Low Temperature Aggregation Transitions in N3 and Y6 Acceptors Enable Double‐Annealing Method That Yields Hierarchical Morphology and Superior Efficiency in Nonfullerene Organic Solar Cells

Abstract

Thermal transition of organic solar cells (OSCs) constituent materials are often insufficiently researched, resulting in trial‐and‐error rather than rational approaches to annealing strategies to improve domain purity to enhance the power conversion efficiency. Despite the potential utility, little is known about the thermal transitions of the modern high‐performance acceptors Y6 and N3. Here, by using an optical method, it is discovered that the acceptor N3 has a clear solid‐state aggregation transition at 82 °C. This unusually low transition not only explains prior optimization protocols, but the transition informs and enables a double‐annealing method that can fine‐tune aggregation and the device morphology. Compared with 16.6% efficiency for PM6:N3:PC71BM control devices, higher efficiency of 17.6% is obtained through the improved protocol. Morphology characterization with x‐ray scattering methods reveals the formation of a multilength scale morphology. Moreover, the double‐annealing method is illustrated and easily transferred and validated with Y6‐based devices, using the transition of Y6 at 102 °C. As a result, the PCE improved from 16.0% to 16.8%. Design of high‐performance acceptors with yet lower aggregation transitions might be required for OSCs to successfully transition to low thermal budget industrial processing methods where annealing temperatures on plastic substrates have to be kept low. A clear solid‐state aggregation transition of the acceptor N3 is discovered, which enables a double‐annealing method that can fine‐tune aggregation and morphology. Compared with 16.6% efficiency for PM6:N3:PC71BM‐control devices, a higher efficiency of 17.6% is obtained through the improved protocol. The results point out a molecular design and engineering conundrum to achieve simultaneously low annealing temperatures, high efficiency, and stability.

Full text

www.afm-journal.de
© 2020 Wiley-VCH GmbH
2005011 (1 of 10)
Full PaPer
Low Temperature Aggregation Transitions in N3 and Y6
Acceptors Enable Double-Annealing Method That Yields
Hierarchical Morphology and Superior Eciency in
Nonfullerene Organic Solar Cells
Yunpeng Qin, Ye Xu, Zhengxing Peng, Jianhui Hou,* and Harald Ade*
Thermal transition of organic solar cells (OSCs) constituent materials
are often insufficiently researched, resulting in trial-and-error rather
than rational approaches to annealing strategies to improve domain
purity to enhance the power conversion efficiency. Despite the potential
utility, little is known about the thermal transitions of the modern high-
performance acceptors Y6 and N3. Here, by using an optical method,
it is discovered that the acceptor N3 has a clear solid-state aggregation
transition at 82 °C. This unusually low transition not only explains
prior optimization protocols, but the transition informs and enables a
double-annealing method that can fine-tune aggregation and the device
morphology. Compared with 16.6% efficiency for PM6:N3:PC71BM control
devices, higher efficiency of 17.6% is obtained through the improved
protocol. Morphology characterization with x-ray scattering methods
reveals the formation of a multilength scale morphology. Moreover,
the double-annealing method is illustrated and easily transferred and
validated with Y6-based devices, using the transition of Y6 at 102 °C.
As a result, the PCE improved from 16.0% to 16.8%. Design of high-
performance acceptors with yet lower aggregation transitions might
be required for OSCs to successfully transition to low thermal budget
industrial processing methods where annealing temperatures on plastic
substrates have to be kept low.
DOI: 10.1002/adfm.202005011
Dr. Y. P. Qin, Z. X. Peng, Prof. H. Ade
Department of Physics and Organic and Carbon Electronics Laboratories
(ORaCEL)
North Carolina State University
Raleigh, NC 27695, USA
E-mail: hwade@ncsu.edu
Y. Xu, Prof. J. H. Hou
State Key Laboratory of Polymer
Physics and Chemistry
Institute of Chemistry
Chinese Academy of Sciences
Beijing 100190, China
E-mail: hjhzlz@iccas.ac.cn
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adfm.202005011.
1. Introduction
Solution-processed organic solar cells
(OSCs) utilizing bulk heterojunction
(BHJ) architecture have attracted con-
siderable attention for their promise of
being low cost and easy to fabricate by
large-scale roll-to-roll processing.[1–5]
A combination of novel materials syn-
thesis, morphology control, and device
optimization boosted the power con-
version eciencies (PCEs) of the state-
of-the-art OSCs to >16% over the past
three decades for multiple material sys-
tems.[6–10] For solution-processed OSCs,
the general optimization procedures of
BHJ comprise the control of the donor–
acceptor ratios,[11] additive use,[12] host
solvent selection,[13,14] and thermal or
solvent annealing.[15–17] For solution-pro-
cessed OSCs, varying process conditions
often lead to distinctly dierent mor-
phology as well as the performance of
OSCs. What is often lacking is a mole-
cular level understanding why process
conditions have the impact they do and
how new materials with targeted prop-
erties have to be designed. Therefore,
further investigations of molecular interactions and thermal
transitions and their impact on device optimization strate-
gies as well as stability are critical for improving and transi-
tioning OSC technology further.
Though a considerable amount of work has been performed
on developing new organic photovoltaic materials to enhance
the PCEs,[11,18–20] device engineering and processing have also
contributed significantly to improving performance.[13,21,22]
Throughout the development of OSCs, much of the device
optimization strategies have been driven by empirical trial-
and-error approaches. Of all the methods mentioned above,
thermal annealing is the most common method for control-
ling the BHJ morphology and improving performance, which
in general improves the molecular packing and charge trans-
port within the system.[15,16] Thermal annealing was first suc-
cessfully applied in the P3HT:PC61BM system by Padinger
et al., who reported eciencies up to 3.5% for films after
thermal annealing at 75°C for 4 min.[23] Improved optimization
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (2 of 10)
protocols for P3HT:PC61BM use an annealing temperature
close to the glass transition temperature (Tg) of PCBM, with
a tendency of the fullerene to form detrimental macroscopic
aggregates due to overannealing.[24] The Tg is also a critical para-
meter even during the casting of the BHJ active layer,[25,26] and
can be used to qualitatively predict the stability of organic semi-
conductor devices.[27,28]
The emergence of nonfullerene small molecule acceptors
(NFAs)[8,10,19,29–34] has enriched, yet complicated matters. The
dierences between NFAs in molecular geometry, stiness,
and size might have a profound eect on Tg and the aggrega-
tion behavior of the NFAs, optimization protocols, and stability
of the active layer.[28] Recently, a series of binary and ternary
eciency records have been achieved with the high-performing
NFAs named Y6 and N3[6,8] (for the full name of materials see
Experimental Section). Even though N3 and Y6 are some of the
most successful and promising acceptors, their thermal transi-
tions and consequent impact on stability and rational optimiza-
tion are unknown due to the relative diculty to observe any
transitions other than melting when using conventional dier-
ential scanning calorimetry (DSC). It is particularly important
to better understand thermal structure–function properties in
general for all organize devices and to rationally accelerate OSC
optimization strategies and technology translation.
