Remarkable self-organization and unusual conductivity behavior in cellulose nanocrystal-PEDOT: PSS nanocomposites

Vol:.(1234567890)
Journal of Materials Science: Materials in Electronics (2019) 30:1390–1399
https://doi.org/10.1007/s10854-018-0409-y
1 3
Remarkable self-organization and unusual conductivity behavior
in cellulose nanocrystal-PEDOT: PSS nanocomposites
Kazi M. Alam1
· Piyush Kar1
· Ujwal K. Thakur1
· Ryan Kisslinger1
· Najia Mahdi1
· Arash Mohammadpour1
·
Payal A. Baheti1,3
· Pawan Kumar1
· Karthik Shankar1,2
Received: 9 September 2018 / Accepted: 17 November 2018 / Published online: 19 December 2018
© Springer Science+Business Media, LLC, part of Springer Nature 2018
Abstract
Aqueous suspensions of cellulose nanocrystals were blended with Poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate)
[PEDOT:PSS], and cast into thin films. The morphology, structure and electrical properties of the resulting nanocomposite
thin films were thoroughly characterized. We found that the CNC–PEDOT:PSS blends self-organize into a layered vertical
stack with a pitch of 100–200 nm while retaining a continuous percolation network for PEDOT. Atomic force microscopy,
dynamic light scattering and multi-angle light scattering measurements confirmed the wrapping of polymer chains around
the rod-like CNCs. The blended films exhibited improved molecular ordering of the PEDOT chains with concomitant
improvement in the carrier mobility. The remarkable self-organization and enhanced structural order enabled the CNC–
PEDOT:PSS blends to exhibit a high conductivity typical of PEDOT:PSS even when the content of the insulating CNCs in
the nanocomposite was as high as 50 wt%.
1 Introduction
Cellulose is the most abundant natural polymer on earth. It
is a biocompatible, biodegradable, non-toxic, low-cost, eas-
ily processable and renewable material. Cellulose nanocrys-
tals (CNCs) are high aspect ratio nanorods derived from
cellulose. In recent years, this renewable nanomaterial has
attracted intense scientific research interest worldwide not
only for fundamental investigations into the properties
of this material but also for potential use in the develop-
ment of functional nanomaterials. Owing to their excellent
mechanical and chemical properties coupled with nanomet-
ric dimensions, CNCs have become a technologically prom-
ising nanomaterial in diverse areas such as material science
[1–3], biomedical devices [4, 5], energy harvesting [6, 7],
photonics [3, 7–10] and electronics [7, 11].
Cellulose nanocrystals have a high elastic modulus, high
optical transparency, low thermal expansion coefficient and
flexible surface chemistry, which are features suitable for the
formation of nanocomposite materials. Compared to con-
ventional composite materials, nanocomposites show unique
properties even for very low filler content. This characteristic
feature is a result of the filler material’s nanoscale dimension
[12]. In the majority of the CNC-based nanocomposite mate-
rials, CNCs have been incorporated as a reinforcing material
owing to its very high Young’s modulus, good dispersion in
hydrophilic systems, high aspect ratio and ability to form
percolated network in the polymer matrix [13]. CNCs are
best suited to form nanocomposites with water-dispersible
polymers since these nanocrystals can form stable colloi-
dal dispersions in water. In addition, using surfactants and
chemical grafting CNCs can be chemically modified to be
dispersed in organic solvents and can thus be incorporated
into a wide range of polymer matrices [1, 13].
Cellulose nanocrystals (CNCs) and cellulose nanofib-
ers (CNFs) are relatively new entrants in the realization of
biodegradable and renewable electronic devices [14]. There
Electronic supplementary material The online version of this
article (https://doi.org/10.1007/s10854-018-0409-y) contains
supplementary material, which is available to authorized users.
