A series of V-shaped small molecule non-fullerene electron acceptors for efficient bulk-heterojunction devices

Article abstract of DOI:10.1016/j.dyepig.2019.107677

Two simple semiconducting acceptor-acceptor₁-donor-acceptor (A-A₁-D-A) modular, small molecule non-fullerene electron acceptors, 2-(4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-yl)phenyl)-3-(6-

Full text of DOI:10.1016/j.dyepig.2019.107677

Accepted Manuscript
A series of V-shaped small molecule non-fullerene electron acceptors for efficient
bulk-heterojunction devices
Pedada S. Rao, Vishal G. More, Asha D. Jangale, Sheshanath V. Bhosale, Rajesh S.
Bhosale, Avinash L. Puyad, Jing-Yu Chen, Jing-Liang Li, Sidhanath V. Bhosale, Akhil
Gupta, Ganesh D. Sharma
PII:
S0143-7208(19)30672-2
DOI:
https://doi.org/10.1016/j.dyepig.2019.107677
Article Number: 107677
Reference:
DYPI 107677
To appear in: Dyes and Pigments
Received Date: 26 March 2019
Revised Date: 20 June 2019
Accepted Date: 26 June 2019
Please cite this article as: Rao PS, More VG, Jangale AD, Bhosale SV, Bhosale RS, Puyad AL, Chen
J-Y, Li J-L, Bhosale SV, Gupta A, Sharma GD, A series of V-shaped small molecule non-fullerene
electron acceptors for efficient bulk-heterojunction devices, Dyes and Pigments (2019), doi: https://
doi.org/10.1016/j.dyepig.2019.107677.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
A series of V-shaped small molecule non-fullerene electron acceptors for
efficient bulk-heterojunction devices
a,b
c
c
c,*
Pedada S. Rao , Vishal G. More , Asha D. Jangale , Sheshanath V. Bhosale , Rajesh S.
a,1
d
e
e
a,b,*
Bhosale , Avinash L. Puyad , Jing-Yu Chen , Jing-Liang Li , Sidhanath V. Bhosale
,
f,*,2
g,*
Akhil Gupta , Ganesh D. Sharma
a
Polymers and Functional Materials Division and CSIR-Indian Institute of Chemical
Technology, Hyderabad 500007, Telangana, India; Email: bhosale@iict.res.in
b
Academy of Scientific and Innovative Research (AcSIR), Ghaziabad 201002, India
c
School of Chemical Sciences, Goa University, Taleigao Plateau 403206, Goa, India; Email:
svbhosale@unigoa.ac.in
d
School of Chemical Sciences, Swami Ramanand Teerth Marathwada University, Nanded
4
31606, Maharashtra, India.
e
Institute for Frontier Materials, Deakin University, Waurn Ponds, Victoria 3216 Australia.
ARC Centre of Excellence in Exciton Science, School of Chemistry, Bio21 Molecular
f
Science and Biotechnology Institute, University of Melbourne, Parkville, Victoria 3052
Australia; E-mail: akhilgk15@gmail.com
g
Department of Physics, The LNM Institute of Information Technology, Jamdoli 302031,
Jaipur, India; gdsharma273@gmail.com
Current Addresses
1
2
Department of Chemistry, Indrashil University, Kadi, Mehsana 382470 Gujarat, India.
School of Science, Faculty of Science, Engineering and Technology, Swinburne University
of Technology, Hawthorn, Victoria 3122 Australia. E-mail: akhilgk15@gmail.com; Tel: +61
02 131 118.
4

ACCEPTED MANUSCRIPT
Abstract
Two simple semiconducting acceptor-acceptor -donor-acceptor (A-A -D-A) modular, small
1
1
molecule non-fullerene electron acceptors, 2-(4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-
tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-yl)phenyl)-3-(6-((4-(7-hexadecyl-1,3,6,8-
tetraoxo-3,6,7,8-tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-yl)phenyl)ethynyl)-9-octyl-
9
2
H-carbazol-3-yl)buta-1,3-diene-1,1,4,4-tetracarbonitrile (NDICz-5) and 2-(4-(3,3-dicyano-
-(4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-
yl)phenyl)-1-(6-((4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-
tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-yl)phenyl)ethynyl)-9-octyl-9H-carbazol-3-
yl)allylidene)cyclohexa-2,5-dien-1-ylidene)malononitrile (NDICz-6), were designed,
synthesized and characterized for application in solution-processable bulk-heterojunction
solar cells. The optoelectronic and photovoltaic properties of NDICz-5 and NDICz-6 were
directly compared with those of a structural analogue, 7,7'-(((9-octyl-9H-carbazole-3,6-
diyl)bis(ethyne-2,1-diyl))bis(4,1-phenylene))bis(2-hexadecylbenzo[lmn][3,8]phenanthroline-
1
,3,6,8(2H,7H)-tetraone) (NDICz-4), which was designed based on an A-D-A format. All of
these new materials were designed to be V-shaped and comprised an electron rich carbazole
donor core (D) together with electron deficient naphthalene diimide terminal core (A). In the
simple A-D-A system, naphthalene diimide was the terminal functionality, whereas in the A-
A -D-A system, tetracyanoethylene (TCNE) and tetracyanoquinodimethane (TCNQ) derived
1
functionalities were incorporated as A units by keeping D/A units constant. The inclusion of
1
A was mainly done to induce cross-conjugation within the molecular backbone and hence to
1
improve light-harvesting. The physical and optoelectronic properties were characterized by
ultraviolet-visible, thermogravimetric analysis, photo-electron spectroscopy in air and cyclic
voltammetry. These new materials exhibited energy levels complementing those of the
conventional donor polymer poly(3-hexyl thiophene) (P3HT). Solution-processable bulk-

ACCEPTED MANUSCRIPT
heterojunction devices were fabricated with NDICz-4, NDICz-5 and NDICz-6 as non-
fullerene electron acceptors. Studies on the photovoltaic properties revealed that the best
P3HT: NDI-Cz6-based device showed an impressive enhanced power conversion efficiency
of 7.58%, an increase of around two-fold with respect to as-cast blend, after solvent vapor
annealing using carbon disulfide. Not only are the results among the best in the current
literature using the conventional donor polymer P3HT, but clearly demonstrate that A-A -D-
1
A type, V-shaped small molecules are promising non-fullerene acceptors in the research field
of organic solar cells.
Keywords: Non-fullerene; V-shaped; solution-processable; tetracyanoethylene; carbazole;
tetracyanoquinodimethane
1
. Introduction
The constant demand for the development of renewable energy technologies is a pressing
need for society. The natural resources such as wind, sunlight and water can provide a
definitive answer. However, technologies based on harvesting solar energy at an affordable
price are highly promising. The technologies such as organic (otherwise termed bulk-
heterojunction) solar cells are the most studied, well understood and widely accepted
strategies given the advantages, such as lightweight, flexibility and low-cost, they offer [1–8].
The development of such devices has seen advancements in the areas of new material design
and fabrication strategies. In terms of materials’ advancement, non-fullerene acceptors have
been developed over the past few years as they are seen to hold multiple advantages such as
strong light-harvesting, easily tunable energy levels, improved morphological stability, and
facile synthetic manipulations. The current literature indicates that a variety of structural

