nature energy

Article https://doi.org/10.1038/s41560-024-01557-z

Non-fullerene acceptor with asymmetric

structure and phenyl-substituted alkyl side
chain for 20.2% efficiency organic solar cells

Received: 28 June 2023 Yuanyuan Jiang1,2,4, Shaoming Sun1,2,4, Renjie Xu1,2,4, Feng Liu 1
,
Xiaodan Miao1,2, Guangliu Ran 3, Kerui Liu1,2, Yuanping Yi 1,
Accepted: 20 May 2024
Wenkai Zhang 3 & Xiaozhang Zhu 1,2
Published online: xx xx xxxx

Check for updates For organic solar cells (OSCs), bridging the gap with Shockley–Queisser
limit necessitates simultaneously reducing the energy loss for a high
open-circuit voltage, improving light utilization for enhanced short-circuit
current density and maintaining ideal nanomorphology with a high fill
factor through molecular design and device engineering. Here we design
and synthesize an asymmetric non-fullerene acceptor (Z8) featuring
tethered phenyl groups to establish an alloy acceptor in ternary OSCs. The
asymmetric structure minimizes non-radiative energy loss and charge
recombination owing to delocalized excitons. The phenyl-substituted alkyl
side chain impacts on the intermolecular interactions, improving the f­i l­m
n­an­omorphology with efficient exciton dissociation and reduced charge
recombination. We demonstrate OSCs with an efficiency of 20.2% (certified
19.8%) based on the D18:Z8:L8-BO ternary blend. Through theoretical
calculations, we examine the overall distribution of photon and carrier
losses and analyse the potential for improvement on open-circuit voltage,
short-circuit current density and fill factor, providing rational guidance for
further development of the OSC performance.

The power conversion efficiency (PCE) of organic solar cells (OSCs)
is mainly determined by parameters: open-circuit voltage (VOC),
short-circuit current density (JSC) and fill factor (FF), following
PCE = VOC × JSC × FF/Pin, where Pin is the incident light intensity. Over
the past decades, great efforts have been consistently devoted to
improving the PCE of OSCs, including the innovation of new polymer
donors1–4, electron acceptors5–10 and device engineering11–17. The mile-
stone achievements can be attributed to the rapid development of
novel fused-ring non-fullerene electron acceptors (NFAs) since 2015
(refs. 6,10,17–22). Currently, achieving highly efficient single-junction
OSCs with PCE exceeding 20% remains challenging, which necessitates
simultaneous improvements in VOC, JSC and FF. NFA-based OSCs usually
show relatively large voltage loss in the range of 0.50–0.80 V, which is
far beyond the ideal value of 0.25–0.30 V given by the Shockley–Queis-
ser (S-Q) theory23, ascribed to the radiative and non-radiative energy
losses below bandgap. Recent studies indicate the A-DA’D-A-type of
Y-analogues can form the three-dimensional charge transport chan-
nels with low energetic disorder24–26, enabling the hybridization of
localized exciton (LE) and charge transfer (CT) states to mitigate the
non-radiative recombination loss27,28. Although voltage loss has long
been considered as a limiting factor for OSCs, the current loss is an
equally critical problem that requires immediate attention29. The loss

Beijing National Laboratory for Molecular Sciences, CAS Key Laboratory for Organic Solids, Institute of Chemistry, Chinese Academy of Sciences,
1

Beijing, China. 2School of Chemical Sciences, University of Chinese Academy of Sciences, Beijing, China. 3Department of Physics and Applied Optics
Beijing Area Major Laboratory, Center for Advanced Quantum Studies, Beijing Normal University, Beijing, China. 4These authors contributed equally:
Yuanyuan Jiang, Shaoming Sun, Renjie Xu. e-mail: liufeng112@iccas.ac.cn; xzzhu@iccas.ac.cn

Nature Energy
Article
of incident photons is derived from insufficient internal quantum effi-
ciency (IQE), parasitic absorption of interlayers and electrodes, and the
reflection from the device. For state-of-the-art OSCs, the charge genera-
tion efficiency is mainly susceptible to the highest occupied molecular
orbital (HOMO) energy offset (ΔEHOMO), and achieving efficient charge
generation with maximized JSC under minimal ΔEHOMO value has been
considered as a practical approach to realizing high-performance
OSCs30. FF is another key factor affecting the PCE of photovoltaics.
The non-ideal FF is, complicatedly, subject to the relatively low carrier
mobility and affected by the blend nanomorphology, suffering inter-
facial and bulk charge recombination31,32. To simultaneously improve
the VOC, JSC and FF of single-junction OSCs, constructing highly efficient
ternary organic solar cells (TOSCs) by introducing a third component
into the binary host system is a promising strategy11,33–36. Blending
multi-component active materials produces a broader absorption
window, resulting in desired light harvesting. Moreover, the guest
component can fine-tune the electronic properties and affect the nano-
morphology of the blend films, resulting in efficient charge generation
and transport, even with reduced voltage loss37–40. However, the PCE of
TOSCs with all parameters simultaneously improved is rarely reported
due to the trade-off among VOC, JSC and FF.
In this Article, we design asymmetric NFA (Z8) featuring phenyl-
substitution side chains that exhibits desirable optoelectronic proper-
ties, such as high photoluminescence quantum yield (PLQY) and delo-
calized excitons, increasing the probability to simultaneously achieve
a low non-radiative energy loss and an efficient charge generation. The
tethered phenyl groups in Z8 can participate in the intramolecular
and intermolecular interaction, affecting the acceptor:acceptor and
donor:acceptor intermolecular interaction. The two NFAs Z8 and
L8-BO can form an alloy acceptor, which helps to form a favourable
nanomorphology by blending with polymer donor D18, including suit-
able phase separation, vertical component distribution and enhanced
crystallinity. We achieve a high PCE of 20.2% in D18:Z8:L8-BO-based
single-junction OSCs. The photovoltaic parameters (VOC, JSC and
FF) achieved maximal gain in ternary devices benefitting from the
improved charge extraction and reduced charge recombination, result-
ing in efficient photon and carrier management.
Non-fullerene design and properties
Aromatic additives have been widely utilized to fine-tune the nano-
morphology of NFA-based OSCs, which can reduce the intermolec-
ular absorption energy, leave relatively sufficient free volumes and
decrease nucleation sites, thus mitigating over-agglomeration of
NFAs and improving molecular stacking with enhanced crystallin-
ity41. Moreover, recent works also demonstrated that asymmetric
NFAs present a promising potential in constructing high-performance
TOSCs via modifying the nanomorphology of blend films42. Here, we
designed NFAs 2,2′-((2Z,2′Z)-((12,13-bis(2-phenethyl-6-phenylhexyl)-
3,9-diundecyl-12,13-dihydro-[1,2,5]thiadiazolo[3,4-e]thieno[2′,3′:4′,5′]-
thieno[2′,3′:4,5]pyrrolo[3,2-g]thieno[2′,3′:4,5]thieno[3,2-b]indole-2,
10-diyl)bis(methaneyl-ylidene))-bis(6,7-difluoro-3-oxo-2,3-dihydro-
1H-cyclopenta[b]naphthalene-2,1-diylidene))dimalononitrile (Z7) with
symmetric structure and 2-((Z)-2-((10-(((Z)-1-(dicyanomethylene)-
5,6-difluoro-3-oxo-1,3-dihydro-2H-inden-2-ylidene)methyl)-12,13-bis
(2-phenethyl-6-phenylhexyl)-3,9-diundecyl-12,13-dihydro-[1,2,5]
thiadiazolo[3,4-e]thieno[2′,3′:4′,5′]thieno[2′,3′:4,5]pyrrolo[3,2-g]
thieno[2′,3′:4,5]thieno[3,2-b]indol-2-yl)methylene)-6,7-difluoro-3-o
xo-2,3-dihydro-1H-cyclopenta[b]-naphthalen-1-ylidene)malononitrile
(Z8) with asymmetric terminal groups by introducing four aromatic
substituents (phenyl groups) on the inner side chains of the conjugated
backbone to finely modulate the aggregation behaviour. The molecular
structures are shown in Fig. 1a. The detailed synthetic routes for NFAs
Z7 and Z8 are depicted in Supplementary Scheme 1.
The photoelectric properties of Z7 and Z8 are exhibited in Fig. 1b,c
and Supplementary Figs. 1–7. The maximum absorption peaks for Z7,
Nature Energy
https://doi.org/10.1038/s41560-024-01557-z
Z8 and 2,2′-((2Z,2′Z)-((3,9-bis(2-butyloctyl)-12,13-bis(2-ethylhexyl)-
12,13-dihydro-[1,2,5]thiadiazolo[3,4-e]thieno-[2′,3′: 4′,5′]
thieno[2′,3′:4,5]pyrrolo[3,2-g]thieno[2′,3′:4,5]thieno[3,2-b]
indole-2,10-diyl)bis(methaneyl-ylidene))bis(5,6-difluoro-3-oxo-
2,3-dihydro-1H-indene-2,1-diylidene))dimalononitrile (L8-BO) are
located at 761, 746 and 731 nm in dilute chloroform solution, and 832,
814 and 796 nm in neat films, respectively. Compared with L8-BO, Z7
and Z8 deliver red-shifted absorption spectrum, which can broaden
the light-harvesting window for higher JSC in photovoltaic devices. The
energy levels of the investigated materials were checked by ultravio-
let photoelectron spectra and inverse photoemission spectroscopy
measurements, and the relevant data are summarized in Supplemen-
tary Table 1. The HOMO and the lowest unoccupied molecular orbital
(LUMO) energy levels are −5.60 and −3.96 eV for Z7, −5.68 and −3.91 eV
for Z8, and −5.72 and −3.84 eV for L8-BO, respectively. For the blend
acceptors Z7:L8-BO and Z8:L8-BO, it is expected that they will exhibit
middling energy levels between these of L8-BO and Z-type acceptors,
considering the formation of alloy acceptors. From Supplementary
Figs. 6 and 7, it can be observed that Z7:L8-BO (0.3:0.9) and Z8:L8-BO
(0.3:0.9) deliver LUMO energy levels of −3.88 and −3.84 eV and HOMO
energy levels of −5.72 and −5.72 eV, respectively. The variation of energy
levels agrees well with the density functional theory (DFT) calculations
(Supplementary Fig. 4). Z-type acceptors exhibit more delocalized
orbitals, which would facilitate exciton dissociation25.
The molecular dynamics (MD) simulations of Z8 neat film
were conducted to reveal the effect of phenyl-substituted alkyl side
chain on intermolecular interactions (Supplementary Fig. 8). Sup-
plementary Figs. 9–11 show the neighbour atom distribution of the
tethered phenyl groups and the possible π–π stacking dimers, con-
taining the phenyl-mediated interactions. A relatively high ratio of
phenyl-mediated π–π stacking was obtained, and the tethered phe-
nyl groups trend to interact with the electron-withdrawing parts of
the NFAs. Moreover, the phenyl-alkyl groups are expected to restrict
molecular motion via phenyl-mediated intermolecular interactions,
leading to a suppressed electron–phonon coupling and a reduced
non-radiative energy loss43, as discussed below. The miscibility between
the active layer materials has a significant influence on the resulting
blend nanomorphology. Therefore, contact angle measurements were
performed to investigate the miscibility between materials (Supple-
mentary Fig. 14), and the related data are presented in Supplementary
Table 2. The Flory–Huggins interaction parameter (χ) for the donor–
acceptor pairs (D18:Z7, D18:Z8 and D18:L8-BO) give low χ values of
~0.4 K, implying the good miscibility between donor and acceptors,
which is beneficial to supplying sufficient donor–acceptor interfaces
for exciton dissociation. Additionally, the low χ values for Z7:L8-BO
(0.002 K) and Z8:L8-BO (0.004 K) generate great promise for forma-
tion of alloy acceptors.
Photovoltaic performance
To achieve complementary absorption and matched energy level, in
this work, poly[(2,6-(4,8-bis(5-(2-ethylhexyl-3-fluoro)thiophen-2-yl)-
benzo[1,2-b:4,5-b’]dithiophene))-alt-5,5′-(5,8-bis(4-(2-butyl-octyl)
thiophen-2-yl)dithieno[3′,2′:3,4;2′,3′:5,6]benzo[1,2-c][1,2,5]thiadia-
zole)] (D18) was selected as polymer donor, and the corresponding
absorption spectrum and energy level diagram are shown in Fig. 1b,c.
The detailed device optimizations, including D:A ratio and film
thickness, are displayed in Supplementary Tables 3 and 4 and Sup-
plementary Fig. 15. The optimal PCE values for D18:Z7-, D18:Z8- and
D18:L8-BO-based binary solar cells are 17.6%, 18.6% and 18.6%, respec-
tively (Fig. 1d and Table 1). Compared with D18:L8-BO, the D18:Z7 and
D18:Z8 binary OSCs show higher JSC of 26.5 mA cm−2 and 26.7 mA cm−2,
respectively, attributed to a broader absorption range, as evidenced
by the external quantum efficiency (EQE) spectra (Fig. 1e). Because
of the high-lying LUMO energy level in L8-BO, D18:L8-BO displays a
higher VOC of 0.92 V, surpassing the VOC values in D18:Z7 (0.88 V) and
Article https://doi.org/10.1038/s41560-024-01557-z