Herein, we report thermal transitions of N3 at 82°C and Y6 at
102°C as observed by a UV–vis method.[35] These transitions are
lower than thermal transitions previously observed with DSC in
other NFAs such as EH-IDTBR (≈120°C),[28] ITIC (≈200°C),[28]
and IEICO-4F (≈200°C).[16] Guided by the knowledge of these
transitions, we developed a double-annealing method to fabri-
cate PM6:N3:PC71BM based device that can fine-tune the mor-
phology and performance, with a short annealing just below
and subsequently annealing above this transition yielding the
best results. We employ grazing incidence wide-angle X- ray
scattering (GIWAXS)[36–38] and resonant soft X-ray scattering
(R-SoXS)[39–44] techniques to systematically characterize the
molecular packing and morphology of the blends. In contrast
to the single-annealed systems, we observed a hierarchical mor-
phology with improved “domain purity” and longer coherence
length in the double-annealed systems. Moreover, we were able
to establish the relation of the multilength scale morphology to
the device physics. As a result of using the double-annealing
method and forming the multilength scale morphology, a PCE
of 17.6% was obtained, which is among the best values in the
field of OSCs overall and a record for the particular materials
system used.[8] Our results point out a molecular design and
engineering conundrum in how to achieve simultaneously low
annealing temperatures, high stability, and high eciency.
2. Results and Discussions
2.1. Basic Properties and Photovoltaic Performance
The molecule structures of the materials used in the blend are
presented in Figure 1. The UV–vis absorption spectra of PM6,
N3, and PC71BM in the film state are shown in Figure S1 (Sup-
porting Information). The PM6 exhibits its main absorption in
the 400–700nm range. N3 exhibits a strong and broad absorp-
tion over 600–950 nm that complements absorption by PM6
and is one of the most promising acceptors whose basic aggre-
gation properties need to be understood.
In order to guide our double-annealing protocols, we initially
measured the absorption spectra of the N3 thin films with the
increasing thermal annealing temperature (see Figure 2a,b).
Upon annealing, the absorbance increases in intensity with
an apparent red-shift of the edge. However, when normalized
(Figure2b), the edge is not changing, yet the peak location is
slightly changing. This implies several interesting details. The
sample is inhomogeneous and the spectra reflect a population
distribution of aggregate characteristics (e.g., size, strength,
type). Within that distribution, the as-cast film has already
Figure 1. Schematic illustration of double-annealing method and chemical structures of the model material system (PM6, N3, Y6, and PC71BM).
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (3 of 10)
N3 molecules/aggregates whose inter- and/or intramolecular
optical coupling has reached intrinsic characteristics that
cannot be further changed with increased temperature. One
possibility is that the aggregate size has reached the intrinsic
coherence length of the exciton and further increases in aggre-
gate size would not change the optical properties. The spectral
changes therefore reflect the evolutions in the populations
characterized by dierences in intra-molecular (e.g., conforma-
tions) and intermolecular interactions (e.g., packing) as well as
the aggregate size and exciton coherence length.
For a quantitative analysis, we utilize the same method as
reported by Root et al.,[35] and use a deviation metric (DMT)
that is as the sum of the squared deviation in the absorbance
between as-cast and annealed films
∑
λλ
[]
() ()
≡−
λ
λ
DM
TRTT
2
min
max
II
(1)
where λ is the wavelength, and λmin and λmax are the lower and
upper bounds of the optical sweep, IRT(λ) and IT(λ) are the nor-
malized absorption intensities of the as-cast (room tempera-
ture) and annealed films, respectively. As shown in Figure2c,
when plotting the deviation metric against annealing tempera-
ture for upper and lower bounds of 1000 and 350nm respec-
tively, a clear transition is observed. Using two linear fits to the
two branches, the transition was determined to be 82 ±1 °C. It
has been suggested that the transition observed with UV–vis is
Tg,[35] but the values of the optical measurements of polymers
do not always agree with rheology and dynamic mechanical
analysis characterization, which are also impacted and compli-
cated by the side chains. Since optical properties are controlled
by the core of the NFA or backbone of a polymer, the transition
observed is likely the Tg of the core, with side-chains having
a Tg often below 0 °C. Furthermore, we are unable to observe
an aggregation transition of PM6 by the same method over
the T range probed (Figure S2b, Supporting Information), as
its absorption spectrum barely changes when annealed. Simi-
larly, standard DSC does not observe any thermal transitions
for PM6 (see Figure S3, Supporting Information). PM6 is likely
a very sti polymer with limited ability to rearrange backbone
segments once solidified. The dierent thermal and aggrega-
tion characteristics of PM6 and N3 allow us to selectively study
the impact of N3 aggregation on device performance, Further-
more, the observed thermal transition does not change signifi-
cantly for PM6:N3 binary and PM6:N3:PC71BM ternary blends
(Figure S2, Supporting Information), providing uniform guid-
ance for any devices optimization protocol.
Given the clear aggregation transition of N3, we explore
its relevance to device optimization by annealing below and
subsequently above this transition of N3 and comparing such
double-annealed devices to single annealed reference devices.