ACCEPTED MANUSCRIPT
arrangements have been used to design efficient non-fullerene acceptors, and to improve
power conversion efficiency (PCE) in particular [9–37].
In 2016, Hou et al. reported 12% PCE for bulk-heterojunction (BHJ) solar cells based
on methyl-modified dicycanovinylindan-1-one non-fullerene acceptor [38]. The same
group recently achieved a PCE of 13% based on devices fabricated using a
combination of PBDB-T-SF and IT-4F [39]. Such literature clearly indicates that one
can easily tune optical and electrochemical properties of small molecule non-fullerene
electron acceptors using organic synthetic strategies [40–42]. Particularly, non-
fullerene acceptors based on donor–acceptor (D–A) subunits are attractive due to their
feasibility to tune electronic energy levels, absorption properties, solubility in common
organic solvents and self-assembly behaviour. Among the reported non-fullerene
acceptors based on D–A format, target structures using naphthalene diimide (NDI)
unit are very promising given the capacity of such targets to accept electrons.
Moreover, the advantages of using NDI in a given target can be realised in terms of its
interesting physical and electronic properties, strong electron-withdrawing strength,
and the capacity to tune its solubility via incorporating lipophilic chains on nitrogen
atoms [43–44]. Keeping such criteria in mind, researchers have developed a number of
NDI-based target materials, and in particular non-fullerene acceptors, for organic solar
cell applications. Such targets have been generated by functionalization of NDI unit
either at imide position or at the central core.
Marder et al. reported pyrrole[3,2-
as electron-transport materials for organic field-effect transistor (OFET) applications
45]. Patil et al. reported the OPV performance of DPP-NDI-DPP derivative as non-
d:4,5-d′]bisthiazole-bridged bis(naphthalene diimide)s
[
fullerene electron-acceptor and achieved a PCE of 1.02% [46]. Russell and co-workers
synthesized NDI-based dimer, where NDIs were linked with vinyl bridge, which

ACCEPTED MANUSCRIPT
exhibited 2.41% PCE [47]. Kim et al. have demonstrated an encouraging PCE of
3
.01% with an open circuit voltage (Voc) of 0.75 V using NDI derivative NDICN-TVT
[
48]. The NDI core has further been modified with the use of a number of electron
withdrawing groups, such as rhodanine, 1,3-indanedione, 2-methoxyethyl-2-
cyanoacetate and cyanopyridone, so as to improve device performance [49–50]. We
have recently demonstrated the use of TPA-NDI-C triad as a D-A-A system with
60
spectacular PCEs of 3.03% and 4.85%, respectively, before and after annealing [51].
Samuel et al. reported NDI-based imide substituted derivatives as electron acceptor
materials for BHJ applications [52]. However, less attention has been given to the
design and develop of those NDI derivatives which boast functionalization at the
imide position.
Our group recently reported a PCE of 2.3% for a non-fullerene electron acceptor based
on central carbazole (Cz) and terminal diketopyrrolopyrrole (DPP) units [53]. It is well
documented that Czs are extensively utilized as electron donor materials, providing
good charge transport abilities for various applications. Würthner et al. reported core-
extension of NDI with two annulated indole rings leading to carbazolo[2,3-
b
]carbazole
2
diimides (CbDIs) and its application in OFETs with mobility values up to 0.2 cm /(V
s) [54]. Li and co-workers also demonstrated Cz-based polymers and their
photovoltaic properties [55]. On the other hand, tetracyanoethylene (TCNE) and
tetracyanoquinodimethane (TCNQ) units are very strong electron accepting building
blocks with high electron mobility that have been extensively employed to
functionalize acceptor molecules [56]. Small molecule non-fullerene electron
acceptors comprising TCNE and TCNQ units have not been extensively investigated,
with the exceptions reported by Mishra and co-workers, which showed impressive
PCEs [57]. In 2017, we developed TPA-based small molecule non-fullerene electron

ACCEPTED MANUSCRIPT
acceptors containing TCNE and TCNQ functionalities for BHJ solar cell applications,
with good to moderate efficiencies of 2.28% and 3.75% [58].
Taking into account of advantages exhibited by NDI moiety and the literature precedence
where NDI, Cz, and TCNE and TCNQ units have been used for organic photovoltaic
applications, herein we report the design, synthesis and characterization of three novel non-
fullerene
phenylene))bis(2-hexadecylbenzo[lmn][3,8]phenanthroline-1,3,6,8(2H,7H)-tetraone)
NDICz-4), 2-(4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-
acceptors,
7,7'-(((9-octyl-9H-carbazole-3,6-diyl)bis(ethyne-2,1-diyl))bis(4,1-
(
tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-yl)phenyl)-3-(6-((4-(7-hexadecyl-1,3,6,8-
tetraoxo-3,6,7,8-tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-yl)phenyl)ethynyl)-9-octyl-
9
2
H-carbazol-3-yl)buta-1,3-diene-1,1,4,4-tetracarbonitrile (NDICz-5) and 2-(4-(3,3-dicyano-
-(4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-
yl)phenyl)-1-(6-((4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-
tetrahydrobenzo[lmn][3,8]phenanthrolin-2(1H)-yl)phenyl)ethynyl)-9-octyl-9H-carbazol-3-
yl)allylidene)cyclohexa-2,5-dien-1-ylidene)malononitrile (NDICz-6), where NDICz-4 is
based on an A-D-A format, and NDICz-5 and NDICz-6 are based on A-A -D-A modular
1
design. We introduced TCNE and TCNQ functionalities in NDICz-5 and NDICz-6,
respectively, which served as A units so as to generate the latter module. The common donor
1
core (D) was Cz and the terminal acceptor unit (A) was NDI. It was observed that the
introduction of TCNE and TCNQ functionalities in A-A -D-A system resulted in strong
1
intramolecular charge transfer transition when compared with A-D-A system. The
introduction of TCNQ in particular exhibited a broader absorption profile which can be
ascribed to the stronger electron withdrawing nature of TCNQ relative to TCNE. Solution-
processable BHJ devices were fabricated with NDICz-4, NDICz-5 and NDICz-6 as non-
fullerene electron acceptors. Studies on the photovoltaic properties revealed that the best

ACCEPTED MANUSCRIPT
P3HT: NDICz-6-based device showed an impressive PCE of 7.58% after solvent vapor
annealing (SVA) using carbon disulfide. The increase in efficiency after SVA treatment was
around two-fold when compared with as-cast blend as well as P3HT: NDICz-4-based
devices. To the best of our knowledge, there are no reports on the conjoint use of Cz and NDI
functionalities where such functionalities have been used to design V-shaped non-fullerene
acceptors comprising either TCNE or TCNQ units. Our results clearly indicate that these A-
A -D-A-based small molecules are potential acceptors for efficient organic solar cells using
1
the commercially available donor polymer P3HT and can further be explored with a variety
of donors. The molecular structures of the investigated materials are represented in Fig. 1.
>
Fig. 1<
2
. Experimental Details
2
.1. Materials, methods and device details
All the reagents and chemicals used, for example naphthalene dianhydride, hexadecane
amine, carbazole, 1-bromooctane, sodium hydride, -iodosuccinimide, triethylamine,
N
copper iodide, (trimethylsilyl)acetylene, tetrakis(triphenylphosphine)palladium(0),
TCNE and TCNQ, to name a few, unless otherwise specified, were purchased from Sigma-
Aldrich Co., Bengaluru, Karnataka, India, and were used without any purification. AR grade
solvents were purchased from Finar chemicals Limited, India and used as received. The
intermediates and target materials were purified using silica gel chromatography.
FT-IR spectra were measured on a Thermo Nicolet Nexus 670 spectrometer using KBr
1
13
1
13
pellets. H and C NMR spectra were recorded at 400 or 500 MHz as indicated. H and C
chemical shifts were calibrated using residual non-deuterated solvent as an internal reference
and are reported in parts per million (δ) relative to tetramethylsilane (δ = 0). ESI-MS data

