This document is the Accepted Manuscript version of a Published Work that appeared in final form in
ACS Appl. Mater. Interfaces 2015, 7, 1078−1086 © American Chemical Society after peer review
and technical editing by the publisher. To access the final edited and published work see
http://pubs.acs.org/doi/pdf/10.1021/am505197b

Influence of the molecular weight and size
dispersion of the electroluminescent polymer on the
performance of air stable Hybrid Light Emitting
Diodes

Eugenia Martinez-Ferrero,*a Souren Grigorianb, James W. Ryanc,†, Werther Cambarau,c and
Emilio Palomaresc,d
a

b

Fundació CETEMMSA, Avda. Ernest Lluch 36, E-08302 Mataró (Spain).

Institute of Physics, University of Siegen, Walter-Flex-Str. 3, D-57068 Siegen (Germany).
c

Chemical Research Institute of Catalonia (ICIQ), Avda. Païssos Catalans 16, E-43007
Tarragona (Spain).
d

Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona (Spain).

1

ABSTRACT: The influence of the chain length and the molecular weight distribution of the
electroluminescent polymer on the carrier transport properties and morphology of air stable
Hybrid Light Emitting Diodes is reported. It is found that variations between diverse as-received
commercial batches play a major role in the performance of the devices, whose maximum
luminance can differ up to two orders of magnitude. Through complementary opto-electronic,
structural and morphological characterization techniques we provide insights into the
relationship between charge dynamics and the structure of polymeric electroluminescent
materials. The carrier dynamics are found to be dominated by both the polymeric chain length
and the hole transport, which in turn, is dependent on the concentration of trap states.
Furthermore, the chain length is seen to affect the morphology of the active layer.

KEYWORDS: electroluminescent polymers, hybrid light emitting diodes, F8BT, molecular
weight, polidispersity index, charge carrier mobility

INTRODUCTION
The performance of optoelectronic devices relies heavily on the properties of the semiconducting
polymers that make up the active layer. The packing of the molecular chains affects the charge
transport in such a manner that polaron or exciton separation, charge transport and recombination
depend on the resulting morphology.1-3 This has been reported in significant detail for many
organic electronic devices, in particular organic solar cells and organic transistors,4 where it has

2

been demonstrated that the behaviour of the polymer depends on its molecular weight, which can
be difficult to control and can lead to large differences in device performance.5 Furthermore, the
chain length and its size distribution are also key factors that determine the packing of the
polymer chains and their orientation towards the electrodes.6 Indeed, the distance between the
aromatic groups of adjacent chains also varies with the polymeric Mw resulting in different
charge mobilities.7-9 On the other hand, the polidispersity index (PDI) is ubiquitous, particularly
in commercial grade samples, adding variability to the reproduction of the devices.
In this work, we explore the impact of the variations of the molecular weight and the PDI of
commercial electroluminescent polymer. Despite numerous investigations of polymeric
properties on the performance of organic field effect transistors and solar cell applications,10-13
this topic has not been explored with the same intensity in the field of organic light emitting
diodes (OLEDs).14-16 Therefore, we have chosen these types of devices due to their potential as
light sources and the major influence that the electroluminescent layer has on the properties of
the device. However, despite their impressive efficiencies, one significant drawback of OLEDs is
the use of low work function electrodes and organic transport layers that are sensitive to moisture
and oxygen and require costly protective layers.17 Therefore, it is desirable to replace the organic
materials by stable transition metal oxides as charge transport layers. In this way, the robustness,
transparency and electronic properties of the inorganic materials are then combined with the
possibilities offered by the organic lumophores giving rise to Hybrid Light Emitting Diodes (see
Scheme 1). These devices, which have shown similar performances to those of the best published
polymeric OLEDS,18 are formed by sandwiching organic electroluminescent polymers between
an electron transport layer (ETL) such as ZnO, TiO2 or ZrO2 and a MoO3 hole transport layer
(HTL) topped with an inert high work function gold anode.19-21 In addition, the low cost and the

3

facile preparation of the metal oxides permit the reduction of production costs and the fabrication
of devices in air by solution methods. F8BT (poly-9,9’-dioctylfluorene-alt-benzothiadiazole,
Scheme 1b) has been used as an electroluminescent polymer since it is widely used in organic
electronics and its LUMO and HOMO levels matches well with the conduction bands of the
charge transport layers (Scheme 1c).

Scheme 1. (a) Description of the HyLED multilayered structure, (b) the chemical structure of the
F8BT polymer and (c) the energy band diagram of the device.
In order to focus exclusively on the variability of the semiconducting polymer, we have
acquired different commercial batches of F8BT and tested them on devices with identical
configurations. Depending on the batch, the maximum luminance achieved by the diodes can
differ by up to two orders of magnitude and significant differences in the current-voltage
characteristics are observed. To determine the key properties controlling the device performance,
a detailed study on the morphology, surface and optical properties of the active layer has been
carried out. In addition, we have also probed the charge carrier dynamics of the devices under

4

working conditions to obtain more detailed information on the effect of polymer structure on
device performance.
EXPERIMENTAL SECTION
F8BT samples were acquired from different suppliers and their molecular weights measured by
gel permeation chromatography against polystyrene standards (Table 1). The reported purities
are 99.99% and metal contents, when indicated by the provider, below 84 ppm. Zinc acetate
dihydrate, methanol, xylene and MoO3 were purchased from Sigma Aldrich and used as
received.
Device preparation and characterization. The ZnO layers were prepared by spray pyrolysis
deposition from a precursor solution of zinc acetate dihydrate in methanol (80 g/L) onto prepatterned ITO glass substrates (10 Ω/ sq, obtained from Psiotec Ltd.) at 450 °C followed by
annealing at 450 ºC for 30 min. The electroluminescent F8BT layer was spin coated from a
xylene solution and annealed at 155 °C for 30 min in a nitrogen rich atmosphere. The solution
concentration was adjusted to get film thicknesses of 100 nm. Finally, the samples were
transferred to a thermal evaporation chamber where the MoO3 (15 nm) and Au (100 nm, Kurt
Lesker) layers were deposited through a shadow mask under high vacuum (1 x 10-7 mbar), which
defined an active area of 0.25 cm2. The as-prepared devices were transferred out of the glove box
and measured in air without any form of encapsulation. The devices were characterized with a
Thornlabs silicon photodiode combined with a Keithley 2400 source controlled by in-house
software.
Hole only devices where prepared and characterized following the same procedure than the
bipolar diodes. The ZnO layer is substituted by 50 nm of PEDOT:PSS which is deposited by spin

