Synthesis, characterization and photovoltaic performance of novel glass-forming perylenediimide derivatives

Article abstract of DOI:10.1016/j.orgel.2016.04.025

Novel glass-forming perylene-3,4,9,10-tetracarboxylic diimide (PDI) derivatives were synthesized from 1,7-dibromoperylene-3,4,9,10-tetracarboxylic diimides by nucleophilic substitution with a glass-forming mexylaminotriazine unit, characterized, and their

Full text of DOI:10.1016/j.orgel.2016.04.025

Organic Electronics 34 (2016) 146e156
Contents lists available at ScienceDirect
Organic Electronics
journal homepage: www.elsevier.com/locate/orgel
Synthesis, characterization and photovoltaic performance of novel
glass-forming perylenediimide derivatives
,
, **
Tham Adhikari a, Zahra Ghoshouni Rahami b, Jean-Michel Nunzi c *, Olivier Lebel b
a Department of Chemistry, Queen's University, Kingston, K7L 3N6, Canada
b Department of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, ON, K7K 7B4, Canada
c Department of Chemistry, Department of Physics, Engineering Physics and Astronomy, Queen's University, Kingston, K7L 3N6, Canada
a r t i c l e i n f o
a b s t r a c t
Article history:
Novel glass-forming perylene-3,4,9,10-tetracarboxylic diimide (PDI) derivatives were synthesized from
1,7-dibromoperylene-3,4,9,10-tetracarboxylic diimides by nucleophilic substitution with a glass-forming
mexylaminotriazine unit, characterized, and their performance as electron acceptors in bulk hetero-
junction photovoltaic cells with P3HT as donor were studied. Imide groups (octyl and 2,6-
diisopropylphenyl) and bay substituents (bromo or pyrrolidinyl) were used to study the impact of ste-
rics and electronics on device performance. The HOMO and LUMO levels of the materials were deter-
mined using cyclic voltammetry. The morphology and packing behavior of molecules in films of blends
with P3HT were studied using Atomic Force Microscopy and X-ray diffraction, and in all four cases, the
PDI derivatives remain in the amorphous phase, while the P3HT portion of the blends crystallizes. The
devices gave photovoltaic performances ranging from 0.2 to 0.6%, and while the bay substituents showed
negligible impact on device performance, switching from bulky 2,6-diisopropylphenyl imide groups to
linear octyl chains improved the efficiency of the devices by 36%, current density by 66% and fill factor by
16%. The performances observed for devices incorporating these PDI glasses are comparable to that of
devices with crystalline PDI acceptors.
Received 11 December 2015
Received in revised form
9 March 2016
Accepted 13 April 2016
Available online 22 April 2016
Keywords:
Photovoltaic cells
Bulk heterojunction
Molecular glass
Fullerene-free acceptor
Perylenediimide
Aggregation
© 2016 Elsevier B.V. All rights reserved.
1. Introduction
commercialization. Currently, organic PV cells have been demon-
strated to exhibit power conversion efficiency above 9% [5,6]. The
Due to the availability of carbon feedstocks and flexible organic
synthetic methodologies, organic materials have become extremely
attractive for photovoltaic applications [1e3]. Depending on the
solubility of organic materials, organic PV cells can be fabricated
either by vacuum or by printing technologies such as spin-coating,
drop-casting, doctor blading, inkjet printing, screen printing and
roll-to-roll printing. Tang et al. reported in 1986 a p-n hetero-
junction device with a power conversion efficiency of 1% using
copper phthalocyanine and a perylene-3,4,9,10-tetracarboxyl-bis-
benzimidazole (PTCBI) electron donor and acceptor, respectively
[4]. In recent years, great progress has been achieved in the
development of solution-processed bulk heterojunction PV cells,
and the technology has reached the early stages of
high efficiency of bulk heterojunctions is due to exciton dissocia-
tion at the interface between the electron donor and acceptor
materials in the intimately mixed system and followed by efficient
charge separation and transport to the electrodes by diffusion of
charges in the interpenetrating donor and acceptor network,
thereby facilitating an efficient charge separation due to large
interfacial area [7e10]. So far, various organic materials have been
developed as electron donor or p-type materials, which have
improved properties such as high absorption, hole mobility, and
adequately tuned HOMO and LUMO levels [11e13]. However, the
development of electron acceptors or n-type materials is far behind
that of donor materials. The dominant acceptors in bulk hetero-
junction PV cells are fullerene derivatives due to their high electron
mobility, large electron affinity, and capacity to pack efficiently to
form a percolated tridimensional supramolecular network. How-
ever, fullerenes absorb light only weakly in the visible region and
play only a limited role in the light harvesting process. In addition,
their high synthesis costs and limited tunability of energy levels
hinder the commercial success of fullerene-based organic PV cells
* Corresponding author.
** Corresponding author.
E-mail addresses: nunzijm@queensu.ca (J.-M. Nunzi), Olivier.Lebel@rmc.ca
(O. Lebel).
http://dx.doi.org/10.1016/j.orgel.2016.04.025
1566-1199/© 2016 Elsevier B.V. All rights reserved.

T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
147
[9,10,14]. As the development of optimal electron acceptor mate-
rials is equally important to that of donor materials for the
improvement of device performances, it is crucial to develop
optimal non-fullerene acceptors for organic PV cells with improved
performances [15,16].
the chromophore with minimal perturbation. This approach was
demonstrated with a series of azobenzene dyes, with which glass-
forming adducts were generated using a simple, efficient and high-
yielding procedure [39,40].
In the present work, four glass-forming PDI derivatives con-
taining mexylaminotriazine groups in the bay position were syn-
thesized, characterized, and incorporated into organic PV cells with
P3HT as donor material. The PDI glasses synthesized feature two
different imide groups, 2,6-diisopropylphenyl and n-octyl, and two
different bay substituents, bromo and pyrrolidinyl, to study the
impact of steric bulk and electronics on device performance. All
four acceptor materials investigated showed the ability to sponta-
neously form stable glasses that did not crystallize upon heating or
operation of the devices, and all materials showed photovoltaic
activity, with device efficiencies with P3HT as donor ranging from
0.2 to 0.6%. These values are comparable to most values previously
reported for PV cells using P3HT and PDI derivatives containing a
single perylene moiety [10,41], and are reasonably close to the
highest values reached so far for such systems (close to 1%) [21,42].
Interestingly, the performances of the devices has been found to be
independent on the electronics of the acceptor materials, with both
bromo and pyrrolidinyl series giving similar results. On the other
hand, the octyl imide series showed efficiency 3 times greater than
their sterically hindered 2,6-diisopropylphenyl imide analogues.
This is believed to be due to the 2,6-diisopropylphenyl groups
shielding the perylene core and preventing efficient stacking of the
perylene moieties, thereby impeding electron transport. While the
P3HT portion of the bulk heterojunction is semi-crystalline, the PDI
domains remain amorphous, and the results reported herein show
that the electron transport over long range required for current
generation is possible in the amorphous state and yields perfor-
mances comparable to that of crystalline materials.
3,4,9,10-Perylenetetracarboxylic diimide (PDI) derivatives are
promising candidates as acceptors in organic photovoltaics. Not
only do PDI derivatives display high electron mobilities [17], but
they also possess high molar absorption coefficients in the visible
range [18]. Moreover, they can be synthesized in large scale in
relatively economic fashion [10]. The LUMO of parent 3,4,9,0-
perylenetetracarboxylic diimide is similar to that of PCBM [19].
Their optoelectronic properties can be easily tuned by tailoring the
substituents on the imide groups or on the bay positions
[14,20e22]. In addition, the low-lying HOMO levels of PDI de-
rivatives can facilitate the hole transfer process from the PDI mol-
ecules to the donor material after absorption of light [10]. All these
properties make PDI derivatives an extremely attractive class of
materials for use in organic photovoltaics as electron acceptor
materials.
As the poly-aromatic core in PDI derivatives is planar, as
opposed to spherical for fullerene derivatives, the performance of
the devices is expected to depend on proper packing of the per-
ylene moieties so that the
p-orbitals can interact with each other
[10]. A common problem with small molecules is their tendency to
rapidly crystallize [23,24], which limits their usefulness in appli-
cations involving thin films by yielding films of poor quality, with
polycrystalline topologies constituted of small domains separated
by grain boundaries that limit the long-range transport of electrons
by acting as electron traps [25]. Annealing conditions must thus be
screened thoroughly to find the conditions that will give optimal
film morphologies and limit grain boundaries.