a b c
S
N N

C11H23 S S
C11H23

Normalized absorption intensity

CN D18 Z7 Z8 L8-BO
S
N N NC
1.0 D18 EVAC
NC S CN Z7
Z8
O Ph Ph O
0.8 L8-BO
F

F Ph Ph
–4.57 eV –4.85 eV –4.95 eV –4.94 eV
Z8 F
0.6
F

N
S
N
LUMO –2.88 eV
0.4
C11H23 S S –3.96 eV –3.91 eV –3.84 eV
C11H23
CN NC
S
N N
0.2
NC S CN EF
–0.68 eV
O Ph HOMO –5.25 eV –0.75 eV –0.73 eV –0.78 eV
Ph O
0 –5.60 eV –5.68 eV –5.72 eV
400 600 800 1,000
Ph Ph
Z7
F
F Wavelength (nm)
F
F

d e f
5 100
30 20
Current density (mA cm−2)

Current density (mA cm−2)

Power density (mW cm−2)

0
D18:Z7 80
−5 D18:Z8
D18:L8-BO
D18:Z7:L8-BO 60 20
EQE (%)

−10 D18:Z8:L8-BO D18:Z7

−15 D18:Z8 10
40 D18:L8-BO
D18:Z7:L8-BO 10
−20 D18:Z8:L8-BO
20 Certified result
−25 PCE = 19.8%

−30 0 0 0
0 0.5 1.0 400 600 800 1,000 0 0.2 0.4 0.6 0.8 1.0
Voltage (V) Wavelength (nm) Voltage (V)

g h VOC (V) D18:Z7

0.93 D18:Z8
0.30 ∆E1 ∆E2 ∆E3 D18:L8-BO
D18:Z7:L8-BO
0.25
Energy loss (eV)

0.90 D18:Z8:L8-BO
JSC 27.5 FF
0.20 (mA cm–2)
27.0
26.5 0.80
26.0
25.5 0.87
0.96 0.76
0.84

0.05
0.98
0.88

0 1.00
0.92
Pcoll Pdiss
BO
BO
O
Z8
Z7

-B
8:
8:

8-
8-
L8
D1
D1

:L
:L
8:

Z8
Z7
D1

8:

8:
D1

D1

Fig. 1 | Materials and device performance. a, Chemical structures of Z7 and Z8. energy level, respectively. d, Characteristic J–V curves. e, Corresponding EQE
Different colours contribute to enhancing the distinction between molecular curves based on D18:Z7, D18:Z8, D18:L8-BO, D18:Z7:L8-BO and D18:Z8:L8-BO
structures: grey, electron-donating units; blue/yellow, terminal electron- devices. The D:A weight ratios are fixed at 1:1.2 for binary devices and 1:0.3:0.9
withdrawing units; pink, phenyl groups. b, Normalized absorption intensity for for ternary devices. f, Certified J–V and power density curves based on the
pure D18, Z7, Z8 and L8-BO films. c, The energy level diagram of D18, Z7, Z8 and optimal D18:Z8:L8-BO device, obtained from the NIM. g, The detailed energy
L8-BO neat films. EVAC, vacuum energy; EF, Fermi energy level; LUMO, the LUMO loss for the binary and ternary OSCs. h, Parameter comparison for the binary
energy level; HOMO, the HOMO energy level. The values from top to bottom and ternary OSCs.
represent the work function, LUMO energy level, injection barrier and HOMO

Table 1 | Photovoltaic parameters

Active layer VOC (V) JSC (mA cm−2) FF (%) PCE (%) Eloss (eV) ∆E3 (eV)

D18:Z7 (1:1.2)
a
0.88 (0.87 ± 0.003) 26.5 (26.2 ± 0.2) 76.1 (76.3 ± 0.6) 17.6 (17.4 ± 0.1) 0.50 0.18
D18:Z8a (1:1.2) 0.89 (0.88 ± 0.003) 26.7 (26.6 ± 0.3) 78.4 (77.8 ± 0.8) 18.6 (18.2 ± 0.2) 0.51 0.19
D18:L8-BOa (1:1.2) 0.92 (0.92 ± 0.003) 25.9 (25.3 ± 0.3) 78.5 (78.9 ± 0.7) 18.6 (18.3 ± 0.2) 0.53 0.20
D18:Z7:L8-BOb (1:0.3:0.9) 0.91 (0.91 ± 0.004) 26.5 (26.1 ± 0.4) 78.3 (78.1 ± 0.8) 18.9 (18.6 ± 0.2) 0.52 0.19
D18:Z8:L8-BO (1:0.3:0.9) b
0.92 (0.92 ± 0.003) 27.2 (27.2 ± 0.3) 80.8 (79.4 ± 0.8) 20.2 (19.8 ± 0.3) 0.51 0.19
D18:Z8:L8-BOc (1:0.3:0.9) 0.90 27.0 81.0 19.8 – –
a
The average values with standard deviation were calculated from 30 devices. bThe average values with standard deviation were calculated from 50 devices. cThe certified PCE from NIM, China.
Summarized photovoltaic parameters of binary and ternary OSCs under AM 1.5G illumination, 100 mW cm–2.

D18:Z8 (0.89 V). Such a phenomenon inspired us to combine Z-type which achieved the high PCE of 20.2%, with a high VOC of 0.92 V, JSC of
acceptors with L8-BO to construct ternary devices, thus simultaneously 27.2 mA cm−2 and FF of 80.8%. The higher JSC and FF values of the ter-
achieving desired VOC and JSC. Inspiringly, an all-round enhancement of nary blends are correlated with the broadened light absorption, more
photovoltaic parameters was observed in D18:Z8:L8-BO (1:0.3:0.9), efficient charge generation, transport and collection properties as

Nature Energy
Article
discussed below. The maintained high VOC of 0.92 V in D18:Z8:L8-BO
can be ascribed to the reduced energy loss. However, with further addi-
tion of Z8 into the D18:L8-BO host system, both VOC and FF exhibit a
decreased trend, leading to inferior efficiency of 18.9%. Supplemen-
tary Fig. 16 shows that the Z8-based ternary exhibits a simultaneous
reduction in Vloss, Jloss and FFloss, where Vloss, Jloss and FFloss are defined as
the difference between realistic and ideal parameters, providing an
explanation for the inspiring improvement in efficiency. It is notewor-
thy that the PCE of 20.2% is the highest ever reported for single-junction
OSCs, and a certified PCE of 19.8% obtained from the National Institute
of Metrology (NIM), China, as depicted in Fig. 1f and Supplementary
Fig. 17. Besides, the maximum power point tracking measurements were
performed for D18:Z8-, D18:L8-BO- and D18:Z8:L8-BO-based devices,
as shown in Supplementary Fig. 18. After ~500 h of photo-ageing, the
D18:Z8:L8-BO-based device shows superior stability compared with
the corresponding binary devices, retaining 83% of its original effi-
ciency, indicating that the ternary strategy can effectively improve the
stability of OSCs. Supplementary Table 5 and Supplementary Fig. 19
demonstrate the versatility of the ternary strategy in other OSC systems
(D18:Z8:BTP-eC9, D18:Z8:N3 and D18:Z8:Y6), all of which achieved
excellent photovoltaic performance.
Figure 1e illustrates the corresponding EQE response spectra for
the binary and ternary OSCs. Compared with the D18:L8-BO-based
binary device, the D18:Z8:L8-BO-based ternary devices exhibit a
broader absorption range and enhanced EQE values above 80%
across 600–800 nm, contributing to the increased JSC. Because of
the unique bulk-heterojunction (BHJ) construction in OSCs, the
photon-to-electron conversion efficiency is subject to exciton disso-
ciation and charge collection. Thus, the charge dissociation probability
(Pdiss) and charge collection probability (Pcoll) for all OSCs were calcu-
lated from the photo-induced current density against effective voltage
plots (Supplementary Fig. 20). The D18:Z8:L8-BO-based devices exhibit
the highest Pdiss of 91.4% and Pcoll of 99.6% (Fig. 1h), enabling a high
charge generation probability and consequently yielding the best JSC
and FF. In contrast to relevant binary cells, the Z8-based ternary OSCs
simultaneously improved VOC, JSC and FF, producing a high PCE over
20%, underlining the significance of the ternary strategy.
Energy loss analysis
To gain insight into the high VOC obtained in Z8-based ternary OSCs,
a detailed energy loss (Eloss) analysis was performed (Fig. 1g), and
the relevant data are summarized in Supplementary Table 6. The
Eloss (Eloss = Eg − qVOC) for D18:Z7, D18:Z8 and D18:L8-BO binary OSCs
is calculated to be 0.50, 0.51 and 0.53 eV, respectively, and the values
for D18:Z7:L8-BO and D18:Z8:L8-BO are 0.52 and 0.51 eV, respectively,
where two TOSCs exhibited lower Eloss than that of D18:L8-BO-based
binary OSCs34,44. According to the detailed-balance theory, Eloss can
be divided into three parts: Eloss = q(∆VSQ + ∆Vr + ∆Vnr) = ∆E1 + ∆E2 + ∆E3,
where ∆E1 is the radiative loss above the bandgap derived from the
S-Q limit and ∆E2 and ∆E3 are the radiative and non-radiative energy
losses below the bandgap, respectively.
∆E1 is inevitable for any kind of photovoltaics, determined
only by the bandgap. The ∆E1 values for the investigated devices
are 0.26–0.27 eV. The binary OSCs exhibit the same ∆E2 values of
0.06 eV, whereas Z8-based TOSCs present a reduced ∆E2 value
of 0.05 eV, which can be ascribed to the decreased energetic disorder
of band tail states. In Supplementary Fig. 21b–f, D18:Z8:L8-BO exhibits
the lowest Urbach energy (EU) value of 22.8 meV, implying that the
introduction of Z8 into D18:L8-BO can effectively reduce energetic dis-
order, not only allowing the decrease in Eloss but also helping to reduce
trap density with restricted charge recombination24. The impedance
and capacitance–voltage measurements were performed to evaluate
the trap density (Nt) in ternary and binary OSCs. As shown in Supple-
mentary Fig. 22, ternary OSCs deliver reduced Nt and D18:Z8:L8-BO-
based ternary cells exhibit the lowest Nt value of 6.33 × 1021 m−3,
Nature Energy
https://doi.org/10.1038/s41560-024-01557-z
contributing to the reduction in energy loss and charge recombination.
Given the fast non-radiative decay rate during the exciton relaxation
process, minimizing ∆E3 has been regarded as a key and challeng-
ing issue in reducing the total Eloss. D18:Z7 and D18:Z8 binary devices
exhibit smaller ∆E3 values of 0.18 and 0.19 eV than that of D18:L8-BO
(0.20 eV), which is consistent with the higher PLQY of Z7 and Z8 (Sup-
plementary Table 7). To our delight, D18:Z7:L8-BO and D18:Z8:L8-BO
ternary devices maintained high electroluminescence with suppressed
non-radiative recombination loss, leading to low ∆E3 values of 0.19 eV.
From Supplementary Figs. 24–26, it can be observed that Z-type accep-
tors exhibit more evident delocalized singlet exciton (DSE) signals,
which are well maintained in their corresponding binary and ternary
blends. The DSE signals can offer an additional pathway (LE-DSE-CS)
for exciton dissociation, where CS refers to charge-separated state,
which is beneficial to reducing CT-mediated recombination loss35.
Charge dynamics
To deeply understand the improved JSC and FF in ternary OSCs, it is
necessary to investigate the charge dynamics process, including exci-
ton generation, dissociation, charge transport and collection. In gen-
eral, NFA-based OSCs show ultrafast electron transfer because of the
large LUMO energy offset (>0.3 eV) between donors and acceptors;
thus, the determining factor for charge transfer efficiency lies in the
hole transfer. Transient absorption (TA) spectroscopy was applied to
investigate the exciton dissociation process in the binary and ternary
systems, in which only the acceptor can be excited with a low-energy
pump beam at 800 nm. The positive signals at 590 nm can be assigned
to the ground state bleaching (GSB) of D18, where no GSB signals were
observed from the neat acceptor films (Fig. 2a and Supplementary
Fig. 27). Therefore, the hole transfer process can be assessed by fitting
the donor GSB (590 nm) traces with a double exponential function,
as displayed in Fig. 2c and Supplementary Table 8. For binary blends,
the Z-type acceptors exhibit faster exciton diffusion and exciton dis-
sociation compared with those of L8-BO. Additionally, the incorpo-
ration of Z-type acceptors can accelerate the hole transfer process.
To quantitatively evaluate the hole transfer efficiency for binary and
ternary blends27, the photoluminescence quenching experiments were
performed (Supplementary Fig. 30). The highest quenching efficiency
of 97.3% in the D18:Z8:L8-BO ternary blend indicates superior charge
transfer probability. These phenomena demonstrated that the intro-
duction of Z8 can effectively mitigate the geminate recombination
and yield efficient charge generation, allowing the increase in JSC value.
Transient photocurrent (TPC), transient photovoltage (TPV) and
charge extraction (CE) measurements were conducted to assess the
charge recombination kinetics in the binary and ternary cells45. From
the TPC results (Supplementary Fig. 31), Z8-based ternary solar cell
exhibits rapid rise and decay processes and possesses a shorter charge
extraction time, suggesting rapid charge generation and extraction
with fewer charge traps. Figure 2d plots the charge density against VOC,
and the largest charge carrier density (nCE) in D18:Z8:L8-BO implies
the most efficient charge generation. Moreover, the ternary OSCs
yield a longer carrier lifetime (τ) than those of the corresponding
binary devices (Fig. 2e), which is related to reduced charge recombi-
nation. The charge recombination can be evaluated using the equation
1 1
k= ∝ , where k is the recombination coefficient that can
RτnCE τnCE
reflect the overall recombination occurring in blend, and R is
constant, representing the recombination order. Compared with binary
OSCs, ternary OSCs deliver lower k values under varied charge densities
(Fig. 2f), implying suppressed charge recombination in ternary devices.
It is widely accepted that the FF of OSCs is determined by the
competing process of charge extraction and charge recombination
at a low internal electric field. The recombination mechanism can be
revealed by the relationship between VOC and light intensity (Plight)46.
As shown in Supplementary Fig. 20b, all the investigated binary and
Article https://doi.org/10.1038/s41560-024-01557-z

