RES EARCH

SOLAR CELLS the perovskite layer on micrometric Si pyra-

mids, resulting in perovskite/c-Si tandems in
Interface passivation for 31.25%-efficient which both rear and front sides were textured
(16). Although these tandem cells had a high
perovskite/silicon tandem solar cells photocurrent thanks to the front-side pyrami-
dal texture, nonradiative recombination losses
Xin Yu Chin1,2*, Deniz Turkay1, Julian A. Steele3,4,5, Saba Tabean6,7, Santhana Eswara6, were substantial. One challenge is that most
Mounir Mensi8, Peter Fiala1, Christian M. Wolff1, Adriana Paracchino2, Kerem Artuk1, top-surface passivation approaches reported
Daniel Jacobs1, Quentin Guesnay1, Florent Sahli2, Gaëlle Andreatta2, Mathieu Boccard1, to date are not directly applicable to micro-
Quentin Jeangros1,2, Christophe Ballif1,2 metric textures because they involve depositing
nanometric organic layers from liquid solutions
Silicon solar cells are approaching their theoretical efficiency limit of 29%. This limitation can be (17–19). These processing routes usually yield
exceeded with advanced device architectures, where two or more solar cells are stacked to improve the nonconformal (incomplete) coatings on such
harvesting of solar energy. In this work, we devise a tandem device with a perovskite layer conformally surface textures. Building on our previous
coated on a silicon bottom cell featuring micrometric pyramids—the industry standard—to improve its work (16), we demonstrate enhanced tandem
photocurrent. Using an additive in the processing sequence, we regulate the perovskite crystallization performance using phosphonic acids in two
process and alleviate recombination losses occurring at the perovskite top surface interfacing the different roles to passivate interfacial defects:
electron-selective contact [buckminsterfullerene (C60)]. We demonstrate a device with an active area of first, as a hole-transport layer (HTL), [4-(3,6-
1.17 square centimeters, reaching a certified power conversion efficiency of 31.25%. dimethyl-9H-carbazol-9-yl)butyl]phosphonic acid