An alternative approach consists in designing PDI derivatives (or
electron acceptor materials in general) that form thin films that
remain amorphous [26,27]. While amorphous solids possess no
long-range periodical order, intermolecular interactions, such as
2. Experimental
2.1. Materials
the
p-stacking of aromatic moieties, can nonetheless be present
and regulate to a certain degree the organization at the molecular
2-Methylamino-4-mexylamino-6-(4-mercaptophenyl)amino-
level, with
boundaries.
a
continuous material that is exempt of grain
1,3,5-triazine_(1)
[37],
N,N0-bis(2,6-diisopropylphenyl)-2,7-
dibromo-3,4,9,0-perylenetetracarboxylic diimide (2a) [43], and
N,N0-dioctyl-2,7-dibromo-3,4,9,0-perylenetetracarboxylic diimide
(2b) [44], were prepared according to the literature. Pyrrolidine
was purchased from Sigma-Aldrich, and all other solvents and re-
agents were purchased from Caledon Labs. All reagents were used
without further purification. Regioregular poly (3-hexylthiophene-
2,5-diyl (P3HT) used as electron donor was purchased from Rieke
Metals. Molybdenum oxide used as hole transport layer was pur-
chased from Alfa Aesar. ZnO as electron transporting material was
synthesized by the sol-gel method reported in the literature [45].
The Indium tin oxide (ITO) glasses used as substrates were pur-
chased from Luminescence Technology Corporation, which has ITO
As with the majority of small molecules, most PDI derivatives
readily crystallize [26,27]. However, it was demonstrated that
strategic molecular design can hinder the process of crystallization,
resulting in molecular materials that can remain in the amorphous
state for extended (sometimes indefinite) periods of time. This class
of materials are known as molecular glasses, or amorphous mo-
lecular materials, and combine the monodisperse nature of small
molecules with the processability of polymers [28e30]. As typi-
cally, compounds designed to form glasses possess bulky, non-
planar and irregular shapes, and generally interact weakly with
neighbouring molecules [31], designing PDI derivatives that can
form glasses despite their planar perylene core and that can
interact reliably by pi-stacking in the amorphous state is expected
to be challenging. However, approaches where glass-forming de-
rivatives of chromophores are synthesized by functionalization
with peripheral groups have been reported in the literature
[32e35]. Mexylaminotriazines are one such family of glass-forming
materials that usually show outstanding glass-forming ability, high
kinetic stability towards crystallization, and glass transition tem-
perature (Tg) values that can be modulated by tuning their mo-
lecular structures [36e38]. Glass-forming mexylaminotriazine
derivatives can bear functional groups that can be bonded in a
covalent fashion to chromophores, thereby yielding adducts that
can remain indefinitely in the amorphous state at operating tem-
peratures, while preserving the optical and electronic properties of
film thickness about 135 15 nm, and sheet resistance is 15 Osqꢀ1
.
Glass transition temperatures were determined with a TA In-
struments 2010 Differential Scanning Calorimeter calibrated with
indium at a heating rate of 5 ꢁC/min from 30 to 250 ꢁC. Values were
reported as the half-height of the heat capacity change averaged
over two heating runs. For compounds 3a and 4a, values were
averaged over heating runs on two different samples because of
thermal degradation. FTIR spectra were acquired with thin films
cast from CH2Cl2 on KBr windows using a Perkin-Elmer Spectrum
65 spectrometer. UVevisible absorption spectra were acquired
using a Hewlett-Packard 8453 spectrometer or a Olid® HP8452
Diode Array Spectrometer. NMR spectra were acquired on either a
400 MHz Bruker AV400 or on a 300 MHz Varian Oxford spec-
trometer. Luminescence was measured with a USB2000-Ocean

148
T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
Optics spectrometer. XRD were obtained on P3HT/PDI films using
an Xpert Pro Philips powder X-ray diffractometer.
residue was redissolved in minimal AcOEt, insolubles were
removed by filtration, and the filtrate was poured into hexanes. The
precipitated product was collected by filtration, washed with
hexanes and allowed to dry at 60 ꢁC in an oven to yield 0.485 g pure
green compound 4a (0.429 mmol, 98%). Tg 186 ꢁC; IR (CH2Cl2/KBr)
3915, 3846, 3821, 3739, 3689, 3659, 3621, 3551, 3349, 3193, 3070,
2961, 2930, 2915, 2868, 1700, 1667, 1606, 1593, 1579, 1556, 1496,
1423, 1403, 1381, 1363, 1336,1290, 1239, 1201, 1181, 1152, 1132, 1119,
1057, 1014, 965, 940, 917, 838, 809, 795, 773, 753, 739, 714, 691, 668,
2.2. Synthesis
2.2.1. Synthesis of compound 3a
A mixture of tetrahydrofuran (800 mL) and K2CO3 (11.1 g,
80.0 mmol) in a round-bottomed flask was stirred and deoxygen-
ated under nitrogen atmosphere for 30 min. Next, compound 2a
(14.0 g, 16.0 mmol) and glass 1 (7.05 g, 20.0 mmol) were added and
the reaction mixture was stirred at room temperature for 3 days.
THF was removed under reduced pressure and the crude product
was purified on a short silica pad eluting with 100% dichloro-
methane initially to remove unreacted staring material. After col-
lecting the fractions containing the starting material, the eluent
was switched to DCM/EtOAc (80:20). 12.6 g of pure dark red
product was obtained (11.1 mmol, 69%). Tg 206 ꢁC; FTIR (CH2Cl2/
KBr) 3420, 3346, 3195, 3102, 3065, 3027, 2964, 2930, 2869, 1709,
1669, 1621, 1586, 1559, 1538, 1518, 1497, 1457, 1444, 1430, 1413,
1388,1363,1335,1306,1242,1198,1181,1148,1094,1057,1040,1013,
995, 969, 937, 919, 885, 859, 838, 809, 793, 770, 749, 741, 714, 698,
601, 582, 537 cmꢀ1; 1H NMR (400 MHz, DMSO-d6, 363.15 K):
d 8.99
(d, 3J ¼ 8.6 Hz,1H), 8.96 (br s, 1H), 8.65 (d, 3J ¼ 8.1 Hz, 1H), 8.59 (s,
1H), 8.53 (d, 3J ¼ 8.6 Hz, 2H), 8.40 (s, 1H), 7.88 (d, 3J ¼ 8.6 Hz, 2H),
7.65 (d, 3J ¼ 8.11 Hz,1H), 7.47 (m, 2H), 7.41 (d [3],J ¼ 8.6 Hz, 2H), 7.35
(d, 3J ¼ 7.8 Hz, 2H), 7.31 (s, 2H), 7.29 (d, 3J ¼ 7.8 Hz, 2H), 6.61 (br s,
1H), 6.53 (s, 1H), 3.36 (br s, 4H), 2.85 (d, 3J ¼ 4.3 Hz, 3H), 2.78 (m,
3J ¼ 6.8 Hz, 2H), 2.68 (m, 3J ¼ 6.8 Hz, 2H), 2.14 (s, 6H), 2.06 (br s, 4H),
1.13 (d, 3J ¼ 6.8 Hz, 6H), 1.11 (d, 3J ¼ 6.8 Hz, 6H), 1.08 (d, 3J ¼ 6.8 Hz,
6H), 1.04 (d, 3J ¼ 6.8 Hz, 6H); 13C NMR (75 MHz, DMSO-d6): 165.9,
163.9, 163.6, 163.48, 163.41, 163.04, 162.8, 148.2, 145.4, 145.2, 141.9,
139.8, 137, 135.7, 134.6, 133.5, 131.4, 130.9, 130.8, 130.2, 129.3, 129.1,
128.4,126.1, 125.5, 124.2, 123.7, 123.1, 122.2, 121.9, 121.7, 121.5, 120.5,
119.9, 117.7, 116.9, 114.1, 52.6, 28.5, 27.2, 25.3, 20.9 ppm; UVeVis
(CH2Cl2): lmax (ε) 435 nm (13 700), 664 nm (22 400); HRMS
(MALDI, MHþ) calcd. for C70H68N9O4S (m/e): 1130.5115, found:
1130.5124.
