Phosphorescent light-emitting diodes using triscarbazole/bis(oxadiazole) hosts: Comparison of homopolymer blends and random and block copolymers

Article abstract of DOI:10.1039/c4tc01079e

Examples of blends of carbazole- and bis(oxadiazole)benzene-based side-chain polymers have recently been reported to be efficient host materials for phosphorescent emitters in organic light-emitting diodes. Here, the properties and performance of a physical blend of polynorbornene homopolymers with triscarbazole and bis(oxadiazole)benzene side chains are compared to those of random and block copolymers of the corresponding triscarbazole- and bis(oxadiazole)benzene-functionalized monomers. Green-emitting devices in which the blend is used a host for Ir(ppy)₃ are significantly more efficient than those based on copolymers. Differential scanning calorimetry and solid-state NMR data show that there is no macroscale separation between the two polymers in the blend. The NMR data suggest that there are significant differences in the dimensionality and characteristic length of nanoscale domain structures in the block copolymer and the blend. Use of Ir(pppy)₃ in place of Ir(ppy)₃ leads to even more efficient light-emitting diodes, with external quantum efficiencies of up to ca. 21% (at 100 cd m^-2). This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4tc01079e

Journal of
Materials Chemistry C
View Article Online
View Journal | View Issue
PAPER
Phosphorescent light-emitting diodes using
triscarbazole/bis(oxadiazole) hosts: comparison of
homopolymer blends and random and block
copolymers†
Cite this: J. Mater. Chem. C, 2014, 2,
6743
Xuyang He,a Dengke Cai,b Dun-Yen Kang,‡a Wojciech Haske,b Yadong Zhang,a
Carlos A. Zuniga,a Benjamin H. Wunsch,a Stephen Barlow,a Johannes Leisen,a
c
b
b
*
*
David Bucknall, Bernard Kippelen and Seth R. Marder
Examples of blends of carbazole- and bis(oxadiazole)benzene-based side-chain polymers have recently
been reported to be efficient host materials for phosphorescent emitters in organic light-emitting
diodes. Here, the properties and performance of a physical blend of polynorbornene homopolymers
with triscarbazole and bis(oxadiazole)benzene side chains are compared to those of random and block
copolymers of the corresponding triscarbazole- and bis(oxadiazole)benzene-functionalized monomers.
Green-emitting devices in which the blend is used a host for Ir(ppy)3 are significantly more efficient than
those based on copolymers. Differential scanning calorimetry and solid-state NMR data show that there
is no macroscale separation between the two polymers in the blend. The NMR data suggest that there
are significant differences in the dimensionality and characteristic length of nanoscale domain structures
in the block copolymer and the blend. Use of Ir(pppy)3 in place of Ir(ppy)3 leads to even more efficient
light-emitting diodes, with external quantum efficiencies of up to ca. 21% (at 100 cd mꢀ2).
Received 23rd May 2014
Accepted 24th June 2014
DOI: 10.1039/c4tc01079e
www.rsc.org/MaterialsC
Hosts with ambipolar charge-transporting ability are poten-
tially advantageous over unipolar host materials in providing a
Introduction
broader recombination zone in the emissive layer, thereby
reducing the possibility of triplet–triplet quenching.3 In some
cases the use of ambipolar hosts can facilitate simpler device
structures.4 A widely used approach to ambipolar materials is to
combine electron-transporting and hole-transporting moieties5
either through a covalent linker – in small-molecules or as
copolymers – or as a physical blend – either of small molecules,
a small molecule and a polymer, or of different polymers.
Carbazoles and oxadiazole derivatives are among the most
widely used groups for hole- and electron-transporting host
components, respectively. There are various ways to incorporate
both moieties: as blends of polymeric carbazole derivatives,
such as poly(N-vinylcarbazole) (PVK), with small-molecule or
polymeric oxadiazole derivatives,4a,4b,4e,6 as random and block
copolymers of carbazole- and oxadiazole-based monomers,7 and
as carbazole-backbone polymers with oxadiazole pendants.8
Molecular hosts have also been used in which carbazole and
oxadiazole moieties are connected in an either conjugated or
non-conjugated manner.9 We have reported that physical
blends of carbazole- and oxadiazole-based polymers can be used
as solution-processible hosts for green-emitting iridium phos-
phors in efficient PHOLEDs. Devices using crosslinked poly-
TPD-F as a hole-transport layer, a blend of poly(N-vinyl-
carbazole) (PVK), poly(norbornene)s with 1,3-bis(5-phenyl-1,3,4-
The emissive layers of phosphorescent organic light-emitting
diodes (PHOLEDs) typically incorporate heavy-metal emitters,
oen iridium(III) complexes1 diluted in a suitable host at rela-
tively low concentration to reduce the incidence of triplet–
triplet annihilation.2 In general, the triplet and singlet excited-
state energies of the hosts should be higher than the corre-
sponding energies of the emitters to favor forward energy
transfer to the emitter and minimize back energy transfer from
the emitter to the host. In addition, it is desirable that the glass-
transition temperature of the host be sufficiently high to
maintain a stable morphology of the emissive layer in the solid-
state device at operational temperatures.
aCenter for Organic Photonics and Electronics and School of Chemistry and
Biochemistry, Georgia Institute of Technology, Atlanta, GA 30332-0400, USA. E-mail:
seth.marder@chemistry.gatech.edu
bCenter for Organic Photonics and Electronics and School of Electrical and Computer
Engineering, USA. E-mail: bernard.kippelen@ece.gatech.edu
cCenter for Organic Photonics and Electronics and School of Materials Science and
Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0245, USA
† Electronic supplementary information (ESI) available: Synthesis of monomers
and of Sty2-TCz, solid-state NMR spectra and details of spin-diffusion
experiments, table of absorption/uorescence maxima, gures showing OLED
performance. See DOI: 10.1039/c4tc01079e
‡ Present address: Department of Chemical Engineering, National Taiwan
University, Taipei, Taiwan.