ACCEPTED MANUSCRIPT
was taken on Shimadzu lab solutions. For MALDI-TOF measurements, a Schimadzu Biotech
Axima performance spectroscopic instrument were used. Differential scanning calorimetry
(DSC) analysis was performed using a TA Instruments Q-100 DSC with a heating rate and a
cooling rate of 10 °C per minute. Thermogravimetric analysis (TGA) measurement was
performed using a TA instruments Q-500 TGA under flowing nitrogen gas at a heating rate
o
of 10 C per minute. The UV-vis absorption spectra were recorded in a 1 cm path-length
quartz cell using Shimadzu UV-1800 spectrophotometer. Fluorescence spectra were recorded
using Shimadzu model R-6000 spectrofluorometer. Cyclic voltammetry measurements were
recorded under nitrogen in a deoxygenated solution of tetra-n-butylammonium
hexafluorophosphate (0.1 M) in dichloromethane or 1,2-dichlorobenzene using a PowerLab
ML160 potentiostat interfaced via a PowerLab 4/20 controller to a PC running E-Chem for
Windows version 1.5.2. Cyclic voltammograms (potential scan rate of 50 mV/s) and all
measurements were carried out at room temperature using a three-electrode configuration
consisting of a platinum working electrode; a platinum wire auxiliary electrode; and a
nonaqueous Ag/AgCl reference electrode. PESA measurements were recorded using a Riken
Keiki AC-2 PESA spectrometer with a power number of 0.5. Samples for PESA were
prepared on ITO cleaned glass substrates and were run using a power setting of 10 nW.
Organic solar cells were fabricated using the conventional architecture, ITO/PEDOT:
PSS/P3HT: NDICz-4 or NDICz-5 or NDICz-6 /PFN /Al. Indium tin oxide (ITO) coated
glass substrates were cleaned sequentially by 10 minutes ultrasonication in detergent,
deionized water, acetone and isopropyl alcohol followed by drying in a nitrogen flow. The
substrates were further dried at ambient conditions. The PEDOT: PSS solution was filtered
and spin coated onto the pre-cleaned ITO substrates at 3500 rpm for 30 s and then annealed at
1
20 °C for 10 minutes. The blends of P3HT and acceptors (different weight ratios of donor
and acceptor, total concentration 14 mg/mL) were dissolved in chloroform (CF). The

ACCEPTED MANUSCRIPT
solutions were spin coated atop the PEDOT:PSS surface at 2000 rpm for 60 s and the samples
were dried in a vacuum oven at 40 °C. A thin layer of PFN was spin coated on top of the
active layer, which was dissolved in methanol and mixed with acetic acid with a
concentration of 0.5 mg/mL and then spin coated at a rpm of 2000 to get a thin film of about
1
0 nm. Finally, the aluminium (Al) electrode was deposited on top of the active layer by
–
5
thermal evaporation under a vacuum of 10 Torr. The current-voltage characteristics of the
devices were measured using a computer-controlled Keithley 2400 source meter under
2
stimulated AM 1.5 G at an intensity of 100 mW/cm provided by a solar simulator. The
incident photon to current conversion efficiency of the devices was measured by illuminating
the device through the light source and the monochromator, and the resulting current was
measured using a Keithley electrometer.
2
.2. Synthetic details
>
Scheme 1<
Compound NDICz-4 was synthesized by reacting 3,6-diethynyl-9-octyl-9H-carbazole
compound 4) with 2-hexadecyl-7-(4-iodophenyl)benzo[lmn][3,8]phenanthroline-
,3,6,8(2H,7H)-tetraone (compound 5) using Sonogashira coupling reaction in dry
(
1
tetrahydrofuran (THF) (Scheme 1). Compounds NDICz-5 and NDICz-6 were synthesized by
reacting NDICz-4 with TCNE and TCNQ, respectively (Scheme 2). The final compounds
were purified using silica gel chromatography, characterized spectrally, and their
optoelectronic properties were investigated. The materials were prepared in moderate to high
yields and were found to be highly soluble in a variety of conventional organic solvents such
as chlorobenzene, chloroform, dichloromethane and toluene. High solubility of organic

ACCEPTED MANUSCRIPT
materials is an essential criterion for the fabrication of solution-processable organic
photovoltaic devices and newly designed materials NDICz-4, NDICz-5 and NDICz-6 fulfil
this criterion. Compound 5 was synthesized following the procedures reported in the
literature (J Org Chem 2013;78:5227–38).
>
Scheme 2<
2
.2.1. 9-Octyl-9H-carbazole (1)
Carbazole (3 g, 17.9 mmol) and 1-bromooctane (9.3 mL, 53.8 mmol) were dissolved in dry
DMF (40 mL) in a 100 mL round bottom flask with magnetic stirring bar. Sodium hydride
(1.29 g, 53.8 mmol, 60 % dispersion in mineral oil) was added slowly at room temperature
and the resulting suspension was refluxed overnight. After completion of reaction, excess
NaH was quenched by adding methanol dropwise. The resulting reaction mixture was poured
on ice and extracted with dichloromethane. The organic layer was separated, washed with
water twice followed by brine and recovered to get a viscous brown liquid which was
purified by column chromatography on silica gel using hexane as an eluent; yield = 4.55 g,
1
9
1 %. H NMR (500 MHz, CDCl ) δ = 8.10 (d, J = 7.7 Hz, 2H), 7.47 (t, J = 8.2 Hz, 2H), 7.41
3
(d, J = 8.2 Hz, 2H), 7.23 (t, J = 7.7 Hz, 2 H), 4.31 (t, J = 7.3 Hz, 2H), 1.89–1.84 (m, 2H)
1
3
1
1
2
.40–1.23 (m, 10H), 0.86 (t, J = 6.7 Hz, 3H); C NMR (100 MHz, CDCl ) δ = 140.77,
3
25.70, 122.74, 120.48, 118.53, 108.80, 43.17, 31.98, 29.91, 29.55, 29.36, 29.12, 27.48,
+
2.80, 14.27; ESI-MS (m/z) [M+H] : 280.25.
2
.2.2. 3,6-Diiodo-9-octyl-9H-carbazole (2)
Compound 1 (3 g, 10.7 mmol) was dissolved in a mixture of chloroform and acetic acid (30
mL; 1: 1) and the reaction vessel was covered with aluminum foil. N-iodosuccinimide (4.85

ACCEPTED MANUSCRIPT
g, 21.5 mmol) was added and the reaction mixture was stirred for 12 h. After the
consumption of starting material, the reaction mixture was diluted with chloroform (50 mL)
and water (100 mL). The organic layer was separated, washed with water followed by 20%
sodium thiosulphate solution, dried over anhydrous MgSO , and recovered to crude product
4
which was purified by silica gel column chromatography (eluent: hexane); yield = 5.24 g, 92
1
%
. H NMR (500 MHz, CDCl ) δ = 8.31 (d, J = 1.6 Hz, 2H), 7.70 (dd, J = 8.6, 1.7 Hz, 2H),
3
7
1
1
.16 (dd, J = 8.5, 3.2 Hz, 2H), 4.20 (t, J = 7.2 Hz, 2H), 1.87–1.75 (m, 2H), 1.41–1.16 (m,
1
3
0H), 0.85 (t, J = 7.0 Hz, 3H); C NMR (125 MHz, CDCl ) δ = 139.62, 134.26, 129.79,
3
24.00, 110.92, 81.66, 43.57, 31.89, 29.33, 29.15, 28.84, 27.24, 22.61, 14.09.
2
.2.3. 9-Octyl-3,6-bis((trimethylsilyl)ethynyl)-9H-carbazole (3)
A mixture of dry THF and triethylamine (30 mL; 1: 1) was degassed for 1 h, and compound 2
2 g, 3.76 mmol) and copper iodide (35 mg, 0.18 mmol) were added. To the resulting reaction
(
mixture, trimethylsilylacetylene (TMSA) (2.61 mL, 18.8 mmol) and catalytic amount of
bis(triphenylphosphine)palladium(II) dichloride (5% mol) were added, and the reaction was
stirred at room temperature for 48h. The reaction mixture was filtered through Celite,
concentrated under vacuum, and the crude product was dissolved in dichloromethane and
washed with water (× 3) followed by brine. The organic layer was dried over anhydrous
MgSO and recovered to get crude brown liquid which was purified by silica gel column
4
chromatography (eluent: hexane) to afford compound 3 as a yellow viscous oil; yield = 1.58
1
g, 89 %. H NMR (500 MHz, CDCl ) δ = 8.19 (d, J = 1.1 Hz, 2H), 7.58–7.55 (dd, J = 1.6 Hz,
3
J = 1.5 Hz, 2H), 7.29 (d, J = 8.5 Hz, 2H), 4.24 (t, J = 7.2 Hz, 2H), 1.84–1.80 (m, 2H), 1.31–
1
3
1
1
2
.26 (m, 10H), 0.86 (t, J = 7.0 Hz, 3H), 0.29 (s, 18H); C NMR (100 MHz, CDCl ) δ =
3
40.37, 129.82, 124.54, 122.15, 113.65, 108.64, 106.22, 92.04, 43.07, 31.59, 29.15, 29.00,
+
8.76, 27.08, 22.44, 20.13, 13.91; ESI-MS (m/z) [M+Na] : 495.

