, 2202076 (2022).
tion and device performance of (AgCu)(InGa)Se2 absorber layers: 151 T. P. Weiss, R. Carron, M. H. Wolter, J. Löckinger, E. Avancini, S.
 S. Siebentritt, et al. Sci. China-Phys. Mech. Astron. January (2023) Vol. 66 No. 1 217301-15

Siebentritt, S. Buecheler, and A. N. Tiwari, Sci. Tech. Adv. Mater. zuka, R. Muller, D. Lackner, F. Dimroth, and T. Sugaya, IEEE J.
20, 313 (2019). Photovol. 12, 639 (2022).
152 T. T. Werner, G. M. Mudd, and S. M. Jowitt, Ore Geol. Rev. 86, 939 173 T. Feurer, R. Carron, G. Torres Sevilla, F. Fu, S. Pisoni, Y. E. Ro-
(2017). manyuk, S. Buecheler, and A. N. Tiwari, Adv. Energy Mater. 9,
153 O. Lundberg, M. Bodegård, J. Malmström, and L. Stolt, Prog. Pho- 1901428 (2019).
tovolt.-Res. Appl. 11, 77 (2003). 174 S. Nishiwaki, S. Siebentritt, P. Walk, and M. C. Lux-Steiner, Prog.
154 G. Baraldi, R. Caballero, C. A. Kaufmann, and J. Gonzalo, Energy Photovolt-Res. Appl. 11, 243 (2003).
Procedia 10, 38 (2011). 175 A. Onno, N. Rodkey, A. Asgharzadeh, S. Manzoor, Z. J. Yu, F. Toor,
155 N. Dahan, Z. Jehl, T. Hildebrandt, J. J. Greffet, J. F. Guillemoles, D. and Z. C. Holman, Joule 4, 580 (2020).
Lincot, and N. Naghavi, J. Appl. Phys. 112, 094902 (2012). 176 W. Witte, D. Hariskos, S. Paetel, M. Maiberg, S. Zahedi-Azad, P.
156 J. G. Mutitu, U. Obahiagbon, S. Shi, W. Shafarman, and D. W. Pistor, H. Kempa, R. Scheer, D. Hauschild, V. van Maris, M. Blan-
Prather, IEEE J. Photovol. 4, 948 (2014). kenship, L. Weinhardt, C. Heske, J. Keutgen, O. Cojocaru-Mirédin,
157 C. van Lare, G. Yin, A. Polman, and M. Schmid, ACS Nano 9, 9603 E. Ghorbani, K. Albe, X. Jin, R. Schneider, D. Gerthsen, A. Niko-
(2015). laeva, J. A. Marquez-Prieto, M. Krause, S. Schäfer, D. Abou-Ras, T.
158 F. Mollica, M. Jubault, F. Donsanti, A. Loubat, M. Bouttemy, A. Unold, R. Mainz, J. Seeger, F. Wilhelmi, M. Hetterich, M. Schwei-
Etcheberry, and N. Naghavi, Thin Solid Films 633, 202 (2017). ger, B. Dimmler, and M. Powalla, CIGS Devices with Increased
159 B. Bissig, R. Carron, L. Greuter, S. Nishiwaki, E. Avancini, C. Bandgap Energy: Results of the EFFCIS Project, EU PVSEC Pro-
Andres, T. Feurer, S. Buecheler, and A. N. Tiwari, Prog. Photovolt.- gramme Online, 2021, https://www.photovoltaic-conference.com/
Res. Appl. 26, 894 (2018). programme/conference.html.
160 B. Vermang, V. Fjällström, J. Pettersson, P. Salomé, and M. Edoff, 177 H. Hiroi, Y. Iwata, S. Adachi, H. Sugimoto, and A. Yamada, IEEE J.
Sol. Energy Mater. Sol. Cells 117, 505 (2013). Photovol. 6, 760 (2016).
161 E. Jarzembowski, B. Fuhrmann, H. Leipner, W. Fränzel, and R. 178 S. Shukla, M. Sood, D. Adeleye, S. Peedle, G. Kusch, D. Dahliah, M.
Scheer, Thin Solid Films 633, 61 (2017). Melchiorre, G. M. Rignanese, G. Hautier, R. Oliver, and S. Sie-
162 F. Mollica, J. Goffard, M. Jubault, F. Donsanti, S. Collin, A. Cattoni, bentritt, Joule 5, 1816 (2021).
L. Lombez, and N. Naghavi, in Comparative study of patterned TiO2 179 F. Larsson, N. S. Nilsson, J. Keller, C. Frisk, V. Kosyak, M. Edoff,
and Al2O3 layers as passivated back-contact for ultra-thin Cu(In, Ga)Se2 and T. Törndahl, Prog. Photovolt.-Res. Appl. 25, 755 (2017).