a b c
–3 –3
∆T/T(10 ) ∆T/T(10 )

3 Z8
Z8 1.0 3 D18:Z8:L8-BO 1.0
10 10
3

Time delay (ps)

Time delay (ps)

10
2 0 10
2
Donor 0

∆T/T (10–3)
CS
GSB ESA GSB Acceptor 2
1 −1.0 1 ESA −1.0
10 10 Donor GSB
1 D18:Z7
D18:Z8
D18:L8-BO
Acceptor D18:Z7:L8-BO
1 1 GSB D18:Z8:L8-BO
0
0 0
600 700 800 900 600 700 800 900 −1 0 1 10 100 1,000
Wavelength (nm) Wavelength (nm) Time delay (ps)

d e f
8
D18:Z7
Charge density (×1016 cm−3)

D18:Z7 D18:Z7
6 D18:Z8 D18:Z8 D18:Z8
D18:L8-BO D18:L8-BO D18:L8-BO

Carrier lifetime (µs)

D18:Z7:L8-BO 10 D18:Z7:L8-BO D18:Z7:L8-BO
4 D18:Z8:L8-BO D18:Z8:L8-BO D18:Z8:L8-BO

k (m3 s−1)
−17
10
2

−18
10
0.65 0.70 0.75 0.80 0.85 0.90 0.95 1 2 3 4 5 6 1 2 3 4 5 6
16 −3
VOC (V) Charge density (×10 cm ) Charge density (×1016 cm−3)

g 0 –15
h i
10 10 1.5 80
Ldiff
kex (×107)/krec (×107) (s−1)

1.0
Ldr
0.5 60 900
kL/γ (m3 s−1)

Ldiff (nm)

Ldr (nm)
γpre

10
–1
0.5 40 600
1.0
γpre 10
–16 1.5 kex 20 300
kL
γ 2.0 krec
–2
10 2.5 0 0
BO

BO
BO

BO

BO

BO
O

O
Z8

Z8

Z8
Z7

Z7

Z7
-B

-B

-B
8:

8:

8:
8:

8:

8:
8-

8-
8-

8-

8-

8-
L8

L8

L8
D1

D1

D1
D1

D1

D1
:L
:L

:L

:L

:L
:L
8:

8:

8:
Z8

Z8
Z7

Z7

Z8

Z7
D1

D1

D1
8:

8:

8:
8:

8:

8:
D1

D1

D1
D1

D1

D1
Fig. 2 | Charge dynamics analysis. a,b, Femtosecond TA spectra of Z8 neat reduction factor (γpre), Langevin recombination coefficient (kL) and bimolecular
film (a) and D18:Z8:L8-BO blend film (b). ESA, excited-state absorption; recombination coefficient (γ) of the devices. h, Bimolecular recombination
T, transmittance. c, The trace kinetic of donor GSB signal probed at 590 nm. rate (krec) and charge extraction rate (kex) for the binary and ternary OSCs. i, The
d, Charge density versus VOC. e,f, Carrier lifetime (e) and recombination rate exciton diffusion length (Ldiff ) and charge drift length (Ldr) of the devices.
coefficient (f) as a function of the charge density of the devices. g, The Langevin

ternary cells are mainly subjected to bimolecular recombination. in D18:Z8:L8-BO-based OSCs, it can be concluded that the incorpora-
Therefore, the FF–θ analysis model was introduced to quantitatively tion of Z8 is favourable for reducing charge recombination, both in
analyse the improvement of FF in the optimal TOSCs47. The ratio of geminate and non-geminate recombination, as well as elevating charge
krec transport and extraction.
extraction to recombination, θ, can be defined as θ ∝ , where krec
kex

and kex are the bimolecular recombination and charge extraction rate, Morphology analysis
respectively. krec is proportional to the bimolecular recombination The aggregation behaviour of polymer donor and NFAs in the blend
constant (γ) that obeys the equation of γ = γprekL, in which γpre and kL are film is of vital importance for achieving high photovoltaic performance
the Langevin reduction factor and the classic Langevin recombination in OSCs, and the nanomorphology of the neat and blend films was
constant, respectively (the related data are displayed in Supplemen- systematically investigated. The surface morphology for different
tary Table 9)48. As illustrated in Fig. 2g, D18:Z8:LB-BO cells have small binary and ternary blends was evaluated by atomic force microscopy
γpre and γ values of 0.026 and 2.31 × 10−17 m3 s−1, respectively, leading (AFM) as displayed in Fig. 3a and Supplementary Figs. 38 and 39. The
to minimal carrier loss. Figure 2h displays the kex and krec variations in smaller roughness (Rq) value in the D18:Z8:L8-BO-based blend indicates
ternary and binary OSCs. All the devices give larger kex values of ~107 s−1 a more uniform surface achieved by adding Z8 as the third component,
than that of krec (~104 s−1), guaranteeing efficient charge collection. promoting the formation of superior ohmic contact with reduced series
D18:Z8:L8-BO-based TOSCs exhibit the highest kex value of 1.33 × 107 s−1 resistance. Transmission electron microscopy (TEM) measurement
and the lowest krec value of 1.45 × 104 s−1, allowing faster charge extrac- was performed to study the bulk texture (Supplementary Fig. 40).
tion with suppressed charge recombination, thus leading to a desired JSC A relatively obvious fibre structure is visible in the D18:Z8 and D18:Z7
and FF gain. The FF-versus-θ plot is provided in Supplementary Fig. 37, blend films, but it is absent in the D18:L8-BO blend. For ternary blends,
and a reduced θ is accompanied by the improvement of FF. In addition, the D18:Z8:L8-BO possesses a relatively high-quality fibril texture,
Z8-based TOSCs show superior charge transport properties with the indicating that the addition of Z8 can better enhance the crystallinity.
highest Ldiff (69.9 nm) and Ldr (1070 nm) values (Fig. 2i), contributing The fibril networks can be clearly observed by using the photo-induced
to decreasing recombination loss49. Combining the higher JSC and FF force microscopy measurements (Supplementary Fig. 42). Both donors

Nature Energy
Article https://doi.org/10.1038/s41560-024-01557-z

a b
D18:Z8
2.0 D18:Z8 D18:L8-BO D18:Z8:L8-BO
Rq = 1.13 nm

Intensity (a.u.)
1.6 3.5

qz (Å−1)
3.0
1.0 2.5
0.6 2.0
0
0.5 0 –0.5 –1.0 –1.5 –2.0 0.5 0 –0.5 –1.0 –1.5 –2.0 0.5 0 –0.5 –1.0 –1.5 –2.0
4.0 nm –4.0 nm 8.0° –8.0°
qxy (Å−1) qxy (Å−1) qxy (Å−1)
D18:L8-BO
Rq = 1.17 nm
c 10
9 d 19
(010) (010)
8
(001) 18
10 D18:Z7

Crystal coherence length (Å)

7 17
10
D18:Z8 16
6
10

Intensity (a.u.)
88 (100)
4.0 nm –4.0 nm 7.0° –7.0° 5
10 D18:L8-BO 84
D18:Z8:L8-BO 4
10 80
Rq = 0.96 nm
3
D18:Z7:L8-BO 76
10
80 (001)
2
10 70

1
D18:Z8:L8-BO 60
10
(100) 50
10
0
40
1 2

Z8

BO

BO
Z7
3.3 nm –3.3 nm 10.0° –10.0°

-B
8:
q (Å−1)

8:

8-

8-
L8
D1
D1

:L

:L
8:

Z7

Z8
D1

8:

8:
D1

Fig. 3 | Morphology characterizations. a, AFM (3 × 3 μm) height (left) and
phase (right) images of the optimized D18:Z8, D18:L8-BO and D18:Z8:L8-BO.
b, Two-dimensional GIWAXS patterns for the D18:Z8, D18:L8-BO and D18:Z8:
l8-BO blends. c, The in-plane (IP) and OOP line cut profiles of the 2D GIWAXS
D1
based on the blend films (the solid lines represent the OOP direction and the
dashed lines represent the IP direction). d, Crystal coherence length (CCL) of
(010), (100) and (001) diffraction peaks in the binary and ternary blend films.

and acceptors can form a fibril structure, and they entangled together
to construct an interpenetrating bicontinuous nanostructure, facili-
tating suitable phase separation and efficient charge transport. This
favourable nanotexture can reduce both geminate and non-geminate
recombination and, thus, yielding higher JSC and FF.
The crystalline behaviours of different films were assessed by
the grazing-incidence wide-angle X-ray scattering (GIWAXS). Supple-
mentary Figs. 43–45 and Fig. 3b,d display 2D GIWAXS patterns for neat
and blend films, and the detailed data are summarized in Supplemen-
tary Tables 10 and 11. All the acceptors exhibit a preferential face-on
orientation, in which strong π–π stacking diffraction (010) peaks in
the out-of-plane (OOP) direction are clearly seen, benefitting to the
formation of vertical charge transport50. L8-BO neat film exhibits a
broad π–π stacking peak with a shoulder peak located at 1.51 Å−1 (OOP),
suggesting the presence of amorphous content7, while Z7 and Z8 neat
films exhibit stronger diffraction intensity and narrower π–π stacking
peaks, implying less disordered states and parasitic crystallinity. Thus,
the incorporation of Z-type acceptors into D18:L8-BO contributes to
enhancing crystallinity as evidenced by the stronger (010) and (100) dif-
fraction intensity and enlarged CCLs (Fig. 3d). Besides, the (001) peak at
~0.5 Å−1 can be assigned to the characteristic peak of D18 (Supplemen-
tary Fig. 44). The larger CCL of the (001) peak in Z-type acceptor-based
binary and ternary blends indicates the favourable crystallinity of
D18 (Fig. 3d). The enhanced crystallinity in D18:Z8:L8-BO contributes
to improving charge transport property, leading to higher and more
balanced charge carrier mobilities (Supplementary Figs. 34–36 and
Supplementary Table 12).
The grazing-incidence small-angle X-ray scattering (GISAXS) meas-
urements were performed to quantitatively analyse the phase separa-
tion of blend films, and the relevant data are shown in Supplementary
Figs. 46–47 and Supplementary Table 13. Compared with D18:L8-BO
film, Z-type blends exhibit a larger values of acceptor domain size (2Rg),
which can be ascribed to the higher crystallinity of Z-type acceptor.
The amorphous intermixing phase (ζ) sizes for D18:Z7, D18:Z8 and
D18:L8-BO blends are 30.0, 25.5 and 37.1 nm, respectively. Adding
Z8 into D18:L8-BO blend produces a smaller intermixing domain
size (31.8 nm), while an increased ζ value of 45.1 nm is observed in
D18:Z7:L8-BO-based ternary blends, which may lead to the more charge
recombination. Moreover, the incorporation of Z8 into D18:L8-BO
blends can improve the acceptor domain size (26.7 nm), thus enhanc-
ing electron transport property. The time-of-flight secondary ion
mass spectrometry (TOF-SIMS) was applied to investigate the vertical
component distribution in blend films. As displayed in Supplementary
Fig. 48, Z8-based ternary film can maintain a favourable vertical dis-
tribution, while the addition of Z7 into D18:L8-BO blends is unable to
retain the desired component variation along the vertical direction.
Therefore, we deduced that the incorporation of Z8 into D18:L8-BO
blends can reduce the amorphous intermixing phase, enlarge the
acceptor domain and form a desired vertical distribution, leading to
decreased charge recombination, improved charge transport and
collection, thus realizing higher JSC and FF.
To elucidate the correlation among molecular structure, blend
morphology and device performance, a deep insight into the inter-
molecular interactions was further studied by using MD simulations.
Supplementary Fig. 49 shows the detailed simulation processes and
the related data are summarized in Supplementary Tables 14–16. As
illustrated in Supplementary Fig. 53, the electron-withdrawing units
(dithieno[3′,2′:3,4;2′,3′:5,6]benzo[1,2-c][1,2,5]thiadiazole) of polymer
donor D18 trend to interact with the 2-(5,6-difluoro-3-oxo-2,3-dihy-
dro-1H-inden-1-ylidene)malononitrile (IC-2F)/2-(6,7-difluoro-3-oxo-
2,3-dihydro-1H-cyclopenta[b]naphthalen-1-ylidene)malononitrile
(ITN-2F) terminals of NFAs. The respective acceptor–acceptor (A–A)
dimers and donor–acceptor (D–A) pairs were extracted from the corre-
sponding MD simulation, and the relevant exciton binding energies (Eb)
of excited and CT states were calculated, as displayed in Supplementary
Figs. 54–55 and Supplementary Table 14. For a given type of A–A dimer,

Nature Energy
 Article https://doi.org/10.1038/s41560-024-01557-z

a c e ηSQ
100 50 1.0
2.0 ∆E1 Reflectance 0.9
∆E2 Parasitic absorption
55
Si (ref. )
0.8

v*f (VOC × FF/VOC,SQ × FFSQ)

Current density (mA cm–2)
80 ∆E3 IQE loss 40 Perovskite (ref. )
54
0.7
1.6
qVOC
Light percentage (%)

ITO JSC 0.8

JSC,SQ D18:Z8:L8-BO (this work) 0.6
PEDOT 58
0.5

Energy (eV)
PBQx-TF:eC9-2Cl:F-BTA-3 (ref. )
60 qVOC,SQ 30
1.2
24
Simulated EQE PM6:Y11 (ref. ) 11
D18:PM6:L8-BO (ref. )
0.4
PDINN 0.6 57
PM6:IT4F (ref. )
56
0.3
40 Ag 20 PTB7-Th:ATT-9 (ref. )
0.8 0.2
Reflectance 24 0.1
23 PM6:PCBM (ref. )
Transmittance 0.4 PTB7:PCBM (ref. )
<0.1
20 0.4 10
Test EQE
23
P3HT:PCBM (ref. )
0 0 0 0.2
400 600 800 1,000 Eg JSQ 0.4 0.6 0.8 1.0
Wavelength (nm) j/(JSC/JSC,SQ)
b d Eg (eV) f S-Q limit
PCE (%) at 1.43 eV
2
|E| 2.1 1.6 1.2 1.0
33.7 ∆E3
3.10 33.1%
9.5 6.4 ∆E2
Ag FFloss
300 Jloss (par. abs. & R)
2.71 0.5
PDINN 18.6 15.5 27.6
12.5 12.5 IQEloss 27.3%
25.9%
0.4 23.7%
2.03
Position (nm)

21.6

PCE (%)
200 21.0%
∆Vnr (V)

D18:Z8:L8-BO 20.2%
0.3
15.5 24.6
PEDOT:PSS 1.35 15.5
0.2 27.3
100 ITO
0.68 9.46
0.1 21.6 30.7 27.6
Photon loss Carrier loss
0 0 3.40
400 600 800 1,000 600 800 1,000 1,200
Wavelength (nm) Wavelength (nm)
Fig. 4 | Theoretical evaluation. a,b, Light percentage distribution (a) and optical represents the ideal PCE at the bandgap and ∆Vnr of D18:Z8:L8-BO. e, Fraction
simulation on electric field intensity distribution (b) across the D18:Z8:L8-BO of S-Q limit (at their corresponding bandgap) for photon loss (j = JSC/JSC,SQ) and
device with a conventional structure of ITO/PEDOT:PSS/D18:Z8:L8-BO/PDINN/ carrier loss (v*f = VOC/VOC,SQ × FF/FFSQ) achieved by typical and state-of-the-
Ag. Simulated EQE assumes 100% IQE. c, The detailed voltage loss and current art OSCs11,23,24,54,55,68–70. f, Detailed various factors contributing to PCE loss in
loss for the D18:Z8:L8-BO-based device, VOC,SQ and JSC,SQ, are the maximum D18:Z8:L8-BO-based OSC on approaching the S-Q limit efficiency. The arrow
voltage and current density according to the S-Q limit. d, The detailed balanced indicates the direction for performance improvement. Jloss(par.abs.&R) represents the
limit of PCE under different bandgap wavelengths and ∆Vnr. The red star photon loss caused by parasitic absorption and device reflection.

the Z-type acceptors show the relatively smaller Eb values compared
with L8-BO, generating a higher probability of direct photogeneration
in acceptor domain35,51. The MD simulation of ternary blends is depicted
in Supplementary Fig. 56, from which more comprehensive insights
can be gained and the detailed insights are provided in Supplementary
Figs. 57–59 and Supplementary Tables 15 and 16. Supplementary Fig. 57
reveals that the tethered phenyl groups on the Z-type acceptor can
interact with polymer donor, thereby enhancing the D–A intermolecu-
lar interaction. The average intermolecular binding energies (ΔE) are
−3.903, −4.076 and −3.499 kJ mol−1 for D18:Z7, D18:Z8 and D18:L8-BO,
respectively, indicating a stronger D–A interaction in D18:Z8 pairs.
Supplementary Fig. 58 visualizes the interacting atoms between the
host donor and the host/guest acceptors in two ternary blends. The
D18:Z8:L8-BO-based ternary blend exhibits a higher number of D–A
(guest) pairs in the region with a large number of interacting atoms
compared to that of the D18:Z7:L8-BO-based blend, implying a higher
probability to forming D18:Z8 pairs with stronger intermolecular
interaction, which contributes to facilitating charge transfer process.
Compared with the D18:L8-BO-based binary blend, the incorporation
of Z-type acceptors in ternary leads to an enhanced ratio of D–A pairs
(Supplementary Fig. 59), implying more D/A interfaces and thereby
promoting exciton dissociation.
Considering the superior photoluminescence properties and the
more delocalized exciton in asymmetric NFA Z8, the formation of D18/
Z8 interfaces is expected to minimize non-radiative energy loss and
geminate charge recombination. Based on the experimental results
and MD simulations, constructing a highly efficient TOSC necessitates
a guest acceptor with high PLQY and delocalized exciton as well as high
FFs exist in the relevant binary OSCs. A strong D–A interaction between
host donor and guest acceptor and a good miscibility between the guest
and host acceptors are necessary to form favourable nanomorphology,
thus enhancing the optoelectronic process. All these factors contribute
to achieving an overall improvement in OSC performance.
Theoretical evaluation
A theoretic calculation was performed to evaluate the photovoltaic
performance reliability of Z8-based TOSCs. For photovoltaic materi-
als with a given bandgap, the realistic efficiency is inferior to the S-Q
limit, which is mainly caused by the non-ideal charge carrier collec-
tion and recombination and then leads to current and voltage losses.
Polman et al. proposed two characteristic parameters52, j (j = JSC/JSC,SQ)
and v*f (v*f = VOC/VOC,SQ × FF/FFSQ), to evaluate photovoltaic systems,
which is also performed in this study. Optical simulation based on the
transfer matrix method was carried out to evaluate the current loss
in D18:Z8:L8-BO ternary OSC53. Figure 4a,b illustrates the light per-
centage distribution in different layers and electric field distribution
across positions in the Z8-based ternary OSCs. Light absorbed by the
active layer accounts for a current density of 28.2 mA cm−2, which cor-
responds to the theoretical EQE if all absorbed photons are transferred
into photocurrent (IQE 100%). The experimental integrated JSC value
from the EQE test is 26.3 mA cm−2, indicating that the average IQE of our
OSC is 93%, which is consistent with the IQE test result (Supplementary
Fig. 63). According to the S-Q model (Fig. 4c), for Z8-based ternary OSC
with a bandgap of 1.43 eV (867 nm), the maximum theoretical current
density (JSC,SQ) can reach 31.8 mA cm−2. A high j value of over 85% sug-
gests the highest light management among the state-of-the-art OSCs.