668 cmꢀ1 1H NMR (400 MHz, DMSO-d6, 363 K):
; d 9.64 (d,
3J ¼ 8.3 Hz,1H), 9.04 (br s,1H), 8.89 (m, 2H), 8.84 (d, 3J ¼ 8.3 Hz,1H),
8.70 (d, 3J ¼ 8.1 Hz, 1H), 8.56 (br s, 1H), 8.38 (br s, 1H), 7.93 (d,
3J ¼ 8.6 Hz, 2H), 7.47 (m, 4H), 7.36 (d, 3J ¼ 7.6 Hz, 2H), 7.32 (s, 2H),
7.30 (d, 3J ¼ 7.8 Hz, 2H), 6.63 (br s, 1H), 6.57 (s, 1H), 2.85 (s, 3H), 2.80
(m, 3J ¼ 6.8 Hz, 2H), 2.71 (m, 3J ¼ 6.8 Hz, 2H), 2.18 (s, 6H), 1.12 (d,
3J ¼ 6.8 Hz, 6H), 1.11 (d, 3J ¼ 6.8 Hz, 6H), 1.08 (d, 3J ¼ 6.8 Hz, 6H), 1.04
2.2.4. Synthesis of compound 4b
Compound 4b was synthesized from precursor 2b and glass 1
following a procedure similar to that used for analogue 4a. The
crude product was purified on a short silica pad using AcOEt/
Hexanes 1:1 as eluent. Yield: 82%; Tg 86 ꢁC; FTIR (CH2Cl2/KBr) 3333,
3181, 2954, 2923, 2852, 1692, 1653, 1590, 1579, 1554, 1500, 1431,
1415, 1397, 1346, 1334, 1286, 1238, 1178, 1127, 1099, 1032, 863, 837,
(d, 3J ¼ 6.8 Hz, 6H) ppm; 13C NMR (75 MHz, DMSO-d6):
d 166.4,
164.5, 164.4, 164.1, 163.2, 163.2, 162.6, 145.9, 145.8, 143.3, 141.4,
140.2, 137.6, 137.5, 135.9, 133.2, 132.9, 132.2, 130.9, 129.8, 129.4,
129.2, 128.7, 128.4, 127.5, 126.0, 124.3, 123.7, 122.8, 122.6, 122.2,
122.0, 121.1, 120.5, 118.3, 28.9, 27.7, 24.2, 21.5 ppm; UVeVis
(CH2Cl2): lmax (ε) 548 nm (24 200); HRMS (MALDI, MHþ) calcd. for
806, 749, 708, 658 cmꢀ1 1H NMR (400 MHz, CDCl3):
; d 8.49 (br m,
C
66H60BrN8O4S (m/e): 1139.3642, found: 1139.3649.
1H), 8.46 (d, 3J ¼ 7.8 Hz, 1H), 8.38 (s, 1H), 8.34 (d, 3J ¼ 7.8 Hz, 1H),
8.15 (br m, 1H), 7.55 (br s, 3H), 7.34 (d, 3J ¼ 7.8 Hz, 1H), 7.25 (br s,
2H), 7.10 (br s, 2H), 6.99 (br s, 1H), 6.55 (s, 1H), 5.36 (br s, 1H), 4.12
(br m, 2H), 4.00 (br m, 2H), 3.61 (br s, 1H), 2.76 (s, 3H), 2.67 (br s,
1H), 2.14 (s, 6H), 1.99 (br s, 1H), 1.89 (br s, 1H), 1.67 (br m, 2H), 1.58
(br m, 2H), 1.20 (m, 20H), 0.80 (m, 6H) ppm; 13C NMR (75 MHz,
2.2.2. Synthesis of compound 3b
Compound 3b was synthesized from precursor 2b and glass 1
following a procedure similar to that used for analogue 3a. The
crude product was purified on a short silica pad using AcOEt/
Hexanes 1:1 as eluent. Yield: 68%; Tg 91 ꢁC; FTIR (CH2Cl2/KBr) 3341,
3187, 3085, 2954, 2925, 2869, 2853, 1698, 1659, 1588, 1553, 1518,
1498, 1433, 1415, 1395, 1349, 1333, 1303, 1238, 1178, 1097, 1036, 833,
CDCl3):
d 166.5, 164.1, 163.9, 163.8, 163.6, 163.2, 148.1, 140.7, 138.3,
136.4, 135.6, 135.1, 134.2, 133.0, 132.7, 132.0, 131.8, 129.4, 129.3,
128.8, 128.4, 127.1, 125.8, 125.3, 124.8, 123.9, 122.8, 122.0, 121.2,
120.7, 120.6, 120.4, 118.3, 115.8, 52.6, 40.7, 31.8, 29.4, 29.2, 28.1, 27.7,
27.2, 25.8, 22.6, 21.4, 14.1 ppm; UVeVis (CH2Cl2): lmax (ε) 433 nm
(11 400), 656 nm (15 800); HRMS (MALDI, MHþ) calcd. for
806, 745, 693, 654 cmꢀ1 1H NMR (400 MHz, CDCl3):
; d 9.46 (d,
3J ¼ 7.8 Hz, 1H), 8.87 (s, 1H), 8.73 (m, 1H), 8.65 (d, 3J ¼ 7.8 Hz, 1H),
8.57 (m, 1H), 8.29 (m, 1H), 7.68 (br s, 2H), 7.37 (m, 2H), 7.37 (br s,
1H), 7.16 (s, 1H), 7.06 (s, 2H), 6.64 (s, 1H), 5.27 (br s, 1H), 4.20 (br m,
2H), 4.07 (br m, 2H), 2.97 (s, 3H), 2.23 (s, 6H), 1.76 (br m, 2H), 1.65
(br m, 2H), 1.30 (m, 20H), 0.90 (m, 6H) ppm; 13C NMR (75 MHz,
C62H68N9O4S (m/e): 1034.5109, found: 1034.5135.
2.2.5. Sample one-pot procedure: synthesis of compound 4a
In a round-bottomed flask equipped with a magnetic stirrer, a
suspension of K2CO3 (1.59 g, 11.5 mmol) in THF (50 mL) was
degassed by sparging with nitrogen for 15 min. N,N0-Bis(2,6-
diisopropylphenyl)-2,7-dibromo-3,4,9,0-perylenetetracarboxylic
diimide 2a (2.00 g, 2.30 mmol) and 2-methylamino-4-
mexylamino-6-(4-mercaptophenyl)amino-1,3,5-triazine 1 (1.01 g,
2.88 mmol) were added simultaneously, and the mixture was
stirred 48 h at ambient temperature under nitrogen atmosphere, at
which point pyrrolidine (5 mL) was added and stirring was
continued another 48 h. The volatiles were evaporated under vac-
uum while maintaining the water bath temperature below 50 ꢁC,
the residue was redissolved in CH2Cl2 and purified by filtration on a
short silica pad with AcOEt/hexanes (3:7 then 7:3) to yield, after
complete drying, 1.66 g compound 4a (1.47 mmol, 64%). Spectral
data was identical to that obtained for compound 4a prepared from
pure compound 3a.
CDCl3):
d 166.2, 166.1, 163.8, 162.9, 162.8, 162.5, 162.3, 141.5, 141.2,
140.5, 139.0, 138.1, 137.6, 135.6, 132.8, 132.6, 132.4, 131.9, 130.9,
130.0, 129.8, 129.1, 128.7, 128.3, 127.9, 126.6, 125.2, 124.6, 123.8,
123.5, 123.2, 122.6, 122.0, 121.6, 121.3, 120.7, 120.0, 117.9, 40.7, 31.8,
29.3, 28.0, 27.6, 27.2, 22.7, 21.4, 14.1 ppm; UVeVis (CH2Cl2): lmax (ε)
548 nm (17,600); HRMS (MALDI, MHþ) calcd. for C58H60BrN8O4S
(m/e): 1045.3620, found: 1045.3662.
2.2.3. Synthesis of compound 4a
In a round-bottomed flask equipped with a magnetic stirrer, a
mixture of pyrrolidine (2.5 mL), THF (10 mL) and K2CO3 (0.078 g,
0.50 mmol) was degassed by sparging with nitrogen for 15 min.
Compound 3a (0.500 g, 0.440 mmol) was added, then the reaction
mixture was stirred at ambient temperature for 24 h under nitro-
gen atmosphere. The volatiles were removed under reduced pres-
sure while keeping the water bath temperature under 50 ꢁC. The

T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
149
2.3. Cyclic voltammetry
Cyclic voltammograms were recorded using an EG&G Model
263 potentiostat connected with three electrodes: working elec-
trode (glassy carbon-carbon), reference electrode (Ag/AgCl (in 3 M
NaCl) and counter or auxiliary electrode (Pt) in a 0.1 M solution of
tetrabutylammonium hexafluorophosphate in acetonitrile at a scan
rate of 100 mVꢀ1 with N2 gas bubbling. Ferrocene was used as in-
ternal standard.