We choose the condition for the discussion presented here
those systems with 80 °C as the initial annealing temperature
and include the annealing temperature of 100 °C, 120 °C, and
150 °C for the second step (A full set of conditions explored
are presented in the Supporting Information). As shown in
Figure 2d, with the increase of the annealing temperature,
the UV–vis spectra of the blend thin films had a significant
Figure 2. a) UV–vis absorption spectra. b) Normalized UV–vis absorption spectra of N3 thin film with the increasing thermal annealing temperature.
c) Evolution of the deviation metric as a function of annealing temperature, showing a distinct increase at the low T transition of N3(82 ± 1 °C).
d) UV–vis absorption spectra of PM6:N3:PC71BM thin film with the increasing thermal annealing temperature. d) The change of peak position with the
increasing thermal annealing temperature of the thin films.
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (4 of 10)
average red-shift, the peak position moved from 818 nm
(80 °C) to 834 nm (150 °C) (see insert Figure2d). This shift
is attributed to the formation of ordered aggregates from the
disordered volume fraction, which is generally accompanied by
an improvement in the charge transport properties of the thin
film.
The OSCs were fabricated with a conventional configura-
tion of indium tin oxide (ITO)/PEDOT:PSS(4083)/active layer/
PFN-Br/Al (see Figure 1), and the general device processing
conditions were similar to the previous work.[8] The current
density–voltage (J–V) curves of the OSCs under the illumina-
tion of AM 1.5G 100 mW cm−2 are shown in Figure 3a and
Figure S4a (Supporting Information), and the corresponding
photovoltaic parameters, including that of the best devices
and the statistical data of forty cells are summarized in
Table 1 and Table S1 (Supporting Information). Henceforth,
we will refer to the sample prepared by a single annealing
step only as “single-annealed” whereas the films prepared
by the initial annealing step followed by another higher tem-
perature annealing step as “double-annealed” with the under-
standing that the corresponding two annealing times were
5 and 10min, respectively, single annealing for 10min. As shown
in Figure3a and Table1, when we used a single annealing step
at 100°C for 10 min, a PCE of 16.6% was achieved with a VOC
of 0.85V, a JSC of 26.05mA cm−2 and the FF of 0.75. This is
consistent with the reported state-of-the-art performance for
this ternary.[8] Meanwhile, for the double-annealing method,
the annealed photoactive films were prepared by annealing
PM6:N3:PC71BM devices at 100 °C, 120 °C, or 150°C following
the initial annealing at 80°C. Overall, the dierences observed
are not dramatic, but self-consistent and significant when
averaged over 40 devices are considered. Furthermore, the
double-annealed devices always outperform the corresponding
single-annealed devices irrespective of whether the initial or
final annealing temperature matches the single annealing tem-
perature. Overall, an excellent PCE of 17.6% was achieved for
the 80°C/120°C double-annealed system, with a VOC of 0.84V,
a JSC of 26.85mA cm−2 and the FF of 0.78.
The external quantum eciency (EQE) spectra of the above-
mentioned devices are displayed in Figure 3b, and the EQE
results are consistent with the J–V measurements. In addition,
as shown in Figure S4b (Supporting Information), a red-shift of
the response range is observed for higher annealing tempera-
ture, which is in agreement with the results obtained from the
UV–vis spectra. These results agree well with the higher JSC
for the higher annealing temperature obtained from the J–V
measurement. Furthermore, we studied a range of dierent
annealing time and temperature of the double-annealed and
single-annealed devices to better understand the significance of
the N3 transition at 82 °C, which were shown in Tables S2–S4
(Supporting Information). Furthermore, We also applied a third
annealing step and tried several temperatures for the best per-
formance devices, which was shown in Table S5 (Supporting
Information). When we used a lower annealing tempera-
ture (100 °Cand 80 °C)for the third annealing step, the PCE
changed little, but when we used a higher annealing temper-
ature (120 °C) for the third annealing step, the PCE degrade.
This is similar to the results achieved when considering longer
annealing times. The double-annealing condition at 80 °C for
5 min and then a higher T (100 °C, 120 °C, and 150 °C for
10 min) showed the best performance. Furthermore, we have
measured the stability of the optimal devices, which stored in
an N2-atmosphere, the devices showed good stability. After we
stored for 500 h, the devices remain at about 94% of the initial
PCE (see Figure S5 in the Supporting Information). Finally, we
choose these conditions for the systematic discussion presented
here that includes underannealed, optimally annealed, and
overannealed data sets.
Figure 3. a) J–V and b) EQE curves of champion OSCs based on dierent annealing method.
Table 1. Photovoltaic parameters of the PM6:N3:PC71BM-based OSCs with the varied annealing methods, under the illumination of AM 1.5G,
100mW cm−2.
Annealing conditionsa) VOC [V] JSC [mA cm−2]FF PCE [%]
100 °C 0.84±0.01(0.85) 25.90±0.15(26.05) 0.75±0.06(0.75) 16.6±0.07(16.6)
80°C/120 °C 0.84±0.01(0.84) 26.76±0.11(26.85) 0.78±0.04(0.78) 17.6±0.04(17.6)
a)The statistical values were obtained from 40 devices for each of the fabrication conditions.
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (5 of 10)
2.2. Domain Spacings and Composition Variations
at Two Length Scales
The PCE improvements of the double-annealed devices are
mostly attributed to the increase of FF and JSC, which can
generally be aected by the morphology, the interlayer used,
and the film thickness. Since the same device configuration is
utilized and the thickness of all these samples is very similar
(100 ± 5nm), we can conclude that the improvement of FF and
JSC in the double/single-annealed devices are mainly due to the
more favorable nanomorphology.