Nature Energy
Article
The 15% loss of incident photons comes from three parts: (1) IQE loss,
(2) device reflection loss and (3) parasitic absorption of interlayers and
electrodes, as displayed in Fig. 4a,c.
Since we attributed the efficiency loss from both the voltage and
current sides, based on our best-performance TOSC, attention needs to
be paid to several possible approaches to realize a higher PCE towards
the theoretical S-Q limit for the future development of OSCs. The VOC
of our ternary cell is 0.92 V, which accounts for 79% of the theoretical
limit of VOC,SQ (1.16 V). Considering that the radiative loss (∆E1 and ∆E2) is
inevitable for a solar cell at a certain bandgap due to mechanisms such
as dissipative process, carrier cooling and entropy-related étendue
expansion, non-radiative recombination loss (∆E3 or q∆Vnr) becomes
the most influential and the most promising factor to overcome in
voltage loss, since multiple ways to suppress ∆Vnr are focused by many
works recently. Here, we simulated the theoretical FF and PCE limit on
the basis of a detailed-balance model applying different ∆Vnr values in
Fig. 4d and Supplementary Fig. 64. Our ternary OSC exhibits ∆Vnr of
0.19 V at the bandgap wavelength of 867 nm, in which the theoretical
limits of FF and PCE are 88.1% and 27.3%, respectively, leaving a remark-
able improvement room for further development.
Figure 4e summarizes the light management (j) and carrier man-
agement (v*f) ability towards the detailed-balance S-Q limit of the
D18:Z8:L8-BO ternary solar cell along with some typical OSCs, one
of the state-of-art perovskite54 and single-crystal silicon (c-Si)55 pho-
tovoltaics. The classic fullerene-based OSCs suffer from a high rate
of non-radiative recombination, thus undergoing very large voltage
loss and can hardly exceed 40% of its PCE limit. As the development of
non-fullerene acceptors and the iteration of new polymer donors, the
PCE of OSCs has grown rapidly and is gradually closing the gap to the
theoretical limit, especially in the direction of carrier management. In
this work, a high level of carrier management was maintained, and the
remarkable increase in light management was simultaneously realized
in Z8-based ternary OSCs. The D18:Z8:L8-BO ternary solar cell performs
71.6% in carrier management (79.3% for v and 90.3% for f), and 85.5% in
light management, which is one of the top values among OSCs. Such an
achievement points out that the process towards S-Q limited efficiency
needs the simoultaneous management of light and carrier. Compared
with the c-Si counterpart with over 97% of j values, OSCs still lag behind
in device engineering and optical modulation to reduce reflection and
parasitic absorption to a minimum. Compared with perovskite coun-
terparts with 89% of v*f, OSCs perform at almost the same level in FF
but have an intrinsic disadvantage in ∆Vnr due to the narrower bandgap.
However, the difference is narrowing as the active layer materials are
developing rapidly. More and more high-performance OSCs with low
∆Vnr (even less than 0.18 V) have been reported. Future development
of OSCs needs to focus on both voltage and current, including further
material revolution and device engineering, to achieve higher PCE. On
the one hand, device engineering, including modification on optical
modulation layer and interlayers/electrodes, contributes to minimiz-
ing reflection and parasitic absorption. On the other hand, developing
new materials with excellent mobility and emission properties offers
the benefit of wiping out carrier recombination loss. Thus, to advance
the performance of OSCs, device engineering and material innovation
should be efficiently united to eliminate light and carrier losses.
Figure 4f gives the potential improvement of PCE in different
factors based on existing D18:Z8:L8-BO ternary solar cells towards
the S-Q limit. By overcoming the IQEloss through possible material
optimization, the PCE would increase to 21.0%. By carefully designing
the device structure and applying optical modulations to reduce reflec-
tion and parasitic absorption, which is done well in c-Si photovoltaics,
the PCE could reach 23.7% to its maximum JSC,SQ value. FF is affected by
both materials and devices, and reaching an ideal FFSQ would increase
the PCE to 25.9%. Considering the voltage loss part, the classical S-Q
theory assumes full absorption above the bandgap and zero below
the bandgap, so ΔE2 does not exist in the original S-Q model. However,
Nature Energy
https://doi.org/10.1038/s41560-024-01557-z
absorbance expansion to the subgap region often occurs due to the
existence of CT states and energetic disorder in realistic non-ideal
systems, and for high-performance OSCs, the CT states are very close
to the bandgap and subgap absorption is also low, resulting in a small
ΔE2 of around 0.05 eV. If ΔE2 is totally overcome in future material sys-
tems, a PCE of 27.3% can be achieved. ∆Vnr counts for a large part of PCE
loss; however, OSCs are developing rapidly with reduced ∆Vnr. A PCE
of 33.1% can be expected if no non-radiative recombination loss exists
in a solar cell, which is close to the ultimate S-Q limit. The final gap is
due to the difference in bandgap where PCESQ has two local maxima of
33.7% (1.34 eV, 930 nm) and 33.5% (1.15 eV, 1,080 nm).
Conclusions
In this work, alloyed acceptors consisting of two NFAs Z8 and L8-BO in
ternary OSCs contributed to the reduced photon and carrier losses, and
a high PCE of 20.2% (certified PCE of 19.8%) is achieved in single-junction
OSCs. The tethered phenyl groups in the asymmetric Z-type accep-
tor can not only endow NFA with high PLQY and delocalized exciton
but also contribute to facilitate charge transfer and minimize charge
recombination in ternary blend to simultaneously achieve ideal VOC, JSC
and FF. The overall distribution of photon and carrier losses for OSCs is
presented through detailed theoretical calculation, and the potential
for improvement on each device parameter is also analysed, which
indicates that the material and device revolution is essential to fur-
ther minimize photon and carrier losses. These findings highlight that
both photon loss and carrier loss should gain equal attention from the
community, which could push the photovoltaic performance of OSCs
towards the efficiency limit.
Methods
Materials
All reactions involving air or moisture-sensitive compounds were
carried out in a dry reaction vessel under a positive pressure of nitro-
gen. Unless stated otherwise, starting materials were obtained from
Adamas, Energy Chemical or J&K and were used without further puri-
fication. Anhydrous tetrahydrofuran (THF) and toluene were distilled
over Na/benzophenone before use.
Compound 2 was synthesized using iridium-catalysed
4-phenylbutan-1-ol 1 as the raw material, followed by bromina-
tion with an excess amount of PBr3 to afford the key intermediate
((3-bromomethyl)heptane-1,7-diyl)dibenzene 3. Compound 5 was
synthesized by N-alkylation of compound 4, which was prepared by
double intramolecular Cadogan reductive cyclization of the dini-
trobenzothiadiazole precursor based on reported procedures, with
((3-bromomethyl)heptane-1,7-diyl)dibenzene 3 using potassium
carbonate as the base in 69% yield. The dialdehyde compound 6 was
efficiently obtained from compound 5 using the Vilsmeier–Haack reac-
tion, followed by Knoevenagel condensation with 2-(6,7-difluoro-3-oxo-
2,3-dihydro-1H-cyclopenta[b]naphthalen-1-ylidene)malononitrile
(ITN-2F), and the target molecule Z7 was obtained in 80% yield as a black
solid. The asymmetric NFA Z8 was efficiently synthesized by a one-pot
reaction in 35% yield, where the 2-(5,6-difluoro-3-oxo-2,3-dihydro-1H-i
nden-1-ylidene)malononitrile (IC-2F) and ITN-2F were sequentially
reacted with dialdehyde compound 6.
Compound 2: an oven-dried Schlenk flask was loaded with
4-phenylbutan-1-ol 1 (900 mg, 6 mmol), [Cp*IrCl2]2 (96 mg, 0.12 mmol)
and t-BuOK (269 mg, 2.4 mmol) in anhydrous 1,4-dioxane (10 ml) under
N2 atmosphere. The reaction solution was stirred and refluxed for 24 h
under dark. The resulting mixture was extracted with dichloromethane
(DCM) and purified by column chromatography to give 250 mg of com-
pound 2 in 30% yield. 1H NMR (400 MHz, CDCl3): δ 7.34–7.23 (m, 4H),
7.18 (m, 6H), 3.58 (d, J = 4.5, 2H), 2.63 (m, 4H), 1.61 (m, 5H), 1.39 (m, 4H).
Compound 3: compound 2 (282 mg, 1 mmol) and PBr3 (1080 mg,
4 mmol) were dissolved in dichloromethane (10 ml), and the mixture
was stirred at room temperature for 6 h. The resulting solution was
Article
extracted with DCM and purified on a silica-gel column chromatogra-
phy to give 310 mg compound 3 in 90% yield. 1H NMR (400 MHz, CDCl3):
δ 7.32–7.25 (m, 4H), 7.23–7.15 (m, 6H), 3.54–3.42 (m, 2H), 2.70–2.51
(m, 4H), 1.79–1.56 (m, 5H), 1.50–1.24 (m, 4H). 13C NMR (100 MHz, CDCl3):
δ 142.49, 142.02, 128.39, 128.38, 128.31, 128.27, 125.85, 125.67, 77.32,
77.00, 76.68, 39.11, 38.85, 35.78, 34.38, 32.84, 32.30, 31.47, 26.11.
Compound 5: the cyclized intermediate 4 (373 mg, 0.5 mmol) and
compound 3 (688 mg, 2 mmol) were mixed with KI (42 mg, 0.25 mmol)
and K2CO3 (414 mg, 3 mmol) and then dissolved in DMF (10 ml) under
N2 atmosphere. The reaction mixture was stirred at 80 °C overnight.
The resulting solution was extracted with DCM and washed with water
followed by brine. The crude mixture was purified by silica-gel column
chromatography, and 440 mg compound 3 was obtained in 69% yield.
1
H NMR (400 MHz, CDCl3): δ 7.19 (t, J = 7.3, 4H), 7.12 (t, J = 7.1, 2H), 7.01
(s, 2H), 6.98–6.90 (m, 10H), 6.65–6.55 (m, 4H), 4.68–4.44 (m, 4H), 2.83
(t, J = 7.6, 4H), 2.38–2.18 (m, 8H), 2.16–2.04 (m, 2H), 1.91–1.84 (m, 4H),
1.51–0.99 (m, 46H), 0.98–0.78 (m, 12H). 13C NMR (100 MHz, CDCl3): δ
147.58, 142.29, 142.18, 141.29, 137.18, 136.87, 131.46, 128.23, 128.16, 128.05,
127.84, 125.62, 125.56, 123.53, 122.88, 119.38, 111.75, 54.94, 38.16, 35.47,
32.23, 31.93, 31.74, 31.29, 30.10, 29.70, 29.64, 29.56, 29.47, 29.36, 28.87,
25.00, 22.69, 14.11. MS (MALDI-TOF) calculated for C80H98N4S5 [M]+
1,274.64, found 1,274.64.
Compound 6: to a solution of compound 5 (255 mg, 0.2 mmol)
in dry THF was dropped n-BuLi/n-hexane (0.38 ml, 0.6 mmol, 1.6 M)
at −78 °C, followed by stirring for 1 h. Then, N,N-dimethylformamide
was added and reacted for another 1 h. After warming to room tem-
perature, the reaction mixture was extracted with DCM and purified
by silica-gel column chromatography to give 210 mg of compound 6
in 79% yield. 1H NMR (400 MHz, CDCl3): δ 10.15 (s, 2H), 7.20 (t, J = 7.4,
4H), 7.13 (t, J = 7.3, 2H), 6.96 (d, J = 7.3, 4H), 6.90 (m, 6H), 6.56 (m, 4H),
4.58–4.43 (m, 4H), 3.29–3.09 (m, 4H), 2.43–2.13 (m, 8H), 2.10–1.84 (m,
7H), 1.53–0.96 (m, 36H), 0.96–0.77 (m, 12H). 13C NMR (100 MHz, CDCl3):
δ 181.65, 147.41, 146.71, 143.22, 141.93, 140.78, 137.08, 136.86, 136.83,
132.59, 129.44, 128.23, 128.18, 128.02, 127.75, 127.35, 127.33, 125.69,
112.55, 112.53, 55.18, 37.97, 35.42, 31.89, 31.52, 31.27, 30.34, 30.04, 29.68,
29.63, 29.59, 29.52, 29.36, 29.32, 28.16, 27.90, 25.00, 22.67, 14.10. MS
(MALDI-TOF) calculated for C82H98N4O2S5 [M]+ 1,330.63, found 1,330.63.
Z7
Compound 6 (247.8 mg, 0.186 mmol, 1 equiv.), 2-(5,6-difluoro-3-oxo-
2,3-dihydro-1H-inden-1-ylidene)malononitrile (208 mg, 0.4 mol,
4 equiv.) and pyridine (0.1 ml) were added in chloroform (5 ml).
The reaction was stirred at 65 °C overnight. After cooling to room
temperature, the solvent was removed and the residue was purified
by column chromatography to yield a dark-blue solid (262.4 mg, 76%).
1
H NMR (400 MHz, CDCl3) δ 9.19 (s, 2H), 9.05 (s, 2H), 8.25 (d, J = 6.9 Hz,
2H), 7.72 (m, 9.8 Hz, 4H), 7.20 (m, 4H), 7.13 (d, J = 2.6 Hz, 2H), 7.07
(d, J = 7.4 Hz, 2H), 7.03 (d, J = 7.4 Hz, 2H), 6.96–6.85 (m, 6H), 6.63
(dd, 4H), 4.86–4.57 (m, 4H), 3.18 (m, 4H), 2.63–2.25 (m, 8H), 2.13
(s, 2H), 1.86 (m, 4H), 1.32 (m, 48H), 0.88 (m, 7.3 Hz, 6H). 13C NMR
(101 MHz, CDCl3) δ 187.78, 160.21, 153.89, 147.46, 145.35, 142.23, 142.18,
140.92, 140.86, 137.86, 136.15, 135.69, 135.36, 134.08, 133.78, 133.44,