Downloaded from https://www.science.org at Argonne National Lab on July 07, 2023

(Me-4PACz) (9) and second, as a perovskite ad-

T
ditive in the form of 2,3,4,5,6-pentafluoro-
he deployment of photovoltaics (PV) can Metal halide perovskites combine several benzylphosphonic acid (FBPAc) (20). Together
be accelerated by maximizing the elec- key properties for effective multijunction PV, with a micrometric texture standard in the c-Si
trical power produced per unit area be- including a high absorption coefficient with a PV industry and an optimization of the layers
cause the cost distribution of a PV system sharp absorption edge (5), ambipolar charge located at the front of the tandem for further
is now dominated by the balance of transport with long diffusion lengths (6, 7), optical gains, these passivation strategies
system components (such as the mounting and a compositionally tunable bandgap en- mitigate voltage losses and result in tandem
system, wiring, manpower, and inverters) ergy (Eg) (8). A thin-film perovskite solar cell device efficiencies >30%, with an accredited
rather than by the cost of the PV panels. can be deposited directly on the front side of value of 31.25%.
This balance of system costs scales roughly a c-Si cell to lower thermalization losses and
with the installed area and favors PV tech- extend the range of achievable PCE to >30% Identification and mitigation of voltage losses
nologies with a high power-to–panel area (3). The performance potential of monolithic in perovskite top cells
ratio. However, the record power conversion two-terminal tandem architectures is illus- To assess loss pathways triggered by non-
efficiency (PCE) of crystalline silicon (c-Si) trated by reported PCEs as high as 33.7% for radiative recombination, we measured the
solar cells of 26.8% (1) is approaching the 1-cm2 illuminated areas (1). Most of the high- photoluminescence quantum yield of different
theoretical limit of 29.5% (2). The only ex- efficiency tandem cells reported to date use a layer stacks representative of our p-i-n perov-
perimentally validated approach to overcome Si wafer in which the front surface is mech- skite solar cells used in tandem cells. We
this PCE limitation under 1-sun illumination anically or chemically polished (9) or features quantified the internal voltage potential (in
conditions consists of combining several com- an adapted submicrometric texture smaller relative terms), or quasi–Fermi-level splitting
plementary photoactive materials (meaning than the perovskite layer thickness (typically (QFLS), created within the cesium formamidi-
multiple junctions) in a single device (3). Among from 500 nm to 1 mm) (10–13). This planar or nium lead iodide bromide [Cs0.18FA0.82Pb(I,
the different types of multijunction designs nanotextured front-side topography—wafers Br)3] absorber under light excitation (Fig. 1A).
reported to date, the combination of c-Si with used in the PV industry are usually etched to Flat glass/indium tin oxide (ITO) substrates
a metal halide perovskite in a tandem solar form pyramids of a few micrometers in height— were first coated with a HTL, either 2,2′,7,7′-
cell has been the focus of intense research enables the deposition of pinhole-free perov- tetra(N,N-di-tolyl)amino-9,9-spiro-bifluorene
efforts given the potential for both high PCE skite films by standard solution-based pro- (spiro-TTB), N4,N4,N4’’,N4’’-tetra([1,1’-biphenyl]-
and low manufacturing costs (4). cesses. However, this modification comes at 4-yl)-[1,1’:4’,1’’-terphenyl]-4,400-diamine (TaTM),
the expense of optical performance because or Me-4PACz (fig. S1). These three materials
the front side of the tandem is flat, and also were selected on the basis of their compati-
1
Photovoltaics and Thin-Film Electronics Laboratory (PV- when using a submicrometric Si texture, be- bility with textured surfaces. The first two
Lab), Institute of Microengineering (IMT), Ecole cause the latter is planarized by the non- can be thermally evaporated, and the latter can
Polytechnique Fédérale de Lausanne (EPFL), Neuchâtel, conformal solution-processed perovskite film. self-assemble on ITO to yield a conformal coat-
Switzerland. 2Centre Suisse d'Électronique et de
Microtechnique (CSEM), Neuchâtel, Switzerland. 3cMACS,
Consequently, these cell designs exhibit more ing on textured surfaces (21). We then de-
Department of Microbial and Molecular Systems, KU Leuven, reflection losses at the front of the tandem posited a perovskite thin film using a hybrid
3001 Leuven, Belgium. 4Australian Institute for because of an absence of double-bounce effects two-step deposition method (16). Specifically,
Bioengineering and Nanotechnology, The University of
Queensland, Brisbane, QLD 4072, Australia. 5School of
(14). Overall, a pyramidal texture at the front of a CsBr/PbI2 template was coevaporated on either
Mathematics and Physics, The University of Queensland, the tandem device limits reflection losses flat glass substrates or glass/ITO/HTL stacks
Brisbane, QLD 4072, Australia. 6Advanced Instrumentation because it enables the absorption of reflected before spin coating an organo-halide solution
for Nano-Analytics (AINA), Luxembourg Institute of Science
light in neighboring pyramids, whereas the consisting of FABr and FAI and, in some cases,
and Technology (LIST), Materials Research and Technology
Department, 41, Rue du Brill, L-4422 Belvaux, Luxembourg. presence of texture on both sides of the Si wafer with FBPAc. A thermal annealing step in an
7
University of Luxembourg, 2 Avenue de l’Université, improves the trapping of infrared light (15). ambient atmosphere at 150°C converted the
L-4365 Esch-sur-Alzette, Luxembourg. 8Institute of Chemical We previously reported a hybrid two-step films into a perovskite phase (see the experi-
Sciences and Engineering (ISIC), École Polytechnique
Fédérale de Lausanne, Sion, Switzerland. deposition method combining thermal evapo- mental section for details). In view of making
*Corresponding author. Email: xinyu.chin@csem.ch ration and spin coating to conformally coat tandems, a perovskite composition with an

Chin et al., Science 381, 59–63 (2023) 7 July 2023 1 of 4

RES EARCH | R E S E A R C H A R T I C L E

optical Eg of ~1.70 eV was selected (fig. S2).