5

coating and annealed under nitrogen at 120 ºC for 30 min. The subsequent layers are deposited as
described above.
Film characterization. Thicknesses were measured by an Ambios Technology XP-1
profilometer and confirmed by X-ray reflectometry. Atomic Force Microscopy measurements
were done in tapping mode on a Molecular Imaging model Pico SPM II while images were
analyzed using WSxM software.22 Grazing incident X-ray diffraction (GIXD) studies were
performed using a high-resolution diffractometer Seifert XRD 3003 operating at 8 keV and
equipped with a point detector. Out-of-plane GIXD profiles were measured at the fixed incident
angle α = 0.17º being higher than the critical angle of the polymer film and lower than the
critical angle of the substrate. The surface free energy was estimated by the Owens-WendtRabel-Kaelble (OWRK) method from the contact angle measurements done on the polymer films
by means of a Krüss DSA100. Differential scanning calorimetry (DSC) and thermogravimetric
analysis (TGA) were performed using a Mettler-Toledo TGA-DSC 1 under nitrogen. DSC
experiments were carried out in two consecutive heating cycles from 30 to 300 and cooling back
to 30 ºC at a rate of 10 ºC/min while TGA consisted in a single measurement from 30 to 500 ºC
at a rate of 20 ºC/min. UV-Visible absorption was measured in a Perkin Elmer Lambda 950
while fluorescence measurements were recorded using an Aminco-Bowman Series 2
luminescence spectrometer. Time correlated single photon counting (TCSPC) experiments were
carried out after recording 1000 counts per second (CPS) with a Lifespec picosecond
fluorescence lifetime spectrophotometer from Edinburgh Instruments© equipped with lasers as
excitation sources. The instrument response was always shorter than 440 ps measured at full
width half maximum (FWHM). For transient electroluminescence measurements (TEL), the
devices were electrically excited by a fast pulse generator Thurlby Thandar Instruments TGP110,

6

which permitted accurate control of rise and decay times. Single square-wave voltage pulses
between 4 and 7 V were applied during 100 microseconds. The electroluminescence response
was detected using a PM Hamamatsu H6780-01 photomultiplier tube detector connected to an
amplifier from Hamamatsu C9693. A Teknotronix TDS 2022 oscilloscope was used to record the
voltage, time and electroluminescence.
All the measurements were carried out at ambient conditions without using encapsulating layers.

RESULTS
The commercial samples employed in this study are summarized in Table 1. The Hybrid Light
Emitting Diodes (HyLEDs) were structured as ITO/ZnO/F8BT/MoO3/Au and all the layers were
deposited under the same conditions to permit a fair comparison of the behavior of the different
polymers.
Table 1. Description of the F8BT batches employed in this work. PDI accounts for the
polidispersity index.

Sample

Mn (g/mol) Mw (g/mol)

PDI

Supplier

F15k

5897

15686

2.7

Sigma Aldrich

F23k

4107

23000

5.6

American Dye Source

F27k

12273

27000

2.2

1-Material

F35k

14583

35000

2.4

American Dye Source

F43k

17133

43517

2.5

Sigma Aldrich

F60k

31579

60000

1.9

QCR Solutions

F73k

22812

73000

3.2

American Dye Source

7

F116k

34118

116000

3.4

American Dye Source

The diodes were characterized by measuring the evolution of light emission with applied bias
(Figure 1). The devices present turn on voltages lower than 3 V and reach the maximum
luminance below 10 V (see Table 2 and Table S1 on the supplementary information). The best
luminance values are achieved in devices containing the F73k sample while poorer behaviors of
up to two orders of magnitude lower are observed when F15k and F60k are employed. It is
interesting to note that the turn-on voltages for the samples F15k, F60k and F73k deviate from
the mean estimated value of the other samples. This is rather surprising since turn on voltages are
related to the electron injection barrier which is defined as the difference between the ITO
workfunction and the metal oxide conduction band edge.18

Figure

1.

Luminance

(a)

and

current

density

(b)

vs

applied

bias

for

the

ITO/ZnO/F8BT/MoO3/Au devices prepared with F8BT of different Mw and PDI.

8

Figure 1b shows the evolution of the density current vs. applied bias (Figure 1b) from a linear
dependence that turns exponential at voltages closer to the built-in voltage (Vbi), pointing to the
change from a diffusion dominated current into a space charge limited regime.23 This fact arises
from the energy band alignment of the different components of the device, described in Scheme
1c. The MoO3 of the HTL forms an ohmic contact with the electroluminescent polymer allowing
the injected holes to drift from the electrode to the organic-inorganic interface where, due to the
energy mismatch between the valence band (VB) of the ETL and the HOMO of the polymer,
accumulate leading to a built in field. The field assists the injected electrons on the other
electrode to overpass the electron injection barrier between the conduction band (CB) of the ETL
and the LUMO of the F8BT.18 The charge transport at voltages higher than Vbi is thus bulk
limited and is best described as pseudo-hole only space charge limited current, since the
predominant carriers are holes.24-26 The J-V curve can be modeled using the Mott-Gurney
equation combined with field dependent mobility (equation 1):