ACCEPTED MANUSCRIPT
2
.2.4. 3,6-Diethynyl-9-octyl-9H-carbazole (4)
Compound 3 (1.5 g, 3.17 mmol) was dissolved in a solvent mixture of MeOH: THF (v/v, 1:1)
and finely powdered K CO (2.19 g, 15.8 mmol) was added. The reaction mixture was stirred
2
3
at room temperature for 5 hr and the solvent was evaporated under reduced pressure. The
crude solid was dissolved in diethyl ether and washed with water followed by brine. The
organic layer was evaporated and the crude mass was subjected to column chromatography
1
on silica (eluent: hexane) to afford 4 as a light-yellow solid; yield = 0.95 g, 92 %. H NMR
(
500 MHz, CDCl ) δ = 8.22 (d, J = 1.1 Hz, 2H), 7.60 (dd, J = 8.5, 1.5 Hz, 2H), 7.34 (d, J =
3
8
.5 Hz, 2H), 4.27 (t, J = 7.3 Hz, 2H), 3.08 (s, 2H), 1.88–1.80 (m, 2H), 1.58 (s, 2H), 1.38–1.18
1
3
(
m, 10H), 0.86 (t, J = 7.0 Hz, 3H); C NMR (125 MHz, CDCl ) δ = 140.69, 130.15, 124.79,
3
1
22.28, 112.71, 108.97, 84.76, 75.46, 43.37, 31.77, 29.32, 29.15, 28.94, 27.25, 22.61, 14.07;
+
+
ESI-MS (m/z) [M+H] : 328 (30%) and [M+NH ] : 346 (100%).
4
2
.2.5.
7,7'-(((9-Octyl-9H-carbazole-3,6-diyl)bis(ethyne-2,1-diyl))bis(4,1-phenylene))bis(2-
hexadec ylbenzo[lmn][3,8]phenanthroline-1,3,6,8(2H,7H)-tetraone) (NDICz-4)
Compounds 4 (0.3 g, 0.91 mmol) and 5 (1.3 g, 1.9 mmol) were dissolved in a solvent mixture
of dry THF: N,N-diisopropylethylamine (DIPEA) (30 mL; 1: 1) and the resulting solution
was degassed using nitrogen gas. Copper iodide (87 mg, 0.045 mmol) and
tetrakis(triphenylphosphine)palladium(0) (52 mg, 5 mol%) were added to the reaction
mixture. The resulting suspension was refluxed for 24 h. The completion of the reaction was
monitored by thin layer chromatography and was worked-up using the same procedure
reported for compound 3 to afford NDICz-4 as a light brown solid (column in DCM: hexane
-1
7
0: 30); yield = 0.93 g, 70 %. FT-IR (KBr, cm ): 3421, 3081, 2921, 2850, 2208, 1709, 1664,

ACCEPTED MANUSCRIPT
1
1
8
8
2
580, 1509, 1451, 1342, 1243, 1186, 766; H NMR (400 MHz, CDCl ) δ = 8.83–8.77 (m,
3
H), 8.30 (d, J = 1.2 Hz, 2H), 7.80–7.74 (m, 4H), 7.69 (dd, J = 8.5, 1.5 Hz, 2H), 7.40 (d, J =
.6 Hz, 2H), 7.37–7.32 (m, 4H), 4.31 (t, J = 7.2 Hz, 2H), 4.25–4.18 (m, 4H), 1.94–1.86 (m,
1
3
H), 1.82–1.71 (m, 4H), 1.49–1.23 (m, 62H), 0.88 (t, J = 6.9 Hz, 9H); C NMR (100 MHz,
CDCl ) δ = 162.90, 162.73, 140.48, 133.87, 132.57, 131.37, 130.96, 129.84, 128.66, 126.96,
3
1
4
1
1
8
26.79, 126.57, 125.00, 124.25, 122.39, 113.69, 109.06, 91.85, 87.24, 77.36, 77.04, 76.73,
1.11, 31.95, 31.79, 29.73, 29.58, 29.38, 29.19, 28.99, 28.13, 27.30, 27.14, 22.72, 22.63,
+
4.15, 14.10; MALDI-TOF: calculated for C H N O [M] (m/z) 1456.89; found =
9
6
105
5
8
456.95. Elemental analysis calculated for C H N O (%): C 79.14; H 7.26; N 4.81; O
9
6
105
5
8
.79; found C 79.10; H 7.23; N 4.78; O 8.75.
2
.2.6. 2-(4-(7-Hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-tetrahydrobenzo[lmn][3,8]phenanthrolin-
(1H) -yl)phenyl)-3-(6-((4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-
2
tetrahydrobenzo[lmn][3,8]phenan throlin-2(1H)-yl)phenyl)ethynyl)-9-octyl-9H-carbazol-3-
yl)buta-1,3-diene-1,1,4,4-tetracarbonitrile (NDICz-5)
To a solution of NDICz-4 (100 mg, 0.068 mmol) in dry THF (5 mL) was added TCNE (87
mg, 0.68 mmol) and the resulting solution was refluxed for 12 h. The progress of reaction
was monitored by TLC. After completion of reaction the solvent was evaporated off and the
resulting dark brown mass was purified by column chromatography (silica gel; hexane: DCM
-1
1
: 9) to afford titled NDICz-5 as a cherry solid (97 mg, 90%); FT-IR (KBr, cm ): 3449,
1
2
922, 2851, 2222, 1711, 1666, 1582, 1509, 1452, 1337, 1245, 1186, 880, 813, 766; H NMR
(
500 MHz, CDCl ) δ = 8.82–8.77 (m, 8H), 8.54 (d, J = 2.0 Hz, 1H), 8.37 (d, J = 1.0 Hz, 1H),
3
8
7
1
.06 (dd, J = 8.9, 2.0 Hz, 1H), 8.02–7.96 (m, 2H), 7.80–7.73 (m, 3H), 7.61–7.55 (m, 3H),
.46 (d, J = 8.5 Hz, 1H), 7.38–7.32 (m, 2H), 4.36 (t, J = 7.2 Hz, 2H), 4.25–4.16 (m, 4H),
1
3
.93–1.91 (m, 2H), 1.79–1.71 (m, 4H), 1.49–1.20 (m, 62H), 0.89–0.86 (m, 9H); C NMR