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. C, 2014, 2, 6743–6751 | 6743

View Article Online
Journal of Materials Chemistry C
Paper
oxadiazol-2-yl)benzene side chains (I), and Ir(ppy)3 or Ir(pppy)3 (1-co-2) and diblock copolymers (1-b-2) and present photo-
phosphors (see Fig. 1 for structures) as the emissive layer, and a physical and solid-state NMR data that show differences
vacuum-deposited BCP electron-transport layer exhibited EQEs between these different approaches to host materials.
approaching 13%.6c More recently, we have also achieved
similar performance using a blend of substituted polystyrenes
with 3,6-bis(carbazol-9-yl)carbazole (“triscarbazole”) and 1,3-
Experimental section
1H NMR and 13C{1H} NMR spectra were recorded on a 400 MHz
bis(5-phenyl-1,3,4-oxadiazol-2-yl)benzene side chains (II and III,
Fig. 1) as the host for Ir(pppy)3, in conjunction with hole-
Bruker DMX spectrometer and were referenced using the
transport layers based on copolymers of styrene monomers
residual solvent proton and carbon signals. Elemental analyses
substituted with triscarbazole and oxetane or benzocyclobutene
were carried out by Atlantic Microlab Inc. Mass spectra (EI) were
crosslinkable side chains, and a BCP electron-transport layer.10
recorded on a Micromass Autospace mass spectrometer and
Here we directly compare the performance as a host of a related
Applied Biosystems 4700 Proteomics Analyzer mass spectrom-
blend of polymers (poly-1/poly-2; Scheme 1) with that of random
eter. The molecular weights of the polymers were estimated in
tetrahydrofuran (THF) by using gel permeation chromatog-
raphy using a Waters 1515 pump, a Waters 2489 UV-vis detector,
a Styragel HR 5E THF 4.6 ꢁ 300 mm column, and linear poly-
(styrene) standards. Differential scanning calorimetry (DSC)
data were ꢀo1btained using a TA DSCQ200 at a heating rate of
ꢂ
ꢂ
10 C min from 40 to 300 C under a nitrogen atmosphere.
Thermogravimetric analysis (TGA) was performed on ꢀa1
NETZSCH STA 449C instrument at a heating rate of 20 ꢂC min
from 20 to 500 ꢂC. UV-vis absorption spectra were measured on
a Hewlett-Packard 8453 spectrometer. Fluorescence spectra
were recorded on a Fluorolog III ISA spectrouorimeter.
1H spin diffusion experiments and simulations
The samples for the 1H spin diffusion experiments were
prepared from drop-casting the as-synthesized polymers dis-
solved in chlorobenzene with a concentration of 10 mg mLꢀ1
into a glass vial. The vials with solution were placed in a
ꢂ
ꢂ
convection oven at 50 C for 12 h followed by at 120 C for 6 h
yielding solid-state polymer samples. Solid-state samples (ca. 50
mg portions) obtained in this way were used for NMR experi-
ments. The experiments were performed using a Bruker DSX
300 NMR spectrometer with a statiꢂc H/X broadband probe.
Experiments were conducted at 190 C, at which temperature
the oxadiazole domain exhibits signicant large-angle molec-
ular motions with frequencies larger than >ca. 10 kHz in
comparison to the rigid triscarbazole domain according to the
spin-spin relaxation results, which showed two distinct relaxa-
tion times assigned to two domains. The conduct of the spin-
diffusion experiment follows closely the experimental approach
as described by Mellinger et al.11 Briey, the magnetization of
the mobile poly-1 domain is selected through the repeated
application of a rf-pulse train consisting of 12 p/2-pulses
(dipolar lter), which is followed by a p-pulse to align magne-
tization along the z-axis, a mixing time and a readout pulse
followed by the detection. A good selection was achieved for n ¼
4 pulse trains, where individual pulses with a duration of 3.6 ms
were separated by a delay of 20 ms. IDF, the amount of magne-
tization corresponding to poly-1, was determined by integration
over the corresponding narrow peak. Contribution of the
longitudinal relaxation T1 were eliminated by conducting the
same experiment without the dipolar lter (i.e. n ¼ 0) measuring
an intensity I0DF. The (normalized) ratio IDF/ID0 F as displayed in
Fig. 5 will then provide a reliable measurement of the
Fig. 1 Structures of some polymers previously used in blends as hosts,
along with those of phosphors and transport materials discussed in this
paper.
6744 | J. Mater. Chem. C, 2014, 2, 6743–6751
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry C
Scheme 1 Synthesis and polymerization of 1 and 2.
magnetization leaving the poly-1 phase through spin diffusion.
Spin-diffusion coefficients were estimated from measured T2-
relaxation constants using the calibration curves.11 T2-relaxation
data were measured using a regular spin-echo sequence. This
resulted in a biexponential decay, providing the relaxation
constants for both phases (details are presented in the ESI†).
The 1H spin diffusion between two domains was modeled by
Fick's law of diffusion:12
vM
¼ ꢀDVM
(1)
vt
where M is the normalized magnetization (a.u.), t is time (ms),
and D is the spin diffusion coefficient (nm2 msꢀ1). The advan-
tage of this approach is that it can be readily expanded for the
study of more complex geometries. For the domain geometry,
two line segments, two concentric squares, and two concentric
cubics were regarded as the representative geometries for the
one-, two-, and three-dimensional diffusion process respec-
tively. Fick's law was applied as the governing equation for the
spin diffusion with the known diffusion coefficients obtained
from the spin-spin relaxation experiments. Continuous
Fig. 2 Absorption (solid squares) and emission (open squares, exci-
tation wavelength of 280 nm) spectra of poly-1 (black), poly-2 (olive),
poly-1-co-2 (orange), poly-1-b-2 (blue), and 1 : 1 poly-1/poly-2 (red)
in CH2Cl2.
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. C, 2014, 2, 6743–6751 | 6745

View Article Online
Journal of Materials Chemistry C
Paper
Fig. 3 Absorption (dotted lines) and emission (solid lines, excitation
wavelength of 280 nm) spectra of poly-1 (olive), poly-2 (black), poly-1-
co-2 (orange), poly-1-b-2 (blue), and 1 : 1 poly-1/poly-2 (red) in the
solid state.