2.4. Atomic force microscopy
The morphology of nanocomposite P3HT/PDI films was charac-
terized using an Ambios multimode Atomic force Microscope (AFM)
in tapping mode with 300 KHz resonant frequency cantilever.
2.5. Device fabrication and characterization
The ITO substrates were either patterned with 0.1 M HCl and
zinc dust for 10 min and thoroughly washed with water, or pre-
patterned as purchased. The patterned ITO substrates were
cleaned consecutively in an ultrasonic bath with detergent powder,
distilled water, acetone and isopropyl alcohol for 10 min each and
finally dried with air. The substrates were heated at approximately
100 ꢁC for 5 min. The dried substrates were further cleaned with a
plasma cleaner for 15 min. ZnO precursor gel was spin-coated at a
rate of 3000 rpm for 40 s under ambient conditions. The glass
substrates coated with ZnO were transferred on a hotplate and
baked at 150 ꢁC for 30 min. The thickness of ZnO was approximately
30 nm. The blend solutions of P3HT and PDI glasses 3e4 were
prepared in dichlorobenzene in a fixed ratio (1:1,1:2 and 2:1) by
weight and stirred for 24 h at 50 ꢁC. The solution was filtered
Fig. 1. Device structure and energy level alignment of device components.
through a 0.45 mm pore sized poly(tetrafluoroethylene) (PTFE) fil-
ter. The active layer was spin-coated at 1500 rpm and annealed at
110 ꢁC for 10 min. The thickness of the active layer was measured to
range from 150 to 180 nm with a Sloan Dektak II profilometer. The
top and bottom electrodes of the devices were cleaned with a
cotton stick soaked with chloroform to make contacts. Finally, 5 nm
of MoO3 and 100 nm of Ag were deposited with deposition rates of
0.5 A/s and 1.5 A/s, respectively, on all the devices using a physical
vapour deposition system (PVD) under high vacuum at a pressure
of 1 ꢂ 10ꢀ6 mBar. The active area of the device was 0.2 cm2. The
device structure and energy alignment is shown in Fig. 1. Current
densityevoltage (JeV) measurements were carried out using a
Keithley 4200-SCS in the dark and under illumination. All the
photovoltaic parameters of the inverted cells were measured at
ambient conditions using a Xenon light with an intensity of
100 mW/cm2 calibrated with an AM 1.5 solar simulator [46].
Mobility measurements were performed with diodes of com-
pounds 3e4 that were prepared in a similar fashion, using solutions
of pure PDI glasses 3e4 instead of P3HT/PDI-glass blends. The
electron mobility for each diode was calculated using the Mott-
Gurney space-charge-limited current (SCLC) equation [49].
solubility in most organic solvents, and they can be further func-
tionalized by substitution of the bromo groups at the 2-and 7-
positions. Among possible nucleophiles, thiolates and secondary
amines are the most appealing to functionalize precursors 2a-b,
because the reactions can be carried out under mild conditions, with
roughly stoichiometric amounts of reagent, and with some control
over the degree of substitution on the PDI. As a glass with a thiol
group has already been reported by our group (glass 1) and can be
easily synthesized in high yield from simple precursors, it consti-
tuted the most elegant solution for the synthesis of glass-forming
PDI derivatives. Furthermore, PDI derivatives with amino bay sub-
stituents undergo a significant shift in their absorption, while sul-
fide substituents impact the electronic properties of the conjugated
system in a more subtle fashion, thereby allowing a higher
tunability through the introduction of additional substituents.
Dibromo precursors 2a-b could be substituted only at the 2-
position of the PDI core with thiol-functionalized glass 1 in the
presence of K2CO3 with a slight excess of glass 1 in THF at ambient
temperature to give monoadducts 3a-b as major products, along
with small quantities of unreacted precursors 2a-b and the corre-
sponding bis-adducts, which are only sparingly soluble in most
organic solvents (and therefore not promising candidates for
solution-processed PV cells). Both products could be easily purified
by filtration on a short silica pad with AcOEt/CH2Cl2 (1:4) for
compound 3a, or with AcOEt/Hexanes (1:1) for compound 3b, to
give after drying pure products 3a-b in 69% and 68% yield,
respectively (Scheme 1).1
3. Results and discussion
3.1. Synthesis of PDI glasses 3e4
Initial attempts to functionalize the imide groups of the PDI core
with glass-inducing mexylaminotriazine moieties were unsuc-
cessful, as the products showed very low solubility in any solvent.
For these reasons, N,N0-Bis(2,6-diisopropylphenyl)-2,7-dibromo-
3,4,9,0- perylenetetracarboxylic diimide 2a and N,N0-dioctyl-2,7-
dibromo-3,4,9,0-perylenetetracarboxylic diimide 2b were selected
as precursors because they are known to show appreciable
1
Marvin was used for drawing, displaying and characterizing chemical struc-
tures, substructures and reactions, Marvin 15.1.12, 2015, ChemAxon (www.
chemaxon.com).

150
T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
Scheme 1. Synthesis of PDI glasses 3e4.
As compounds 3a-b are still substituted with a bromo group,
units previously reported, compounds 3e4 readily form glassy
solids upon drying from solution that remain amorphous over time.
Differential Scanning Calorimetry (DSC) confirmed that compounds
other nucleophiles can be introduced at the 7-position of the per-
ylene ring system to tune its electronics, and its absorption range.
An amino substituent was selected because these strongly electron-
donating groups are known to result into strongly red-shifted ab-
sorption maxima (more than 100 nm). These two series of acceptor
materials can harvest different portions of the visible range to
complement absorption from donor materials. Analogues 4a-b
substituted with a pyrrolidinyl group were synthesized from
compounds 3a-b by reaction with pyrrolidine in THF at ambient
temperature in 82e98% yields (Scheme 1). Alternatively, com-
pounds 4a-b can also be synthesized directly from precursors 2a-b
in a one-pot procedure where both bromo groups are displaced
sequentially with glass 1 and pyrrolidine, followed by a single pu-
rification step by chromatography.
3e4
indeed
show
glass
transitions.
Bulky
bis(2,6-
diisopropylphenyl) imides 3a and 4a show very high Tg values
ranging from 186 ꢁC to 206 ꢁC (Table 1), while more flexible dioctyl
imides 3b and 4b show Tg values between 86 and 91 ꢁC (sample
DSC curves for all four compounds are shown in Fig. S1, Supporting
Information).
The structural features that impacts Tg the most in this series are
Table 1
Glass transition temperature (Tg) values for PDI-
functionalized glasses 3e4.
Compound
Tg (ꢁC)
3a
3b
4a
4b
206
91
196
86
3.2. Physical properties of PDI glasses 3e4
As with other dyes functionalized with mexylaminotriazine

T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
151
No crystallization was observed upon heating for compounds
3e4, but in the case of glasses 3a and 4a, heating the samples past
their Tg resulted in partial decomposition, which was confirmed by
1H NMR spectroscopy. The DSC measurements thus had to be
performed on different samples to confirm the Tg values, as they
decreased steadily upon repeated heating cycles. Compounds 3b
and 4b could be safely heated up to 200 ꢁC, and did not show any
signs of crystallization upon heating. While the absence of unde-
sirable crystallization on heating could not be verified reliably for
compounds 3a and 4a, they undergo glass transition at tempera-
tures so high that crystallization will not occur under standard
operating conditions. To confirm this, thin films of glasses 3e4 were
prepared by spin coating from CH2Cl2 solution, and annealed at
110 ꢁC for 10 min. The films were then studied by atomic force
microscopy (AFM) (Fig. S2) and X-ray diffraction (XRD) (Fig. S3),
and compared to films before annealing. In all cases, spin-coating
yielded amorphous and topologically smooth films, and annealing
did not result in any changes in either the phase or the topology of
the films.
As with all PDI derivatives, glasses 3e4 absorb strongly in the
visible range. UVevisible absorption spectra of compounds 3e4
were thus recorded in CH2Cl2 solution (Fig. 2a). Absorption maxima
and extinction coefficients are reported in Table 2. It can be
observed that the presence of the mexylaminotriazine moieties
bonded to the PDI through a sulfide group results in a slight red
shift, while amino-substituted analogues 4a-b instead shows two
absorption bands, one near 430 nm, and a larger one near 660 nm,
which is characteristic of PDI dyes with amino bay substituents.
Unlike precursors 2a-b, however, no significant fluorescence could
be observed from compounds 3e4 in solution, possibly a result of
self-quenching by the triazine group.
