1604603 (1 of 8)
Panchromatic Sequentially Cast Ternary Polymer Solar Cells
Masoud Ghasemi, Long Ye, Qianqian Zhang, Liang Yan, Joo-Hyun Kim, Omar Awartani,
Wei You, Abay Gadisa, and Harald Ade*
M. Ghasemi, Dr. L. Ye, Dr. J.-H. Kim, Dr. O. Awartani,
Dr. A. Gadisa, Prof. H. Ade
Department of Physics and ORaCEL
North Carolina State University
Raleigh, NC 27695, USA
E-mail: harald_ade@ncsu.edu
Q. Zhang, Dr. L. Yan, Prof. W. You
Department of Chemistry
University of North Carolina at Chapel Hill
Chapel Hill, NC 27599, USA
DOI: 10.1002/adma.201604603
morphological outcomes associated with the traditional ternary
blend. Our method gives rise to a favorable geometry and is
simpler than lamination, but yields similar layered structures.
While sequential casting has previously been used for the
fabrication of binary PSCs,
[29–34]
it has never been utilized for
casting stacks of different binary bulk heterojunction (BHJ)
films, let alone for materials that form mechanical alloys. Our
sequential casting approach has prevented mechanical alloying
and improved device performance by deliberately selecting a
difficult-to-dissolve binary BHJ with a crystalline donor as the
bottom layer.
We studied the ternary systems of a superb, high-efficiency
middle-bandgap polymer, poly(benzodithiophene-fluorinated
benzotriazole) (i.e., FTAZ),
[3,35–38]
and a promising and exten-
sively studied low-bandgap polymer, poly(diketopyrrolopyrrole-
terthiophene) (i.e., PDPP3T).
[39–41]
Although optically well
matched, these two polymers in a conventional ternary blend
with a fullerene molecule, [6]-phenyl C
71
butyric acid methyl
ester (PC
71
BM), only gave poor photovoltaic device performance
(Figure S1, Supporting Information). This is due to mechanical
alloying of the two donor polymers, as inferred from heat-flux
characteristics measured by differential scanning calorimetry
(DSC) as discussed below. Fortunately, we can overcome such
material-induced limitations by fabricating sequentially cast
ternary (SeCaT) solar cells based on these three components
(FTAZ, PDPP3T, and PC
71
BM). Schematics of conventional
and SeCaT solar cells, energy levels, and the chemical struc-
tures of the donor polymers and the electron-acceptor molecule
(PC
71
BM) employed in this study are shown in Figure 1a–d.
Taking advantage of the hard-to-dissolve, stable aggrega-
tion behavior of PDPP3T, we successfully fabricated SeCaT
devices with a vertical phase-segregated morphology. Specifi-
cally, the semicrystalline PDPP3T:PC
71
BM BHJ blend was first
spin-cast on top of a conducting substrate from warm solution
(100–110 °C). It provided a stable (i.e., solvent-resistant) bottom
layer onto which the FTAZ:PC
71
BM blend is cast from room-
temperature solution without the need for an orthogonal sol-
vent. Since holes move slightly better from PDPP3T to FTAZ
than in the other direction as shown via transport measure-
ments in bilayer, hole-only diodes (see Figure S2, Supporting
Information), an inverted device architecture is used.
Current-density–voltage (J–V) characteristics of the SeCaT,
conventional ternary, and corresponding binary single-layer
PSCs are shown in Figure 2a, and the respective photo-
voltaics parameters and thicknesses are summarized in
Table 1. To assure a fair and simple comparison and keeping
the extracting field relatively similar by having similar overall
thicknesses, the SeCaT films were fabricated in such a way that
the top high-performance, medium-bandgap FTAZ layer has
the same absorption coefficient as its corresponding binary
Most recent improvements in the power conversion efficiency
(PCE) of polymer solar cells (PSCs) have been attained through
a better understanding of the material structure-property rela-
tionships, which have resulted in the synthesis of well-designed
polymers and molecules with enhanced structural and elec-
tronic properties.