133.33, 132.51, 132.43, 130.94, 128.32, 128.30, 128.25, 128.09, 128.03,
127.90, 127.82, 125.82, 125.78, 125.50, 123.32, 122.28, 116.54, 116.37, 116.05,
115.88, 115.73, 115.48, 113.81, 77.35, 77.04, 76.72, 67.26, 55.74, 38.69, 38.24,
35.72, 35.68, 32.09, 31.94, 31.66, 31.55, 31.50, 31.25, 30.51, 30.32, 29.92,
29.84, 29.71, 29.66, 29.58, 29.50, 29.38, 27.90, 25.75, 25.51, 22.72, 22.67,
19.21, 14.16, 11.45. MS (MALDI-TOF) calculated for C114H106F4N8O2S5 [M]+
1,855.70, found 1,855.70.
Z8
Compound 6 (133.2 mg, 0.1 mmol, 1 equiv.), 2-(5,6-difluoro-3-oxo-
2,3-dihydro-1H-inden-1-ylidene)malononitrile (23.1 mg, 0.1 mol,
1 equiv.), 2-(6,7-difluoro-3-oxo-2,3-dihydro-1H-cyclopenta[b]
naphthalen-1-ylidene)malononitrile (28.1 mg, 0.1 mol, 1 equiv.) and
Nature Energy
https://doi.org/10.1038/s41560-024-01557-z
pyridine (0.1 ml) was added in chloroform (5 ml). The reaction was
stirred at 65 °C overnight. After cooling to room temperature, the
solvent was removed. The residue was purified by column chroma-
tography to yield a dark-blue solid (63.2 mg, 35%). 1H NMR (400 MHz,
CDCl3) δ 9.23 (s, 1H), 9.14 (s, 1H), 9.09 (s, 1H), 8.56 (dd, J = 9.6, 6.4 Hz,
1H), 8.25 (d, J = 2.1 Hz, 1H), 7.79 (dd, J = 9.2, 7.7 Hz, 1H), 7.75–7.60 (m,
2H), 7.23–7.15 (m, 4H), 7.15–7.08 (m, 2H), 7.07–6.96 (m, 4H), 6.96–6.83
(m, 6H), 6.69–6.61 (m, 2H), 6.61–6.54 (m, 2H), 4.79–4.53 (m, 4H), 3.22
(d, J = 8.2 Hz, 4H), 2.55–2.20 (m, 8H), 2.11 (s, 2H), 1.89 (t, J = 7.8 Hz, 4H),
1.35–1.19 (m, 48H), 0.87 (m, 6H). 13C NMR (101 MHz, CDCl3) δ 187.79,
186.12, 160.26, 158.85, 153.92, 153.80, 147.45, 145.33, 145.18, 142.17, 142.14,
142.10, 142.07, 140.87, 140.82, 137.81, 136.12, 135.75, 135.37, 135.21, 134.06,
133.72, 133.44, 133.26, 132.51, 130.94, 130.76, 128.30, 128.28, 128.26,
128.23, 128.19, 128.09, 128.07, 128.04, 128.02, 127.87, 127.80, 125.77,
122.30, 120.08, 115.72, 115.47, 114.98, 114.57, 113.76, 68.81, 67.26, 55.71,
38.14, 35.68, 35.65, 35.57, 35.54, 32.01, 31.94, 31.61, 31.48, 31.45, 31.33,
31.27, 30.42, 30.25, 29.92, 29.85, 29.70, 29.66, 29.57, 29.49, 29.38, 25.45,
25.26, 22.72, 14.16. MS (MALDI-TOF) calculated for C110H104F4N8O2S5 [M]+
1,805.68, found 1,805.68.
Structural and optoelectronic characterization
1
H and 13C spectra were measured with Bruker Avance 400 spectrom-
eters. Chemical shifts for hydrogens are reported in parts per million
(ppm, scale) downfield from tetramethylsilane and are referenced to
the residual protons in the NMR solvent (CDCl3: 7.26).13C NMR spectra
were recorded at 100 MHz. Chemical shifts for carbons are reported in
ppm (scale) downfield from tetramethylsilane and are referenced to the
carbon resonance of the solvent (CDCl3: 77.2). The data are presented as
follows: chemical shift, multiplicity (s, singlet; d, doublet; t, triplet; m,
multiplet and/or multiple resonances; br, broad), coupling constant in
hertz (Hz) and integration. MALDI measurements were performed with
a MALDI-FT 9.4 T, Bruker solariX, or MALDI-TOF MS Bruker Autoflex
III. UV–visible was recorded with Jasco V-570 spectrometers. Cyclic
voltammetry was performed with a CHI620D potentiostat. All meas-
urements were carried out in a one-compartment cell under a nitrogen
atmosphere, equipped with a glassy-carbon electrode, a platinum
counter-electrode and an Ag/Ag+ reference electrode with a scan rate of
100 mV s−1. The supporting electrolyte was a 0.1 mol l−1 dichlorometh-
ane solution of tetrabutylammonium perchlorate. All potentials were
corrected against Fc/Fc+. Single crystals were grown through slow
diffusion of ethanol to its chloroform solution. Low-energy inverse
photoemission spectroscopy (LEIPS) measurement was performed
on a customized ULVAC-PHI LEIPS instrument with Bremsstrahlung
isochromatic mode. UV photoelectron spectroscopy measurement
was performed by AXIS ULTRA DLD instrument of Kratos company.
The UV light source used is non-monochromatic He I, and the energy of
He I light source is 21.22 eV. The basic vacuum of the analysis chamber
is 3.0 × 10−8 Torr, and the bias voltage applied during the test is −9 V.
Device fabrication and characterization
The device was equipped with a conventional construction of glass/ITO/
PEDOT:PSS/D:A/PDINN/Ag, where ITO is indium tin oxide, PEDOT:PSS
is poly(3,4-ethylenedioxythiophene):polystyrene sulfonate and PDINN
is N,N′-bis{3-[3-(dimethylamino)propylamino]propyl}perylene-3,4,9,
10-tetracarboxylic diimide. The weight ratio of D18:acceptor is 1:1.2,
and chloroform was applied as a solvent. The D18:acceptor blend
solution with a total concentration of 10 mg ml−1 was heated at 100 °C
till completely dissolved. Then the blend solution was spin-coated on
PEDOT:PSS, and then the blend was heated at 120 °C for 10 min. The
PDINN in ethyl alcohol (1 mg ml−1) was spin-coated on the active layer
by 3,000 rpm. The Ag (100 nm) was deposed onto the substrates by
vacuum evaporation. Shadow masks were used to define the OSC
active area (0.04 cm2) of the devices. The current density–voltage (J–V)
characteristics of unencapsulated photovoltaic devices were measured
under N2 using a Keysight B2912A Precision Source/Measure unit.
Article
The illumination of AM 1.5G (100 mW cm−2) was achieved by a XES-
70S1 solar simulator (SAN-EI Electric Co., Ltd., AAA grade,
70 mm × 70 mm photo-beam size). The EQE was determined using
certified incident photon to current conversion efficiency equipment
from Enlitech.
Morphological characterization
AFM images of the thin films were obtained on a NanoscopeIIIa AFM
(Digital Instruments) operating platform in tapping mode. TEM obser-
vation was performed on JEOL 2200 FS at 160 kV accelerating voltage.
The samples for electron microscopy were prepared by dissolving
the PEDOT:PSS layer using water and transferring the floating active
layer to the TEM grids. The GIWAXS characterization of the thin films
was performed at the Advanced Light Source on the beamline of 7.3.3.
Samples were prepared under device conditions on the Si/PEDOT:PSS
substrates. GISAXS measurements were carried out with a Xeuss 2.0
SAXS laboratory beamline using a Cu X-ray source (8.05 keV, 1.54 Å)
and a Pilatus3R 300K detector. The samples were prepared on sili-
con wafer, and the incidence angle was 0.2°. The data were fitted by
the Debye–Anderson–Brumberger model and a fractal-like network
model. TOF-SIMS measurement was conducted using a TOF-SIMS
5–100 instrument (ION-TOF GmbH), where a 1 keV Cs+ ion beam was
used for erosion and a 30 keV Bi13+ pulsed primary ion beam was used
for the analysis. The samples were prepared by spin-coating the active
layers on the PEDOT:PSS layer. The corresponding blend samples were
thermally annealed for 10 min.
FTPS-EQE and EQEEL measurements
Fourier transform photocurrent spectroscopy (FTPS)-EQE was meas-
ured using an integrated system (PECT-600, Enlitech), where the
photocurrent was amplified and modulated by a lock-in instrument.
Electroluminescent spectroscopy (EQEEL) measurements were per-
formed by applying external voltage/current sources through the
devices (ELCT-3010, Enlitech). EQEEL measurements were performed
for all devices according to the optimal device preparation conditions.
TPC, TPV and CE measurements
The TPC, TPV and CE measurements were performed on an all-in-one
characterization platform Paios developed by Fluxim AG.
PLQY measurements
PLQY measurements were conducted utilizing an FLS980 Fluorimeter,
equipped with a LabSpheres integrating sphere and a near-infrared
photomultiplier tube (NIR-PMT, model R5509-72). Subsequent data
analysis and calculation were carried out using a specialized PLQY
software package.
Quantum chemical calculations
The geometry optimization of the molecules is performed by DFT cal-
culations at the b3lyp/6-311g(d,p) level. All the alkyl groups are replaced
by methyl groups to reduce the computational costs. The calculations
were conducted using the polarizable continuum model, taking the
dielectric constant for chloroform (ε = 4.7113) as reference. All the
DFT calculations were carried out using the Gaussian 16 package56.
VMD (version 1.9.3) was used to plot the images of molecular orbitals
(isosurface 0.02) and dipole moments57. Multiwfn software is used for
the calculation of the orbital delocalization index and the analysis of
dipole moment of molecular fragments58.
MD simulations
All atomistic MD simulations were performed using the Gromacs–2019.3
software package59. For all the studied systems, the atom types and the
intra- and intermolecular interaction parameters were taken from
the general AMBER force field (GAFF)60. The atomic partial charges
were obtained by DFT calculations at the B3LYP/cc-PVTZ level61,62, and
Nature Energy
https://doi.org/10.1038/s41560-024-01557-z
fitted with the restrained electrostatic potential method. Moreover,
the missing torsion potentials were reparametrized according to DFT
calculations. The simulations were carried out with three-dimensional
periodic boundary conditions; the leap-frog integrator with a time
step of 1.0 fs was selected. A spherical cut-off of 1.2 nm for the sum-
mation of van der Waals interactions and short-range Coulomb inter-
actions and the particle-mesh Ewald method for solving long-range
Coulomb interactions was used. These thin films were built and imi-
tated using the following procedure: (1) randomly placing molecules
in a large box to generate an initial geometry (D18:L8-BO, D18:Z7 and
D18:Z8 thin films were obtained by adding 100 D18 molecules and 315
L8-BO, 251 Z7 and 258 Z8 molecules, respectively; D18:Z7:L8-BO and
D18:Z8:L8-BO thin films were obtained by adding 100 D18 molecules,
237 L8-BO molecules and extra 63 Z7 or 65 Z8 molecules according
to the experimental D:A weight ratio); (2) 5 ns of simulation at 600 K
and 100 bar to quickly bring molecules close together; (3) 10 ns of
simulation at 600 K and 1 bar, then cooling down to 300 K in 3 ns; (4)
50 ns of equilibration at 300 K and 1 bar. The velocity rescaling thermal
thermostat63 and the Berendsen barostat64 under the NPT ensemble
were applied to control the temperature and pressure, respectively.
However, for the final 10 ns of equilibration, the Nosé–Hoover thermo-
stat65,66 and Parrinello–Rahman barostat67 were used to obtain better
equilibrium conformations.
Reporting summary
Further information on research design is available in the Nature
Portfolio Reporting Summary linked to this article.
Data availability
The authors declare all data supporting the findings of this study are
available within the manuscript and Supplementary Information.
Source data are provided with this paper.
References
1. Yao, H. et al. Molecular design of benzodithiophene-based
organic photovoltaic materials. Chem. Rev. 116, 7397–7457 (2016).
2. Guo, X. et al. Polymer solar cells with enhanced fill factors. Nat.
Photon. 7, 825–833 (2013).
3. Liang, Y. et al. For the bright future—bulk heterojunction polymer
solar cells with power conversion efficiency of 7.4%. Adv. Mater.
22, E135–E138 (2010).
4. Vohra, V. et al. Efficient inverted polymer solar cells employing
favourable molecular orientation. Nat. Photon. 9, 403–408 (2015).
5. Yuan, J. et al. Enabling low voltage losses and high photocurrent in
fullerene-free organic photovoltaics. Nat. Commun. 10, 570 (2019).
6. Yuan, J. et al. Single-junction organic solar cell with over 15%
efficiency using fused-ring acceptor with electron-deficient core.
Joule 3, 1140–1151 (2019).
7. Li, C. et al. Non-fullerene acceptors with branched side chains
and improved molecular packing to exceed 18% efficiency in
organic solar cells. Nat. Energy 6, 605–613 (2021).
8. Liang, Y. et al. Organic solar cells using oligomer acceptors for
improved stability and efficiency. Nat. Energy 7, 1180–1190 (2022).
9. Baran, D. et al. Reducing the efficiency–stability–cost gap of
organic photovoltaics with highly efficient and stable small
molecule acceptor ternary solar cells. Nat. Mater. 16, 363–369
(2017).
10. Lin, Y. et al. An electron acceptor challenging fullerenes for
efficient polymer solar cells. Adv. Mater. 27, 1170–1174 (2015).