The phase separation morphologies of the photoactive
layers in the devices are monitored by atomic force microscopy
(AFM). As shown in Figure S6 (Supporting Information), the
double-annealed films (Figure S6d–f, Supporting Information)
exhibit reduced surface roughness compared with the single-
annealed films (Figure S6a–c, Supporting Information), indi-
cating possibly smaller aggregates or smaller domains in the
double annealed devices. As AFM provides limited information
about the bulk morphology, we then examined the mesoscale
structures of the blend films using R-SoXS technique at beam-
line 11.0.1.2 of the Advanced Light Source.[39–43] A resonant
energy of 283.8eV was used to provide highly enhanced mate-
rial contrast and reduced fluorescence background between the
donor and acceptor. As shown in Figure 4, the R-SoXS profiles
for the double-annealed films show distinct two-length scale
phase separations with two log-normal distributions associated
with larger and smaller domains, while the single-annealed
films only exhibit as single log-normal domain distribution are
larger length scales. We list the quantitative results for multi-
peak fitting for low-q and high-q peaks in Table 2. Herein, we
choose the double-annealing condition at 80 °C/120 °C and
the single-annealing condition of 120 °C for comparison. For
the double-annealed films at 80 °C/120 °C, two long-normal
peaks are observed, which are located at q= 0.16 and 0.40 nm−1,
corresponding to the characteristic length scales of 39 and
16 nm, respectively. The multilength scale structure always
leads to improved eciency in the devices by establishing e-
cient exciton dissociation and enhanced charge collection.[45,46]
Here, the presence of the small domains associated with the
Figure 4. a–c) Peak fits to the circularly averaged R-SoXS profiles obtained at 283.8eV with one log-normal components for single-annealed films.
d–f) Profiles obtained with two log-normal components for double-annealed films. g–i) R-SoXS profile with double- and single-annealed blend films.
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (6 of 10)
characteristic length of 16 nm likely enables ecient charge
creation and minimized the charge recombination, as observed
previously.[45–47] In contrast, for the single-annealed films at
120 °C, only one log-normal peak (q= 0.16 nm−1), corresponding
to the characteristic length scale of 39 nm, is observed. It is
noteworthy that the R-SoXS profile of double-annealed devices
is more complex than that of the single-annealed devices and as
a result, the higher JSC and FF for the double-annealed devices
may be directly related to the multilength scale morphology
that includes the smaller domains.
In addition to length scales, the composition variations
(monotonically related to domain purity) are important and we
analyze the integrated scattering intensity (ISI) of the devices.
The average composition variation over the length scales
probed is proportional to the square root of the normalized
ISI. The ISI values (over the entire q range of the measure-
ment) were calculated from the scattering profiles to evaluate
the average relative composition variations of the domain
overall length scales for the blends. The higher the ISI over
the full q range probed, the higher σ (standard deviation) and
the purer the phases. Higher purity often correlates to high
performance as long as the percolation threshold for electron
transport is not crossed.[45,46] Here we use σ of the 80°C/120°C
double-annealed film as a reference and assign the value of 1
(see Table 2). The relative σ for the double-annealed systems
(80°C/100°C, 80 °C/120°C, and 80°C/150°C) are measured
as 0.74, 1, and 0.92, respectively. Furthermore, as shown in
Table2, the σ’s of the double-annealed systems are consistently
higher than the single-annealed systems, viz. 0.69, 0.76, and
0.84 (100°C, 120°C, and 150 °C). We will discuss the implica-
tions below.
We note that the domain size as captured by the long period
changes very little with the second annealing temperature (see
Table2). The small domains get patterned by the initial anneal
at 80 °C. The larger domains get patterned by the casting
dynamics and kinetics. The annealing at the second, higher
temperatures leads to very little coarsening or charges in size
distribution. It only changes the purity of the domains across
all length scales.
2.3. Molecular Packing, Texture, and Coherence Lengths
Previous research revealed that the domain purity and coher-
ence length (CL) of the molecular packing is often related to
the charge mobilities and recombination.[40,41,48] Here, we
investigate the CL by performing GIWAXS to reveal the mole-
cular packing and the texture of the active layer and its rela-
tion to the higher FF obtained by the double-annealed systems.
As observed from the GIWAXS patterns and 1D profiles in
Figure 5 and their analysis in Table S6 (Supporting Informa-
tion), scattering peaks of blend films at q values of 1.78 Å−1 in
the out-of-plane direction with a corresponding (100) in the
in-plane indicate face-on orientation in both double-annealed
Table 2. R-SoXS analysis results for PM6:N3:PC71BM-based OSCs with
the varied annealing conditions.
Annealing
conditions
Long period (Low-q peak)
[nm]
Long period (high-q peak)
[nm]
Relative σ
100 °C45 – 0.69
80°C/100 °C45 17 0.74
120 °C39 – 0.76
80°C/120 °C39 16 1.00
150 °C42 – 0.84
80°C/150 °C42 14 0.92
Figure 5. a) 2D GIWAXS patterns. b) 1D GIWAXS out-of-plane and c) In-plane of the dierent annealing conditions.
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (7 of 10)
and single-annealed samples. In addition, the corresponding
out-of-plane (010) π–π coherence lengths of single-annealed
and double-annealed blend films were extracted via peak fitting
by using the full width at half-maximum of the (010) stacking
peaks (see Table S6, Supporting Information). We find that the
CLs of the double-annealed systems are consistently larger than
the single-annealed systems (see Figure 6a). Such increase in
coherence length has been observed previously to frequently
correlate to device performance and charge mobility, even for
electrons in PCBM-based systems.[48–51] Furthermore, as shown
in Figure6b, the device FF has a strong correlation with relative
σ of the composition, and the highest relative σ is obtained in
the highest FF and eciency device processed at the double-
annealed device.
2.4. Charge Extraction and Recombination
The hole and electron mobility (μh and μe) of the blend films
processed with the double-annealing method were measured by
using the space-charge-limited-current (SCLC) method.[52] The
resultant μh and μe are summarized in Figure S7 and Table S7
(Supporting Information). A clear dierence in the μh and μe
of these blends is observed. The double-annealed system of
80°C/120°C device has the highest μh (2.21 × 10−4 cm2 V-1 s-1)
and μe (8.74 × 10−4 cm2 V-1 s-1). As before, the double-annealed
devices have consistently better mobility than the single-
annealed devices.
To unravel the dierences in recombination mechanism
between the devices, we analyze the semi-logarithmic plot of
VOC as a function of the light intensity (see Figure S8 in the
Supporting Information), which shows a linear relationship
with a slope of nkT/q (1 < n< 2), where k is the Boltzmann’s
constant, q is the elementary charge, n is a scaling factor and
T is Kelvin temperature.[53] The purely bimolecular recom-
bination is identified with a dependence of VOC on light
intensity with a slope of kT/q, while the trap-assisted recom-
bination is identified with a slope of 2 kT/q (T= 300 K). As
shown in Figure S8 (Supporting Information), the depend-
ence of VOC on the light intensity for the 80°C/120°C double-
annealed film had the lowest scaling factor of n= 1.02. The
other systems had stronger trap-assisted recombination due
to the lower domain purity compared to the 80 °C/120 °C
double-annealed devices. Again, the double-annealed devices
have less bimolecular recombination than the single-annealed
devices.