J sclc =

 V
9
V2
εµ0 3 exp  β

8
L
 L

(1)

where ε is the ε0·εr permittivity of the F8BT polymer, µo is the hole zero field mobility, V is the
effective applied bias (the applied bias corrected by the built-in potential), L is the thickness of
the active layer and β is the field-effect mobility coefficient (also known as Poole-Frenkel
factor). The parameters obtained from the fitting of the J-V curves, listed in Table 2 and shown in
the supplementary information, indicate that the room temperature hole mobility µo amounts to
~10-5 cm2/V·s for all the samples except for F15k, F43k and F60k. The β coefficient, which
depends upon trap depths, is estimated to be 10-4 cm1/2/V1/2, except for F15k, F35k and F43k
which is one order of magnitude lower suggesting an increased influence of traps in the charge

9

carrier mobility. Comparison with parameters obtained from hole-only devices made of
ITO/PEDOT:PSS/F8BT/MoO3/Au when the hole injection takes place from the MoO3 (see Table
S2 and Figure S2 in the supplementary information), indicate that the difference is less than onefold and probably arises from the beneficial influence of the ZnO layer.
Independent measurements of the hole mobility under working conditions are obtained by
transient electroluminescence (TEL) in which the time dependence of the rise of
electroluminescence is recorded after application of square voltage pulses of different amplitude.
The delay time (td) between the voltage start and the onset of light emission is a direct
measurement of the transit time of the faster charge carriers (holes, vide supra) towards the
recombination zone. Therefore, TEL curves can be well described by the equations used for
hole-only single layer OLEDs in which holes drift faster towards the interface dominating the
early stage of the rising of electroluminescence.27
Table 2. Optoelectronic data of the devices obtained from the L-V curve, the fitting of the JV
curve and TEL measurements.

Von

Lmax

µ0

β

Vtrap

(V)a

(cd/m2)b

(cm2/V·s)c

(cm1/2/V1/2)c

(V)d

F15k

2.8

26

1.1 x10-6

4.1 x10-3

2.9

F23k

2.3

1440

4.3 x10-5

1.3 x10-4

2.7

F27k

2.1

1340

1.0 x10-5

1.0 x10-4

2.3

F35k

2.4

890

1.8 x10-5

1.2 x10-3

4.1

F43k

2.3

224

1.9 x10-7

3.9 x10-3

4.9

F60k

2.7

28

5.9 x10-7

3.9 x10-4

---

Sample

10

F73k

1.8

6400

2.8 x10-5

9.0 x10-4

3.2

F116k

2.3

530

9.7 x10-5

7.3 x10-4

2.5

a

The turn-on voltage is defined as the applied bias when the luminance reaches 0.01 cd/m2; b
maximum luminance; c data obtained from fitting the JV curves; d data obtained from TEL
measurements.

Once the holes reach the steady state, which occurs at times close to td, further rise of
electroluminescence is then due to electron flow in the bulk. Even at this stage, hole density
exceeds the electron concentration and is not affected by charge recombination. Consequently,
we can estimate the hole mobility from the electroluminescence output using the equation of the
drift mobility in a electric field (Equation 2):

L2
µh =
Vtd

(2)

where L accounts for the thickness of the active layer, V is the effective applied bias (V-Vbi) and
td is the time delay. Figure 2a illustrates the linear dependence of the as-estimated µh with the
electric field, except for F60k that could not be measured due to its low luminance. The variation
between samples corroborates the previous observations of Figure 1b in the sense that charge
dynamics are influenced by the molecular weight and PDI.

11

Figure 2. Hole mobility as a function of the square root of the electric field (a) and variation of
the hole mobility with the free hole density (b). Solid lines indicate linear fitting, dashed lines
indicate power law fitting.
Further information can be obtained following the development made by Rihani et al. who
estimated the free hole density (pe) considering the Langevin charge recombination coefficient
(R) and the recombination current, Jr.27 Briefly, R can be estimated from the hole mobility
through the equation 3.

R=

qµh

ε

(3)

where q stands for the elementary charge, µh is the hole mobility and ε is the permittivity of
F8BT. Furthermore, the recombination current is directly proportional to the light emission.
Therefore, from the evolution of light emission with time we can estimate the time constant τ
(equation 4), which is inversely proportional to the free hole density, pe. Equation 5 demonstrates
this relationship where EL refers to the electroluminescent signal at a given time t and a is the
value of the intersection at time 0:
12

t
 EL(t → ∞) − EL(t) 
ln 
=a−

τ
EL(t → ∞)


pe =

τ −1
R

(4)
(5)

The as-calculated free hole densities (equation 5) depend linearly with the applied bias so
extrapolation to lower bias indicates that pe vanishes at certain voltages pointing to the presence
of traps in the bulk (Vtrap, see Table 2). Therefore, the low charge carrier density at lower applied
bias is due to the fact that the empty traps are becoming filled. As the applied voltage increases,
the number of empty traps decreases and the number of free holes increases, resulting in
enhanced current densities. A close inspection reveals that Vtrap is similar to Vturn-on suggesting
that, at lower bias, the dominant mechanism is non-radiative trap assisted recombination in
agreement with previous observations.23 The exception is found for the samples F35k, F43K and
to less extent F73k that show higher Vtrap than the rest. Again, comparison of this data with that
obtained from the J-V curve reveals that these samples posses’ high β values confirming the
increased influence of the traps on the mobility values.
On the other hand, the variation of µh with charge carrier density follows a power law trend, in
agreement with the conduct observed for other polymers at charge carrier densities higher than
1016 cm-3.28 The dependence of the field-effect mobility on charge carrier density has been
related to the width of the exponential density of states; therefore, the stronger dependence of the
mobility on the charge carrier density in samples F35k and F27k are indicative of larger
energetic disorder.28 Moreover, the analysis of the charge dynamics suggests a major influence
of traps on the performance of the diodes. The device F43k display lower charge carrier mobility
reflected in reduced free hole density and current density together with high values of β and Vtrap.