11. Zhu, L. et al. Single-junction organic solar cells with over 19%
efficiency enabled by a refined double-fibril network morphology.
Nat. Mater. 21, 656–663 (2022).
12. Fu, J. et al. 19.31% binary organic solar cell and low non-radiative
recombination enabled by non-monotonic intermediate state
transition. Nat. Commun. 14, 1760 (2023).
Article
13. Wang, D. H. et al. Sequential processing: control of nanomorpho­
logy in bulk heterojunction solar cells. Nano Lett. 11, 3163–3168
(2011).
14. He, Z. et al. Enhanced power-conversion efficiency in polymer
solar cells using an inverted device structure. Nat. Photon. 6,
591–595 (2012).
15. Jiang, Y. et al. An alcohol-dispersed conducting polymer complex
for fully printable organic solar cells with improved stability. Nat.
Energy 7, 352–359 (2022).
16. Meng, L. et al. Organic and solution-processed tandem solar cells
with 17.3% efficiency. Science 361, eaat2612 (2018).
17. Zheng, Z. et al. Tandem organic solar cell with 20.2% efficiency.
Joule 6, 171–184 (2021).
18. Holliday, S. et al. A rhodanine flanked nonfullerene acceptor for
solution-processed organic photovoltaics. J. Am. Chem. Soc. 137,
898–904 (2015).
19. Yang, Y. The original design principles of the Y series nonfullerene
acceptors, from Y1 to Y6. ACS Nano 15, 18679–18682 (2021).
20. Yuan, J., Zou, Y. & Jun Yuan The history and development of Y6.
Org. Electron. 102, 106436 (2022).
21. Cui, Y., Xu, Y. & Hou, J. High efficiency and more functions bring
a bright future for organic photovoltaic cells. Sci. Bull. 67,
1300–1303 (2022).
22. Yi, J., Zhang, G., Yu, H. & He, Y. Advantages, challenges and
molecular design of different material types used in organic solar
cells. Nat. Rev. Mater. 9, 46–62 (2024).
23. Yao, J. et al. Quantifying losses in open-circuit voltage in solution-
processable solar cells. Phys. Rev. Appl. 4, 014020 (2015).
24. Liu, S. et al. High-efficiency organic solar cells with low
non-radiative recombination loss and low energetic disorder. Nat.
Photon. 14, 300–305 (2020).
25. Zhang, G. et al. Delocalization of exciton and electron
wavefunction in non-fullerene acceptor molecules enables
efficient organic solar cells. Nat. Commun. 11, 3943 (2020).
26. Zhu, X., Zhang, G., Zhang, J., Yip, H.-L. & Hu, B. Self-stimulated
dissociation in non-fullerene organic bulk-heterojunction solar
cells. Joule 4, 2443–2457 (2020).
27. Chen, X.-K. et al. A unified description of non-radiative voltage
losses in organic solar cells. Nat. Energy 6, 799–806 (2021).
28. Eisner, F. D. et al. Hybridization of local exciton and
charge-transfer states reduces nonradiative voltage losses in
organic solar cells. J. Am. Chem. Soc. 141, 6362–6374 (2019).
29. Almora, O. et al. Device performance of emerging photovoltaic
materials (version 2). Adv. Energy Mater. 11, 2102526 (2021).
30. Classen, A. et al. The role of exciton lifetime for charge generation
in organic solar cells at negligible energy-level offsets. Nat.
Energy 5, 711–719 (2020).
31. Li, S. et al. Refined molecular microstructure and optimized carrier
management of multicomponent organic photovoltaics toward
19.3% certified efficiency. Energy Environ. Sci. 16, 2262–2273 (2023).
32. Proctor, C. M., Kim, C., Neher, D. & Nguyen, T. Nongeminate
recombination and charge transport limitations in
diketopyrrolopyrrole‐based solution‐processed small molecule
solar cells. Adv. Funct. Mater. 23, 3584–3594 (2013).
33. Lu, L., Kelly, M. A., You, W. & Yu, L. Status and prospects for ternary
organic photovoltaics. Nat. Photon. 9, 491–500 (2015).
34. Zuo, L. et al. Dilution effect for highly efficient
multiple-component organic solar cells. Nat. Nanotechnol. 17,
53–60 (2022).
35. Jiang, K. et al. Suppressed recombination loss in organic
photovoltaics adopting a planar–mixed heterojunction
architecture. Nat. Energy 7, 1076–1086 (2022).
36. Chen, T. et al. Compromising charge generation and recombi­
nation of organic photovoltaics with mixed diluent strategy for
certified 19.4% efficiency. Adv. Mater. 35, 2300400 (2023).
Nature Energy
https://doi.org/10.1038/s41560-024-01557-z
37. Gasparini, N., Salleo, A., McCulloch, I. & Baran, D. The role of the
third component in ternary organic solar cells. Nat. Rev. Mater. 4,
229–242 (2019).
38. Günther, M. et al. Models and mechanisms of ternary organic
solar cells. Nat. Rev. Mater. 8, 456–471 (2023).
39. Xu, X., Li, Y. & Peng, Q. Ternary blend organic solar cells:
understanding the morphology from recent progress. Adv. Mater.
34, 2107476 (2022).
40. Zhou, Z. et al. High-efficiency small-molecule ternary solar cells
with a hierarchical morphology enabled by synergizing fullerene
and non-fullerene acceptors. Nat. Energy 3, 952–959 (2018).
41. Yu, R. et al. Design and application of volatilizable solid additives
in non-fullerene organic solar cells. Nat. Commun. 9, 4645 (2018).
42. Zhan, L. et al. Desired open-circuit voltage increase enables
efficiencies approaching 19% in symmetric-asymmetric molecule
ternary organic photovoltaics. Joule 6, 662–675 (2022).
43. Jiang, Y. et al. Suppressing electron–phonon coupling in
organic photovoltaics for high-efficiency power conversion.
Nat. Commun. 14, 5079 (2023).
44. Vandewal, K., Tvingstedt, K., Gadisa, A., Inganas, O. & Manca, J. V.
On the origin of the open-circuit voltage of polymer-fullerene
solar cells. Nat. Mater. 8, 904–909 (2009).
45. Maurano, A. et al. Transient optoelectronic analysis of charge
carrier losses in a selenophene/fullerene blend solar cell. J. Phys.
Chem. C 115, 5947–5957 (2011).
46. Koster, L. J. A., Kemerink, M., Wienk, M. M., Maturová, K. &
Janssen, R. A. J. Quantifying bimolecular recombination losses in
organic bulk heterojunction solar cells. Adv. Mater. 23, 1670–1674
(2011).
47. Bartesaghi, D. et al. Competition between recombination and
extraction of free charges determines the fill factor of organic
solar cells. Nat. Commun. 6, 7083 (2015).
48. Wetzelaer, G. A. H., Kaap, N. J. V., der, Koster, L. J. A. &
Blom, P. W. M. Quantifying bimolecular recombination in organic
solar cells in steady state. Adv. Energy Mater. 3, 1130–1134 (2013).
49. Neher, D., Kniepert, J., Elimelech, A. & Koster, L. J. A. A new
figure of merit for organic solar cells with transport-limited
photocurrents. Sci. Rep. 6, 24861 (2016).
50. Rivnay, J., Mannsfeld, S. C. B., Miller, C. E., Salleo, A. & Toney, M. F.
Quantitative determination of organic semiconductor microstruc­
ture from the molecular to device scale. Chem. Rev. 112, 5488–5519
(2012).
51. Wang, R. et al. Charge separation from an intra-moiety
intermediate state in the high-performance PM6:Y6 organic
photovoltaic blend. J. Am. Chem. Soc. 142, 12751–12759 (2020).
52. Polman, A., Knight, M., Garnett, E. C., Ehrler, B. & Sinke, W. C.
Photovoltaic materials: present efficiencies and future challenges.
Science 352, aad4424 (2016).
53. Katsidis, C. C. & Siapkas, D. I. General transfer-matrix method for
optical multilayer systems with coherent, partially coherent, and
incoherent interference. Appl. Opt. 41, 3978 (2002).
54. Peng, W. et al. Reducing nonradiative recombination in perovskite
solar cells with a porous insulator contact. Science 379, 683–690
(2023).
55. Haase, F. et al. Laser contact openings for local poly-Si-metal
contacts enabling 26.1%-efficient POLO-IBC solar cells. Sol.
Energy Mater. Sol. Cells 186, 184–193 (2018).
56. Frisch, M. J. et al. Gaussian 16 Rev. A.03 (Gaussian, Inc., 2016).
57. Humphrey, W., Dalke, A. & Schulten, K. VMD: visual molecular
dynamics. J. Mol. Graph. 14, 33–38 (1996).
58. Lu, T. & Chen, F. Multiwfn: a multifunctional wavefunction
analyzer. J. Comput. Chem. 33, 580–592 (2012).
59. Abraham, M. J. et al. GROMACS: high performance molecular
simulations through multi-level parallelism from laptops to
supercomputers. SoftwareX 1–2, 19–25 (2015).
Article
60. Wang, J., Wolf, R. M., Caldwell, J. W., Kollman, P. A. & Case, D. A.
Development and testing of a general amber force field. J. Comput.
Chem. 25, 1157–1174 (2004).
61. Fox, T. & Kollman, P. A. Application of the RESP methodology in
the parametrization of organic solvents. J. Phys. Chem. B 102,
8070–8079 (1998).
62. Bayly, C. I., Cieplak, P., Cornell, W. & Kollman, P. A. A well-behaved
electrostatic potential based method using charge restraints
for deriving atomic charges: the RESP model. J. Phys. Chem. 97,
10269–10280 (1993).
63. Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through
velocity rescaling. J. Chem. Phys. 126, 014101 (2007).
64. Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F.,
DiNola, A. & Haak, J. R. Molecular dynamics with coupling to an
external bath. J. Chem. Phys. 81, 3684–3690 (1984).
65. Nosé, S. A unified formulation of the constant temperature
molecular dynamics methods. J. Chem. Phys. 81, 511–519 (1984).
66. Hoover, W. G. Canonical dynamics: equilibrium phase–space
distributions. Phys. Rev. A Gen. Phys. 31, 1695–1697 (1985).
67. Parrinello, M. & Rahman, A. Polymorphic transitions in single
crystals: a new molecular dynamics method. J. Appl. Phys. 52,
7182–7190 (1981).
68. Liu, W. et al. Theory-guided material design enabling high-
performance multifunctional semitransparent organic photo­
voltaics without optical modulations. Adv. Mater. 34, 2200337
(2022).
69. Li, W. et al. A high-efficiency organic solar cell enabled by
the strong intramolecular electron push–pull effect of the
nonfullerene acceptor. Adv. Mater. 30, 17071170 (2018).
70. Cui, Y. et al. Single-junction organic photovoltaic cell with 19%
efficiency. Adv. Mater. 33, 2102420 (2021).
Acknowledgements
This work was supported by the Beijing Natural Science
Foundation (Z230019), the National Key R&D Program of China
(2019YFA0705900), the Strategic Priority Research Program of the
Chinese Academy of Sciences (XDB0520202), the International
Partnership Program of the Chinese Academy of Sciences
(027GJHZ2022036GC) and National Natural Science Foundation of
China (52225305, 22175187 and 22375119). The authors thank C. Zhu,
L. Meng and Y. Li for their help with the TPC, TPV and CE measurements,
and M. Xie and Z. Wei for their help with the FTPS-EQE and EQEEL
measurements, and X. Pu for his help with photo-induced force
Nature Energy
https://doi.org/10.1038/s41560-024-01557-z
microscopy measurement, and Y. Ouyang and C. Zhang for their
help in TA measurements, and X. Chen, Q. Luo and C. Ma for their
help in MMP measurements and Y. Fu and X. Lu for their help in
GISAXS measurements.
Author contributions
X.Z. and F.L. conceived the study. Y.J. and F.L. fabricated the solar cells
and performed the related measurements. S.S. performed theoretical
calculations. R.X. and K.L. synthesized the NFAs Z7 and Z8. G.R. and
W.Z. performed the TA spectra analysis. X.Z. supervised the project.
X.M. and Y.Y. performed MD simulation. The manuscript was mainly
written by Y.J., S.S., F.L. and X.Z. All the authors contributed to the data
analysis and discussion.
Competing interests
The authors declare that they have no competing interests.
Additional information
Supplementary information The online version
contains supplementary material available at
https://doi.org/10.1038/s41560-024-01557-z.
Correspondence and requests for materials should be addressed to
Feng Liu or Xiaozhang Zhu.
Peer review information Nature Energy thanks Yanming Sun,
Hin-Lap Yip and the other, anonymous, reviewer(s) for their
contribution to the peer review of this work.
Reprints and permissions information is available at
www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Springer Nature or its licensor (e.g. a society or other partner) holds
exclusive rights to this article under a publishing agreement with
the author(s) or other rightsholder(s); author self-archiving of the
accepted manuscript version of this article is solely governed by the
terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to Springer Nature Limited
2024
 nature research | reporting checklist for manuscripts on photovoltaics research
Corresponding author(s): Xiaozhang Zhu
Initial submission Revised version Final submission