[1–5]
However, most of these high-performance
organic materials lack the ability to absorb a wide range of
photon energies. In line with the requirement of panchromatic
absorption for high current generation, low-bandgap mate-
rials,
[4,6,7]
ternary blends,
[8–10]
and tandem
[11–13]
device struc-
tures have been employed to address the optical limitations
of current state-of-the art materials. Of these, ternary PSCs
that consist of two donors and one acceptor (or two acceptors
and one donor) have been considered the simplest strategy
to broaden the optical absorption range in PSCs,
[14–16]
if the
selected donors and acceptors have complementary absorp-
tion. Unfortunately, even though a few breakthroughs have
been achieved,
[8,17,18]
many of the ternary devices are limited by
low fill factor (FF) and/or low short-circuit current (J
SC
) after
adding more than ≈15–20% of the third component.
[19–23]
Many
systems with excellent optical and electronic matches often fail
to deliver their promises.
[24,25]
Due to the complexity of mate-
rial interactions (e.g., miscibility/alloying) in ternary systems
and the lack of appropriate tools to accurately study or predict
these interactions, the most common practice for investigating
ternary systems has largely been based on trial and error. In
general, achieving favorable morphology is the limiting factor
even in binary systems,
[26–28]
and the complex and often unfa-
vorable morphology of the ternary device only exacerbates the
issue of morphology optimization.
Here, we utilize two donor polymers that have shown excel-
lent photovoltaic performance in binary systems with fullerene,
have ideal complementary absorption properties, but fail as a
conventional ternary device due to polymer–polymer mechan-
ical alloying on account of a negative Flory–Huggins interac-
tion parameter (
χ
) between these two polymers. This is the
first time that a polymer–polymer
χ
has been accurately meas-
ured for PSC materials.
[25]
We demonstrate a unique sequen-
tial deposition strategy that circumvents these detrimental
Adv. Mater. 2017, 29, 1604603
(2 of 8) 1604603
blend film (Figure 2b) with the thin film of PDPP3T:PC
71
BM
added for extra absorption. Indeed, the SeCaT devices (ZnO/
PDPP3T:PC
71
BM/FTAZ:PC
71
BM) show significantly improved
performance over the FTAZ reference cell and better overall
performance compared to either binary counterparts. This is
primarily due to excellent contributions of both donors to the
short-circuit current (J
SC
), as shown in the external quantum
efficiency (EQE) spectra (Figure 2b), moderated by an open-
circuit voltage (V
OC
) that is in between the V
OC
of the binary
single-layer devices. The EQE reveals the vibronic peaks of
PDPP3T and FTAZ, indicating that both donor polymers are
aggregated in the binary and SeCaT devices. Furthermore,
the photocurrent contribution of each polymer in SeCaT
devices nearly matches the photocurrent generated in the cor-
responding binary-blend solar cells (Figure 2c), indicating effi-
cient hole transfer from PDPP3T to FTAZ (for details see the
Supporting Information). The EQE of the SeCaT device in the
absorption region of the PDPP3T is particularly good, given
that the absorbance is only ≈60% that of the binary.
To correlate the effect of different processing conditions
with the device structures and materials interactions, resonant
soft-X-ray scattering (R-SoXS) was carried out to quantitatively
examine the lateral domain-size distribution within the active
layer.
[42]
In R-SoXS, tuning the incident photon energy makes
it possible to probe the material contrast between donor-rich
domains and acceptor-rich domains. The Lorentz-corrected
circular averaged R-SoXS profiles of the conventional ternary,
and SeCaT films are depicted in Figure 2c, supplemented
with data from ZnO/FTAZ:PC
71
BM/PDPP3T:PC
71
BM, and
binary reference devices. The R-SoXS profiles were acquired at
284.2 eV, which is an energy below the carbon K-edge, to opti-
mize the polymer-rich domains and the PC
71
BM-rich domains
contrast over the mass thickness contrast
[43,44]
and avoid radia-
tion damage.
[45]
Since PC
71
BM dominates the contrast func-
tion, R-SoXS primarily maps the spatial correlations between
the PC
71
BM-rich domains even in the ternary devices. Overall,
the SeCaT film with PDPP3T:PC
71
BM as the bottom (front)
BHJ layer exhibits domain spacing (28 nm) comparable to the
domain spacing of PDPP3T:PC
71
BM binary films (29 nm).
This indicates that the bottom PDPP3T:PC
71
BM is not much
disturbed by the subsequent casting of the FTAZ:PC
71
BM.
Domain spacings of conventional ternary (39–40 nm) and
ZnO/FTAZ:PC
71
BM/PDPP3T:PC
71
BM SeCaT (35 nm) indicate
only slightly larger phase separation in these films.