ACCEPTED MANUSCRIPT
(
100 MHz, CDCl ) δ = 165.74, 162.89, 162.67, 162.49, 144.20,140.88,139.86, 134.18,
3
1
1
1
2
32.57, 131.88, 131.62, 131.41, 130.99, 130.62, 130.41, 128.69, 127.64, 127.21, 127.01,
26.49, 126.03, 124.75, 124.40, 123.82, 123.47, 122.39, 121.76, 115.78, 113.09, 112.20,
11.60, 111.15, 110.68, 109.92, 90.80, 88.27, 81.93, 88.21, 81.51, 43.87, 41.07, 31.99, 31.71,
9.65, 29.51, 29.32, 29.07, 28.97, 28.06, 27.20, 27.07, 22.66, 22.55, 14.09, 14.03; MALDI-
+
+
TOF: calculated for C H N O [M] (m/z) 1584.98; found = 1584.16; [M+H] = 1586.15;
1
02 105
9
8
+
[
M+2H] = 1587.20. Elemental analysis calculated for C H N O (%): C 77.29; H 6.68; N
102 105 9 8
7
.95; O 8.08; found C 77.24; H 6.64; N 7.92; O 8.03.
2
.2.7.
2-(4-(3,3-Dicyano-2-(4-(7-hexadecyl-1,3,6,8-tetraoxo-3,6,7,8-
enanthrolin-2(1H)-yl)phenyl)-1-(6-((4-(7-hexadecyl-1,3,6,8-
tetrahydrobenzo[lmn][3,8]ph
tetraoxo-3,6,7,8-tetrahydrobenzo [lmn][3,8]phenanthrolin-2(1H)-yl)phenyl)ethynyl)-9-octyl-
H-carbazol-3-yl)allylidene)cyclohex a-2,5-dien-1-ylidene)malononitrile (NDICz-6)
9
Compounds NDICz-4 (100 mg, 0.068 mmol) and TCNQ (139 mg, 0.68 mmol) were
suspended in 1,2- dichlorobenzene and the resulting reaction mixture was stirred at 120 °C.
The progress of reaction was monitored by TLC. The solvent was evaporated under reduced
pressure and the crude residue was purified by column chromatography (silica gel; eluent:
-1
DCM) to afford NDICz-6 as a bluish-black solid (82 mg, 72%); FT-IR (KBr, cm ): 3448,
1
2
921, 2850, 2209, 1710, 1669, 1582, 1452, 1339, 1245, 1190, 766; H NMR (400 MHz,
CDCl ) δ = 8.78 (dd, J = 18.6, 7.3 Hz, 8H), 8.35 (s, 1H), 8.10 (s, 1H), 7.93 (d, J = 8.6 Hz,
3
2
H), 7.74 (t, J = 7.4 Hz, 3H), 7.58 (dd, J = 9.6, 1.9 Hz, 1H), 7.53–7.42 (m, 5H), 7.41–7.32
(m, 4H), 7.21 (dd, J = 9.6, 1.9 Hz, 1H), 4.34 (t, 2H), 4.20 (dd, J = 17.2, 9.6 Hz, 4H), 1.91 (d,
1
3
J = 7.3 Hz, 2H), 1.82–1.68 (m, 4H), 1.49–1.19 (m, 62H), 0.87 (td, J = 6.8, 3.0 Hz, 9H); C
NMR (100 MHz, CDCl ) δ = 170.24, 162.90, 162.69, 162.53, 162.48, 153.83, 150.92,
3
1
42.60, 140.84, 139.17, 135.75, 134.31, 134.06, 133.84, 132.52, 131.59, 131.42, 130.99,

ACCEPTED MANUSCRIPT
1
1
8
2
30.83, 130.59, 130.34, 129.63, 128.67, 127.25, 127.04, 126.81, 126.51, 125.99, 124.57,
23.70, 122.38, 115.01, 114.03, 113.74, 113.69, 112.54, 112.22, 110.30, 109.67, 91.19,
8.25, 87.88, 84.78, 43.73, 41.30, 31.94, 31.90, 31.74, 29.67, 29.33, 29.12, 29.01, 28.07,
+
7.26, 27.08, 22.67,14.093; MALDI-TOF: calculated for C H N O [M] (m/z) 1661.08;
1
08 109
9
8
+
found = 1661.93, [M+H] = 1662.95. Elemental analysis calculated for C H N O (%): C
1
08 109
9
8
7
3
3
8.09; H 6.61; N 7.59; O 7.71; found C 78.05; H 6.57; N 7.53; O 7.69.
. Results and discussion
.1. Design and synthetic strategies
We designed these novel molecules to be V-shaped, and by combining electron rich and
electron poor subunits, such as Cz and NDI, respectively, presuming it might help to improve
charge transfer transition and in turn electron accepting properties. To achieve this, firstly we
developed NDICz-4 based on an A-D-A format comprising NDI and Cz functionalities
interlinked with acetylenic bond. Then the reaction of NDICz-4 was carried out with TCNE
and TCNQ to afford NDICz-5 and NDICz-6, respectively, thus generating A-A -D-A format.
1
The successful synthesis of NDICz-4 was carried out using the Sonogashira-Hagihara
coupling reaction where intermediates 4 and 5 were reacted using tetrakis
(triphenylphosphine)palladium(0) catalyst. The synthesis of intermediates 4 and 5 was carried
out using the literature procedures [59]. All of the synthetic details are depicted in Scheme 1.
For the synthesis of NDICz-5 and NDICz-6, the [2 + 2] cycloaddition (CA) reaction
followed by retro-electrocyclization (RE) reaction of NDICz-4 with TCNE and TCNQ was
conducted, respectively, according to the procedure reported in the literature (Scheme 2)
[
57,58]. This reaction strategy is highly successful due to the simplicity of transformation and
generally provides higher yields. For example, NDICz-5 was obtained in 90% yield when the

ACCEPTED MANUSCRIPT
reaction of NDICz-4 was carried out with TCNE in THF at 66 °C. We observed that only one
acetylene subunit in NDICz-4 was reacted with TCNE. To check the reactivity of second
acetylene subunit, we performed the reaction of NDICz-5 with TCNE in THF at 80 °C in a
sealed tube for 2 days. No transformation was observed from NDICz-5 to NDICz-7 (Scheme
3
), thus implying that Cz subunit of NDICz-5 experiences electron deficiency which in turn
stops a further [2 + 2] CA-RE reaction with TCNE. This suggests that it is important to
balance the donor-acceptor subunits to proceed a further [2 + 2] CA-RE reaction. As we
successfully synthesized NDICz-5 using THF as the reaction solvent, we thought to employ
the same reaction conditions to generate NDICz-6. But surprisingly, the reaction didn’t occur
in the presence of THF, thus, prompting us to evaluate the effect of solvent and temperature
on the reactivity of NDICz-4 towards TCNQ. Dichlorobenzene was found to be the most
suitable solvent. The [2 + 2] CA-RE reaction of NDICz-4 with TCNQ was conducted in
dichlorobenzene at 120 °C for 24 h affording NDICz-6 as a dark green solid in 72% yield
(
Scheme 2). An effort to react NDICz-6 with an excess of TCNQ in dichlorobenzene at 140
C for 3 days did not yield NDICz-8 (Scheme 3). Thus, we realized that not only the donor-
acceptor subunits, but the TCNQ moiety renders the reduction of alkyne reactivity.
°
>
Scheme 3<
3
.2. Thermal properties
The thermogravimetric and differential scanning calorimetry (DSC) analyses were carried out
to investigate thermal properties of NDICz-4, NDICz-5 and NDICz-6. Under nitrogen
atmosphere, the thermogravimetric analysis (TGA) of all the three compounds NDICz-4,
NDICz-5 and NDICz-6 experienced 5% weight loss at 426 °C, 460 °C and 375 °C,
respectively, suggesting that the materials are thermally stable (Figures S1, S2 and S3 in