* Kazi M. Alam
kmalam@ualberta.ca
* Karthik Shankar
kshankar@ualberta.ca
1
Department of Electrical & Computer Engineering,
University of Alberta, Edmonton, AB T6G 1H9, Canada
2
National Research Council National Institute
for Nanotechnology, Edmonton, AB T6G 2M9, Canada
3
Department of Applied Chemistry, Sardar Vallabhbhai
National Institute of Technology (SVNIT), Surat 395007,
India

1391Journal of Materials Science: Materials in Electronics (2019) 30:1390–1399
1 3
have been a few recent studies of transparent nanocellulose
papers for their potential use as substrates in flexible and
biodegradable electronic devices. Nanocellulose based trans-
parent papers have many advantages over plastic substrates,
such as sustainability, recyclability, low-cost preparation,
light weight and flexibility that allows large scale roll-to-roll
fabrication similar to newsprint [14]. Recently nanocellulose
has been used in electrochromic devices for the realization
of deformable displays [15]. The potential of nanocellulose
has been explored in some more recent studies such as touch
sensors [16, 17], organic solar cells [18], thin film transistors
[19–21] and organic light emitting diodes [22, 23]. Smart
actuators were realized using single-walled carbon nanotube/
polyaniline coated cellulose electroactive paper [24] and by
mixing cellulose with multi-walled carbon nanotubes [25].
As mentioned previously, there are a large number of
publications where CNCs have been used to reinforce poly-
mer matrices [1, 2]. On the other hand, experimental studies
on CNC based conductive organic polymer nanocomposites
for electronic applications or cellulosic electronic materials
in general, are scarce. Wang et al. formed highly conduc-
tive core–shell CNC@poly(dopamine)@Ag nanoparticles
and proposed their use in conductive adhesives [26]. Hao
et al. synthesized a textile consisting of polypyrrole grown
on knitted cellulose fabric with an electrical sheet conduct-
ance of 303 Ω/□ [27]. Karimi et al. fabricated graphene
paper through drop-casting of an ink they developed which
consisted of a mixture of 300 nm graphene platelets and
ethylcellulose [28]. In this work, we opted to explore the
properties and applications of a nanocomposite material
comprised of CNCs blended with the widely used conduc-
tive conjugated polymer Poly(3,4-ethylenedioxythiophene
):poly(styrene sulfonate) or PEDOT:PSS. The negatively
charged saturated PSS is a polymer surfactant, which helps
positively charged conjugated PEDOT to disperse in water
and other solvents [29]. In its role as the hole transport
layer, PEDOT:PSS constitutes a near essential component
of organic light emitting diodes and organic solar cells due
to its high transparency, flexibility, wide conductivity range
dependent on synthetic conditions and high work function
[29]. Tkalya et al. reported a nanocomposite consisting of
cellulose nanowhiskers, polystyrene (latex) spheres and
PEDOT:PSS with an extremely low percolation threshold
for PEDOT:PSS [30]. However, the maximum loading of
CNCs used was 5 wt% and the low percolation threshold
was speculatively attributed to the templating effect of the
cellulose nanowhiskers. In this work, the typical loading of
CNCs was 50 wt% and blends with CNC loading as high as
90 wt% were also examined. The blended nanocomposites
showed no loss in conductivity even for CNC loading as
high as 50 wt%. In addition, we report a number of hitherto
unreported observations—(i) unusual self-organization of
CNC–PEDOT:PSS blended films into vertical stacks with a
pitch of 100–200 nm (ii) improved local molecular ordering
of PEDOT chains in the blends and (iii) clear evidence for
polymer chains wrapping around the CNCs.
2 Experimental section
2.1 Materials
Cellulose nanocrystals were supplied by Innotech Alberta
(Edmonton, Canada), and typical SEM images of the CNCs
used are shown in Fig. 1. These CNCs were H2SO4 hydro-
lyzed and extracted from wood pulp. Elemental analysis
using CHNS testing indicated that C, H and S content are
40.88%, 6.11% and 1.17% respectively. There were, as
expected, no measurable quantities of nitrogen. Water-dis-
persed conductive grade PEDOT:PSS was purchased from
Sigma-Aldrich, Canada.