Dynamic secondary-ion mass spectroscopy (DSIMS) was
further used to characterize the vertical composition gradients
of the SeCaT and conventional ternary PSCs. The analysis and
details of the DSIMS spectra are described in the Supporting
Adv. Mater. 2017, 29, 1604603
Figure 1. a) Schematic of conventional ternary and SeCaT solar cells with ZnO as electron and MoO
3
as hole transport layers, creating an inverted
device architecture. Schematic of energy levels of b) conventional ternary and c) SeCaT with vertically segregated morphology, which provides suitable
pathways for electron and, more importantly, hole charge transport. d) The chemical structures of the donor polymers and the fullerene acceptor.
1604603 (3 of 8)
Information. As displayed in Figure 3a, a vertically segregated,
layered structure is achieved when FTAZ:PC
71
BM was cast on
top of the PDPP3T:PC
71
BM bottom layer. It is worth noting
that the presence of C
9
cluster ions is associated with PC
71
BM
domains and the DSIMS results of SeCaT PSCs reveal a uni-
form vertical distribution of PC
71
BM. Given that PCBM phase-
separates from either polymer, one can infer the presence of
continuous charge pathway for the photogenerated electrons
in the vertical direction throughout the device, which means
that after exciton dissociation at the D/A interface, the elec-
tron transport can happen in a network of PC
71
BM domains,
while holes can transfer from the PDPP3T to the FTAZ at the
intermixed polymer–polymer region of the SeCaT active layer.
Conventional ternary films with PDPP3T:FTAZ weight ratios
of 0.2:0.8, 0.8:0.2 (Figure S5, Supporting Information), and
0.5:0.5 (Figure 3b) were also characterized by DSIMS, with
Adv. Mater. 2017, 29, 1604603
Figure 2. a) J–V curves, b) UV–vis absorption spectra of SeCaT and binary PSCs, and c) external quantum efficiency (EQE). d) Lorentz-corrected
and thickness-normalized circular averaged resonant soft-X-ray scattering (R-SoXS) profiles of ZnO/FTAZ:PC
71
BM, ZnO/PDPP3T:PC
71
BM, ZnO/
PDPP3T:PC
71
BM/FTAZ:PC
71
BM, ZnO/FTAZ:PC
71
BM/PDPP3T:PC
71
BM, (PDPP3T:FTAZ = 0.2:0.8):PC
71
BM, and (PDPP3T:FTAZ = 0.8:0.2):PC
71
BM films,
at 284.2 eV.
Table 1. Device performance of the binary, conventional ternary, and SeCaT PSCs.
Device
V
OC
[V]
J
SC
[mA cm
−2
]
FF
[%]
PCE avg./max.
[%]
Thickness
[nm]
PDPP3T:PC
71
BM Binary 0.64 14.40 67.05 6.23/6.39 80
FTAZ:PC
71
BM Binary 0.75 10.39 73.82 5.78/5.86 75
(PDPP3T:FTAZ = 0.2:0.8):PC
71
BM
Conventional ternary 0.63 10.40 64.77 4.27/4.40 82
(PDPP3T:FTAZ = 0.5:0.5):PC
71
BM
Conventional ternary 0.63 11.29 66.73 4.78/4.81 78
(PDPP3T:FTAZ = 0.8:0.2):PC
71
BM
Conventional ternary 0.63 12.01 65.29 4.98/5.09 90
PDPP3T:PC
71
BM/FTAZ:PC
71
BM SeCaT 0.69 15.67 61.86 6.64/6.73 105
(4 of 8) 1604603
uniform depth profiles observed for all three films. For fur-
ther morphological clarification, grazing-incidence wide-angle
X-ray scattering (GI-WAXS) was employed. The 2D GI-WAXS
patterns and 1D profiles of binary, ternary blends, and SeCaT
films (Figure S6, Supporting Information) reveal relatively
well-defined scattering features similar to the PDPP3T:PC
71
BM
blend, indicating higher molecular ordering of PDPP3T com-
pared to FTAZ.
To understand the differences in the DSIMS profiles and
performance, we used DSC to determine the crystallinity and
polymer–polymer interaction parameter (
χ
). The latter pro-
vides important information associated with the miscibility of
polymer–polymer systems.
[46–48]
Based on the thermograms in
Figure 4a, PDPP3T is a semicrystalline polymer with a melting
point
m
0
()
T
of about 295 °C, which is consistent with previous
reports.