ACCEPTED MANUSCRIPT
Supporting Information (SI)). The DSC traces of NDICz-4 and NDICz-5 are shown in
Figures S4 and S5 (SI), respectively, suggesting the melting point temperatures of 221.72 °C
and 184.95 °C. On the other hand, NDICz-6 demonstrated crystalline behavior at 261.54 °C,
thus, indicating that it exhibits stronger intermolecular interactions in its solid state (Fig. S6;
SI). These analyses revealed these materials to be thermally stable, meaning that the BHJ
devices can be annealed at a higher temperature, e.g. > 100 °C), if required.
3
.3. Photophysical properties
The optical absorption spectra of NDICz-4, NDICz-5 and NDICz-6 in chloroform solution
and as a thin solid film are shown in Fig. 2. NDICz-4 showed absorption peaks at 314 nm
and 344 nm (shoulder peak) with two major peaks at 360 nm and 380 nm. NDICz-5
displayed absorption peaks at 344 nm (shoulder peak), 360 nm, 380 nm and one broad peak
at 475 nm. NDICz-6 bearing TCNQ functionality exhibited absorption peaks at 344 nm
(
shoulder peak), two major peaks at 360 nm and 380 nm and two broad peaks at 450 nm and
00 nm. The absorption peaks of NDICz-4, NDICz-5 and NDICz-6 appeared at ~400 nm
can be assigned to π → π* transitions (S0 → S2) of the NDI moieties. The peaks for NDICz-
appeared at 475 nm, and for NDICz-6 at 450 nm and 600 nm can be ascribed to strong
6
5
intramolecular charge transfer (ICT) transitions from the donor Cz subunit to the acceptor
NDI and the additional A moieties. The absorption peaks for NDICz-4 in thin film form
1
exhibited peaks at 315 nm and 357 nm, whereas for NDICz-5, the peaks appeared as one
shoulder peak at 304 nm and two prominent peaks at 361 nm and 383 nm together with a
broad peak at 463 nm. As a thin film, NDICz-6 exhibited peaks at 319 nm (shoulder peak),
two peaks at 363 nm and 383 nm, and a broad peak at 600 nm. The broad absorption maxima
at 463 nm and 600 nm suggest that thin films of NDICz-5 and NDICz-6 possess effective π-
π stacking in their solid state and have an ability to exhibit superior light-harvesting

ACCEPTED MANUSCRIPT
properties when compared with their structural analogue NDICz-4. These results clearly
indicate that incorporation of additional acceptor units, TCNE and TCNQ, can indeed be
utilized for absorbing a larger amount of the solar spectrum.
>
Fig. 2<
The estimated optical band gaps of NDICz-4, NDICz-5 and NDICz-6 using the onset of film
absorption were 3.02 eV, 2.13 eV and 1.59 eV, respectively, and are depicted in Table S1
(ESI). The optical band gaps of NDICz-5 and NDICz-6 are narrower than that of NDICz-4,
suggesting their strong electron accepting capacity due to the incorporation of TCNE and
TCNQ functionalities that connect the Cz and NDI subunits of the molecular backbone. Thus,
our molecular design strategy can be useful for controlling the absorption properties together
with manipulating energy levels and plausibly increasing photovoltaics performance. The
normalized absorption spectra of the blends of NDICz-4, NDICz-5 and NDICz-6 with P3HT
(D: A 1: 1.5) are given in Fig. 3. According to the blend film spectra shown in Fig. 3, it is
apparent that the blend films have more favorable optical absorption throughout the entire
visible region. Therefore, the use of a commercially available donor polymer, such as P3HT,
is indeed helpful to carry out OPV study and to observe initial performance. We also
investigated the steady state emission properties of NDICz-4, NDICz-5 and NDICz-6 in
their chloroform solutions. The emission spectra are shown in Fig. S7, SI.
>
Fig. 3<
3
.4. Computational Calculations

ACCEPTED MANUSCRIPT
The geometrical optimization of NDICz-4, NDICz-5 and NDICz-6 with truncated alkyl
chains was performed at the B3LYP/6-31G(d) level of theory using the Gaussian 09 ab
initio/DFT quantum chemical simulation package [60]. The frequency calculations were
carried out to ensure the structures of molecule to be real. The obtained geometries of
NDICz-4, NDICz-5 and NDICz-6 at the B3LYP/6-31G(d) level were subjected to the time-
dependent density functional theory (TD-DFT) studies using M062X/6-31G(d) [61]. The
frontier molecular orbitals (FMOs) calculated at the M062X/6-31G(d) level of theory and
generated using Avogadro are given in Fig. 4 and Fig. S8, SI [62]. In case of NDICz-4, the
HOMO electron density resided over Cz and acetylene-linked phenyl subunits, whereas the
LUMO was localized over the terminal NDI unit. The former was true for both NDICz-5 and
NDICz-6, given the donating strength of Cz functionality. The LUMO density for NDICz-5
was mainly residing on terminal NDI unit, but for NDICz-6, there was a strong contribution
from the TCNQ functionality. This is logical given the accepting strength of TCNQ and its
role in tuning the energy levels, though the levels L+1 and L+2 inhabited on NDI unit.
Overall, there was a good segregation between the HOMO and the LUMO densities, thus,
generating an ideal environment for the ICT transition over the whole molecular backbone.
The obtained TD-DFT results at the M062X/6-31G(d) level of theory for NDICz-4, NDICz-
5
and NDICz-6 are depicted in Table S2 (SI). The computed absorption spectra shown in
Figures S9, S10 and S11 (SI) indicated transition peaks which are mainly described as the
HOMO → LUMO transition. Moreover, the enhancement of ICT transition in NDICz-6
followed by NDICz-5 and NDICz-4 was further confirmed by evaluating the dipole moment
values using the M062X/6-31G(d) level of theory (see Table S3; SI).

ACCEPTED MANUSCRIPT
>
Fig. 4<
3
.5. Energy levels and electrochemical investigation
As such, the experimental estimation of the HOMO energies was carried out using photo
electron spectroscopy in air (PESA) and the LUMO energies were calculated by adding the
optical band gap (film spectra) to the HOMO values (for PESA curves see Fig. S12 SI). The
LUMO levels of NDICz-5 and NDICz-6 were deepest when compared with the LUMO level
of NDICz-4. This suggests the influence of strong accepting functionality, of which TCNQ in
NDICz-6 is an example, in tuning the overall energy levels. A similar drop of the LUMOs
was also observed while conducting DFT calculations (Theoretical LUMOs of NDICz-4,
NDICz-5 and NDICz-6 were 2.63 eV, 2.71 eV and 3.20 eV, respectively), a result that
provides a good support to our experimental findings. These theoretical and experimental
findings strongly indicate that the A-A -D-A format really plays a crucial role for density
1
segregation, efficient light-harvesting, and tuning optical energy levels when compared with
the A-D-A format. The electrochemistry experiments revealed that the investigated materials
undergo reversible reduction cycles, thus suggesting their suitability as n-type materials in
BHJ devices (for cyclic voltammograms, see Fig. S13; SI). The HOMO and the LUMO
energy levels of NDICz-4, NDICz-5 and NDICz-6 were estimated from the onsets of the
oxidation and reduction potentials (Table S1), respectively, including the standard value of
ferrocene to be -4.8 eV. It was observed that the energy levels deduced from the cyclic
voltammetry experiments were found to follow a similar pattern as was determined using
PESA.
3
.6. Photovoltaic properties