The UVevisible spectra of thin films of glasses 3e4 prepared by
spin coating from CH2Cl2 solution were recorded (Fig. 2b). Their
absorption maxima are listed in Table 2. The absorption bands for
monoadducts 3a and 4a underwent a slight red shift, which is likely
a consequence of a higher-polarity environment relative to solu-
tion. Compounds 3b and 4b, on the other hand, showed larger
absorption shifts, from 548 to 564 nm for compound 3b, and from
433 to 443 nm for compound 4b. Moreover, the absorption bands of
compounds 3b-4b showed a more important broadening between
solution and solid state than those of compounds 3a-4a.
Fig. 2. UVeVisible spectra of PDI glass derivatives 3e4. a) 50
thin solid films spin-coated from CH2Cl2.
mM solution in CH2Cl2; b)
the nature of the imide groups. Derivatives 3a-4a with rigid and
bulky 2,6-diisopropylphenyl groups showed Tg values over 100 ꢁC
higher than their analogues with flexible octyl chains 3b-4b, which
is likely attributable to the available degrees of freedom of linear
alkyl chains, considering that the rest of the molecule is relatively
rigid: the perylene ring system is fused, and triazine amino groups
are strongly conjugated. As a result, compound 3a, for instance,
possesses few unhindered rotational degrees of freedom, making it
difficult to generate free volume through molecular motion.
Pyrrolidinyl-substituted glasses 4a and 4b also undergo glass
transition at slightly lower temperatures than their bromo-
substituted precursors 3a-b, because of the higher flexibility of
the pyrrolidine ring.
These observations hint towards more efficient formation of p-
stacked aggregates in the solid state. Compounds 3a-4a possess
extremely bulky 2,6-diisopropylphenyl imide substituents, which
are perpendicular to the PDI ring system because of the isopropyl
groups in the ortho positions. As a result, the PDI core is shielded
from interacting with the PDI moieties from other molecules. On
the other hand, linear octyl chains generate less steric hindrance
around the PDI group, leading to more efficient p-stacking. Again,
no noticeable fluorescence was observed on thin films of dyes 3e4.
The reduction potentials of compounds 3e4 were measured by
cyclic voltammetry. 0.01 M solutions of PDI glasses 3e4 and 0.1 M
n-Bu4NPF6 were prepared in acetonitrile. The electrolytic cell was
previously cleaned with acetonitrile, half filled with 0.1 M n-
Table 2
Absorption bands and HOMO-LUMO energy gaps for PDI-functionalized glasses 3e4 in CH2Cl2 solution (50
m
M) and as solid thin films.
lmax (film) (nm)
Compound
lmax (CH2Cl2) (nm)
ε (CH2Cl2) (cmꢀ1 Mꢀ1
)
EHOMO-LUMO (eV)a
3a
3b
4a
548
548
435
664
433
656
24,200
17,600
13,700
22,400
11,400
15,800
554
564
436
668
443
658
1.96
1.80
1.63
1.61
4b
a
HOMO-LUMO gaps were calculated from EHOMO-LUMO ¼ 1240/lonset
.

152
T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
Table 3
derivatives 3a-b also showed mobilities neraly ten times higher
than pyrrolidinyl analogues 4a-b, despite the latter showing
smaller HOMO-LUMO gaps. It is likely that the larger pyrrolidinyl
groups also negatively impact interactions between the PDI groups
in the solid state, leading to less efficient electron transport.
Reduction potentials and HOMO and LUMO energy levels for compounds 3e4.
LUMO energy levels were measured by cyclic voltammetry, HOMO-LUMO gaps were
measured by UVevisible spectroscopy, and LUMO energy levels were calculated
from the HOMO levels and the energy gaps.
Compound
Ered (V)
EHOMO (eV)a
ELUMO (eV)b
EHOMO-LUMO (eV)c
1/2
3a
3b
4a
4b
ꢀ0.32
ꢀ0.40
ꢀ0.50
ꢀ0.46
ꢀ5.88
ꢀ5.78
ꢀ5.43
ꢀ5.31
ꢀ3.92
ꢀ3.98
ꢀ3.80
ꢀ3.76
1.96
1.80
1.63
1.61
3.3. Deposition and characterization of thin films of P3HT blends
with PDI glasses
Films of P3HT blends with PDI glasses 3e4 were cast in a 1:1
ratio from dichlorobenzene and annealed at 110 ꢁC for 10 min. The
normalized UVeVisible absorption spectra of the thin film blends
of P3HT with PDI glasses 3e4 are shown in Fig. 3 along with the
spectra of both individual components. It can be observed that for
bromo derivatives 3a-b, the contribution of the PDI acceptor to the
absorption of the blend in the visible range is negligible, because of
the overlap between the absorption bands of both components. In
the case of pyrrolidinyl derivatives 4a-b, additional shoulders cor-
responding to the absorption maxima of the PDI derivatives can be
seen in the absorption band of the blend. This was more pro-
nounced for compound 4a, where an additional band near 600 nm
was observed.
a
Calculated from the LUMO levels and the HOMO-LUMO gaps.
Calculated from the reduction potentials measured by cyclic voltammetry with
b
ELUMO ¼ Ered1/2-E1/2(ferrocene) þ 4.8.
c
Taken from Table 2.
Bu4NPF6 and outgassed with bubbling N2. A small amount of
ferrocene was added to it as internal standard and stirred. The
starting potential was higher than the reference electrode potential
to ensure that the species of interest was completely oxidized. The
solution was scanned at a rate of 100 mVꢀ1. Cyclic voltammograms
are shown in Fig. S4 (Supporting Information). In all cases, a single
reversible reduction signal was observed, with compound 3b
showing an additional small oxidation signal. The values for the
reduction potentials are listed in Table 3. From the reduction po-
tentials, the LUMO energy levels for compounds 3e4 could be
calculated by comparison to ferrocene (ꢀ4.8 eV) (Table 3). From the
LUMO energy levels and the values for the HOMO-LUMO gaps
measured by UVevisible spectroscopy, the HOMO levels could be
calculated, and are listed in Table 3. HOMO and LUMO energy levels
for compounds 3e4 are in agreement with published values for PDI
derivatives substituted in the bay area: the HOMO and LUMO en-
ergy levels for 2,7-dibromo PDI derivatives are close to ꢀ6.1
and ꢀ3.9 eV, respectively, while for more electron-rich 2,7-
bis(pyrrolidinyl) analogues, these values are close to ꢀ5.2
and ꢀ3.5 eV [47]. The impact of the imide groups was minimal,
with bis(octyl) derivatives showing slightly higher LUMO levels and
slightly lower HOMO-LUMO gaps, perhaps because of the forma-
tion of aggregates in solution. In comparison, the HOMO and LUMO
levels of PC61BM are ꢀ6.1 and ꢀ3.7 eV [19], which is close to the
LUMO energy levels of pyrrolidinyl derivatives 4a-b.
Diodes incorporating compounds 3e4 were fabricated with the
configuration ITO/ZnO/PDI-glass/MoO3/Ag to measure the electron
mobility of PDI glasses 3e4. Current-voltage curves were generated
for each PDI glass derivative, and are shown in Fig. S5 (Supporting
Information). Electron mobility was calculated using the Mott-
Gurney space-charge-limited current (SCLC) equation [49], and
values are reported in Table 4 (details about the calculations can be
found in Supporting Information and additional data is listed in
Table S1). From Table 4, it can be observed that PDI glasses 3e4
show relatively low mobility values in the 10ꢀ8 to 10ꢀ6 range,
possibly a consequence to the lack of long-range order associated
with the amorphous state. Octyl-functionalized derivatives 3b and
4b were found to display mobilities one order of magnitude higher
than their 2,6-diisopropylphenyl analogues 3a-4a, reinforcing the
The surface morphology of the blend films was studied by AFM.
Fig. 4 shows the surface topographic images for blends of P3HT
with PDI-glass derivatives 3e4 in a 1:1 ratio after annealing at
110 ꢁC for 10 min. The topographic images of the blends of P3HT
with compounds 3e4 all show relatively smooth surfaces with root
mean square roughness ranging from 26 nm for compound 3b to
41 nm for compound 4a. High surface roughness is typically asso-
ciated with the presence of large aggregated features that decrease
exciton dissociation and charge collection in the device and result
in lower photocurrent and efficiency [20]. The surface of blends of
P3HT with dioctyl imide derivatives 3b and 4b are relatively
smoother with root mean square roughness roughly 10 nm lower
than their bis(2,6-diisopropylphenyl) analogues 3a and 4a. PDI
derivatives with pyrrolidinyl bay substituents 4a-b also show
higher roughness compared to their bromo-substituted analogues
3a-b, which is likely due to the higher steric hindrance of the
pyrrolidinyl groups that impacts the crystallization of P3HT in the
blend, resulting in the presence of larger aggregates. Considering
surface topography, octyl derivatives 3b-4b are expected to show
better device performance than analogues 3a-4a due to more
optimal facilitation of charges in the devices. Furthermore, despite
the fact that the LUMO level of compound 4b is closer to that of
P3HT than its bromo analogue 3b, and compound 4b contributes
more to the overall absorption of the blend, compound 3b exhibits
a smoother surface, and thus should show higher efficiency.