Light-intensity dependent JSC measurements were also exam-
ined to investigate the nongeminate bimolecular recombination
losses in the devices (see Figure S8 in the Supporting Infor-
mation).[54,55] The relationship between JSC and incident light
intensity can be described as JSC∝ ligh
t
S
P
. S should be equal to 1
if all dissociated free carriers are collected at the corresponding
electrodes without charge recombination, while S<1 indicates
the presence of some extent of bimolecular recombination. All
of the devices herein showed a linear dependence of current
density on the light intensity in logarithmic coordinates with
a slope of 0.99 for the double-annealed film of 80 °C/120 °C,
while other systems had lower S values, which means the
double-annealed film at 80 °C/120 °C suppresses bimolecular
recombination, but when the temperature is too high, the trend
is broken.
2.5. Thermal Transition of Y6 and the Generality of the
Double-Annealing Method
To validate the power conferred by knowing thermal transitions
and general applicability of the double-annealing method, the
protocol has been successfully transferred to PM6:Y6:PC71BM
within a single afternoon and a single device fabrication run. In
order to guide our double-annealing protocols on this system,
we measured the absorption spectra of the Y6 thin films and
when plotting the deviation metric against annealing tempera-
ture, a clear transition of 102 ± 1 °Cis observed (see Figure 7a).
Furthermore, we also acquired UV–vis data from EH-IDTBR
and a clear transition of 117 ± 1 °C is observed (Figure 7b),
which is consistent with the reported value in the literature.[28]
As a result of this knowledge, we did not explore a larger para-
meter space in temperature for the PM6:Y6:PC71BM system,
but set 100 °C as the first and 120 °C, 140 °C, and 160 °C as the
second annealing temperature in an analogous fashion to the
N3-based optimization where the first annealing step was just
below the thermal transition. As shown in Table S8 (Supporting
Information), when we used a thermal annealing step at 100°C
for 10min, a PCE of 16.0% was achieved, which is consistent
with the reported performance for this ternary,[56] in the mean-
time, when we adopt the double-annealing method, a higher
PCE of 16.8%, a 5% increase compared the single-annealed
Figure 6. a) π–π coherence lengths of the dierent annealing conditions. b) Plot of device FF and relative σ versus the dierent annealing conditions
(C1:100°C and 80°C/100°C; C2:120°C and 80°C/120°C; C3:150°C and 80°C/150°C).
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (8 of 10)
device, was achieved, which agrees well with those obtained
for the PM6:N3:PC71BM-based OSC. Though the thermal
transition of N3 and Y6 are surprisingly dierent, the double-
annealing method improves the eciency of PM6:N3:PC71BM
and PM6:Y6:PC71BM-based OSCs by finely controlling the
annealing protocols that are determined and guided by the
measured aggregation transitions of the materials.
3. Conclusions
The thermal transitions of N3 (82 °C) and Y6 (102 °C) are
lower than those in other modern acceptors such as EH-IDTBR
(≈120 °C), ITIC (≈200 °C), and IEICO-4F (≈200 °C).[16,28] The
clear dierence between N3 and Y6 is somewhat surprising,
given that the only structural dierence between these two
materials is the location of the branching points on the inside
(bay) branched side chains and a slightly longer chain for N3.
The origins of the wide range of transition temperature across
the modern NFAs classes of materials and the underlying
structure–functions are not fully understood. A low transition
can be beneficial or have detrimental impact. Low transition
temperatures would enable annealing of as-printed active layers
in low thermal budget industrial processing on highly flexible
plastic substrates where temperatures should be kept as low
as possible, ideally below 100°C, but definitively below 120°C.
Low Tg materials also yield more flexible and stretchable active
layers.[57,58] In contrast, high Tg leads to more stable films and
devices and many synthesis strategies often aim at making
acceptors stier and more planar to control their aggregation
properties and yield more stable devices.[26–28]
Considering all the results, a self-consistent description
emerges. Guided by the aggregation transition of N3 at 82°C
as observed by UV–vis, the double-annealing method, which
related to anneal below and subsequently above the transition, is
utilized and the devices show distinct multilength scale structure
features and increase the purity of the domains and the coher-
ence length of the molecular packing. Although larger CL and
purer domains favor improved performance, the relative purity
and multi-length scale structure are optimized for 80°C/120°C
devices and are the dominating factors over molecular packing
dierences. Benefiting from these synergistic improved
properties, the 80°C/120°C double-annealed device exhibits the
highest PCE of 17.6% with a higher JSC (26.85 mA cm−2) and
FF (0.78) compared with the single-annealed devices. The pre-
cise underlying molecular and thermodynamic mechanisms
of the morphology optimization need to be further elucidated,
but we surmise that the low temperature annealing patterns the
molecular packing at small length scale, and the second step
changes the overall purity in the mixed domains. Both steps
likely depend on the specific intermolecular interactions, the
nucleation and growth kinetics, and the temperature dependent
diusion coecients and miscibility.[59,60] We articulate the
following hypothesis: Below the thermal transition, diusion
is very limited yet supercooling is enhanced and any aggrega-
tion and crystal nucleation is local, at small length scales and
numerous. This is consistent with the recent observations that
the nucleation density of ITIC crystals is three orders of magni-
tude higher at 160 °C (Tg≈ 180 °C) than at 220 °C.[27] The second
annealing step yields longer distance diusion between the
larger domains that allows the mixed domains to purify towards
the temperature-dependent binodal composition. The dier-
ence in nucleation density in ITIC as a function of temperature,
in conjunction of the emergence of small domains only in the
double-annealed devices indicates that the counter-intuitive first
annealing step below the thermal transition is crucially impor-
tant to achieve phase separation at small lengths scales and
optimized performance. Irrespective of the precise mechanism,
our results point out that optimizing the morphology and per-
formance can require a complex processing path. The demon-
strated double-annealing method provides a guidance as to how
morphologies and device performance can be further optimized
by manipulating nucleation and growth of aggregates. As OSCs
improve toward theoretical limits, all optimization protocols will
yield smaller fractional improvements, yet will become increas-
ingly important in order to achieve the best performance pos-
sible. The successful application of our approach to a second
system indicates that systematic determination of optical aggre-
gation transitions can successfully guide optimization. The
value of such transitions might have been overlooked so far and
more systematic exploration of its utility is advisable.