Some of these stacks were then completed by
a C60 electron-transport layer (ETL) thermally
evaporated in high-vacuum conditions.
The QFLS of bare perovskite layers depo-
sited on glass (left column) remains constant
when incorporating FBPAc (Fig. 1A). When
inserting the HTL (Fig. 1A, hollow diamond),
the addition of FBPAc increases the QFLS
slightly with all three HTLs (+20 meV). Overall,
QFLS losses relative to the glass/perovskite
reference are highest on spiro-TTB (−80 and
−45 meV compared with the glass/perovskite
reference, without and with FBPAc, respec-
tively), followed by TaTm (−65 and −40 meV),
and lowest on Me-4PACz (−25 and −10 meV).
These differences suggest improved surface
passivation with Me-4PACz (9). Without
FBPAc, the addition of the C60 ETL (Fig. 1A,

Downloaded from https://www.science.org at Argonne National Lab on July 07, 2023

filled diamond) on the perovskite leads to
substantial losses in QFLS (−100 to −50 meV
in addition to the HTL-induced losses) caused
by the formation of trap states in the first
monolayer of C60 (22). On the other hand,
the incorporation of FBPAc in the perovskite
film synthesis mitigates some of these ETL-
induced losses, resulting in a +100-meV QFLS
gain compared with FBPAc-free films.
Overall, FBPAc suppresses Pb2+/Pb0 sur-
face defects, as shown by x-ray photoelectron Fig. 1. Reduced nonradiative recombination losses and improved crystallographic properties
spectroscopy (XPS) in Fig. 1B and figs. S3 to S5. when adding FBPAc. (A) QFLS loss of perovskite films (without and with FBPAc) extracted from
The phosphonic group of FBPAc coordinates photoluminescence quantum yield data relative to the glass/perovskite reference. Hollow diamond data
with Pb-related defects as also demonstrated points correspond to the QFLS of perovskite films deposited on different HTLs (spiro-TTB, TaTm, and
elsewhere (20, 23, 24), effectively lowering the Me-4PACz), and filled diamonds relate to films sandwiched by one of these HTLs and a C60 ETL.
population of top-surface nonradiative recom- (B and C) High-resolution XPS spectra of the Pb 4f orbital (B) and XRD patterns of perovskite films
bination states (25). Furthermore, the positive with and without FBPAc (C). a.u., arbitrary units.
impact of FBPAc is mainly observed when C60
is present: FBPAc likely separates the perov-
skite from the C60 layer, reducing the popula- improved solar cell performance achieved with using FBPAc. Overall, this two-step and glob-
tion of deep trap states forming in C60 when it FBPAc (see below). ally slower crystallization kinetics with the
is in direct contact with the perovskite (11, 22). additive likely originates from the interaction
It is important to note that some effects may Modification of the perovskite crystallization between Pb2+ and FBPAc (20, 23, 24). FBPAc
affect the QFLS data presented in this work. kinetics and resulting microstructure slows down the conversion of the CsBr/PbI2
First, the presence of oxygen and humidity To understand the impact of FBPAc on the template to the perovskite phase because of
during the annealing step and subsequent perovskite crystallization process, grazing- competition between FBPAc and other perov-
photoluminescence measurements may lead incidence wide-angle x-ray scattering (GIWAXS) skite precursors (namely FAI and FABr) for a
to a passivation of Pb-related defects (26). was performed in situ at 150°C starting from coordination with Pb2+ (25, 30). Advantageously,
Second, x-ray diffraction (XRD) patterns show a CsBr/PbI2 template freshly spin-coated with this difference in formation kinetics results in
that the addition of FBPAc decreases the FAI/FABr (Fig. 2A and fig. S8). Although the films with apparently larger and more clearly
fraction of PbI2 crystalline phases remaining crystallization onset of the perovskite (110) defined perovskite domains because FBPAc may
at the end of the crystallization process (Fig. 1C and (002) planes (q ~ 10.1 nm−1) (29) is similar inhibit the nucleation of new perovskite grains
and fig. S8)—an effect that should lower the with or without the additive, FBPAc slows down or domains, as shown by top-view scanning
photoluminescence quantum yield and hence the perovskite crystallization rate 100 s into electron microscopy (SEM) images (Fig. 2B).
the QFLS of films featuring the additive (27). the experiment (Fig. 2A). The intensity of these Submicrometer dark regions are observed
Other effects are less likely to explain the QFLS perovskite reflections then continues to in- on perovskite films featuring FBPAc (Fig. 2B).
trends we observed. For example, FBPAc and crease throughout the experiment up to ~400 s— From gas-field ion-source helium and neon
C60 do not appear to chemically interact based an effect not observed without the additive. ion microscopy coupled with secondary-ion
on additional XPS data (fig. S6). Also, the Without FBPAc, the peak intensity stabilizes mass spectrometry (SIMS) (Fig. 2C and fig. S9)
electronic band structure of uncoated perovskite after 150 s and remains at a lower value. This (31), these clusters are found to be rich in F
films treated with the additive shifts by ~0.5 eV modification of the crystallization pathway by and C, which is indicative of the clustering of
because of the strong dipole moment of FBPAc FBPAc is also observed when analyzing PbI2 the FBPAc excess on the perovskite surface.
(fig. S7), in line with literature (28). However, peaks (Fig. 2A). As in Fig. 1C, their evolution Films processed without FBPAc do not show
FBPAc is unlikely to lead to such a conduction confirms a more-complete conversion of the any clustering and feature a lower amount of
band level offset in full devices considering the CsBr/PbI2 template to a perovskite phase when F and C on their surface (Fig. 2, C and D, and