The crystallinity and packing behavior of the P3HT: PDI-glass
blends in a 1:1 ratio were further studied by XRD. The XRD re-
sults for films of blends of P3HT with PDI glasses 3e4 after
annealing at 110 ꢁC for 10 min are shown in Fig. 5. Diffractograms of
the P3HT blends with all four PDI glasses are similar. One promi-
nent diffraction peak was observed for all blends at angle
(2W ¼ 5e7ꢁ) showing the first order structure of P3HT with a
d spacing of 1.7 nm. The pristine films of P3HT: PDI blends exhibit
weak signals as compared to annealed samples. The weak signals
observed for pristine films are due to the low-order structure of
P3HT. After annealing, the diffraction peak is increased in all four-
blend films but more pronounced in films of P3HT with bromo
derivatives 3a-b. Interestingly, in the blends with compound 3a-b,
the P3HT crystalline peak was more intense than in the blend with
pyrrolidinyl analogues 4a-4b, hinting that compound 3b, while
remaining amorphous, promotes crystallization of P3HT in the
blend. The slight increase of peak intensity for P3HT blends with
4a-b as compared to 3a-b suggests little modification in the
hypothesis that the less hindered octyl chains lead to stronger
stacking between perylene moieties. Surprisingly, bromo
p-
Table 4
Electron mobility values for PDI-functionalized glasses 3e4.
Compound
Mobility (cm2 Vꢀ1 sꢀ1
)
3a
3b
4a
4b
1.29 ꢂ 10ꢀ7
4.00 ꢂ 10ꢀ6
6.34 ꢂ 10ꢀ8
3.30 ꢂ 10ꢀ7

T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
153
Fig. 3. UVevis absorption spectra of films of blends of P3HT with PDI glasses 3e4. (a)
3a, (b) 3b, (c) 4a, and (d) 4b.
Fig. 4. AFM topographic images (5
mm ꢂ 5
m
m) for blends of P3HT with PDI glasses
3e4 after annealing at 110 ꢁC for 10 min; (a) 3a, (b) 4a, (c) 3b, and (d) 4b.

154
T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
intimate blends constituted of small domains, rather than fully
segregated biphasic structures.
3.4. Device fabrication and performance
Photovoltaic cells with inverted configurations were fabricated
with blends of P3HT with PDI glasses 3e4 with P3HT: PDI-glass
compositions of 1:2, 1:1 and 2:1 by mass. The blends were
deposited by spin coating from dichlorobenzene solution on ITO
substrates coated with 30 nm of ZnO. After annealing at 110 ꢁC for
10 min, which was determined to be the optimal annealing time,
the active layers were then coated with a layer of MoO3 and an Ag
electrode. Photovoltaic measurements were performed in ambient
condition using light from a Xenon source calibrated with an AM
1.5 solar simulator to reproduce an incident light intensity of
100 mWcmꢀ2. The measurements were carried out at ambient at-
mosphere and the currentevoltage curves for PDI glasses 3e4 are
shown in Fig. 6; characteristics in the dark are given for comparison
in Fig. S6. The power conversion efficiency (PCE) was calculated
from the value of open-circuit voltage (Voc), short-circuit current
density (Jsc) and fill factor (FF) with the following equation:
PCE ¼ Pout/Pin ¼ Jsc ꢂ Voc ꢂ FF/Pin, where Pin represents the power of
incident light (mW/cm2). The fill factor was calculated from the
values of Voc, Jsc, and maximum power (Pmax) according to the
following equation: FF ¼ Pmax/Jsc ꢂ Voc ¼ Jmax ꢂ Vmax/Jsc ꢂ Voc
,
where Jmax and Vmax are the current density and voltage at
maximum power respectively. The photovoltaic parameters of de-
vices fabricated with PDI glasses 3e4 with varying donor-acceptor
ratios are reported in Table 5 (standard deviations are listed in
Table S1).
Devices fabricated with bis(2,6-diisopropylphenyl) imide de-
rivatives 3a-4a showed comparable efficiencies, which increased as
the P3HT fraction increased, to reach maxima of 0.23% and 0.21%
respectively for ratios of 2:1. Further increasing the P3HT:PDI glass
ratio did not result in an improvement of the efficiency. Jsc and FF
values were also higher for devices built with glass 3a, while Voc
was higher for glass 4a, possibly a consequence of the higher LUMO
level of compound 4a. This behavior is consistent with observations
reported in the literature [14]. On the other hand, devices fabricated
with dioctylimide analogues 3b and 4b showed power conversion
efficiencies three times higher, with maximal observed efficiencies
of 0.66% and 0.63%, respectively. Current density was also increased
by values ranging from 66% to 150%, while fill factors were
increased by 16%e19%. Interestingly, for the octyl derivatives,
higher efficiencies were observed with increasing the fraction of
PDI relative to P3HT, which is a trend opposite to that observed for
analogues 3a-4a. The imide groups obviously play an important
role in the performance of the materials as electron acceptors. As
suggested previously, while bulky 2,6-diisopropylphenyl groups
shield the perylene core from individual molecules, the linear octyl
Fig. 5. XRD for the blend films of P3HT with PDI derivatives (a) 3a and 4a (b) 3b and 4b
spin-coated in dichlorobenzene solution. Films were annealed at 110 ꢁC for 10 min.
packing of PDI glass molecules upon improved phase separation. It
is therefore suggested that the pyrolidinyl groups in compounds
4a-b hinder the crystallization of P3HT.
The electron acceptor behavior of PDI glasses 3e4 was further
studied by photoluminescence quenching, which is a characteristic
behavior of acceptor materials in photovoltaic applications [22].
Photoluminescence quenching for representative PDI glasses 3a
and 3b were studied by the deposition of blends of compound 3a
with highly luminescent polymer MDMO-PPV [Poly(2-methoxy-5-
(3070-dimethyloctyloxy)-1,4-phenylene-vinylene], and of com-
pound 3b with P3HT. The photoluminescence (PL) spectra of the
blends were then recorded under a 360 nm LED excitation source.
The PL spectra are depicted in Fig. S5 (Supporting Information). The
emission spectrum of pure MDMO-PPV shows a broad and intense
band in the wavelength range 400e550 nm (Fig. S5a), while P3HT
shows a band in the range 600e750 nm (Fig. S5b). As the con-
centration of PDI-glass is increased, the intensity of PL decreases for
both polymers, and is totally quenched at PDI-glass concentrations
above 8 wt% for MDMO-PPV, while a 10 wt% concentration of PDI-
glass resulted in a twofold decrease in the PL intensity of P3HT. The
quenching of polymer luminescence by PDI glasses 3a and 3b is
attributed to electron transfer from the photo-excited donor poly-
mer to the acceptor PDI molecules, as observed by Nikitenko et al.
in the study of photoluminescence quenching of MDMO-PPV by
fullerene derivatives [48]. The strong PL quenching by the PDI
glasses suggests that in both cases, the two components exist as
chains lead to more optimal
pep stacking of the PDI moieties,
thereby improving electron transfer between acceptor molecules.
For both the 2,6-diisopropylphenyl and octyl series, efficiencies
were comparable between bromo-substituted derivatives 3a-b and
pyrrolidinyl derivatives 4a-b. While pyrrolidinyl analogues 4a-b
better contributed to the overall absorption of the blends and
possess LUMO levels closer to the LUMO level of P3HT, bromo de-
rivatives 3a-b gave films with smoother topographies that are a
result of smaller crystalline P3HT domains. Compound 3b also
favored the crystallization of P3HT, as observed by XRD. The mo-
lecular properties of pyrrolidinyl derivatives 4a-b are counter-
balanced by the superior supramolecular properties of bromo-
substituted compounds 3a-b, leading to similar performances.