[49]
In contrast, FTAZ has no evidence of a melting tran-
sition and is noncrystalline. These results are consistent with
the molecular ordering or respective lack thereof found in
GI-WAXS. The depression of the melting-point temperature
(T
m
) for blended materials, using the Flory–Huggins approxi-
mation,
[50]
can be used to determine the molecular interaction
parameter (
χ
) of the polymers in the presence of a miscible
diluent according to:
[51]
TT
R
H
v
v
11
1
mm
0
f
2
1
2
2
χφ
()
−=−
∆
−
(1)
where the subscripts 1 and 2 identify the amorphous and sem-
icrystalline polymer, respectively; T
m
and
m
0
T
are the melting
points of the mixture and the pure semicrystalline polymer,
respectively; R is the ideal gas constant; v
1
and v
2
are the molar
volumes of the amorphous (v
1,FTAZ
= 903 cm
3
mol
−1
) and the
semicrystalline (v
2,PDPP3T
= 721 cm
3
mol
−1
) polymer; and
φ
is
the volume fraction. Utilizing
Bv RT/
1
χ
=
, which represents
the polymer–polymer interaction that is driven by enthalpy and
substitution into Equation (1), the data can be represented as
shown in Figure 4b, and
χ
= −0.56 can be extracted by a linear
fit. We note that the excellent fit achieved indicates that
χ
has
a negligible entropic component or a D/A ratio dependence.
The critical
χ
, i.e.,
χ
c
, above which phase separation can occur,
is generally positive. In the limit of infinite molecule weight,
χ
c
= 0. The negative interaction parameters observed here indi-
cates strong attractive interactions and the amorphous fractions
of the two polymers form a miscible and thermodynamically
stable mixture (see the Supporting Information for further
details on
χ
calculation).
χ
measurements in the field are
rare,
[52,53]
and this the first time such negative
χ
has been meas-
ured for semiconducting polymers. It means there is no driving
force for a polymer–polymer phase separation and the poly-
mers form a mechanical alloy in which the PC
71
BM is phase
separated. The DSC thermograms of the second heating and
cooling runs of the polymer–polymer and ternary blends repre-
sent similar melting depression and melting enthalpy trends to
those that were observed in the first heating run (see Figure S8,
Supporting Information).
In order to investigate the impact of mechanical alloying on
charge transport,
[54]
the space-charge-limited current method in
a diode configuration was used to measure the hole mobility
(see Figure S8, Supporting Information). For FTAZ concen-
trations that form an alloy, the mobility is lowered by about a
factor of 2. This is consistent with the bilayer diode results that
showed that hole hopping is asymmetric, with holes moving
more easily from PDPP3T to FTAZ than from FTAZ to PDPP3T.
Ideally, the hole from the PDPP3T hops only once along the
highest occupied molecular orbital (HOMO) energy cascade
(see Figure 1) to the FTAZ and a hole from the PC
71
BM hops
once or at most twice and then remains within the FTAZ-rich
phase until it reaches the electrode. Alas, in an FTAZ:PDPP3T
alloy, a hole might be forced to hop back onto the lower energy
HOMO of the PDPP3T, hop via a longer distance to the next
FTAZ, or explore longer and more tortuous FTAZ pathways.
Consequently, a miscible phase, i.e., a mechanical polymer
alloy, is detrimental here for hole transport. This reduces the
performance in the conventional ternary configuration of this
system, which could have been ideal when only considering
the matched optical and electronic properties. We suspect that
transport would be even more impacted if the hopping is more
asymmetric in other material pairs that have a larger HOMO
offset.
To verify the existence of a mechanically alloyed phase in
conventional ternary blends, the ternary films were character-
ized using DSC (Figure S9, Supporting Information). As shown
Adv. Mater. 2017, 29, 1604603
Figure 3. a,b) DSIMS depth profiles of SeCaT devices comprising PDPP3T:PC
71
BM bottom layer and FTAZ:PC
71
BM top layer (a) and conventional
ternary with 0.5:0.5 wt% of PDPP3T:FTAZ (b), with species monitored as indicated. The top surface is at t = 0 s.