Our results point out a molecular design and engineering
conundrum. Unless as cast morphologies can be achieved
to yield high performance, even lower thermal transitions
Figure 7. Evolution of the deviation metric as a function of annealing temperature, showing a low T transition of a) Y6 (102 ± 1 °C),b) EH-IDTBR
(117 ± 1 °C).
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (9 of 10)
as currently available are needed to enable annealing in low
thermal budget industrial processing on highly flexible plastic
substrates, yet many synthesis strategies often aim at making
acceptors stier and more planar to control their aggregation
properties and yield more stable devices. Only further detailed
research understanding of structure-function relationship will
provide rational synthesis and processing strategies that will
simultaneously yield low annealing temperatures, high sta-
bility, and high performance.
4. Experimental Section
Solar Cell Fabrication and Measurement: The OSCs were fabricated
with a conventional configuration of indium tin oxide (ITO)/
PEDOT:PSS(4083)/active layer/PFN-Br/Al. The ITO substrates were first
scrubbed by detergent and then sonicated with deionized water, acetone
and isopropanol subsequently. The glass substrates were treated by
UV-Ozone for 15min before use. PEDOT:PSS (4083) was spin-cast onto
the ITO substrates at 4000 rpm for 30 s and then dried at 150 °C for
15 min in air. The PM6:N3(Y6):PC71BM blends (1:1.2:0.2 weight ratio)
were dissolved in chloroform (the total concentration of blend solutions
was 19.2 mg mL−1), with the addition of 0.5% CN as additive, and
stirred overnight in a nitrogen-filled glove box. The blend solutions
were spincast at 2500 rpm for 30 s on the top of a PEDOT:PSS layer
followed by a thermal annealing step at 100°C for 10min. Then a thin
PFN-Br layer (≈5nm) was coated on the active layer. At last the top
metal electrode was evaporated at ≈1 × 10−6Torr and consisted of an
Al (100nm) layer. J–V characteristics were recorded with a Keithley 2400
source meter under 100mW cm−2 AM 1.5G light. The light is provided by
a Class 3A Solar Simulator and KG5 silicon reference cell.
UV–vis Measurement: All thin films subject to UV−vis absorption
measurement were spin-cast from chloroform solutions directly onto
optically transparent glass. The UV–vis spectra were recorded with
a Cary 60 spectrometer (Agilent) after annealing individual films for
10 min. To minimize optical scattering induced by glass substrates, a
bare glass was put into the reference optical path.
Molecular Packing and Morphology Characterizations: GIWAXS and
R-SoXS measurements were respectively performed at the beamline
7.3.3[38] and beamline 11.0.1.2,[42] respectively, at the Advanced Light
Source (ALS), Lawrence Berkeley National Laboratory, following the
previously established protocols. GIWAXS data were acquired just above
the critical angle (0.13°) of the films with a hard X-ray energy of 10keV,
and Silver Behenate (AgB) was used for geometry calibration. R-SoXS was
performed in a transmission geometry with linearly polarized photons
under high vacuum (1 × 10−7Torr) and a Peltier cooled (−45°C) charge-
coupled device (CCD) (Princeton PI-MTE, 2048 pixels × 2048 pixels) was
used to capture the soft X-ray scattering 2D patterns. The raw 2D X-ray
data were processed with a modified version of NIKA into 1D scattering
profiles I(q).[61] Thicknesses for the samples were measured with a KLA-
Tencor P-15 profilometer.
The full name of the materials used in the manuscript:
PM6: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-(1′,30-di-2-thienyl-50,7′-bis(2-
ethylhexyl)benzo[1′,2′-c:4′,50-c′]dithiophene-4,8-dione),
N3:(2,2′-((2Z,2’Z)-((12,13-bis(3-ethylheptyl)-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(methanylylidene))bis(5,6-difluoro-3-oxo-2,3-dihydro-1H-indene-
2,1-diylidene))dimalononitrile
Y6:2,2′-((2Z,2′Z)-((12,13-bis(2-ethylhexyl)-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(me-thanylylidene))bis(5,6-difluoro-3-oxo-2,3-dihydro-1H-indene-
2,1-diylidene))dima-lononitrile
PC71BM: [6,6]-phenyl-C71-butyric acid methyl ester
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
Acknowledgements
Y.Q., Z.P., and H.A. gratefully acknowledge the support from the
US Oce of Naval Research (ONR, grants no. N000141712204 and
N000142012155) and UNC General Administration Research Opportunity
Initiative grant. Y.X. and J.H. acknowledge financial support from the
NSFC (nos. 91333204 and 51261160496), and the Chinese Academy of
Sciences (no. XDB12030200). X-ray data were acquired at beamlines
11.0.1.2, 7.3.3, and 5.3.2.2 at the ALS, which is supported by the Director,
Oce of Science, Oce of Basic Energy Sciences, of the US Department
of Energy under contract no. DE-AC02-05CH11231. C.Z., E.S., A.H.,
and C.W. of the ALS (DOE) are acknowledged for assisting with the
experimental setup and providing instrument maintenance. C.Ho is
acknowledged for helping with the EQE measurement. M. Ghasemi and
R. Henry are acknowledged for helping with the DSC measurement and
providing some of the DSC data.
Conflict of Interest
The authors declare no conflict of interest.
Keywords
domain purity, double annealing, multilength-scale morphology, organic
solar cells, thermal transition
Received: June 12, 2020
Revised: July 20, 2020
Published online:
[1] G.Yu, J.Gao, J. C. Hummelen, F.Wudl, A. J.Heeger, Science 1995,
270, 1789.
[2] G.Li, V.Shrotriya, J. S.Huang, Y.Yao, T.Moriarty, K.Emery, Y.Yang,
Nat. Mater. 2005, 4, 864.