13

Moreover, the devices with lower current densities (F15k, F43k and F60k) lead to higher
luminance turn-on voltages regardless of the electron injection barrier. In addition, these samples
give poorer luminance.
However, the charge mobility does not correlate with the variations in the Mw or the PDI.
Consequently and in order to gain further information, we have undertaken optical and
morphological studies on polymer films. The absorption spectra, shown in Figure 3a, present the
typical pattern for F8BT with the main peak centered at 465 nm. The fact that the peaks are not
significantly shifted or broadened suggests that the optical band gap does not change regardless
of the Mw and the PDI. Photoemission measurements after excitation at 450 nm, Figure 3b,
show a main signal centered at 536 nm except for F116k whose emission maximum is red shifted
to 541 nm. The intensity of the peaks varies according to their respective PDI suggesting that
higher size dispersion is more beneficial to efficient light emission than higher Mw.

14

Figure 3. Optical characterization of the thin films (a) UV-Vis absorption, (b)
photoluminescence, (c) TCSPC decays (IRF: instrument response function).

15

Time-Correlated Single Photon Counting (TCSPC) records the decay signals as a function of
time after excitation at 405 nm. The experimental points are fitted with the equation 6:

I(t) = I 0 exp [−(t / τ )]

(6)

where I(t) represents the time (t) dependent decay of the fluorescence intensity, Io stands for the
initial intensity values and τ is the lifetime value obtained from the decay, indicating the average
time that the polymer remains in the excited state. Figure 3c shows the decays for all the
samples. F15k, F43k and F60k fall within the response of the instrument (IRF) and thus, the
decays are shorter than 0.44 ns. The rest of the samples show monoexponential decays that can
be fitted to equation 6, resulting in lifetime values that are shown in Table 3.
Table 3. Photophysical parameters of the polymeric films
Sample

λ max Abs(nm)

λ max PL(nm)

τ (ns)

F15k

464

537

< 0.44

F23k

464

536

1.55

F27k

462

532

1.45

F35k

465

537

1.30

F43k

462

536

< 0.44

F60k

465

536

< 0.44

F73k

467

538

1.57

F116k

466

541

1.32

Additionally, we have measured the lifetime of the devices under ambient conditions.
Measurements were done every 24 hours on devices without encapsulation that were stored in
air. The results, which have reached values of 600 hours, do not show significant differences
between the most efficient polymers.

16

In order to gain further insight into the relationship between the molecular weight distribution
and the morphology, we have proceeded with differential scanning calorimetry (DSC). Figure 4
shows the second heating scan, where the first feature above 100 ºC is an endothermic transition
followed by an exothermic peak around 150 ºC. The combinations of the first cooling scan with
the second heating scan indicates that it is due to the transition of the polymer from the glassy to
the rubbery state that takes place alongside a crystallization step.29 The glass transition
temperature (Tg) has been therefore estimated at the onset of the first change of the heat capacity
defined by the point of intersection between the baseline before the thermal effect and the
tangent of the curve (Table 4). A second exothermic transition located around 185 ºC is assigned
to further crystallization. Finally, there is the onset of a third transition at temperatures around
200 ºC. The endothermic nature of the smooth transition points to the occurrence of the melting
process to liquid crystalline phase.

Figure 4. DSC data of the second heating scan for all the polymers.

17

Structural analysis to probe ordering was performed by GIXD measurements at the incident
angle of αi = 0.17º. This incident angle is above the critical angle of the polymer (αc = 0.15º) and
below that of the glass substrate (αc = 0.21º) to permit the penetration of the X-beam through the
depth of the whole film. Figure 5 shows the scattering profiles, which present a peak at small
angles suggesting the presence of similar structural organization of crystalline domains
regardless of the Mw or PDI of the samples (see supporting information Figure S3). The peaks at
higher angles could not be resolved despite long counting time (200s) due to the weak signals
arising from the low thickness of the samples. Therefore, X-ray studies were focused on the most
intense scattering peaks centred at a 2θ position of ≈ 5.2º (see Figure 5). According to the
literature, polyfluorenes usually form lamellas that stack parallel to the substrate in a
configuration described as the β-phase and defined by the π-π stacking, the size of the F8-BT
repeating unit and the interchain spacing.30,9 We have focused on the peak corresponding to the
interchain spacing that lies between 1.66 to 1.69 nm, depending on the sample, which agrees
with the 1.6 nm reported in the literature.30 Moreover, since the estimated length of the alkyl side
chains is 8.75 Å we can deduce that the observed interchain distance corresponds to two
polyfluorene backbones separated by alkyl chains. Similar edge-on orientation with the strongest
(100) peak associated with side chain distances along the out-of-plane direction has been
reported for a large family of thiophenes.31,32 All samples here show these characteristic features
but for three exceptions, the F60k, F73k and F43k samples. The first, F60k, shows a rather broad
distribution shifted to higher 2θ angles with two peak positions suggesting the existence of two
polymorphs where the chains are separated by 1.6 or 1.4 nm. In addition, it confirms the higher
amount of disordered domains for the F60k in comparison to all other samples in the series that
provide narrower peaks. Interestingly, F73k and F43k samples in turn show just one peak (very

18

intense for F73k) located at even higher 2θ angles (see supplementary information) that
correspond to interchain distances of about 1.2 nm. The size of the crystallites is obtained from
the width of the peak in combination with the Scherrer equation (Table 4). The as-calculated size
of the domains ranges from 10 to 17 nm with marginal changes of crystalline domains as a
function of the molecular weight, except for F43k that measures 30 nm.