Fig. 5 (a) Illustration of the key steps in the 1H spin diffusion process.
(b) Experimental (black circle) and simulated (red curve) results of the
normalized magnetization of the mobile phase (poly-2 domain) as a
function of the spin diffusion time.
volume, HNO3 : HCl) for 5 min at 60 ꢂC. The substrates were
cleaned in an ultrasonic bath of detergent water, rinsed with
deionized water, and then cleaned in sequential (20 min each)
ultrasonic baths of deionized water, acetone, and isopropanol.
Nitrogen was used to dry the substrates aer each of the last
three baths. The substrates were treated with an O2 plasma for 3
min prior to the deposition of the hole-transport material.
For the green-emitting devices 35 nm thick lms of poly-
TPD-F13 were then spin coated (60 s at 1500 rpm, acceleration
10 000 rpm sꢀ1) from a solution of poly-TPD-F (10 mg) in
chloroform (1 mL, 99.8% purity, distilled and oxygenated) in a
N2-lled glove box. Aer spin-coating, a rectangular strip of the
layer was removed at the edge of the substrate to expose ITO and
ensure electrical contact to the anode; then, the sample was
dried in vacuum and transferred back into the glove-box, where
it was baked for 15 min at 75 ꢂC on a hot plate, aer which the
hot plate was turned off. The sample was removed from the hot
plate when its temperature was down to 40 ꢂC. Finally the
sample was exposed to 0.7 mW cmꢀ2 of UV illumination for 1
min to crosslink the hole-transport layer.
For the emissive layer, Ir(ppy)3 or Ir(pppy)3 (6 or 12 wt%) was
mixed with the host material, and all materials dissolved in
chlorobenzene (1 mL, 99.8% purity, distilled and then oxygen-
ated). 40–50 nm thick lms were then spin coated (60 s at 1000
rpm, acceleration 10 000 rpm sꢀ1) onto the UV-crosslinked poly-
TPD-F layer. Aer spin-coating, the samples were baked at 75 ꢂC
for 15 min. Chlorobenzene was then used to remove the emis-
sive layer in the area not covered by poly-TPD-F, exposing the
ITO substrate to provide electrical contact to the anode. The
samples were then transferred, under a N2 atmosphere, into a
SPECTROS (Kurt J. Lesker) thermal-deposition system directly
connected to the wet-glove box.
Fig. 4 1 DSC plots (exothermic up) of poly-1, poly-2, poly-1-co-2,
poly-1-b-2, and 1 : 1 poly-1/poly-2 obtained at a heating rate of
10 ꢂC minꢀ1
.
boundary conditions were applied at the interfaces of the two
domains. The normalized magnetization for the oxadiazole
domain was taken as one and the triscarbazole domain was
taken as zero as the initial condition. The transient spin-diffu-
sion equation (eqn (1)) was solved for the given domain sizes
numerically using the widely used soware package COMSOL®.
The domain-average magnetization intensity for the triscarba-
zole domain was then obtained from the numerical solution as
a function of time and then compared with the experimental
results.
OLED fabrication
For the hole-blocking/electron-transport layer, a 40 nm thick
BCP layer was vacuum deposited at a pressure below 2 ꢁ 10ꢀ7
Indium tin oxide (ITO)-coated glass (Colorado Concept Coat-
ings LLC) with a sheet resistivity of ꢃ15 U sqꢀ1 was used as the
substrate for the OLEDs fabrication. The ITO substrates were
patterned with Kapton tape and etched in acid vapor (1 : 3 by
Torr and a rate of 0.4 A sꢀ1. Then, a 2.4 nm layer of lithium
˚
uoride (LiF), as an electron-injection layer, and a 200 nm-thick
6746 | J. Mater. Chem. C, 2014, 2, 6743–6751
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry C
aluminum cathode were vacuum deposited through a shadow polymer. The crude product was redissolved in dichloro-
ꢀ1
mask at a pressure below 3 ꢁ 10ꢀ7 Torr and at rates of 0.15 A s
methane (3 mL) and then reprecipitated into methanol (200
˚
and 2 A sꢀ1, respectively. The shadow mask used for the evap- mL) three times and dried to give an off-white solid (0.775 g,
oration of the metal electrodes yields ve devices with an area of 82%). 1H NMR (400 MHz, CDCl3): d 8.69 (m, br, 1H), 8.12 (m, br,
ca. 0.1 cm2 per substrate. The device testing was performed in 4H), 8.03 (m, br, 6H), 7.50 (m, br, 9H), 7.27 (m, br, 8H), 7.15 (m,
an inert atmosphere and without exposing the devices to air br, 5H), 6.98 (m, br, 1H), 5.18 (m, br, 4H), 4.31 (m, br, 2H), 3.89
˚
immediately following the deposition of the metal cathode.
(m, br, 2H), 2.83–1.53 (m, br, 14H), 1.48–1.04 (m, br, 25H). Anal.
Blue-emitting devices were fabricated in an analogous calcd for C86H81N7O3: C, 81.94; H, 6.48; N, 7.78. Found: C, 81.94;
fashion, except: the ITO was treated with penta- H, 6.29; N, 7.67%.
uorobenzylphosphonic acid (PFBPA)14 as previously
Poly-1-b-2. A solution of the 1st generation Grubbs initiator
described15 prior to depositing the hole-transport material; Sty2- (0.0035 g, 0.0042 mmol) in dry CH2Cl2 (2 mL) was added
TCz (see Fig. 1 for structure, ESI† for synthesis) was used as the dropwise to a solution of 2 (0.2726 g, 0.454 mmol) in dry CH2Cl2
hole-transport material in place of poly-TPD-F; FIrpic was used (4 mL) under nitrogen. Aer 30 min the reaction appeared
as the emitter; and crosslinking was carried out by heating to complete by TLC and the reaction mixture (6 mL) was added to