Reporting Checklist For Solar Cell Manuscripts

In order to improve the reproducibility and transparency of manuscripts related to photovoltaic cells, we request that authors
consider the following points in the preparation of their manuscript. Please supply a response to the checklist alongside your
submitted manuscript, and ensure that the relevant responses are also provided in the main manuscript, methods section or
supplementary information as appropriate. The completed checklist will be shared with reviewers.

` Solar Cell Data

1. Dimensions
State the dimensions of the tested solar cells.
2. Current-voltage characterization
Does the manuscript supply current density - voltage
(J-V) plots in both forward and backward direction?
The dimensions of the tested solar cells is 3.159 mm^2.
The current-voltage (J-V) plots in forward is supplied. Both forward and backward
scans were conducted, which yielded identical results.

Describe the voltage scan conditions for your devices. direction: forward; dwell times: 20 ms; speed: 0.015 V/s

Describe the test environment for your devices. We tested the devices at room temperature under N2 in the glove box.

If a preconditioning protocol was used before the

characterization, describe it.

Does the manuscript state whether the stability of the J-V The long-term stability of D18:Z8:L8-BO-, D18:L8-BO-, D18:Z8-based devices under
characteristic has been checked with a) stabilization of maximum power point (MPP) tracking were performed with continuous
photocurrent at maximum power point or b) maximum illumination in N2-filled chamber, as shown in Supplementary Figure 18.
power point data?
Please see ref. 7 for further details.

3. Hysteresis
Did you observe hysteresis or any other unusual
behaviour during the characterization of the solar cells?
4. Efficiency
Does the manuscript provides external quantum
efficiency (EQE) or incident photons to current efficiency
(IPCE) data for the devices?
No hysteresis or other unusual behaviour was observed during the characterization
of the solar cells.
External Quantum Efficiency (EQE) data for the devices are provided in Figure 1e
and calculated short-circuit current densities are comparable to that derived from
J-V plots.

Was the integrated response under the standard The integrated response under the standard reference spectrum is compared to
reference spectrum compared to the response measured the response measured under the simulator.
under the simulator?
For tandem solar cells, does the manuscript provide the Describe where this information can be found in the text OR state why this
bias illumination intensity and the bias voltage used for characterization has not been performed/is not applicable.
each subcell?

5. Calibration
State the light source and the reference cell or sensor
used for the characterization.
The reference cell was calibrated and certified.
Was a spectral mismatch calculation between the
Light source: solar simulator (XES-70S1, 7 × 7 cm^2 beam size; SAN-EI Electric Co.
Ltd.) coupled with AM 1.5G solar spectrum filters; Reference cell: SRC-1000-TC-QZ,
2 × 2 cm^2; VLSI Standards Inc.
The reference cell was calibrated and certified before the test.
June 2017

reference cell and the devices under test performed?

6. Mask/aperture
State the size of the mask/aperture used during testing. A mask of 3.159 mm^2 was used in NIM.

Does the measured short-circuit current density of the No, a JSC of 27.0 mA cm-2 was obtained using a mask of 3.159 mm^2.
devices vary with the mask/aperture area?

1
7. Performance certification
Name any independent certification laboratories that The performance of the devices is confirmed from National Institute of Metrology,

nature research | reporting checklist for manuscripts on photovoltaics research

have confirmed the photovoltaic performance of your China (NIM). The detailed data is provided in Figure 1f and Supplementary
devices. information.

You have provided a copy of any certificate(s) in the Supplementary Information.

8. Statistics
How many solar cells were tested in the study? We tested more than 30 devices in this study.

Does the manuscript include a statistical analysis of the We have a statistical analysis of the performance included in the manuscript. Table
performance? 1 shows the best and average efficiencies with standard deviation.

9. Long-term stability analysis

Clearly describe the type of stability analysis performed The long-term stability of optimal D18:Z8:L8-BO devices under maximum power
and the bias conditions used during this characterization. point (MPP) tracking was performed with continuous illumination in N2-filled
chamber, as shown in Supplementary Figure 18.

` Further reading
1. Shrotriya, V. et al. Accurate measurement and characterization of organic solar cells. Adv. Funct. Mater. 16, 2016–2023 (2006).
2. Dennler, G. et al. The value of values. Mat. Today 10, 56 (2007).
3. Cravino, A., Schilinsky, P. & Brabec, C. J. Characterization of organic solar cells: the importance of device layout. Adv. Funct. Mater. 17, 3906–3910 (2007).
4. Reese, M. O. et al. Consensus stability testing protocols for organic photovoltaic materials and devices. Sol. Energ. Mat. Sol. C 95, 1253–1267 (2011).
5. Snaith H. .J. The perils of solar cell efficiency measurements. Nat. Photon. 6, 337–340 (2012).
6. Luber, E. J. & Buriak, J. M. Reporting performance in organic photovoltaic devices. ACS Nano 7, 4708–4714 (2013).
7. Snaith, H. J. et al. Anomalous hysteresis in perovskite solar cells. J. Phys. Chem. Lett. 5, 1511–1515 (2014).
8. Grätzel M. The light and shade of perovskite solar cells. Nat. Mat. 13, 838–842 (2014).
9. Zimmermann E. et al. Erroneous efficiency reports harm organic solar cell research. Nat. Photon. 8, 669–672 (2014).
10. Beard M.C., Luther J.M. & Nozik A.J. The promise and challenge of nanostructured solar cells. Nat. Nanotech. 9, 951–954 (2014).
11. Timmreck, R. et al. Characterization of tandem organic solar cells. Nat. Photon. 9, 478–479 (2015).

A number of international committees develop industry standards on the characterization of photovoltaic technologies (for example ASTM-E44 and
IEC-TC 82), which can provide guidance for academic research.

June 2017