ACCEPTED MANUSCRIPT
The photovoltaic performance of NDICz-4, NDICz-5 and NDICz-6 was investigated by
fabricating solution-processable BHJ devices using the conventional and simple
configuration ITO/PEDOT: PSS/active layer/PFN/Al and with P3HT as a donor. We also
fabricated P3HT: PC BM devices so as to realize the validity of our testing strategy. The
7
1
performance of BHJ devices was optimized by adjusting the blend ratio of donor to acceptor
and by solvent vapor annealing (SVA) technique. It was found that the donor to acceptor
weight ratio of 1: 1.5, respectively, was crucial for optimum performance. The devices
incorporating NDICz-4, NDICz-5 and NDICz-6 as non-fullerene acceptors yielded moderate
outcome and the efficiency numbers reached 2.32 % 1.82 % and 3.83 %, respectively.
However, we preceded the SVA treatment of active layers in order to observe an
improvement of bulk film morphology and hence, device performance. We used carbon
disulfide vapors for SVA treatment before the deposition of PFN/Al cathode layer. After
SVA treatment, we undoubtedly observed a notable change in the blend film morphology
that, as per the literature and our own understanding, led to an improved device performance.
That being said, we conducted atomic force microscopy (AFM) analysis of the pre- and post-
annealed blend surfaces. The AFM analysis clearly indicated that the post-annealed blend
surfaces show lesser roughness, well-plaited donor and acceptor domains and phase-
separated nanoscale morphology. The annealed blend surfaces showed surface roughness of
1
.98 nm (P3HT: NDICz-4), 2.12 nm (P3HT: NDICz-5) and 1.71 nm (P3HT: NDICz-6),
thereby suggesting that all of the newly developed non-fullerene acceptors are highly
miscible with their donor counterpart i.e. P3HT. After the optimization of solvent annealing
time (40 seconds), P3HT: NDICz-4 (1: 1.5), P3HT: NDICz-5 (1:1.5) and P3HT: NDICz-6
(1:1.5) blends showed overall PCE of 3.94 %, 4.94 % and 7.58 %, respectively. It is apparent
that the superior morphological properties of NDICz-6-based blend have had a greater effect
on efficiency when compared with NDICz-4 and NDICz-5-based blends. In contrast, the

ACCEPTED MANUSCRIPT
maximum PCE reached 3.08% for a controlled device based on P3HT: PC BM. For current–
71
voltage (J–V) curves and detailed device parameters, see Fig. 5 and Table 1, respectively. The
actual surface morphologies of the blend films of NDICz-4, NDICz-5 and NDICz-6 (1: 1.5
w/w) are shown in Fig. 6. Fig. S14 (SI) represents the distribution of cell parameters for the
champion cell based on P3HT and NDICz-6 (P3HT: NDICz-6 (1: 1.5); SVA-treated films).
>
Fig. 5<
>Table 1<
Not only did we physically observe the smoothness of blend films, which were free from
perceptible projections, lumps and indentations, but also discern that there may be a structural
reorganization in the active layer (based on AFM analysis) after SVA treatment. The SVA
treated active layers were preceded for optical absorption analysis. The optical absorption
spectra not only revealed the broadening of the shoulder peak but were slightly red-shifted
when compared with the pre-treated blends (Fig. 7). Both features were prominent in case of
NDICz-6-based blend when compared with NDICz-4 and NDICz-5-based blends. The slight
red-shift and the broadness of spectra do point out a structural rearrangement and hence,
improved device performance where the improvement is mainly due to an increase in the
short-circuit current density. Furthermore, to examine the well intermixing of donor and
acceptor domains, we conducted transmission electron microscopy (TEM) analysis. The
TEM analysis did indicate fine, flat textures for all the blends (see Fig. S15, SI), albeit the
appearance and fineness of P3HT: NDICz-6 blend was superior to other two blends. The
TEM images of our champion blend (P3HT: NDICz-6 1: 1.5; pre- and post-SVA treated)
suggested that the post-SVA treated surface is a surface which indicates some sort of

ACCEPTED MANUSCRIPT
crystallinity – a result corroborating AFM analysis – when compared with the unannealed
blend surface (Fig. S16, SI). Not only did the image indicate a surface with more desirable
morphology (post-SVA treated), but advocate that such regularized blend surfaces usually
result in relatively higher values of J and FF, as is the case of NDICz-6 blend. Furthermore,
sc
we conducted the wide-angle X-ray diffraction (WAXRD) analysis in order to observe any
surface crystallinity in the annealed blend. The presence of peaks in NDICz-6-based
annealed blend does suggest some sort of crystallinity, a result that corroborates with AFM
and TEM analysis, when compared with the unannealed blend surface (for WAXRD spectra,
see Fig. S17, SI). This further validates the improved performance of annealed blend surfaces
and suggests that the SVA treatment is helpful for the investigated class of V-shaped, non-
fullerene electron acceptors to achieve higher device outcomes.
>
>
Fig. 6<
Fig. 7<
The devices based on NDICz-4 and NDICz-5 showed the highest and the lowest values of
Voc, respectively, which in fact is in agreement with the LUMO values of these acceptors,
given the V is directly proportional to the energy difference between the HOMO of donor
oc
and the LUMO of acceptor. The difference for P3HT: NDICz-4 is 1.74 eV, whereas for
P3HT: NDICz-5, it is 1.08 eV. The value of J is highest for P3HT: NDICz-6 (11.64
sc
2
2
mA/cm ) and lowest for P3HT: NDICz-4 (6.16 mA/cm ). The latter value may be attributed
to the insufficient driving force (0.26 eV) for exciton dissociation for P3HT: NDICz-4 active
layer, since the driving force for exciton dissociation is related with the energy difference
between the LUMOs of donor and acceptor employed in the active layer which should be
higher than the exciton binding energy (~0.3–0.5 eV). This driving force for P3HT: NDICz-5

ACCEPTED MANUSCRIPT
and P3HT: NDICz-6 is about 0.97 eV and 0.73 eV, respectively, and is larger than the
threshold driving force, leading to the higher values of J when compared with NDICz-4-
sc
based devices.
To ascertain the accuracy of J-V measurements, incident photon to current conversion
efficiencies (IPCEs) were measured and are shown in Fig. 5b. The devices based on P3HT:
NDICz-4 and P3HT: NDICz-5 displayed IPCE spectra up to 600 nm with two bands, i.e. in
shorter (300–400 nm) and longer (400–600 nm) wavelength regions, corresponding to the
photocurrent generation due to the absorption of photons by non-fullerene acceptor and
P3HT, respectively. It can be seen from Fig. 5b that the devices based on P3HT: NDICz-5
showed higher IPCE and broader response when compared with the devices based on P3HT:
NDICz-4. This may be attributed to the broader absorption of NDICz-5, thereby contributing
to an efficient exciton generation and producing more photocurrent. In case of P3HT:
NDICz-6, the IPCE spectrum is much broader together with higher value when compared
with other two combinations. The J values calculated by integrating the IPCE data with the
sc
2
2
2
AM 1.5G solar spectrum are 6.84 mA/cm , 11.41 mA/cm and 13.52 mA/cm for the devices
based on P3HT: NDICz-4, P3HT: NDICz-5 and P3HT: NDICz-6, respectively, which agree
with the values obtained from J−V measurements.
In order to get information about photoinduced charge transfer in active layers, we also
measured the photoluminescence (PL) spectra of pristine P3HT and blended films as shown
in Fig. S18 (SI; excitation wavelength = 520 nm). The pristine P3HT film displayed a strong
emission peak at 656 nm which was quenched after mixing with the investigated acceptors.
The PL quenching was 70%, 87% and 92% for NDICz-4, NDICz-5 and NDICz-6,
respectively. These values follow the J and device performance trends. The lowest value of
sc