Chin et al., Science 381, 59–63 (2023) 7 July 2023 2 of 4

RES EARCH | R E S E A R C H A R T I C L E

port across the perovskite/C60 interface. Wash-

ing away the FBPAc with an orthogonal solvent
mixture results in perovskite cells with a simi-
lar Voc to that of untreated films, which con-
firms that FBPAc affects primarily recombination
at the perovskite top surface (fig. S16).
We then adapted the perovskite deposition
conditions to produce 1-cm2 tandem cells on
silicon heterojunction bottom cells that had
pyramids of 2 to 3 mm on both sides of the
wafer (Fig. 3A) (32, 33). Compared with our
previous tandem baseline (16), the perovskite
layer was made thicker (from ~650 nm to
~1 mm) to reach current matching conditions
with an Eg of ~1.7 eV (1.63 eV previously; Fig.
3B). For that purpose, we deposited a CsBr/
PbI2 template of 660 nm and adapted the
spin-coating parameters to convert the tem-
plate into a perovskite phase. The front elec-

Downloaded from https://www.science.org at Argonne National Lab on July 07, 2023

trode was made transparent by switching to
a stack made of C60, SnOx, indium zinc oxide
(IZO), an Ag grid, and a MgF2 antireflective
coating (see materials and methods and sup-
plementary text for details).
A Voc of 1.9 V is obtained with either TaTM
or Me-4PACz (fig. S17), which is consistent
with single-junction cell results (perovskite
top cells in fig. S14; c-Si bottom cell in fig. S18).
The FF of TaTm-based tandems is lower than
Fig. 2. Impact of FBPAc on the perovskite crystallization and resulting microstructure. (A) GIWAXS with Me-4PACz (<70% compared with >75%),
data showing the evolutions of the (110) and (002) planes of the perovskite phase (indicated as PK in the likely because of the low hole mobility in
figure) and of the (001) planes of the PbI2 phase during the perovskite crystallization process at 150°C. undoped TaTm (34). We also investigated the
(B) Secondary-electron SEM images highlighting the difference in surface morphologies when adding FBPAc, impact of a slight change in the perovskite Eg
with arrows showing FBPAc clusters. (C) SIMS maps of the 19F− and 12C2− signals. The bright and dark by changing the FABr:FAI molar ratio from
regions represent higher and lower, respectively, signal counts. (D) Corresponding histograms of the 19F− 2 to 1.75 and 1.5 (corresponding to a Eg of 1.69,
and 12C2− SIMS signals shown in (C). 1.68, and 1.67 eV, respectively). A PCE >30%
could be achieved with several of these perov-
skite compositions, with slight variations in
fig. S9). Depth profiling combined with XPS Perovskite single junctions and perovskite/c-Si Jsc and FF values depending on the top cell
reveals that FBPAc is confined to the top perov- tandem solar cells Eg and the current-limiting subcell (figs. S17
skite surface (figs. S10 to S13). The bulky FBPAc These perovskite films were incorporated and S19). These tandems remained stable for
molecules are expelled toward the top surface into ~0.49-cm2 glass/ITO/HTL/perovskite/ >360 days when stored at room temperature in
during the perovskite crystallization process. C60/bathocuproine (BCP) (or SnOx)/Cu (or Ag) N2 (fig. S20). A selection of these unencap-
The large reduction in QFLS losses suggests single junctions to validate the improvements sulated cells were then operated in ambient at
that FBPAc is also present in between the induced by FBPAc with current-voltage (J-V) their maximum power point at 65°C under a
surface clusters. This hypothesis stems from measurements (1000 W/m2, AM1.5G spectrum, 1-sun illumination (~20 to 30% relative hu-