Fig. 1 Field emission scanning electron microscope (FESEM) images
of blend morphology—(a), (c) and (e) are top-view images of pristine
CNC films, pristine PEDOT:PSS films and CNC–PEDOT:PSS nano-
composite films while (b), (d) and (f): cross-sectional images of (a),
(c) and (e) respectively. In f the blend is seen to self-organize into lay-
ered stacks with a pitch of ca. 100–200 nm

1392 Journal of Materials Science: Materials in Electronics (2019) 30:1390–1399
1 3
2.2 Nanocomposite processing
Cellulose nanocrystals have a strong self-association ten-
dency because of the presence of surface hydroxyl groups
which promote aggregation, detrimental to the fabrication
of nanocomposites with polymers. Acid hydrolysis is the
most common strategy to achieve homogenous dispersion
in solution. Hydrochloric acid hydrolyzed CNCs result
in unstable dispersion [12]. We have used sulfuric acid
(H2SO4) hydrolyzed CNCs and water soluble PEDOT:PSS
to fabricate thin and thick nanocomposite films. H2SO4 is
the most commonly used hydrolyzing agent for CNCs as
it reacts with the surface hydroxyl groups of CNCs via an
esterification process that results in grafting of anionic sul-
fate ester groups. These negatively charged groups create a
negative electrostatic layer around the CNC surface which
facilitates high dispersion in water [31]. CNCs were dis-
solved in DI water using a high power probe ultrasonica-
tor (Ultrasonics FS-450N ultrasonic processor, amplitude
50% of 450 W, 20 kHz, probe diameter 3 mm) for 1 h
in an ice-bath to prevent desulfation caused by heating
of the suspension. Next, CNC and PEDOT:PSS solutions
were mixed with different weight ratios and then stirred for
30 min followed by ultrasonication at room temperature
in an ultrasonic bath for 30 min. Thick (~ 80 µm) and thin
films (~ 100 nm) of pristine CNC, pristine PEDOT:PSS
and blended nanocomposite were prepared by drop cast-
ing and spin coating methods respectively. Both types of
films were fabricated via casting-evaporation process, that
allows the solvent to evaporate. All the films were formed
on glass at room temperature. Prior to casting, glass sub-
strates were cleaned in acetone and isopropanol separately
followed by oxygen plasma etching (Oxford NGP80).
2.3 Characterization
2.3.1 Electron microscopy
The surface topographical and cross-sectional images of
pristine CNCs, pristine PEDOT:PSS and blended nanocom-
posite films were obtained using a field emission scanning
electron microscope (Zeiss Sigma FESEM) operated at 3 kV.
2.3.2 Atomic force microscopy
Individual nanoparticle images of the pristine CNCs and
CNCs blended with PEDOT:PSS were acquired with an
atomic force microscope (Bruker Dimension Edge) operat-
ing in air. The samples were prepared by drop casting very
dilute CNCs and blended suspension onto a SiO2 surface. All
the samples were scanned in tapping mode. A silicon AFM
probe (TAP300-G-50) with resonant frequency 300 kHz and
force constant 40 N/m was used to image all the samples.
2.3.3 Dynamic light scattering
Particle size distributions for pristine and blended CNCs
in solution were measured using dynamic light scattering
(DLS) technique (Malvern Nano-ZS DLS). Solutions were
diluted prior to the measurements. All DLS data was ana-
lyzed by Malvern nanosizer software which uses the normal
resolution cumulants analysis method.
2.3.4 Multi‑angle static light scattering
Static light scattering experiments were performed using five
different concentrations for both the CNC and blended solu-
tions. A Brookhaven BI-200SM multi-angle light scattering
system was employed for this experiment.
2.3.5 X‑ray diffractometry
Powder X-ray diffraction patterns of the three different
thick films (pristine CNC, pristine PEDOT:PSS and blend)
were recorded using Bruker D8 X-Ray Diffractometer with
a CuKα, IµSµ radiation source (λ = 1.5406 Å) operating at
50 W at room temperature. The detector for this XRD was a
2-D detector (VANTEC-500).
2.3.6 Electrical conductivity measurements
Electrical resistivity measurements for both thick and thin
films of pristine PEDOT:PSS films and blended films were
performed using a four-point probe (Lucas Pro4 4000) sys-
tem with Keithley2601 as a source meter. Prior to the resis-
tivity measurements, FESEM and ellipsometry (Gaertner
L115S—Ellipsometer) were used to determine the thickness
of the thick and thin films respectively.