ꢂ
200 C for 30 min.
the second monomer 1 (0.3055 g, 0.463 mmol). Aer 30 min the
polymerization was quenched with vinyl ethyl ether (0.2 mL)
Then the reaction mixture was added dropwise to rapidly stirred
acetone (200 mL) to precipitate the polymer. The crude product
was redissolved in CH2Cl2 (3 mL) and then reprecipitated from
Synthesis of polymeric host materials
The synthesis of monomers 1 and 2 are given in the ESI.†
Poly-1. A solution of 1st generation Grubbs catalyst (0.0290 g, acetone 200 mL; this process was repeated twice, but using
0.035 mmol) in dry CH2Cl2 (5 mL) was added dropwise to a methanol (200 mL) in place of acetone, and the precipitate was
solution of 1 (2.5401 g, 3.85 mmol) in dry CH2Cl2 (15 mL) under dried to give an off-white solid (0.454 g, 79%). 1H NMR (400
nitrogen. Aer 40 min the reaction was quenched by vinyl ethyl MHz, CDCl3): d 8.69 (m, br, 1H), 8.19–8.01 (m, br, 11H), 7.60–
ether (1.0 mL). Then the reaction mixture was added dropwise 7.24 (m, br, 13H), 7.13–6.98 (m, br, 9H), 5.27 (m, br, 2H), 5.11
to rapidly stirred methanol (400 mL) to precipitate the polymer. (m, br, 2H), 4.14 (m, br, 2H), 3.95 (m, br, 2H), 2.88–2.15 (m, br,
The crude product was redissolved in dichloromethane (10 mL) 5H), 1.82–1.75 (m, br, 9H), 1.40–0.88 (m, br, 25H). Anal. calcd
and then reprecipitated into acetone (400 mL); this process was for C86H81N7O3: C, 81.94; H, 6.48; N, 7.78. Found: C, 82.20; H,
repeated twice, but using methanol (400 mL) in place of 6.43; N, 7.67%.
acetone, and the precipitate dried to give an off-white solid
1
(2.258 g, 89%). H NMR (400 MHz, CDCl3): d 8.08 (m, br, 2H),
8.02 (m, br, 4H), 7.45 (m, br, 4H), 7.24 (m, br, 8H), 7.14 (m, br,
4H), 5.11 (m, br, 2H), 4.16 (m, br, 2H), 2.70–1.75 (m, br, 7H),
Results and discussion
Synthesis and thermal characterization
1.24–0.93 (m, br, 8H). Anal. calcd for C48H41N3: C, 87.37; H, We chose to use norbornene side-chains due to the potentially
6.26; N, 6.37. Found: C, 86.84; H, 6.23; N, 6.38%.
living nature of ring-opening metathesis polymerization
Poly-2. A solution of 1st generation Grubbs initiator (0.0339 (ROMP),16 which we anticipated would enable homo- and
g, 0.041 mmol) in dry CH2Cl2 (5 mL) was added dropwise to a copolymers to be synthesized with similar molecular weights
stirred solution of 2 (2.5406 g, 4.23 mmol) in CH2Cl2 (15 mL) and would permit the synthesis of well-dened diblock copoly-
under nitrogen. Aer 40 min the reaction was quenched by the mers. The triscarbazole and m-bis(diaryloxadiazole)benzene
additional of vinyl ethyl ether (1.0 mL). The reaction mixture side chains were chosen since in our previous work10 a blend of
was then added dropwise to rapidly stirred methanol (400 mL) polymers containing these moieties was successfully used as
to precipitate the polymer. The crude product was redissolved in polymer blend host for green phosphorescent emitters. A long
dichloromethane (10 mL) and then reprecipitated into acetone exible spacer was introduced between the host moieties and
(400 mL); this process was repeated twice, but using methanol the polymerizable group in an attempt to reduce any inuence
(400 mL) in place of acetone, and the precipitate was dried to of the side chain on the polymerization rate of the monomer
give an off-white solid (2.324 g, 92%). 1H NMR (400 MHz, and hence our ability to obtain random copolymers. As shown
CDCl3): d 8.69 (m, br, 1H), 8.19 (m, br, 2H), 8.00 (m, br, 2H), in Scheme 1, monomers 1 and 2 were synthesized by nucleo-
7.59–7.49 (m, br, 5H), 7.33 (m, br, 1H), 6.98 (m, br, 1H), 5.26 (m, philic substitution reactions of 5-(5-bromopentyl)-norbornene
br, 2H), 3.96 (m, br, 2H), 2.91–1.73 (m, br, 6H), 1.59–0.86 (m, br, (ca. 17 : 3 endo : exo isomer mixture) with triscarbazole and a
18H). Anal. calcd for C38H40N4O3: C, 75.97; H, 6.71; N, 9.33; hydroxyl-functionalized m-bis(diaryloxadiazole)benzene, which
Found: C, 75.91; H, 6.69; N, 9.34%.