ACCEPTED MANUSCRIPT
PL quenching for the P3HT: NDICz-4 blend may be related to the efficient driving force for
this active layer as the LUMO offset is below the threshold value for the exciton dissociation.
To obtain information about the difference in the performance of BHJ devices employing
these non-fullerene acceptors, the relationship between photocurrent (J ) and effective
ph
voltage (V ) was investigated and is reported in Fig. 8 [63]. The J value is defined as the
eff
ph
difference between J and J , where J and J are the current densities under illumination
L
D
L
D
and dark, respectively. The V is estimated from the difference between V and V where V
a
eff
a
o
is the applied voltage and V is the voltage at which J is zero. At low V , the J for devices
o
ph
eff
ph
based on P3HT: NDICz-5 and P3HT: NDICz-6 increase dramatically and reaches a plateau
at high voltage and is subsequently saturated, thus suggesting that free carriers are swept out
efficiently. In comparison, the J of devices based on P3HT: NDICz-4 active layer increases
ph
at a slower rate and is saturated at a high Veff value, which indicates less efficient charge
carrier extraction. This also indicates that the Veff is not high enough to sweep out all the
charge carriers from active layer to the electrodes. Moreover, the saturation photocurrent
(
Jphsat) of P3HT: NDICz-4 devices is lower than the Jphsat of devices based on P3HT: NDICz-
and P3HT: NDICz-6 combinations. The higher Jphsat suggests that the charge carrier
5
generation in the SVA-treated devices is more efficient than the as-cast devices, which in fact
is consistent with their correspondingly higher J values. These observations can be ascribed
sc
to the lower charge recombination after SVA treatment. The exciton dissociation (ηdiss) and
charge collection (ηcoll) efficiencies were estimated from the expressions J /J
at short
ph phsat
circuit and J /J
at maximum power point, respectively. The ηdiss
/
ηcoll values were
.78/0.67, 0.89/0.76 and 0.92/0.78 for the devices based on P3HT: NDICz-4, P3HT: NDICz-
and P3HT: NDICz-6, respectively. The higher values of both the ηdiss and ηcoll for the
ph phsat
0
5
devices based on NDICz-5 and NDICz-6 indicate that exciton dissociation and charge

ACCEPTED MANUSCRIPT
collection is more efficient in these solar cells when compared with the devices based on
P3HT: NDICz-4 combination. In general, at high value of V , the ηdiss and ηcoll efficiencies
eff
approach to unity, the Jphsat mainly depends on the exciton generation rate (Gmax) and is
expressed as Jphsat = qLGmax, where q is the elementary charge and L is the thickness of the
2
8
-3 -1
2
active layer. The Gmax was calculated to be 0.62 × 10 m s (J
= 8.92 mA/cm ),
phsat
2
8
-3 -1
2
28
-3 -1
2
0
.82×10 m s (J
= 12.65 mA/cm ) and 1.02×10 m s (J
= 14.70 mA/cm ) for
phsat
phsat
the devices based on ND1Cz-4, NDICz-5 and NDICz-6, respectively. The enhanced value of
Gmax suggests that more excitons are generated in the blend film of NDICz-6, a result that is
consistent with the higher performance of P3HT: NDICz-6 when compared with other two
combinations.
>
Fig. 8 <
To gain an insight into the effective charge carrier mobilities, the space charge limited current
SCLC) method was applied to get information about the charge transportation in the devices.
(
The electron-only devices, consisting of active layer sandwiched between a ZnO coated ITO
electrode and LiF/Al counter-electrode as the hole-blocking contact, were fabricated as per
the sketch depicted in Fig. S19, SI. We fabricated these devices using NDICz-5 and NDICz-
6
as only these materials gave reasonable OPV performance. From the current density as a
function of voltage data (Fig. S19, SI), the electron mobility in the trap-free SCLC region can
2
3
be estimated using the Mott-Gurney equation, [(J = 9 (εµ)/8 × (V /d ); where ε is the
dielectric constant, µ is the charge-carrier mobility, d is the sample thickness, and V is the
-4
2
applied voltage]. Using this expression, the electron mobilities of the order of 10 cm /Vs
-3
2
and 10 cm /Vs were observed for NDICz-5 and NDICz-6, respectively. It is vital to

ACCEPTED MANUSCRIPT
mention that the mobilities of the obtained order, and P3HT: NDICz-6 in particular, are well
-3
-4
2
-1 -1
within the mobility range exhibited by PC BM (typically ranging 10 –10 cm V s ), and
7
1
are an indication of good device performance. Evidently, the observance of good electron
mobilities further supports the design principle of these new materials.
4
. Conclusions
We have been able to demonstrate the successful use of an A-A -D-A type modular format in
1
BHJ solar cells. The new materials NDICz-5 and NDICz-6 were designed and synthesized
based on the A-A -D-A format and their optoelectronic and photovoltaic properties were
1
directly compared with a structural analogue NDICz-4, which was based on a simple A-D-A
modular design. All the materials were designed to be V-shaped. It was observed that
NDICz-5 and NDICz-6 exhibited superior properties, such as light-harvesting, enhanced
photocurrent density and overall device performance, when compared with NDICz-4.
Overall, NDICz-6 performed better than any of the three materials reported herein. It is
notable to mention that not only are NDICz-4, NDICz-5 and NDICz-6 the first examples in
the literature which comprise promising building blocks, such as carbazole and naphthalene
diimide, but the device parameters outlined herein are among the highest numbers reported in
this class of materials.
Acknowledgements
S.V.B. (IICT) is grateful for the financial support from SERB, SB/S1/IC-009/2014, New
Delhi, India [IICT Communication Number IICT/Pubs./2018/178]. S.V.B. (GU)
acknowledges University Grant Commission (UGC) – Faculty Research Program (India) –
for providing financial support and an award of professorship. P.S.R. acknowledges SRF
fellowship from CSIR, New Delhi. A.L.P acknowledges the use of the Gaussian 09 procured
under the DST-FIST Scheme (Sanction No. FS/FST/PSI-018/2009). R.S.B. acknowledges

ACCEPTED MANUSCRIPT
financial support from the CSIR, New Delhi under the SRA scheme [13(8772)-A)/2015-
Pool]. J. Li acknowledges the Australian Research Council (ARC) for support through a
Future Fellowship project (FT130100057). The CSIRO Division of Manufacturing, Clayton
Victoria, Australia is acknowledged for providing support through a visiting fellow position
for A. G. A. G. acknowledges a vast variety of facilities at Deakin University, RMIT
University and Bio 21 Institute, the University of Melbourne, Melbourne, Victoria Australia.
References
[
1] Yu G, Gao J, Hummelen JC, Wudl F, Heeger AJ. Polymer photovoltaic cells: enhanced
efficiencies via a network of internal donor-acceptor heterojunctions. Science
995;270:1789–91.
2] Morel DL, Ghosh AK, Feng T, Stogryn EL, Purwin PE, Shaw RF, Fishman C. High-
efficiency organic solar cells. Appl Phys Lett 1978;32:495.
3] Li Y. Molecular design of photovoltaic materials for polymer solar cells: toward suitable
electronic energy levels and broad absorption. Acc Chem Res 2012;45:723-3.
4] Ye L, Zhang S, Huo L, Zhang M, Hou J. Molecular design toward highly efficient
1
[
[
[
photovoltaic polymers based on two-dimensional conjugated benzodithiophene. Acc Chem
Res 2014;47:1595–1603.
[
5] Badgujar S, Song CE, Oh S, Shin WS, Moon S-J, Lee J-C, Jung IH, Lee SKJ. Highly
efficient and thermally stable fullerene-free organic solar cells based on a small molecule
donor and acceptor. Mater Chem A 2016;4:16335–40.
[
6] Krebs FC, Espinosa N, Hösel M, Søndergaard RR, Jørgensen M. 25th Anniversary article:
rise to power–OPV‐ based solar parks. Adv Mater 2013;26:29–39.
7] Heeger AJ. 25th anniversary article: bulk heterojunction solar cells: understanding the
mechanism of operation. Adv Mater 2013;26:10–27.
[