solar cells: Proceedings of the in 2016 IEEE 43rd Photovoltaic 180 N. Barreau, A. Thomere, D. Cammilleri, A. Crossay, C. Guillot-
Specialists Conference (PVSC), Portland, 2016. Deudon, A. Lafond, N. Stéphant; D. Lincot, M. T. Caldes, R. Bod-
163 B. Vermang, J. T. Watjen, C. Frisk, V. Fjallstrom, F. Rostvall, M. euxn, and B. Bérenguier, in High efficiency solar cell based on
Edoff, P. Salome, J. Borme, N. Nicoara, and S. Sadewasser, IEEE J. Cu(In,Ga)S2 thin film grown by 3-stage process: Proceedings of the
Photovol. 4, 1644 (2014). in 2020 47th IEEE Photovoltaic Specialists Conference (PVSC),
164 G. Birant, J. de Wild, M. Meuris, J. Poortmans, and B. Vermang, Calgary, 2020.
Appl. Sci. 9, 677 (2019). 181 T. J. Coutts, J. S. Ward, D. L. Young, K. A. Emery, T. A. Gessert, and
165 F. Werner, B. Veith-Wolf, C. Spindler, M. R. Barget, F. Babbe, J. R. Noufi, Prog. Photovolt.-Res. Appl. 11, 359 (2003).
Guillot, J. Schmidt, and S. Siebentritt, Phys. Rev. Appl. 13, 054004 182 T. P. White, N. N. Lal, and K. R. Catchpole, IEEE J. Photovol. 4, 208
(2020). (2014).
166 S. Albrecht, and B. Rech, Nat. Energy 2, 16196 (2017). 183 M. Sood, P. Gnanasambandan, D. Adeleye, S. Shukla, N. Adjeroud,
167 G. E. Eperon, M. T. Hörantner, and H. J. Snaith, Nat. Rev. Chem. 1, R. Leturcq, and S. Siebentritt, J. Phys. Energy 4, 045005 (2022).
0095 (2017). 184 M. Sood, D. Adeleye, S. Shukla, T. Törndahl, A. Hultqvist, and S.
168 F. Lang, M. Jošt, K. Frohna, E. Köhnen, A. Al-Ashouri, A. R. Siebentritt, Faraday Discuss. 239, 328 (2022).
Bowman, T. Bertram, A. B. Morales-Vilches, D. Koushik, E. M. 185 P. O. Grabitz, U. Rau, and J. H. Werner, Thin Solid Films 487, 14
Tennyson, K. Galkowski, G. Landi, M. Creatore, B. Stannowski, C. (2005).
A. Kaufmann, J. Bundesmann, J. Rappich, B. Rech, A. Denker, S. 186 T. P. Weiss, B. Bissig, T. Feurer, R. Carron, S. Buecheler, and A. N.
Albrecht, H. C. Neitzert, N. H. Nickel, and S. D. Stranks, Joule 4, Tiwari, Sci. Rep. 9, 5385 (2019).
1054 (2020). 187 D. Colombara, H. Elanzeery, N. Nicoara, D. Sharma, M. Claro, T.
169 M. Jošt, E. Köhnen, A. Al-Ashouri, T. Bertram, Š. Tomšič, A. Ma- Schwarz, A. Koprek, M. H. Wolter, M. Melchiorre, M. Sood, N.
gomedov, E. Kasparavicius, T. Kodalle, B. Lipovšek, V. Getautis, R. Valle, O. Bondarchuk, F. Babbe, C. Spindler, O. Cojocaru-Miredin,
Schlatmann, C. A. Kaufmann, S. Albrecht, and M. Topič, ACS En- D. Raabe, P. J. Dale, S. Sadewasser, and S. Siebentritt, Nat. Com-
ergy Lett. 7, 1298 (2022). mun. 11, 3634 (2020).
170 S. Gharibzadeh, I. M. Hossain, P. Fassl, B. A. Nejand, T. Abzieher, 188 H. Elanzeery, M. Melchiorre, M. Sood, F. Babbe, F. Werner, G.
M. Schultes, E. Ahlswede, P. Jackson, M. Powalla, S. Schäfer, M. Brammertz, and S. Siebentritt, Phys. Rev. Mater. 3, 055403
Rienäcker, T. Wietler, R. Peibst, U. Lemmer, B. S. Richards, and U. (2019).
W. Paetzold, Adv. Funct. Mater. 30, 1909919 (2020). 189 I. M. Peters, C. D. Rodriguez Gallegos, S. E. Sofia, and T. Buo-
171 M. Nakamura, C. C. Lin, C. Nishiyama, K. Tada, T. Bessho, and H. nassisi, Joule 3, 2732 (2019).
Segawa, ACS Appl. Energy Mater. 5, 8103 (2022). 190 S. E. Sofia, J. P. Mailoa, D. N. Weiss, B. J. Stanbery, T. Buonassisi,
172 K. Makita, Y. Kamikawa, H. Mizuno, R. Oshima, Y. Shoji, S. Ishi- and I. M. Peters, Nat. Energy 3, 387 (2018).