Figure 5. Out-of-plane GIXD in the vicinity of the (100) peak of F8BT films.
Further information has been obtained by AFM studies (Figure 6). It is interesting to point out
the low roughness of the samples ranging between 1.8 and 3.4 nm (see Table 4). The films are all
formed by domains of aggregated crystallites that agree with literature reports about the
polycrystalline nature of films forming well-ordered domains when annealed above Tg.29,33
However, there are clear differences in the local morphology. The pattern in F43k is formed by
long chains while the aggregates are much closer in F15K and F60k. Samples F23k, F35k, F73k
and F116k are arranged in a similar fashion than F43k, whereas the F27k film is rather smooth
but contains aggregates of 200 x 140 nm. It is worthy to note that F60k, which has the lowest
PDI among the samples, forms small aggregates while larger domains are observed in samples
with PDI values above 2.5.

19

Figure 6. AFM topography images (2 µm x 2 µm) of the F8BT films.
The formation of molecular aggregates is known to be strongly affected by the surface energy of
the molecules.34 Consequently, we have estimated the surface free energy of the different
samples through the measurement of the contact angle of water, ethylene glycol and dimethyl
sulfoxide on F8BT films. These values are applied in equation 7 to calculate the dispersive and
polar interactions between F8BT and the liquids.

σ L (1+ cosθ )
2 σ

D
L

P
= σS

P
σL

σ

D
L

D
+ σS

(7)

20

where θ is the contact angle and σ is the surface free energy of the F8BT (S) or the liquid (L).
The total surface energy is calculated from the sum of the dispersive and the polar components
and the results are summarized in Table 4. The results indicate that surface free energy (SFE)
increases with the Mw especially due to the decrease of the polar component. These observations
agree with Wu who defined a linear variation between the surface energy and Mn-2/3 so the
variation of the Mw provides a reliable method to control the surface energy.34,35 However, the
increase of Mw alone cannot explain the differences found in the structural characterization. The
SFE of the chain ends is usually lower than that of the chain; therefore the increment of SFE
values with higher Mw is assigned to the decreased ratio of chain ends to the repeat unit. The
heterogeneity caused by the chain ends in the surface energy can affect the grain boundaries, the
size of the domains and the order of the molecular arrangement.34 Moreover, it could be the
cause of the irregular change in the mobility of semiconducting layers.36 In addition, we also
need to consider the value of the PDI of each sample since it affects the heterogeneity of the
surface as well. For example, on one hand, F60k has the lowest PDI, 1.9, shows the lowest
mobility and forms films containing a broad dispersion of crystallites with a mean size of 9 nm
arranged in a closed pattern resulting in low surface roughness. On the other hand, F23k has the
highest PDI, 5.6, the charge mobility and luminance values are among the highest of the samples
and it forms large and homogenous domains (15.2 nm) resulting in relatively rougher films. In
fact, the SFE values are higher than expected for this Mw and the polar component is
comparable to F43k. So we can deduce that the polymer samples with higher PDI (F23k, F35k,
F43k, F73k and F116k) produce films with bigger domains that form large patterns as seen in the
AFM images. The SFE values correlate with the Mw although some subtle deviations are
observed due mainly to the influence of the PDI on the heterogeneity of the surface. Finally, and

21

although the films present certain surface roughness, it is worthy noticing that the average values
are lower than 4 nm, and as reported by Wang et al., in these cases, the influence of the
topography can be considered negligible.37 On the other hand, the measurements of the contact
angle have been done using the advancing angle, which has been demonstrated to be relatively
less sensitive to surface roughness.38 Therefore, the influence of the surface roughness on the
measurements of the contact angle, and therefore, on the determination of the SFE have not been
taken into account.
Table 4. Temperature of the glass transition (Tg), wettability properties described by the surface
energy (γe) and the dispersive (σD) and polar components (σP), the crystalline size (S) calculated
by GIXS and surface average roughness (Rav) of the F8BT films.
Film

Tg

γe (mJ/m2)

σD (mJ/m2)

σP (mJ/m2)

S (nm)

RAV (nm)

(ºC)
F15k

---

27.5

27.2

0.3

---

2.2

F23k

104

29.9

29.7

0.2

15.2

2.3

F27k

107

28.3

27.9

0.4

16.9

3.1

F35k

109

28.1

27.8

0.3

14.2

3.0

F43k

104

27.9

27.7

0.2

30.6

2.8

F60k

101

29.7

29.5

0.2

9.7

1.8

F73k

109

29.7

29.6

0.1

11.2

2.6

F116k

112

30.4

30.3

0.1

13.3

3.8

DISCUSSION
We have prepared devices with commercial F8BT samples ranging from 15686 to 116000 g/mol
that present a broad dispersion of sizes. The performance of the devices depends on the

22

morphology of the active layer, which in turns depends on the polidispersity index and chain
length of the polymer, as summarized in Table 5. We consider first the samples F43k and F60k.
The estimated charge carrier mobilities and current densities are lower than the rest of the
samples, and consequently, the devices offer poor results with low luminance values and high
turn-on voltages. This trend is also observed in the optical properties where the emission rates
and the lifetime of the photogenerated excitons are also lower. But despite the similar
optoelectronic properties of these two devices, their morphologies differ and so do the reasons
for their poor behavior. F60k is composed of two polymorphs that form small crystalline
domains. The features of the polymorphism have been detected by GIXD and by the width of the
endothermic peak at DSC measurements, which despite its lower PDI displays a broad feature.
The increased number of domains implies the presence of more grain boundaries that act as
traps, which hinder charge transport and photoemission. On the other hand, F43k possess shorter
spacing between polymer chains that can induce the formation of well-ordered nanodomains.
The presence of these crystalline domains has not been detected by optical measurements but the
structural analysis reveals the closer interchain spacing that gives rise to homogeneous domains
ranging up to 30 nm. The decrease of spacing between polymer chains causes a decrease of the
charge mobility. This effect is also seen for F60k where one of the polymorphs shares similar
structural characteristics with F43k. Interestingly, the sample F73k with shorter interlammelae
distance due to interdigitation of the side chains provides relatively high field effect mobility and
the highest luminance. Despite that in samples F43k and F60k such interdigitations cause a
dramatic decrease of the optoelectronics characteristics, the sample F73k reveals the best
performance. The origin of this enhancement can be related to the strong improvement of the
orientation features confirmed by the highest intensity of the out-of-plane GIXD profile.