[3] M. C. Scharber, D. Wuhlbacher, M. Koppe, P. Denk, C. Waldauf,
A. J.Heeger, C. L.Brabec, Adv. Mater. 2006, 18, 789.
[4] J. S.Huang, G.Li, Y.Yang, Adv. Mater. 2008, 20, 415.
[5] S. H. Park, A. Roy, S. Beaupre, S. Cho, N. Coates, J. S. Moon,
D.Moses, M. Leclerc, K. Lee, A. J.Heeger, Nat. Photonics 2009, 3,
297.
[6] J. Yuan, Y. Zhang, L. Zhou, G. Zhang, H.-L. Yip, T.-K. Lau, X. Lu,
C.Zhu, H. Peng, P. A.Johnson, M. Leclerc, Y.Cao, J.Ulanski, Y.Li,
Y.Zou, Joule 2019, 3, 1140.
[7] Y. Lin, B. Adilbekova, Y. Firdaus, E. Yengel, H. Faber, M. Sajjad,
X. Zheng, E. Yarali, A. Seitkhan, O. Bakr, A. El-Labban,
U. Schwingenschlogl, V. Tung, I. McCulloch, F. Laquai,
T.Anthopoulos, Adv. Mater. 2019, 31, 1902965.
[8] K. Jiang, Q. Wei, J. Y. L. Lai, Z. Peng, H. K. Kim, J. Yuan, L. Ye,
H.Ade, Y.Zou, H.Yan, Joule 2019, 3, 3020.
[9] Y.Cui, H.Yao, J.Zhang, T.Zhang, Y.Wang, L.Hong, K.Xian, B.Xu,
S.Zhang, J.Peng, Z. Wei, F.Gao, J.Hou, Nat. Commun. 2019, 10,
2515.
[10] L.Zhan, S.Li, T.-K.Lau, Y.Cui, X.Lu, M.Shi, C.-Z.Li, H.Li, J.Hou,
H.Chen, Energy Environ. Sci. 2020, 13, 635.
[11] E. Wang, L. Hou, Z. Wang, S. Hellström, F. Zhang, O. Inganäs,
M. R.Andersson, Adv. Mater. 2010, 22, 5240.
Adv. Funct. Mater. 2020, 2005011

www.afm-journal.dewww.advancedsciencenews.com
© 2020 Wiley-VCH GmbH
2005011 (10 of 10)
[12] J.Peet, J. Y.Kim, N. E.Coates, W. L. Ma, D. Moses, A. J. Heeger,
G. C.Bazan, Nat. Mater. 2007, 6, 497.
[13] L.Ye, Y.Xiong, Q. Zhang, S. Li, C.Wang, Z.Jiang, J.Hou, W.You,
H.Ade, Adv. Mater. 2018, 30, 1705485.
[14] Y. Qin, L. Ye, S.Zhang, J. Zhu, B. Yang, H.Ade, J.Hou, J. Mater.
Chem. A 2018, 6, 4324.
[15] C. J.Brabec, S.Gowrisanker, J.Halls, D. Laird, S. Jia, S. Williams,
Adv. Mater. 2010, 22, 3839.
[16] Y. Zhu, A. Gadisa, Z. Peng, M. Ghasemi, L. Ye, Z. Xu, S. Zhao,
H.Ade, Adv. Energy Mater. 2019, 9, 1900376.
[17] W.Zhao, D. Qian, S.Zhang, S. Li, O.Inganas, F.Gao, J. Hou, Adv.
Mater. 2016, 28, 4734.
[18] H. Yao, Y. Cui, D. Qian, C. S. Ponseca Jr., A. Honarfar, Y. Xu,
J.Xin, Z.Chen, L. Hong, B. Gao, R. Yu, Y. Zu, W.Ma, P.Chabera,
T.Pullerits, A.Yartsev, F.Gao, J.Hou, J. Am. Chem. Soc. 2019, 141,
7743.
[19] J. Wang, J. Zhang, Y.Xiao, T. Xiao, R. Zhu, C. Yan, Y. Fu, G. Lu,
X.Lu, S. R.Marder, X.Zhan, J. Am. Chem. Soc. 2018, 140, 9140.
[20] Y.Wang, Y.Zhang, N.Qiu, H.Feng, H. Gao, B. Kan, Y.Ma, C.Li,
X.Wan, Y.Chen, Adv. Energy Mater. 2018, 8, 1702870.
[21] Z. Zheng, Q. Hu, S. Zhang, D. Zhang, J. Wang, S. Xie, R. Wang,
Y.Qin, W.Li, L. Hong, N. Liang, F.Liu, Y.Zhang, Z.Wei, Z.Tang,
T. P.Russell, J.Hou, H.Zhou, Adv. Mater. 2018, 30, 1801801.
[22] G. Zhang, J. Zhao, P. C. Y. Chow, K. Jiang, J. Zhang, Z. Zhu,
J.Zhang, F.Huang, H.Yan, Chem. Rev. 2018, 118, 3447.
[23] F.Padinger, R.Rittberger, N.Sariciftci, Adv. Funct. Mater. 2003, 13, 85.
[24] T.Agostinelli, S.Lilliu, J.Labram, M.Quiles, M.Hampton, E.Pires,
J. Rawle, O. Bikondoa, D. Bradley, T. Anthopoulos, J. Nelson,
J.Macdonald, Adv. Funct. Mater. 2011, 21, 1701.
[25] A.Virkar, S.Mannsfeld, Z.Bao, N.Stingelin, Adv. Mater. 2010, 22, 3857.
[26] J. B. C. Lindqvist, C. Feng, S. Gustafsson, O. Bäcke, N. Treat,
C. Bounioux, P. Henriksson, R. Kroon, E. Wang, A. Velasco,
P. Kristiansen, N. Stingelin, O. E, O. Inganäs, M. Andersson,
C.Müller, Adv. Energy Mater. 2014, 4, 1301437.
[27] L. Yu, D. Qian, S. Marina, F. Nugroho, A. Sharma, S. Hultmark,
A. Hofmann, R. Kroon, J. Benduhn, D. Smilgies, K. Vandewal,
M. Andersson, C. Langhammer, J.Martín, F. Gao, C. Müller, ACS
Appl. Mater. Interfaces 2019, 11, 21766.