was, in turn, obtained through a multi-step synthesis (see ESI†
Poly-1-co-2. A solution of the 1st generation Grubbs initiator for details), respectively. The homopolymers, poly-1 and poly-2,
(0.0062 g, 0.0075 mmol) in dry CH2Cl2 (4 mL) was added and two copolymers, poly-1-co-2 and poly-1-b-2, were synthe-
dropwise to a solution of 1 (0.4956 g, 0.752 mmol) and 2(0.4549 sized using 1 mol% of the “rst-generation” Grubbs initiator in
g, 0.757 mmol) in dry CH2Cl2 (8 mL). Aer 4 h the reaction dichloromethane under nitrogen atmosphere at room temper-
appeared complete according to TLC and was quenched with ature. The polymers were puried by repeated dissolution-
vinyl ethyl ether (0.5 mL). Then the reaction mixture was added precipitation processes. All polymers were characterized by
dropwise to rapidly stirred methanol (200 mL) to precipitate the 1H NMR spectroscopy, gel-permeation chromatography (GPC),
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. C, 2014, 2, 6743–6751 | 6747

View Article Online
Journal of Materials Chemistry C
Paper
and elemental analysis. The number- and weight-average charge-transfer complex or exciplex formed between the oxadi-
molecular weights (Mn and Mw respectively) and the poly- azole acceptor and triscarbazole donor side chains. The poor
dispersity indices (PDI ¼ Mw/Mn) are shown in Table 1. The PDIs overlap between emission from the low-lying state and the
range from 1.5 to 1.6 and the average degrees of polymerization absorption spectrum of Ir(ppy)3 (onset of absorption ¼ ca. 500
range from approximately 30 to 60. The decomposition nm) may impair the efficiency of singlet-energy transfer to this
temperatures (Td) and the glass-transition temperatures (Tg) phosphor; moreover, this state may be accompanied by an even
were measured under nitrogen by thermal gravimetric analysis lower lying triplet charge-transfer state, also affecting triplet-
(TGA) and differential scanning calorimetry (DSC), respectively, energy transfer (Ir(ppy)3 adiabatic triplet energy ¼ 2.4 eV18). The
and are also summarized in Table 1: all polymers possess absorption and emission spectra of spin-coated lms of the
relatively high Td values in the vicinity of 400 ꢂC. The glass- polymers were also measured at room temperature and are
transition temperature of poly-1 (120 ꢂC) is much lower than shown in Fig. 3. Oxadiazole-to-triscarbazole energy transfer
that of poly-2 (206 ꢂC), consistent with the fused aromatic rings appears to be more signicant for lms of the copolymers and
of the triscarbazole moieties leading to a less exible structure blends than in solution. The separate low-energy feature seen in
than in the oxadiazole material, where ve aromatic rings are solution for the random copolymer and attributed to a charge-
connected by single bonds in a linear fashion. The glass-tran- transfer state is not discernable in the lm spectrum; however,
sition temperatures of both copolymers and of a 1 : 1 poly-1/ the changes observed between thin-lm emission spectra of
poly-2 blend are intermediate between those of the two homo- poly-1, the blend, poly-1-b-2, and poly-1-co-2 are consistent with
polymers; however, while poly-1-co-2 shows a comparably sharp increasing contributions from a feature at low energy relative to
glass transition to the homopolymers, those for the block that from the triscarbazole moieties, also consistent with a
copolymer and the physical blends are considerably broader.
possible charge-transfer state.
Optical properties
OLED performance
The absorption and emission spectra of the polymers (Fig. 2,
Table S2, ESI†) were measured in dilute CH2Cl2 solution at
room temperature. The absorption and emission of these
polymers with oxadiazole/triscarbazole side chains are similar
to those of previously reported materials containing similar
structural motifs.10,17 As shown in Fig. 2, the absorption spectra
of both random and block copolymers are very similar to the
spectrum of a 1 : 1 blend of the homopolymers, which in turn is
essentially the sum of the spectra of the unblended homopol-
ymers, indicating very little ground-state interaction between
the triscarbazole and oxadiazole moieties in the copolymers.
The emission spectrum of the block copolymer is also similar to
that of the blend in showing features characteristic of both
triscarbazole and bis(oxadiazolyl)benzene motifs. That of the
random polymer is qualitatively different, rstly in that the
oxadiazole-based emission is much weaker than that from the
triscarbazole moiety, suggesting signicant oxadiazole-to-tris-
carbazole energy transfer, as expected if the two groups are
indeed randomly distributed in the polymer, and secondly, in
showing an additional broad emission with a maximum at ca.
475 nm (2.6 eV). This is tentatively attributed to emission from a
A series of green-emitting OLEDs were fabricated using poly-1-
co-2, poly-1-b-2, and a 1 : 1 poly-1/poly-2 blend as hosts for
Ir(ppy)3. The general device structure of the OLEDs was indium
tin oxide (ITO)/poly-TPD-F (35 nm)/host:Ir(ppy)3 (6 or 12 wt%)
(40 nm)/BCP (40 nm)/LiF/Al. Poly-TPD-F13 (Fig. 1), which we
have used extensively in previous work,19 was used as a photo-
crosslinkable hole transporting material, and 2,9-dimethyl-4,7-
diphenyl-1,10-phenanthroline, BCP (Fig. 1), was used as the
electron-transport material. The performance of the OLEDs is
summarized in Table 2, along with that of green- and blue-
emitting devices with the structures ITO/poly-TPD-F (35 nm)/
host:Ir(pppy)3 (6 wt%) (40 nm)/BCP (40 nm)/LiF/Al and ITO/
PFBPA/Sty2-TCz (35 nm)/host:FIrpic (6 wt%) (40 nm)/BCP (40
nm)/LiF/Al, respectively.
All of the Ir(ppy)3 devices gave the characteristic green
emission of phosphorescent guests and possessed turn-on
voltages lower than 8.5 V. Although, as discussed above, the
random and block copolymers show rather different emission
properties in solution and, to a lesser extent, in lms, their
performance as hosts in OLEDs is quite similar (entries 1 and 2
vs. 3 and 4), with their relative efficiency depending on the
phosphor loading. In previous work, more signicant differ-
ences in EQE were observed between block and random
copolymers with triphenylamine and diaryloxadiazole side
chains as hosts for either (tpy)2Iracac or FPt in single-layer
OLEDs, with much higher EQEs being seen for the block poly-
mers.20 The present devices based on the 1 : 1 blend of polymers
(entries 5 and 6), particularly those employing a lower phosphor
loading, exhibit signicantly higher EQEs than the corre-
sponding random (entries 1 and 2) or block copolymer devices
(entries 3 and 4). The maximum EQE of 16.8% observed for the
6 wt% Ir(ppy)3 blend (entry 5) is considerably higher than the
corresponding values reported for analogous devices using
1 : 1.1 w/w PVK/I mixtures as hosts (9.6–10.1%, values that are
Table 1 GPC and thermal data
b
c
Polymer
Mwa/kDa
Mna/kDa
PDI
Td /ꢂC
Tg /ꢂC
Poly-1
Poly-2
Poly-1-co-2
Poly-1-b-2
1 : 1 poly-1/poly-2
35
58
74
62
n/a
23
36
47
38
n/a
1.5
1.6
1.6
1.6
n/a
445
397
423
410
405
206
120
154
144
136
a
Molecular weights estimated by GPC calibrated by polystyrene
b
standards. Measured by TGA, dened at temꢀp1erature at which 5%
c
mass loss is seen at a heatingꢀr1ate of 20 ꢂC min
.
Measured by DSC
at a heating rate of 10 ꢂC min
.