23

Thermogravimetric analysis of F60k reveals an unusual loss of weight between 100 and 360 ºC
that have not seen in the rest of the samples (see supplementary information).
Samples F15K, F23k, F27k, F35k, F73k and F116k display similar morphological features with
homogeneous domains made of crystallites of similar sizes giving rise to moderate surface
roughness. The performance of the devices improves with increasing length of the polymer
chains, and therefore, F15k has low luminance values coincident with lower mobility values.
This trend is usually assigned to the concentration of chain ends that can act as trapping centres.
However, F116k does not produce the best device, as the trend would suggest. Reasons for the
lower than expected performance can be deduced from the red shift of the photoluminescence
peak and a moderate increase in the roughness of the film surface that suggests the formation of
non-emissive aggregates. These aggregates could originate from partial chain entanglements that
limit molecular motion and crystallization despite the annealing treatment.29 Finally, it is worth
mentioning F23k, which behaves better than expected for its Mw. A priori, the chain ends act as
exciton dissociation sites so shorter polymer chains with more chain ends are expected to have
lower emission efficiency, but this effect is mitigated by the broad dispersion of chain lengths in
the sample that contribute to increase the photoluminescent efficiency. As has been reported
before, polydispersity enhances the ability of polymers to self-organize during the annealing
process at optimized temperatures. In fact, devices made of P3HT with low Mw and high
polidispersity have been observed to be more efficient than devices with higher Mw due to the
enhanced crystallinity of the polymer chains.39 On the other hand, the presence of small fractions
of high molecular weight polymers can drastically affect film formation in blends with polymers
of lower Mw, their morphology and carrier dynamics.40 Finally, if we are to conclude on any

24

general trend from the experimental results, high PDI values favor good optical properties and
the formation of large domains, which result in enhanced performance of the devices.

Table 5. Summary of the experimental results observed in each sample.
Sample
F15k

Mw (g/mol)
15686

PDI

Experimental observations

2.7

Shortest polymer chain length and moderate PDI
Poor charge mobility and optical properties
Poor device performance

F23k

23000

5.6

Broad dispersion of chain lengths
Large crystalline domains
Excellent device performance

F27k

27000

2.2

Moderate polidispersity
Formation of large aggregates
Excellent device performance

F35k

35000

2.4

Moderate polidispersity
High values for Vtrap
Good device performance

F43k

43517

2.5

Short interchain distances generating large domains
Poor charge mobility and optical properties
Moderate device performance

1.9

Presence of polymorphs and impurities
Small domains originating grain boundaries
Poor mobility and optical properties
Poor device performance

F60k

60000

F73k

73000

3.2

High molecular weight and polidispersity
Short interchain distances, strong orientation of the
chains
Excellent device performance

F116k

116000

3.4

High molecular weight and polidispersity
Formation of optical non emissive aggregates
Moderate device performance

CONCLUSIONS
In conclusion, this study determines the influence of the molecular weight and polidispersity
index of commercial F8BT in the performance of Hybrid Light Emitting Diodes. Contrary to

25

previous work on this topic made on purified polymers, the commercial batches have broad size
dispersion that affects the performance of the devices. So, despite higher Mw favors charge
mobility, the presence of long polymer chains does not always guarantee the formation of
adequate morphologies that lead to efficient devices. Consequently, it is necessary to consider
the molecular weight, the size distribution and the purity of the batches because of their influence
on the morphology and the final properties of the devices.
We believe that these conclusions will impact several important research fields where polymers
are utilized such as organic solar cells, light emitting diodes and transistors promoting the
industrial development of optoelectronic devices.

ASSOCIATED CONTENT
Supporting Information. Fittings of the JV curve of bipolar and hole-only devices, long range
out-of-plane GIXD data and thermogravimetric analysis of the polymers. This material is
available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*Fax: +34 93 741 9228; Tel: +34 93 741 9100; E-mail: emartinez@cetemmsa.com
Present Addresses
† Present address: Department of Chemistry, The University of Tokyo, 7-3-1 Hongo, Bunkyoku, Tokyo 113-0033 (Japan)
Author Contributions

26

The manuscript was written through contributions of all authors. All authors have given approval
to the final version of the manuscript.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENT
E.M-F. thanks the Spanish MINECO for the Ramon y Cajal fellowship RYC-2010-06787 and
funding through the project MAT2012-31570. C. Carenas help for the acquisition of TEL data is
acknowledged. S.G. is grateful to L. Grodd and C. Kwamen for assistance during X-ray
measurements and to BMBF for financial support through the project Nr05K13PS4. Financial
support from CETEMMSA is appreciated as well.

REFERENCES
(1)

Ballantyne, A. M.; Chen, L.; Dane, J.; Hammant, T.; Braun, F. M.; Heeney, M.;

Duffy, W.; McCulloch, I.; Bradley, D. D. C.; Nelson, J. The Effect of Poly(3-hexylthiophene)
Molecular Weight on Charge Transport and the Performance of Polymer:Fullerene Solar Cells.
Advanced Functional Materials 2008, 18, 2373-2380.
(2)

Salleo, A.; Kline, R. J.; DeLongchamp, D. M.; Chabinyc, M. L. Microstructural

Characterization and Charge Transport in Thin Films of Conjugated Polymers. Advanced
Materials 2010, 22, 3812-3838.