1604603 (5 of 8)
in Figure 4c,d, similar trends of the melting point depression
and melting enthalpy of PDPP3T are reflected in the DSC ther-
mograms. The melting-point depression offset of PDPP3T in
the ternary blends compared to polymer–polymer blends is due
to the additional interaction of PDPP3T with PC
71
BM. The sim-
ilar melting enthalpy of PDPP3T in polymer–polymer and ter-
nary blends (Figure 4d) indicates that the degree of crystallinity
of PDPP3T is primarily only affected by the presence of FTAZ,
not by the PC
71
BM. This indicates that PDPP3T has a much
higher
χ
with the fullerene than with FTAZ. Furthermore, the
melting-transition signature disappears in all systems as the
FTAZ content reaches about 50%. The similarities in melting
enthalpy and melting-point depression between the polymer
binary and ternary samples indicate similar and dominant
polymer–polymer interactions in all blends. Such a high degree
of attractive interaction between the two polymers indeed leads
to unfavorable polymer-alloy morphologies that might con-
tribute to limiting the device performance in the conventional
ternary devices.
Since FTAZ is not a semicrystalline polymer, an FTAZ-
based bottom BHJ film is thus more unstable when casting
a second layer on top in a SeCaT device. This leads to inter-
diffusion and mixing, which is enhanced by the negative
χ
during the period the film is plasticized or dissolved, resulting
in DSIMS depth profiles and performance (see Figure S4 and
S5, Supporting Information) that are very similar to the con-
ventional ternary system. Open questions remain, though. For
example, how much the negative
χ
contributes to this interdif-
fusion and alloying and thus poor performances is currently
unclear and requires further research. Conceptually, a high
χ
that would be deep in the two-phase region should be sig-
nificantly preferable and would lead to more stable “bilayers.”
How common mechanical alloying is in other ternary sys-
tems or even polymer–polymer binary systems is unclear, as
this aspect has not been extensively studied previously.
[54]
Our
results indicate that
χ
measurements in general, and DSC
measurements in particular, could be a useful screening tool
in ternary PSC research. We note that estimates of
χ
from solu-
bility parameters (
χ
=
α
/kT(
δ
1
−
δ
2
)
2
+ 0.34, where
α
is volume
of one lattice segment,
δ
1
and
δ
2
are Hildebrand solubility
parameters of components 1 and 2, respectively, and kT is the
thermal energy),
[55]
would completely fail here as the method
by definition only yields a positive
χ
. Even if Hansen solubility
parameters are used, a negative
χ
is not possible within that
framework. Such failure has been previously shown to occur
when strong hydrogen bonds and polar interaction are pre-
sent.
[56]
Our results and the likely presence of strong directional
forces reinforces prior conclusions that the use of Hansen solu-
bility parameters for PSC applications might be unreliable.
[25,57]
Our results show that the bottom layer needs to be robust
and retain sufficient integrity during casting of the second layer.
Such a stable bottom layer can be provided by materials that
are semicrystalline and are typically cast from a hot solvent. In
this context, it is worth mentioning that there are a number of
Adv. Mater. 2017, 29, 1604603
(a)
3.0
2.5
2.0
1.5
1.0
Heat Flow (W/g)
320300280260240220200180
T (°C)
T
M
(°C) = 295.6
T
M
(°C) = 294.4
T
M
(°C) = 293.6
T
M
(°C) = 290.7
T
M
(°C) = 287.8
T
M
(°C) = 281.2
T
M
(°C) = N/A
T
M
(°C) = N/A
PDPP3T:FTAZ
(100:0)
(85:15)
(80:20)
(70:30)
(60:40)
(50:50)
(40:60)
(0:100)
Exo Up
(b)
(c)
(d)
30
25
20
15
10
5
∆Η
f
(J/g)
0.50.40.30.20.10.0
φ
1
polymer-polymer
ternary
90x10
-6
80
70
60
50
40
30
[
(1/T
M
-1/T
M
0
)/
φ
1
]
(K
-1
)
900x10
-6
800700600500400300
φ
1
/T
m
(K
-1
)
300
295
290
285
280
275
T
M
(°C)
0.50.40.30.20.10.0
φ
1
polymer-polymer
ternary
Figure 4. a) The DSC traces (10 °C min
−1
) of the first run of PDPP3T:FTAZ blends, b)
(1
/1
/)
/
mm
0
1
TT
φ
− against /
1m
T
φ
for PDPP3T:FTAZ blends, c) melting
point, and d) melting enthalpy of PDPP3T in ternary and polymer–polymer blend as a function of volume fraction of FTAZ (
1
φ
).