6748 | J. Mater. Chem. C, 2014, 2, 6743–6751
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry C
Table 2 Performance of OLEDs with triscarbazole/m-bis(diaryloxadiazole)benzene polymer hosts
Vturn-on
EQE (%)
Max. EQE
(%)
Max. current
Max. power efficacy/
Entry
Polymeric host
Wt% guest
(10 cd mꢀ2)/V
at 100 cd mꢀ2
efficacy/cd Aꢀ1
lm Wꢀ1
1
2
3
4
5
6
7
8
Poly-1-co-2
Poly-1-co-2
Poly-1-b-2
Poly-1-b-2
1 : 1 poly-1/poly-2
1 : 1 poly-1/poly-2
1 : 1.5 poly-1/poly-2
1.5 : 1 poly-1/poly-2
1 : 1 poly-1/poly-2
1 : 1.5 poly-1/poly-2
1.5 : 1 poly-1/poly-2
6% Ir(ppy)3
12% Ir(ppy)3
6% Ir(ppy)3
12% Ir(ppy)3
6% Ir(ppy)3
12% Ir(ppy)3
6% Ir(pppy)3
6% Ir(pppy)3
6% FIrpic
7.4
8.5
7.2
6.6
8.0
6.0
6.5
6.0
8.0
9.2
8.2
6.0
6.7
6.8
5.5
15.9
9.0
16.4
21.0
2.5
1.4
2.9
6.2
7.5
7.2
6.6
16.8
9.4
17.2
28.0
2.9
1.4
3.7
23
28
17
16
34
35
50
66
8.6
9.7
6.3
6.7
10
13
22
31
9
10
11
9.9
5.0
12.7
2.8
1.4
3.3
6% FIrpic
6% FIrpic
still higher than those obtained in the present study using approach can also be successfully applied to blue-emitting
copolymers),6c although the turn-on voltage is a little higher in devices.
the present devices and the current efficacy is only a little
higher.
Microstructure of the host lms
It has previously been suggested that random or diblock
Differences in domain size and shapes in polymer blends and in
block copolymers can potentially affect OLED performance in a
variety of ways,18 inuencing charge transport, charge recom-
bination, and, depending on the location of the phosphors
relative to the charge-recombination sites, the efficiency of
energy transfer to the phosphor. To investigate the micro-
structures of poly-1-b-2 and a 1 : 1 poly-1/poly-2 blend, we
investigated lms of the materials using DSC and solid-state
NMR. As discussed and shown above (Fig. 4) the thermal
behavior of the 1 : 1 poly-1/poly-2 blend is similar to that of poly-
1-b-2; in contrast to the homopolymers and the random copoly-
mers, which possess a well-dened glass transitions, these
materials exhibit a much broader less-well-dened glass tran-
sition, potentially indicating a range of different local environ-
ments in these lms. However, there is no evidence of macro-
scale phase separation of the two homopolymers in the blend,
which would be expected to result in observation of two separate
glass transitions.
copolymer ambipolar host systems might be expected to
perform better than analogous homopolymer blends, given the
possibility of obtaining macroscopic (mm-scale) phase separa-
tion in the latter.20 One contributor to our observation of higher
efficiencies in the blend-based devices may be a less signicant
loss of excitons to low-lying triscarbazole/oxadiazole charge-
transfer-type states (see discussion of emission spectra above).
These charge-transfer states presumably require donor/acceptor
proximity and so will be suppressed to some extent by any
donor/acceptor segregation, raising the question of the length
scale over which such separation might occur. Accordingly, we
examine the microstructure of poly-1/poly-2 blends in the
following section.
Entries 7 and 8 in the table demonstrate that even more
efficient devices can be obtained with poly-1/poly-2 host
blends by using Ir(pppy)3 in place of Ir(ppy)3 as the phospho-
rescent emitter. This is consistent with our previous observa-
tions using PVK/I blends, where an increase in the maximum
EQE from 10.1 to 13.6% was seen on replacing Ir(ppy)3 with
Ir(pppy)3 and was attributed to reduced aggregation and
improved miscibility with the host polymers of the bulkier
phosphorescent guest.6c The present devices are signicantly
more efficient than analogous Ir(pppy)3 device in which the
host is a PVK/I blend,6c or devices using a different hole-
transport layer and a II/III blend as host.10 Indeed, the effi-
ciency of 21% (at 100 cd cmꢀ2) obtained for 1.5 : 1 poly-1/poly-
2 approaches the highest value reported to date for a green
PHOLEDs with a solution-processed emissive layer (23.8% at
1000 cd mꢀ2).21
Solid-state 1H spin-diffusion experiments can potentially
yield more quantitative information about the domain size and
geometry in multicomponent systems.12,23 The principle is
illustrated in Fig. 5a. For a two-domain system, the rst step of
1
the H spin diffusion experiments is to selectively remove the
magnetization of one domain through a suitable sequence of rf-
pulses, leaving that of the other phase unaffected due to
differences in either the chemical shis or the conformational
mobility between two domains. The remaining magnetization
will then be transferred throughout the sample via the network
of the dipolar couplings between 1H nuclei in both phases and
the magnetization of either domain will be recorded as a
function of time. Using the 1H spin diffusion coefficients for
each domain (which are usually deduced from spin-spin relax-
Entries 9–11 in Table 2 refer to devices in which a ther-
mally cross-linkable triscarbazole derivative, Sty2-TCz, was
used in place of poly-TPD-F,22 the ITO was pretreated with
pentauorobenzylphosphonic acid to facilitate effective
wetting by Sty2-TCz, and the greenish-blue emitter FIrpic was
used, and thus demonstrate that this polymer-blend host
1
ation experiments), the H spin diffusion can be simulated by
solving Fick's diffusion equations (see Experimental section) for
a given domain size and geometry. By tting the simulated time
dependence of the magnetization intensity with the
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. C, 2014, 2, 6743–6751 | 6749

View Article Online
Journal of Materials Chemistry C
Paper
experimental results, the approximate domain size and geom-
etry can be determined.