[28] M.Ghasemi, H. Hu, Z. Peng, J.Rech, I.Angunawela, J. Carpenter,
S.Stuard, A. Wadsworth, l.McCulloch, W.You, H.Ade, Joule 2019,
3, 1328.
[29] S.Li, W.Liu, M.Shi, J.Mai, T.Lau, J.Wan, X. Lu, C. Li, H.Chen,
Energy Environ. Sci. 2016, 9, 604.
[30] Y.Lin, J.Wang, T.Li, Y.Wu, C.Wang, L.Han, Y.Yao, W.Ma, X.Zhan,
J. Mater. Chem. A 2016, 4, 1486.
[31] W. Ni, M. Li, B. Kan, F. Liu, X. Wan, Q. Zhang, H. Zhang,
T. P.Russell, Y.Chen, Chem. Commun. 2016, 52, 465.
[32] Y.Qin, Y.Chen, Y.Cui, S.Zhang, H.Yao, J.Huang, W.Li, Z.Zheng,
J.Hou, Adv. Mater. 2017, 29, 1606340.
[33] J.Hou, O.Inganäs, R.Friend, F.Gao, Nat. Mater. 2018, 17, 119.
[34] Y.Qin, S.Zhang, Y.Xu, L.Ye, Y.Wu, J.Kong, B.Xu, H.Yao, H.Ade,
J.Hou, Adv. Energy Mater. 2019, 9, 1901823.
[35] S. Root, M. Alkhadra, D. Rodriquez, A. Printz, D. Lipomi, Chem.
Mater. 2017, 29, 2646.
[36] B. A. Collins, J. E. Cochran, H. Yan, E. Gann, C. Hub, R. Fink,
C.Wang, T.Schuettfort, C. R.McNeill, M. L.Chabinyc, H.Ade, Nat.
Mater. 2012, 11, 536.
[37] N.Stribeck, X-Ray Scattering of Soft Matter, Springer, Berlin 2007.
[38] A.Hexemer, W.Bras, J.Glossinger, E.Schaible, E. Gann, R.Kirian,
A.MacDowell, M.Church, B.Rude, H.Padmore, J. Phys.: Conf. Ser.
2010, 247, 012007.
[39] E. Gann, A. T. Young, B. A. Collins, H. Yan, J. Nasiatka,
H. A. Padmore, H. Ade, A. Hexemer, C. Wang, Rev. Sci. Instrum.
2012, 83, 045110.
[40] X.Jiao, L.Ye, H.Ade, Adv. Energy Mater. 2017, 7, 1700084.
[41] B. A.Collins, Z.Li, J. R.Tumbleston, E.Gann, C. R.McNeill, H.Ade,
Adv. Energy Mater. 2013, 3, 65.
[42] J. H. Carpenter, A. Hunt, H. Ade, J. Electron Spectrosc. Relat.
Phenom. 2015, 200, 2.
[43] S. Swaraj, C. Wang, H. Yan, B. Watts, J. Lüning, C. R. McNeill,
H.Ade, Nano Lett. 2010, 10, 2863.
[44] C. Lindqvist, E. Wang, M. R. Andersson, C. Müller, Macromol.
Chem. Phys. 2014, 215, 530.
[45] L. Ye, W. Zhao, S. Li, S. Mukherjee, J. H. Carpenter, O.Awartani,
X.Jiao, J.Hou, H.Ade, Adv. Energy Mater. 2017, 7, 1602000.
[46] S.Mukherjee, X.Jiao, H.Ade, Adv. Energy Mater. 2016, 6, 1600699.
[47] L.Ye, S.Li, X.Liu, S.Zhang, M. Ghasemi, Y.Xiong, J.Hou, H.Ade,
Joule 2019, 3, 443.
[48] H. Hu, K. Jiang, P. Chow, L. Ye, G. Zhang, Z. Li, J. Carpenter,
H.Ade, H.Yan, Adv. Energy Mater. 2018, 8, 1701674.
[49] L. Ye, Y. Xiong, H.Yao, A. Gadisa, H. Zhang, S. Li, M. Ghasemi,
N. Balar, A. Hunt, B. T. O’Connor, J.Hou, H. Ade, Chem. Mater.
2016, 28, 7451.
[50] X. Du, T. Heumueller, W. Gruber, A. Classen, T. Unruh, N. Li,
C.Brabec, Joule 2019, 3, 215.
[51] S.Mukherjee, C. Proctor, G.Bazan, T.Nguyen, H. Ade, Adv. Energy
Mater. 2015, 5, 1500877.
[52] C.Xie, A.Classen, A.Späth, X.Tang, J.Min, M.Meyer, C.Zhang,
N.Li, A.Osvet, R. H.Fink, C. J.Brabec, Adv. Energy Mater. 2018, 8,
1702857.
[53] X.Xu, Z.Li, Z.Wang, K.Li, K.Feng, Q.Peng, Nano Energy 2016, 25,
170.
[54] W. L.Leong, S. R.Cowan, A. J.Heeger, Adv. Energy Mater. 2011, 1,
517.
[55] J.Wu, C.F, Y.Hsiao, F.Chien, P.Chen, M. H. C.Kuo, C.Hsu, ACS
Nano 2011, 5, 959.
[56] R.Yu, H.Yao, Y.Cui, L. Hong, C.He, J.Hou, Adv. Mater. 2019, 31,
1902302.
[57] O. Awartani, B. Lemanski, H. Ro, L. Richter, D. DeLongchamp,
B.O’Connor, Adv. Energy Mater. 2013, 3, 399.
[58] B.O’Connor, C.Haughn, K.An, K.Pipe, M.Shtein, Appl. Phys. Lett.
2008, 93, 223304.
[59] L. Ye, H. Hu, M. Ghasemi, T. Wang, B. Collins, J. Kim, K. Jiang,
J. Carpenter, H. Li, Z. Li, T. McAfee, J. Zhao, X. Chen, J. Lai,
T.Bredas, H.Yan, H.Ade, Nat. Mater. 2018, 17, 253.
[60] L.Ye, B.Collins, X.Jiao, J.Zhao, H.Yan, H.Ade, Adv. Energy Mater.
2018, 8, 1703058.
[61] J.Ilavsky, J. Appl. Crystallogr. 2012, 45, 324.
Adv. Funct. Mater. 2020, 2005011