Analysis of the current voltage characteristics reveals that incor-
poration of PDIs 3b and 4b into P3HT cells does not degrade the

T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
155
Table 5
Photovoltaic parameters of ITO/ZnO/P3HT:PDI-glass/MoO3/Ag bulk-hetero junction
devices with PDI glasses 3e4 under illumination with an intensity of 100 mWcmꢀ2
.
PDI glass
P3HT:PDI
Jsc (mAcmꢀ2
)
Voc (V)
FF (%)
PCE (%)
3a
2:1
1:1
1:2
2:1
1:1
1:2
2:1
1:1
1:2
2:1
1:1
1:2
1.49
0.74
0.69
1.53
2.23
2.48
0.9
0.5
0.5
1.24
2.00
2.33
0.44
0.42
0.42
0.60
0.62
0.62
0.60
0.65
0.55
0.64
0.64
0.64
34.2
45.0
36.9
40.7
43.3
43.0
37
34
29
44
43.4
42.6
0.23
0.14
0.11
0.37
0.60
0.66
0.21
0.10
0.08
0.34
0.55
0.63
3b
4a
4b
electrical characteristics of the blend; in particular, the series and
shunt resistances that are given by the inverse of the slopes of the
curves at zero and positive bias, respectively are independent of the
molar fraction of PDI-glass. That is opposite to the situation with 3a
and 4a, which reveals that octyl-substituted PDIs do not add charge
traps or recombination centers into the cells [49]. Interestingly, the
fact that the acceptor domains are completely amorphous within
the devices does not prevent the molecules to pack in a way that
allows electron transfer over long range, leading to photovoltaic
efficiencies that are in the upper range of reported values for
photovoltaic cells with P3HT and similar PDI derivatives. In com-
parison, reported efficiencies of PV cells with bis(3-pentyl) PDI and
a 1,4,5,8-tetraalkyl analogue have yielded efficiencies of 0.25% and
0.5%, respectively [10,41], while a PDI derivative with acenaph-
thenyl bay substituents gave a PCE of 0.96% [21]. In fact, all reported
instances of PDI derivatives giving significantly higher efficiencies
with P3HT as donor contain two or more PDI moieties bridged by an
aromatic linker, rather than a single PDI unit [50e53].
4. Conclusions
Four glass-forming perylenediimide derivatives containing
mexylaminotriazine groups in the bay position were synthesized,
characterized, and incorporated as electron acceptor materials in
solution-processed organic photovoltaic cells with P3HT as a donor.
It was found that while devices with all four materials showed
photovoltaic behavior, bulky groups at the imide position hindered
the
pep stacking necessary for electron transport. On the other
hand, the bay substituents, used to tune the absorption in the
visible range and the HOMO and LUMO levels, showed little impact
on device performances. Even though the acceptor materials were
completely amorphous in the devices, proper packing of the mol-
ecules allowing descent electron transport could be attained,
leading to devices with power conversion efficiencies reaching
0.66%, which are comparable to performances of other per-
ylenediimide derivatives and P3HT and to earlier results using
fullerene derivatives. Further optimization of the devices will be
undertaken in the future and is expected to yield improved
performance.
Acknowledgements
We acknowledge the Canadian Defence Academic Research
Programme (CDARP) of RMC and the Discovery Grants Program of
Natural Sciences and Engineering Research Council (NSERC)
(NSERC #327116) for supporting this research. We would also like
ꢀ
ꢀ
to thank Dr. Rene Gagnon (Universite de Sherbrooke) for mass
spectroscopy analyses.
Fig. 6. Current-voltage curves of ITO/ZnO/P3HT: PDI-glass/MoO3/Ag with PDI glass (a)
3a, (b) 3b, (c) 4a, and (d) 4b under illumination with an intensity of 100 mW cmꢀ2
.

156
T. Adhikari et al. / Organic Electronics 34 (2016) 146e156
glass-forming bay-substituted perylenediimide derivatives, J. Phys. Chem. B
Appendix A. Supplementary data
114 (2010) 1782e1789.
[28] P. Strohriegl, J.V. Grazulevicius, Charge-transporting molecular glasses, Adv.
Mater. 14 (2002) 1439e1452.
[29] Y. Shirota, H. Kageyama, Charge carrier transporting molecular materials and
their applications in devices, Chem. Rev. 107 (2007) 953e1010.
[30] R. Lygaitis, V. Getautis, J.V. Grazulevicius, Hole-transporting hydrazones,
Chem. Soc. Rev. 37 (2008) 770e788.
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.orgel.2016.04.025.
References
[31] Y. Shirota, Organic materials for electronic and optoelectronic devices,
J. Mater. Chem. 10 (2000) 1e25.
[32] J. Nakano, T. Takahashi, T. Kadota, Y. Shirota, Formation of a surface Relief
grating using a novel azobenzene-based photochromic amorphous molecular
material, Adv. Mater. 14 (2002) 1157e1160.
[33] E. Ishow, B. Lebon, Y. He, X. Wang, L. Bouteiller, L. Galmiche, K. Nakatani,
Structural and photoisomerization cross studies of polar photochromic
monomeric glasses forming surface relief gratings, Chem. Mater. 18 (2006)
1261e1267.
[1] F.C. Krebs, Polymer Photovoltaics : a Practical Approach, vol. 1e4, SPE Press,
Bellingham: Washington, USA, 2008.
[2] P. Kamat, Meeting the clean energy demand: nanostructure architectures for
Solar energy conversion, J. Phys. Chem. C 111 (2007) 2834e2860.
[3] C. Li, H. Wonneberger, Perylene Imides for Organic Photovoltaics: Yesterday,
Today, and Tomorrow, Adv. Mater. 24 (2012) 613e636.
[4] C.W. Tang, Two layer organic photovoltaic cell, Appl. Phys. Lett. 48 (1986)
183e185.
[5] Y. Liu, J. Zhao, Z. Li, C. Mu, W. Ma, H. Hu, K. Jiang, H. Lin, H. Ade, H. Yan, Ag-
gregation and morphology control enables multiple cases of high-efficiency
polymer solar cells, Nat. Commun. 5 (2014) 5293, http://dx.doi.org/10.1038/
ncomms6293.
[6] S.-H. Liao, H.-J. Jhuo, P.-N. Yeh, Y.-S. Cheng, Y.-L. Li, Y.-H. Lee, S. Sharma, S.-
A. Chen, Single junction inverted polymer solar cell reaching power conver-
sion efficiency 10.31% by employing dual-doped zinc oxide nano-film as
[34] D. Nagahama, H. Nakano, Y. Shirota, Synthesis and photochromic response of
a new photochromic amorphous molecular material based on spirooxazine,
Photopolym. Sci. Technol. 21 (2008) 755e757.
[35] K. Traskovskis, I. Mihailovs, A. Tokmakovs, A. Jurgis, V. Kokars, M. Rutkis,
Triphenyl moieties as building blocks for obtaining molecular glasses with
nonlinear optical activity, J. Mater. Chem. 22 (2012) 11268e11276.
[36] J.D. Wuest, O. Lebel, Anarchy in the solid state: structural dependence on
glass-forming ability in triazine-based molecular glasses, Tetrahedron 65
(2009) 7393e7402.
cathode interlayer, Sci. Rep.
srep06813.
4 (2014) 6813, http://dx.doi.org/10.1038/
[7] Z. He, C. Zhong, S. Su, M. Xu, H. Wu, Y. Cao, Enhanced power-conversion ef-
ficiency in polymer solar cells using an inverted device structure, Nat. Pho-
tonics 6 (2012) 591e595.
[37] R.N. Eren, A. Plante, A. Meunier, A. Laventure, Y. Huang, J.G. Briard, K.J. Creber,
C. Pellerin, A. Soldera, O. Lebel, One ring to rule them all: effect of aryl sub-
stitution on glass-forming ability in mexylaminotriazine molecular glasses,
Tetrahedron 68 (2012) 10130e10144.
[38] A. Laventure, A. Soldera, C. Pellerin, O. Lebel, Heads vs. tails: a double-sided
study of the influence of substituents on the glass-forming ability and sta-
bility of aminotriazine molecular glasses, New J. Chem. 37 (2013) 3881e3889.
[8] L. Dou, J. You, Z. Hong, Z. Xu, G. Li, R.A. Street, Y. Yang, 25th anniversary article:
a decade of organic/polymeric photovoltaic research, Adv. Mater. 25 (2013)
6642e6671.
€
[9] Z. Kis¸ nis¸ ci, O.F. Yüksel, M. Kus¸ , Optical properties of perylene-monoimide
(PMI) and perylene-diimide (PDI) organic semiconductor thin films, Synth.