the logarithmic dependency between QFLS and masked with a 0.25-cm2 aperture; fig. S14). midity; ISOS-L2 protocol). The most stable
surface recombination current (and hence non- The open-circuit voltage (Voc) of the devices device reaches a t80 of ~66 hours until 80% of
passivated surface fraction, presuming that improves by 20 to 30 mV when switching its initial PCE is measured (figs. S21 and S22).
FBPAc is reducing the surface recombination from spiro-TTB to TaTm and Me-4PACz and The degradation rate at 25°C of the FBPAc-
current by a few orders of magnitude). If FBPAc by an extra 40 to 50 mV when adding FBPAc, treated tandem stack is similar to that achieved
was present only at the position of the clusters in line with the QFLS results of Fig. 1A. The without FBPAc in our previous study (ISOS-L1
detected by SIMS, meaning that ~90% of the corresponding fill factor (FF) also increases conditions; figs. S21 and S22) (16).
film surface would be free of FBPAc, the dif- on average from ~69 to ~76% when changing A tandem cell featuring Me-4PACz and
ference in QFLS would be limited to ~3 meV the HTL and then to ~78 to 81% with the FBPAc was sent to the National Renewable
[DQFLS = 25.7 meV × ln(1/0.9) ≅ 3 meV]. We additive, confirming that the molecule does not Energy Laboratory (NREL) for certification.
observe a gain in QFLS of >50 meV when C60 lead to the formation of a barrier impeding The certified relative external quantum effi-
is present (Fig. 1A), highlighting that regions electron extraction. Overall, the combination ciency (EQE) shown in Fig. 3C indicates that
between the clusters also feature FBPAc. of Me-4PACz and FBPAc results in a PCE of up the tandem was slightly top cell limited and
Overall, FBPAc passivates Pb-related defects to 19.5%. Varying the concentration of FBPAc that both subcells produced a Jsc > 20 mA/cm2.
and mitigates the formation of trap states in from 2.5 to 7.5 mM in the FABr:FAI solution The stabilized J-V properties of the device were
the C60 layer by segregating on the perovskite does not affect the performance substantially measured using the asymptotic maximum
top surface, with the excess forming clusters (fig. S15), which suggests that the excess of power (Pmax) scan method in standard test
on the surface. FBPAc (Fig. 2B) does not hinder charge trans- conditions (STC) (25°C, 1000 W/m2, AM1.5G

Chin et al., Science 381, 59–63 (2023) 7 July 2023 3 of 4

RES EARCH | R E S E A R C H A R T I C L E

10. B. Chen et al., Joule 4, 850–864 (2020).

11. J. Liu et al., Science 377, 302–306 (2022).
12. P. Tockhorn et al., Nat. Nanotechnol. 17, 1214–1221
(2022).
13. Y. Hou et al., Science 367, 1135–1140 (2020).
14. F. Gota, R. Schmager, A. Farag, U. W. Paetzold, Opt. Express
30, 14172–14188 (2022).
15. D. A. Jacobs et al., J. Phys. Chem. Lett. 10, 3159–3170
(2019).
16. F. Sahli et al., Nat. Mater. 17, 820–826 (2018).
17. C. M. Wolff, P. Caprioglio, M. Stolterfoht, D. Neher, Adv. Mater.
31, e1902762 (2019).
18. J. Peng et al., Adv. Energy Mater. 8, 1801208 (2018).
19. R. Azmi et al., Science 376, 73–77 (2022).
20. E. Akman, A. E. Shalan, F. Sadegh, S. Akin, ChemSusChem 14,
1176–1183 (2021).
21. M. Roß et al., Adv. Energy Mater. 11, 2101460 (2021).
22. J. Warby et al., Adv. Energy Mater. 12, 2103567 (2022).
23. A. A. M. Brown et al., Nanoscale 11, 12370–12380
(2019).
24. J. S. Kim et al., Nature 611, 688–694 (2022).
25. T. Wang, J. D. A. Ng, E. W. Y. Ong, K. H. Khoo, Z. K. Tan,
Adv. Opt. Mater. 10, 2101914 (2021).
26. J. F. Galisteo-López, M. Anaya, M. E. Calvo, H. Míguez, J. Phys.