27

(3)

Cornil, J.; Beljonne, D.; Calbert, J. P.; Brédas, J. L. Interchain Interactions in

Organic π-Conjugated Materials: Impact on Electronic Structure, Optical Response, and Charge
Transport. Advanced Materials 2001, 13, 1053-1067.
(4)

Chu, T.-Y.; Lu, J.; Beaupré, S.; Zhang, Y.; Pouliot, J.-R.; Zhou, J.; Najari, A.;

Leclerc, M.; Tao, Y. Effects of the Molecular Weight and the Side-Chain Length on the
Photovoltaic Performance of

Dithienosilole/Thienopyrrolodione Copolymers.

Advanced

Functional Materials 2012, 22, 2345-2351.
(5)

Kline, R. J.; McGehee, M. D.; Kadnikova, E. N.; Liu, J.; Fréchet, J. M. J.; Toney,

M. F. Dependence of Regioregular Poly(3-hexylthiophene) Film Morphology and Field-Effect
Mobility on Molecular Weight. Macromolecules 2005, 38, 3312-3319.
(6)

Hiorns, R. C.; de Bettignies, R.; Leroy, J.; Bailly, S.; Firon, M.; Sentein, C.;

Khoukh, A.; Preud'homme, H.; Dagron-Lartigau, C. High Molecular Weights, Polydispersities,
and Annealing Temperatures in the Optimization of Bulk-Heterojunction Photovoltaic Cells
Based on Poly(3-hexylthiophene) or Poly(3-butylthiophene). Advanced Functional Materials
2006, 16, 2263-2273.
(7)

Joshi, S.; Pingel, P.; Grigorian, S.; Panzner, T.; Pietsch, U.; Neher, D.; Forster,

M.; Scherf, U. Bimodal Temperature Behavior of Structure and Mobility in High Molecular
Weight P3HT Thin Films. Macromolecules 2009, 42, 4651-4660.
(8)

Joshi, S.; Grigorian, S.; Pietsch, U.; Pingel, P.; Zen, A.; Neher, D.; Scherf, U.

Thickness Dependence of the Crystalline Structure and Hole Mobility in Thin Films of Low
Molecular Weight Poly(3-hexylthiophene). Macromolecules 2008, 41, 6800-6808.
(9)

Donley, C. L.; Zaumseil, J.; Andreasen, J. W.; Nielsen, M. M.; Sirringhaus, H.;

Friend, R. H.; Kim, J.-S. Effects of Packing Structure on the Optoelectronic and Charge

28

Transport Properties in Poly(9,9-di-n-octylfluorene-alt-benzothiadiazole). J. Am. Chem. Soc.
2005, 127, 12890-12899.
(10)

Tumbleston, J. R.; Collins, B. A.; Yang, L.; Stuart, A. C.; Gann, E.; Ma, W.; You,

W.; Ade, H. The Influence of Molecular Orientation on Organic Bulk Heterojunction Solar
Cells. Nature Photonics 2014, 8, 385-391.
(11)

Merari Masillamani, A.; Orgiu, E.; Samori, P. Effect of the Molecular Weight of

the Polymer Gate Dielectric on the Performances of Solution-processed Ambipolar OTFTs.
Journal of Materials Chemistry C 2013, 1, 7725-7730.
(12)

Intemann, J. J.; Yao, K.; Yip, H.-L.; Xu, Y.-X.; Li, Y.-X.; Liang, P.-W.; Ding, F.-

Z.; Li, X.; Jen, A. K. Y. Molecular Weight Effect on the Absorption, Charge Carrier Mobility,
and Photovoltaic Performance of an Indacenodiselenophene-Based Ladder-Type Polymer.
Chemistry of Materials 2013, 25, 3188-3195.
(13)

Fitzner, R.; Mena-Osteritz, E.; Mishra, A.; Schulz, G.; Reinold, E.; Weil, M.;

Körner, C.; Ziehlke, H.; Elschner, C.; Leo, K.; Riede, M.; Pfeiffer, M.; Uhrich, C.; Bäuerle, P.
Correlation of π-Conjugated Oligomer Structure with Film Morphology and Organic Solar Cell
Performance. Journal of the American Chemical Society 2012, 134, 11064-11067.
(14)

Haldi, A.; Kimyonok, A.; Domercq, B.; Hayden, L. E.; Jones, S. C.; Marder, S.

R.; Weck, M.; Kippelen, B. Optimization of Orange-Emitting Electrophosphorescent
Copolymers for Organic Light-Emitting Diodes. Advanced Functional Materials 2008, 18, 30563062.
(15)

Yim, K.-H.; Doherty, W. J.; Salaneck, W. R.; Murphy, C. E.; Friend, R. H.; Kim,

J.-S. Phase-Separated Thin Film Structures for Efficient Polymer Blend Light-Emitting Diodes.
Nano Letters 2010, 10, 385-392.

29

(16)

Weinfurtner, K.-H.; Fujikawa, H.; Tokito, S.; Taga, Y. Highly Efficient Pure Blue

Electroluminescence from Polyfluorene: Influence of the Molecular Weight Distribution on the
Aggregation Tendency. Applied Physics Letters 2000, 76, 2502-2504.
(17)

So, F.; Kondakov, D. Degradation Mechanisms in Small-Molecule and Polymer

Organic Light-Emitting Diodes. Advanced Materials 2010, 22, 3762-3777.
(18)

Sessolo, M.; Bolink, H. J. Hybrid Organic–Inorganic Light-Emitting Diodes.