Met. 194 (2014) 193e197.
[10] V. Kamm, G. Battagliarin, I.A. Howard, W. Pisula, A. Mavrinskiy, C. Li,
K. Müllen, F. Laquai, Polythiophene: perylene diimide solar cells e the impact
of alkyl-substitution on the photovoltaic performance, Adv. Energ. Mater. 1
(2011) 297e302.
[39] R. Kirby, R.G. Sabat, J.-M. Nunzi, O. Lebel, Disperse and disordered:
a
mexylaminotriazine-substituted azobenzene derivative with superior glass
and surface relief grating formation, J. Mater. Chem. C 2 (2014) 841e847.
[40] O.R. Bennani, T.A. Al-Hujran, J.-M. Nunzi, R.G. Sabat, O. Lebel, Surface relief
grating growth in thin films of mexylaminotriazine-functionalized glass-
forming azobenzene derivatives, New J. Chem. 39 (2015) 9162e9170.
[41] X. Guo, L. Bu, Y. Zhao, Z. Xie, Y. Geng, L. Wang, Controlled phase separation for
efficient energy conversion in dye/polymer blend bulk heterojunction
photovoltaic cells, Thin Solid Films 517 (2009) 4654e4657.
[11] D. Gendron, M. Leclerc, New conjugated polymers for plastic solar cells, En-
ergy Environ. Sci. 4 (2011) 1225e1237.
[12] J. Roncali, P. Leriche, P. Blanchard, Molecular materials for organic photovol-
taics: small is beautiful, Adv. Mater. 26 (2014) 3821e3838.
[13] V. Malytskyi, J.-J. Simon, L. Patrone, J.-M. Raimundo, Thiophene-based
pushepull chromophores for small molecule organic solar cells (SMOSCs), RSC
Adv. 5 (2015) 354e397.
[14] W.S. Shin, H.-H. Jeong, M.-K. Kim, S.-H. Jin, M.-R. Kim, J.-K. Lee, J.-W. Lee, Y.-
S. Gal, Effects of functional groups at perylene diimide derivatives on organic
photovoltaic device application, J. Mater. Chem. 16 (2006) 384e390.
[15] Y. Lin, X. Zhan, Non-fullerene acceptors for organic photovoltaics: an
emerging horizon, Mater. Horiz. 1 (2014) 470e488.
[16] A.F. Eftaiha, J.-P. Sun, I.G. Hill, G.C. Welch, Recent advances of non-fullerene,
small molecular acceptors for solution processed bulk heterojunction solar
cells, J. Mater. Chem. A 2 (2014) 1201e1213.
[17] K.N.N. Unni, A.K. Pandey, J.-M. Nunzi, Nechannel organic field-effect tran-
sistors using N,N0 editridecylperylene -3, 4, 9, 10 -tetracarboxylic diimide and
a polymeric dielectric, Chem. Phys. Lett. 407 (2005) 95e99.
[18] A.K. Pandey, K.N.N. Unni, J.-M. Nunzi, Pentacene/Perylene codeposited solar
cells, Thin Solid Films 511e512 (2006) 529e532.
ꢀ
[42] R. Singh, E. Giussani, M.M. Mrόz, F. Di Fonzo, D. Fazzi, J. Cabanillas-Gonzalez,
L. Oldridge, N. Vaenas, A.G. Kontos, P. Falaras, A.C. Grimsdale, J. Jacob,
K. Müllen, P.E. Keivanidis, On the role of aggregation effects in the perfor-
mance of perylene-diimide based solar cells, Org. Electron. 15 (2014)
1347e1361.
_
[43] H. Dinçalp, S¸ . Kızılok, S. Içli, Fluorescent macromolecular perylene diimides
containing pyrene or indole units in bay positions, Dyes Pigment. 86 (2010)
32e41.
[44] M.J. Ahrens, M.J. Tauber, M.R. Wasielewski, Bis(n-octylamino)perylene-3,4:
9,10-bis(dicarboximide)s and Their Radical Cations: Synthesis, Electrochem-
istry, and ENDOR Spectroscopy, J. Org. Chem. 71 (2006) 2107e2114.
[45] B. Gholamkhass, N.M. Kiasari, P. Servati, An efficient inverted organic solar cell
with improved ZnO and gold contact layers, Org. Electron. 13 (2012)
945e953.
[46] D. Pathak, T. Wagner, T. Adhikari, J.-M. Nunzi, Photovoltaic performance of
AgInSe 2 -conjugated polymer hybrid system bulk heterojunction solar cells,
Synth. Met. 199 (2015) 87e92.
[19] J.C. Hummelen, B.W. Knight, F. LePeq, F. Wudl, J. Yao, C.L. Wilkins, Preparation
and characterization of fulleroid and methanofullerene derivatives, J. Org.
Chem. 60 (1995) 532e538.
[47] M. Queste, C. Cadiou, B. Pagoaga, L. Giraudet, N. Hoffmann, Synthesis and
characterization of 1,7-disubstituted and 1,6,7,12-tetrasubstituted
: de-
rivatives, New J. Chem. 34 (2010) 2537e2545.
[20] Y. Lin, Y. Wang, J. Wang, J. Hou, Y. Li, D. Zhu, X. Zhan, An electron acceptor
challenging fullerenes for efficient polymer solar cells, Adv. Mater. 26 (2014)
5137e5142.
[21] E. Kozma, D. Kotowski, S. Luzzati, M. Catellani, F. Bertini, A. Famulari, G. Raos,
Improving the efficiency of P3HT:perylene diimide solar cells via bay-
substitution with fused aromatic rings, RSC Adv. 3 (2013) 9185e9188.
[22] E. Kozma, M. Catellani, Perylene diimides based materials for organic solar
cells, Dyes Pigment. 98 (2013) 160e179.
[48] S. Nikitenko, J. Mayorova, P. Troshin, R. Lyubovskaya, M. Kaplunov, Photo-
luminescence quenching study of composites comprising novel fullerene-
based acceptors and MDMO-PPV, Mol. Cryst. Liq. Cryst. 468 (2007), 239/
[591]-244/[596].
[49] A. Moliton, J.-M. Nunzi, How to model the behaviour of organic photovoltaic
cells, Polym. Intern. 55 (2006) 583e600.
[50] Z. Lu, X. Zhang, C. Zhan, B. Jiang, X. Zhang, L. Chen, J. Yao, Impact of molecular
solvophobicity vs. Solvophilicity on device performances of dimeric perylene
diimide based solution-processed non-fullerene organic solar cells, Phys.
Chem. Chem. Phys. 15 (2013) 11375e11385.
[23] C.A. Angell, Formation of glasses from liquids and biopolymers, Science 267
(1995) 1924e1935.
[24] M.D. Ediger, C.A. Angell, S.R. Nagel, Supercooled liquids and glasses, J. Phys.
Chem. 100 (1996) 13200e13212.
[25] L. Duan, L. Hou, T.-W. Lee, J. Qiao, D. Zhang, G. Dong, L. Wang, Y. Qiu, Solution
processable small molecules for organic light-emitting diodes, J. Mater. Chem.
20 (2010) 6392e6407.
[26] V. Sivamurugan, R. Lygaitis, J.V. Grazulevicius, V. Gaidelis, V. Jankauskas,
S. Valiyaveettil, Effect of substituents on the electron transport properties of
bay substituted perylene diimide derivatives, J. Mater. Chem. 19 (2009)
4268e4275.
[27] V. Sivamurugan, K. Kazlauskas, S. Jursenas, A. Gruodis, J. Simokaitiene,
J.V. Grazulevicius, S. Valiyaveettil, Synthesis and photophysical properties of
[51] B. Jiang, X. Zhang, C. Zhan, Z. Lu, J. Huang, X. Ding, S. He, J. Yao, Benzodi-
thiophene bridged dimeric perylene diimide amphiphiles as efficient
solution-processed non-fullerene small molecules, Polym. Chem. 4 (2013)
4631e4638.
[52] Q. Yan, Y. Zhou, Y. Zheng, J. Pei, D. Zhao, Towards rational design of organic
electron acceptors for photovoltaics: a study based on perylenediimide de-
rivatives, Chem. Sci. 4 (2013) 4389e4394.
[53] J. Lee, R. Singh, D.H. Sin, H.G. Kim, K.C. Song, K. Cho, A nonfullerene small
molecule acceptor with 3D interlocking geometry enabling efficient organic
solar cells, Adv. Mater. 27 (2015), http://dx.doi.org/10.1002/adma.201504010.