Downloaded from https://www.science.org at Argonne National Lab on July 07, 2023

Chem. Lett. 6, 2200–2205 (2015).
27. J. Euvrard, O. Gunawan, D. B. Mitzi, Adv. Energy Mater. 9,
1902706 (2019).
28. I. Lange et al., Adv. Funct. Mater. 24, 7014–7024 (2014).
29. L. Oesinghaus et al., Adv. Mater. Interfaces 3, 1600403
(2016).
30. T. Jeon et al., Adv. Energy Mater. 7, 1602596 (2017).
31. T. Wirtz, O. De Castro, J.-N. Audinot, P. Philipp, Annu. Rev.
Anal. Chem. 12, 523–543 (2019).
32. F. Sahli et al., Adv. Energy Mater. 8, 1701609 (2018).
33. L. Antognini et al., IEEE J. Photovolt. 11, 944–956
(2021).
34. C. Momblona et al., Energy Environ. Sci. 9, 3456–3463
(2016).
35. T. Song, D. J. Friedman, N. Kopidakis, Adv. Energy Mater. 11,
2100728 (2021).

Fig. 3. 31.25%-efficient perovskite/c-Si thanks to the combined action of Me-4PACz and FBPAc AC KNOWLED GME NTS
and a pyramidal texture on the front side of the tandem. (A) Schematic view of the perovskite/c-Si The authors thank P. Wyss and C. Allebé for SHJ wet-chemical
processing and V. Paratte, B. Kamino, T. Wirtz, J. Hofkens, and
tandem stack and corresponding picture of the perovskite top cell on textured Si. The acronyms a-Si:H
M. B. J. Roeffaers for their valuable scientific and technical
and nc-Si:H stand for amorphous and nanocrystalline hydrogenated Si, respectively, and (i/n/p) indicates discussion. The authors also thank the staff of the BL11 NCD-
the doping of the layers. (B) Secondary-electron SEM images of the front side of the tandem. (C to E) SWEET beamline at ALBA Synchrotron for their assistance in
Independently certified EQE (C), asymptotic maximum power (Pmax) scan (D), and current density and power recording the GIWAXS data. Funding: This study was supported by
Swiss National Science Foundation grant 200021_197006 (PAPET);
density tracking around Pmax (E). Swiss Federal Office of Energy grant SI/501804 (INTENT); Swiss
Federal Office of Energy grant SI/502209 (PRESTO); Luxembourg
National Research Fund (FNR) grant PRIDE17/12246511 (PACE);
spectrum; Fig. 3, D and E, and figs. S23 to S26) metric texture to improve optical properties, and Research Foundation - Flanders FWO, grant no. 12Y7221N,
(35). In agreement with data measured in- a high-quality 1-mm-thick perovskite absorber V400622N. C.M.W. thanks the European Commission for funding
house, the cell reaches a Voc of 1.91 V, a Jsc of deposited conformally on this texture using a through a Marie Skłodowska-Curie Fellowship (no. 101033077).
Author contributions: Conceptualization: X.Y.C. and Q.J.
20.47 mA/cm2, and a FF of 79.8%, which re- hybrid two-step method, and phosphonic Experiments: X.Y.C., D.T., J.A.S., S.T., S.E., M.M., P.F., C.M.W., A.P.,
sults in a certified PCE of 31.25% (with an groups at both sides of the absorber for im- K.A., D.J., Q.G., F.S., G.A., M.B., and Q.J. Project administration:
aperture area of 1.1677 cm2). proved interface passivation resulted in a X.Y.C., C.M.W., G.A., M.B., Q.J., and C.B. Supervision: X.Y.C., C.M.
W., Q.J., and C.B. Writing – original draft: X.Y.C. and Q.J. Writing –
tandem with an independently certified PCE
review & editing: X.Y.C., D.T., C.M.W., J.A.S., S.T., S.E., M.M., P.F.,
Conclusions of 31.25%. These results demonstrate how c-Si A.P., K.A., D.J., Q.G., F.S., G.A., M.B., Q.J., and C.B. Competing
We identified and mitigated nonradiative solar cells with a standard industrial micro- interests: X.Y.C. and Q.J. are inventors on a provisional patent
related to the subject matter of this manuscript (EP22210577.7).
recombination losses occurring at the inter- metric texture can be upgraded to increase
The authors declare no other competing interests. Data and
faces of perovskite/c-Si tandems featuring Si their PCE to >30%. materials availability: All data are available in the main text or the
wafers with a micrometric texture—the stan- supplementary materials. License information: Copyright © 2023
dard used in the c-Si PV industry. The use the authors, some rights reserved; exclusive licensee American
RE FERENCES AND NOTES
Association for the Advancement of Science. No claim to original
of Me-4PACz reduces voltage losses at the 1. “Best Research-Cell Efficiency Chart” (NREL, 2023); US government works. https://www.science.org/about/science-
perovskite/HTL interface, whereas the inclu- https://www.nrel.gov/pv/cell-efficiency.html. licenses-journal-article-reuse
2. S. Schäfer, R. Brendel, IEEE J. Photovolt. 8, 1156–1158
sion of FBPAc in the perovskite deposition (2018).
sequence reduces the voltage losses at the 3. L. C. Hirst, N. J. Ekins-Daukes, Prog. Photovolt. Res. Appl. 19, SUPPLEMENTARY MATERIALS
perovskite/C60 ETL interface and leads to 286–293 (2011). science.org/doi/10.1126/science.adg0091
4. F. Fu et al., Adv. Mater. 34, e2106540 (2022). Materials and Methods
more favorable perovskite microstructures
5. S. De Wolf et al., J. Phys. Chem. Lett. 5, 1035–1039 (2014). Supplementary Text
with larger domains. From XPS and SIMS 6. X. Y. Chin, D. Cortecchia, J. Yin, A. Bruno, C. Soci, Nat. Commun. Figs. S1 to S28
imaging, FBPAc is present on the perovskite 6, 7383 (2015). Tables S1 to S4
top surface and coordinates with perovskite 7. S. D. Stranks et al., Science 342, 341–344 (2013). References (36–45)
8. T. Jesper Jacobsson et al., Energy Environ. Sci. 9, 1706–1724
Pb-related defects through its phosphonic acid (2016). Submitted 27 November 2022; accepted 5 May 2023
group. Overall, the combination of a c-Si micro- 9. A. Al-Ashouri et al., Science 370, 1300–1309 (2020). 10.1126/science.adg0091