Advanced Materials 2011, 23, 1829-1845.
(19)

Bolink, H. J.; Coronado, E.; Sessolo, M. White Hybrid Organic-Inorganic Light-

Emitting Diode Using ZnO as the Air-Stable Cathode. Chemistry of Materials 2009, 21, 439441.
(20)

Tokmoldin, N.; Griffiths, N.; Bradley, D. D. C.; Haque, S. A. A Hybrid

Inorganic-Organic Semiconductor Light-Emitting Diode Using ZrO2 as an Electron-Injection
Layer. Advanced Materials 2009, 21, 3475-3478.
(21)

Morii, K.; Kawase, T.; Inoue, S. High Efficiency and Stability in Air of the

Encapsulation-free Hybrid Organic-inorganic Light-emitting Diode. Applied Physics Letters
2008, 92, 213304/213301-213304/213303.
(22)

Horcas, I.; Fernández, R.; Gómez-Rodríguez, J. M.; Colchero, J.; Gómez-Herrero,

J.; Baro, A. M. WSXM: A Software for Scanning Probe Microscopy and a Tool for
Nanotechnology. Review of Scientific Instruments 2007, 78, 013705.
(23)

Wetzelaer, G. A. H.; Kuik, M.; Nicolai, H. T.; Blom, P. W. M. Trap-assisted and

Langevin-type Recombination in Organic Light-emitting Diodes. Physical Review B 2011, 83,
165204.

30

(24)

Kabra, D.; Lu, L. P.; Song, M. H.; Snaith, H. J.; Friend, R. H. Efficient Single-

Layer Polymer Light-Emitting Diodes. Advanced Materials 2010, 22, 3194-3198.
(25)

Lu, L. P.; Kabra, D.; Friend, R. H. Barium Hydroxide as an Interlayer Between

Zinc Oxide and a Luminescent Conjugated Polymer for Light-Emitting Diodes. Advanced
Functional Materials 2012, 22, 4165-4171.
(26)

Lu, L.-P.; Kabra, D.; Johnson, K.; Friend, R. H. Charge-Carrier Balance and

Color Purity in Polyfluorene Polymer Blends for Blue Light-Emitting Diodes. Advanced
Functional Materials 2012, 22, 144-150.
(27)

Rihani, A.; Hassine, L.; Fave, J. L.; Bouchriha, H. Study of the Transient EL

Slow Rise in Single Layer OLEDs. Organic Electronics 2006, 7, 1-7.
(28)

Tanase, C.; Meijer, E. J.; Blom, P. W. M.; de Leeuw, D. M. Unification of the

Hole Transport in Polymeric Field-Effect Transistors and Light-Emitting Diodes. Physical
Review Letters 2003, 91, 216601.
(29)

Banach, M. J.; Friend, R. H.; Sirringhaus, H. Influence of the Molecular Weight

on the Thermotropic Alignment of Thin Liquid Crystalline Polyfluorene Copolymer Films.
Macromolecules 2003, 36, 2838-2844.
(30)

Faria, G. C.; Plivelic, T. S.; Cossiello, R. F.; Souza, A. A.; Atvars, T. D. Z.;

Torriani, I. L.; deAzevedo, E. R. A Multitechnique Study of Structure and Dynamics of
Polyfluorene Cast Films and the Influence on Their Photoluminescence. The Journal of Physical
Chemistry B 2009, 113, 11403-11413.
(31)

Hukic-Markosian, G.; Basel, T.; Singh, S.; Valy Vardeny, Z.; Li, S.; Laird, D.

Study of Photoexcitations in Poly(3-hexylthiophene) for Photovoltaic Applications. Applied
Physics Letters 2012, 100, 213903.

31

(32)

Bao, Z.; Lovinger, A. J. Soluble Regioregular Polythiophene Derivatives as

Semiconducting Materials for Field-Effect Transistors. Chemistry of Materials 1999, 11, 26072612.
(33)

Watts, B.; Schuettfort, T.; McNeill, C. R. Mapping of Domain Orientation and

Molecular Order in Polycrystalline Semiconducting Polymer Films with Soft X-Ray
Microscopy. Advanced Functional Materials 2011, 21, 1122-1131.
(34)

Sun, X.; Liu, Y.; Di, C.-A.; Wen, Y.; Guo, Y.; Zhang, L.; Zhao, Y.; Yu, G.

Interfacial Heterogeneity of Surface Energy in Organic Field-Effect Transistors. Advanced
Materials 2011, 23, 1009-1014.
(35)

Wu, S. Polymer Interface and Adhesion, 1 ed.; Taylor & Francis: New York,

United States of America, 1982; Vol. 1.
(36)

Umeda, T.; Kumaki, D.; Tokito, S. Surface-energy-dependent Field-effect

Mobilities up to 1 cm2/V s for Polymer Thin-film Transistor. Journal of Applied Physics
2009, 105, 024516.
(37)

Wang, C.-F.; Su, Y.-C.; Kuo, S.-W.; Huang, C.-F.; Sheen, Y.-C.; Chang, F.-C.

Low-Surface-Free-Energy Materials Based on Polybenzoxazines. Angewandte Chemie
International Edition 2006, 45, 2248-2251.
(38)

Drelich, J.; Miller, J. D.; Good, R. J. The Effect of Drop (Bubble) Size on

Advancing and Receding Contact Angles for Heterogeneous and Rough Solid Surfaces as
Observed with Sessile-Drop and Captive-Bubble Techniques. Journal of Colloid and Interface
Science 1996, 179, 37-50.

32

(39)

Kim, Y. S.; Lee, Y.; Lee, W.; Park, H.; Han, S.-H.; Lee, S.-H. Effects of

Molecular Weight and Polydispersity of Poly (3-hexylthiophene) in Bulk Heterojunction
Polymer Solar Cells. Current Applied Physics 2010, 10, 329-332.
(40)

Gasperini, A.; Sivula, K. Effects of Molecular Weight on Microstructure and

Carrier Transport in a Semicrystalline Poly(thieno)thiophene. Macromolecules 2013, 46, 93499358.

33

TABLE OF CONTENTS GRAPHIC

34