Conclusion
In our experiments, the selective magnetization was ach-
ieved through the difference in conformational mobility
between the oxadiazole and triscarbazole side chains at
Random and block copolymers of triscarbazole- and bis-
(oxadiazole)benzene-functionalized norbornene monomers
have been compared as solution-processed ambipolar hosts for
Ir(ppy)3 in green-emitting OLEDs to a blend of the corre-
sponding homopolymers. The blend-based OLEDs exhibit
considerably higher external quantum efficiencies than the
diblock and random copolymer devices. DSC and solid-state
NMR experiments indicate that the blend does not undergo
macroscopic phase segregation, but exhibits a nanoscale sepa-
ration, which is characterized by a different domain dimen-
sionality and size to that observed for the block polymer. The
blend approach has led to some of the most efficient green-
phosphorescent OLEDs with solution-processed emissive layers
reported to date.
ꢂ
190 C, as revealed by spin-spin relaxation measurements at
the same temperature, which showed two distinct relaxation
times from the mobile and rigid domains respectively (see
ESI†). The magnetization of the mobile domain, assumed to
be the oxadiazole domain based on the observation of a lower
glass-transition temperature for the corresponding homo-
polymer than for the triscarbazole homopolymer, was then
selected through the dipolar lter for the spin-diffusion
experiments and the amount of this magnetization was
observed as a function of a mixing time, providing insight
into the timescale at which magnetization is transported out
of this domain into the triscarbazole domain. The experi-
mental and the best-t simulated curves of the temporal
evolution of the normalized magnetization in the oxadiazole
domain for the 1 : 1 poly-1/poly-2 blend and for poly-1-b-2 are
summarized in Fig. 5b. The normalized intensities for both
samples equilibrate at values of ca. 0.5, consistent with oxa-
diazole and triscarbazole moities being present in a 1 : 1 ratio
in both cases and having comparable 1H concentrations. The
blend shows a signicantly larger characteristic time for the
1H spin diffusion than the block polymer, which implies that
larger domains are present in the physical blend sample.
Using the spin-diffusion coefficients for the oxadiazole and
triscarbazole domains in the two materials obtained from
spin-spin relaxation experiments (see ESI†), the 1H spin
diffusion processes for these two samples were simulated (as
described in the Experimental section) to obtain information
about the domain geometry and the characteristic lengths.
The data for the diblock copolymer were successfully simu-
lated using a layered structure with one-dimensional spin-
transport characteristics (characteristic length ¼ 1.8 nm),
whereas the blend data were tted using a cubic structure
with three-dimensional spin-transport (characteristic length
¼ 16.4 nm).
Acknowledgements
Support from Solvay S.A. and the Science and Technology
Center Program of the National Science Foundation (DMR-
0120967) is gratefully acknowledged.
Notes and references
1 (a) M. A. Baldo, D. F. O'Brien, Y. You, A. Shoustikov, S. Sibley,
M. E. Thompson and S. R. Forrest, Nature, 1998, 395, 151; (b)
Y. Kawamura, K. Goushi, J. Brooks, J. J. Brown, H. Sasabe and
C. Adachi, Appl. Phys. Lett., 2005, 86, 071104.
2 M. A. Baldo, C. Adachi and S. R. Forrest, Phys. Rev. B:
Condens. Matter Mater. Phys., 2000, 62, 10967.
3 (a) N. C. Giebink and S. R. Forrest, Phys. Rev. B: Condens.
Matter Mater. Phys., 2008, 77, 235215; (b) S. H. Kim, J. Jang,
K. S. Yook and J. Y. Lee, Appl. Phys. Lett., 2008, 92, 023513.
4 (a) C. Adachi, R. C. Kwong and S. Forrest, Org. Electron., 2001,
2, 37; (b) S. Lamansky, P. I. Djurovich, F. Abdel-Razzaq,
S. Garon, D. L. Murphy and M. E. Thompson, J. Appl. Phys.,
2002, 92, 1570; (c) X. Gong, J. C. Ostrowski, D. Moses,
G. C. Bazan and A. J. Heeger, Adv. Funct. Mater., 2003, 13,
439; (d) X. H. Yang and D. Neher, Appl. Phys. Lett., 2004,
84, 2476; (e) X. H. Yang, F. Jaiser, S. Klinger and D. Neher,
Appl. Phys. Lett., 2006, 88, 021107.
It should be noted the obtained domain geometry and
sizes may not represent the exact domain dimensions;
microstructures in polymeric materials are usually irregular,
ill-dened, and locally dependent. Moreover, without a
detailed knowledge of the location of the phosphor mole-
cules within the polymers, or of the orientation of the layered
structures modeled for the block polymer, the NMR data do
not fully explain the differences in OLED performance.
However, the spin-diffusion results clearly suggest qualita-
tive differences in the nanoscale structure of the two
samples. These differences in domain sizes are broadly
consistent with our tentative assignment (see above) of
differences in the emission spectra of lms of the block
copolymer and the blend to the decreased contribution of
charge-transfer-type emission in the latter case. Futhermore,
the spin-diffusion data for the blend are consistent with the
inference of the absence of macroscopic phase separation
from the DSC data.
5 A. Chaskar, H.-F. Chen and K.-T. Wong, Adv. Mater., 2011, 23,
3876.
6 (a) X. H. Yang, D. C. Muller, D. Neher and K. Meerholz, Adv.
Mater., 2006, 18, 948; (b) K. Zhang, Z. Chen, C. L. Yang,
X. W. Zhang, Y. T. Tao, L. Duan, L. Chen, L. Zhu, J. G. Qin
and Y. Cao, J. Mater. Chem., 2007, 17, 3451; (c) S. J. Kim,
Y. D. Zhang, C. Zuniga, S. Barlow, S. R. Marder and
B. Kippelen, Org. Electron., 2011, 12, 492.
7 L. Deng, P. T. Furuta, S. Garon, J. Li, D. Kavulak,
M. E. Thompson and J. M. J. Frechet, Chem. Mater., 2006,
18, 386.
8 K. Zhang, Y. Tao, C. Yang, H. You, Y. Zou, J. Qin and D. Ma,
Chem. Mater., 2008, 20, 7324.