Chin et al., Science 381, 59–63 (2023) 7 July 2023 4 of 4

 Interface passivation for 31.25%-efficient perovskite/silicon tandem solar cells
Xin Yu Chin, Deniz Turkay, Julian A. Steele, Saba Tabean, Santhana Eswara, Mounir Mensi, Peter Fiala, Christian
M. Wolff, Adriana Paracchino, Kerem Artuk, Daniel Jacobs, Quentin Guesnay, Florent Sahli, Galle Andreatta, Mathieu
Boccard, Quentin Jeangros, and Christophe Ballif

Science, 381 (6653), .

DOI: 10.1126/science.adg0091

Downloaded from https://www.science.org at Argonne National Lab on July 07, 2023

Editor’s summary
Two studies show how interfaces between perovskite layers and silicon cells in tandem solar cells can be modified
to improve performance (see the Perspective by De Wolf and Aydin). Mariotti et al. showed that an ionic liquid,
piperazinium iodide, improved band alignment and enhanced charge extraction at the interface of a trihalide perovskite
and a C60 electron-transporting layer by creating a positive dipole. With these modifications, a 2.0-volt open circuit
voltage was achieved in a silicon tandem cell. Chin et al. report the uniform deposition of the perovskite top cell on the
micropyramids of crystalline silicon cells to achieve high photocurrents in tandem solar cells. Two different phosphonic
acids improved the perovskite crystallization process and also minimized recombination losses. These modifications
yielded perovskite/silicon tandem cells with certified power conversion efficiencies of more than 31% for active areas of
at least 1 square centimeter. —PDS

View the article online

https://www.science.org/doi/10.1126/science.adg0091
Permissions
https://www.science.org/help/reprints-and-permissions

Use of this article is subject to the Terms of service

Science (ISSN ) is published by the American Association for the Advancement of Science. 1200 New York Avenue NW, Washington, DC
20005. The title Science is a registered trademark of AAAS.
Copyright © 2023 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim
to original U.S. Government Works