9 (a) K. Brunner, A. van Dijken, H. Borner, J. J. Bastiaansen,
N. M. Kiggen and B. M. Langeveld, J. Am. Chem. Soc., 2004,
6750 | J. Mater. Chem. C, 2014, 2, 6743–6751
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry C
126, 6035; (b) M. Guan, Z. Q. Chen, Z. Q. Bian, Z. W. Liu, 18 M. A. Baldo and S. Forrest, Phys. Rev. B: Condens. Matter
Z. L. Gong, W. Baik, H. J. Lee and C. H. Huang, Org. Mater. Phys., 2000, 62, 10958.
Electron., 2006, 7, 330; (c) M. K. Leung, C. C. Yang, 19 For example, (a) A. Haldi, A. Kimyonok, B. Domercq,
J. H. Lee, H. H. Tsai, C. F. Lin, C. Y. Huang, Y. O. Su and
C. F. Chiu, Org. Lett., 2007, 9, 235; (d) Y. Tao, Q. Wang,
C. Yang, Q. Wang, Z. Zhang, T. Zou, J. Qin and D. Ma,
Angew. Chem., Int. Ed., 2008, 47, 8104; (e) Y. Tao, Q. Wang,
C. Yang, C. Zhong, K. Zhang, J. Qin and D. Ma, Adv. Funct.
Mater., 2010, 20, 304; (f) Y. Zhang, W. Haske, D. Cai,
S. Barlow, B. Kippelen and S. R. Marder, RSC Adv., 2013, 3,
23514.
L. E. Hayden, S. C. Jones, S. R. Marder, M. Weck and
B. Kippelen, Adv. Funct. Mater., 2008, 18, 3056; (b)
X. W. Zhan, A. Haldi, J. S. Yu, T. Kondo, B. Domercq,
J. Y. Cho, S. Barlow, B. Kippelen and S. R. Marder, Polymer,
2009, 50, 397; (c) Y. Zhang, C. Zuniga, S.-J. Kim, D. Cai,
´
S. Barlow, S. Salman, V. Coropceanu, J.-L. Bredas,
B. Kippelen and S. Marder, Chem. Mater., 2011, 23, 4002.
20 B. Ma, B. J. Kim, L. Deng, D. A. Poulsen, M. E. Thompson and
J. M. J. Frechet, Macromolecules, 2007, 40, 8156.
10 C. A. Zuniga, J. Abdallah, W. Haske, Y. D. Zhang,
I. Coropceanu, S. Barlow, B. Kippelen and S. R. Marder, 21 J.-H. Jou, C.-J. Li, S.-M. Shen, S.-H. Peng, Y.-L. Chen,
Adv. Mater., 2013, 25, 1739.
11 F. Mellinger, M. Wilhelm and H. W. Spiess, Macromolecules,
1999, 32, 4686.
Y.-C. Jou, J. H. Hong, C.-L. Chin, J.-J. Shyue, S.-P. Chen,
J.-Y. Li, P.-H. Wang and C.-C. Chen, J. Mater. Chem., 2013,
1, 4201.
12 J. Clauss, K. Schmidt-Rohr and H. W. Spiess, Acta Polym., 22 Chosen because triscarbazole derivatives have been found to
1993, 44, 1.
have higher triplet energies (2.9 eV, M.-H. Tsai, Y.-H. Hong,
C.-H. Chang, H.-C. Su, C.-C. Wu, A. Matoliukstyte, J.
Simokaitiene, S. Grigalevicius, J. V. Grazulevicius and C.-P.
Hsu, Adv. Mater., 2007, 19, 862) than TPD derivatives (2.3
eV, ref. 18), reducing the possibility of triplet energy
transfer from FIrpic (triplet energy ¼ 2.6 eV, C. Adachi, R.
Kwong, P. I. Djurovich, V. Adamovich, M. Baldo, M. E.
Thompson and S. Forrest, Appl. Phys. Lett., 2001, 79, 2082)
to the hole-transport layer. It may be possible to improve
the performance of the blue-emitting devices further by
replacing the BCP (triplet energy ¼ 2.5 eV, ref. 18) with an
electron-transport material having a higher triplet energy.
13 (a) R. D. Hreha, Y. D. Zhang, B. Domercq, N. Larribeau,
J. N. Haddock, B. Kippelen and S. R. Marder, Synthesis,
2002, 1201; (b) B. Domercq, R. D. Hreha, Y.-D. Zhang,
N. Larribeau, J. N. Haddock, C. Schultz, S. R. Marder and
B. Kippelen, Chem. Mater., 2003, 15, 1491.
14 P. Kim, S. C. Jones, P. J. Hotchkiss, J. N. Haddock,
B. Kippelen, S. R. Marder and J. W. Perry, Adv. Mater.,
2007, 19, 1001.
15 A. Sharma, P. J. Hotchkiss, S. R. Marder and B. Kippelen, J.
Appl. Phys., 2009, 105, 084507.
16 (a) W. J. Feast, Makromol. Chem., Macromol. Symp., 1992, 53,
317; (b) R. H. Grubbs, S. J. Miller and G. C. Fu, Acc. Chem. 23 (a) M. Mauri, Y. Thomann, H. Schneider and K. Saalwachter,
Res., 1995, 28, 446; (c) R. R. Schrock, Top. Organomet.
Chem., 1998, 1, 1.
17 (a) P. Zhang, B. Xia, Q. Zhang, B. Yang, M. Li, G. Zhang and
W. Tian, Synth. Met., 2006, 156, 705; (b) S. Zhang, R. Chen,
J. Yin, F. Liu, H. Jiang, N. Shi, Z. An, C. Ma, B. Liu and
W. Huang, Org. Lett., 2010, 12, 3438.
Solid State Nucl. Magn. Reson., 2008, 34, 125; (b)
R. C. Nieuwendaal, H. W. Ro, D. S. Germack, R. J. Kline,
M. F. Toney, C. K. Chan, A. Agrawal, D. Gundlach,
D. L. VanderHart and D. M. Delongchamp, Adv. Funct.
Mater., 2012, 22, 1255.
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. C, 2014, 2, 6743–6751 | 6751