2.3.7 Fourier transform infrared spectroscopy
FTIR spectra were obtained using an Agilent FTS7000
FTIR Imaging System (diamond ATR/attenuated total
reflection) and recorded in absorbance mode in the range
of 1700–3600 cm−1
. Flakes of the pristine CNCs, pristine
PEDOT:PSS and blended thick films (prepared by drop cast
and subsequent casting-evaporation method) were peeled-off
from the glass substrate and transferred onto the diamond
crystal prior to the data collection.
2.3.8 Raman spectroscopy
Raman spectra were collected for pristine CNCs, pristine
PEDOT:PSS and the blended films using a Nicolet Omega

1 3
XR Raman Microscope with an excitation wavelength of
532 nm.
2.3.9 UV–Vis spectroscopy
The optical spectra of pristine CNCs, pristine PEDOT:PSS
and blended solid films were measured with a Perkin Elmer
Lambda 1050 UV–Vis-NIR spectrophotometer equipped
with an integrating sphere.
3 Results and discussion
3.1 Morphology and crystallinity of pristine
and blended films
Field emission scanning electron microscope (FESEM)
images of three thick solid films (pristine CNC, pristine
PEDOT:PSS and blended nanocomposite) are shown in
Fig. 1. Each of the films were formed by drop-casting and
are several micrometers in thickness. Both Fig. 1a and Fig.
S1a in ESM show the mesh-like film morphology formed
by casting CNCs from concentrated aqueous suspensions.
Figure 1c shows the smooth, featureless surface of the pris-
tine PEDOT:PSS thin film. However, the cross-sectional
image of the CNC–PEDOT:PSS blended film (Fig. 1f) is
unusual in that uniform, stacked layers are clearly observ-
able while such stacking is not seen in the pristine CNC and
PEDOT:PSS films. The stacked layers are also somewhat
periodic, exhibiting a pitch of 100–200 nm or roughly the
length of 1–2 CNC nanorods. The top-view FESEM image
of the CNC–PEDOT:PSS blended film (Fig. 1e) indicates
the nanorods to be vertically oriented and covered by the
polymer, which is fully confirmed in the magnified profile
view FESEM image of a blend cast from CNC-dominant
suspension (Fig. S2 in ESM). The use of a high concentra-
tion of CNCs induces the agglomeration of CNCs (clearly
observable in Fig. S2 in ESM) while also highlighting
the characteristic vertically oriented morphology. Taken
together, the results indicate that the blended films segre-
gate into layers of vertically oriented CNCs wrapped with
polymer.
Powder XRD patterns for pristine CNC, pristine
PEDOT:PSS and the blended nanocomposite are shown in
Fig. 2. The diffraction peaks at 15.2°, 16.7°, 22.6° and 34.7°
are present in the pristine CNC and the blend (Fig. 2a, c).
These peaks correspond to (1̄10) , (110), (200) and (040)
planes of cellulose I respectively [32–34]. The two broad
peaks at 17.4° and 26° in the pristine PEDOT:PSS sam-
ple (Fig. 2b) represent agglomerated polymer chains [35,
36]. The disappearance of two broad peaks of PEDOT:PSS
in the blended composite is indicative of the absence of
long-range order in the packing of the polymer chains. This
result is in agreement with the cross-sectional images of
the films in Fig. 1 where in the blended composite has a
stacked layered structure in contrast to the small agglomer-
ated regions seen in the cross-sectional electron micrograph
of the PEDOT:PSS sample (Fig. 1d). The diffractogram of
the blended sample resembles the pristine CNC peaks. No
noticeable change in the peak positions, broadening or inten-
sity was observed. This implies that CNCs preserved their
native crystallinity in the nanocomposite and a good mixing
of the constituent materials occurred in the nanocomposite.
3.2 Nanoparticle size
Light scattering is a powerful and non-intrusive tool to
study macromolecules and their interactions in solution
without the need for immobilization or labeling [37].
Dynamic light scattering (DLS) is a widely used method
to estimate particle size in a dispersed medium. Particles
undergo Brownian motion in solution and DLS measures
the translational diffusion coefficient. The light scattering
intensity fluctuates because of this motion and this is ana-
lyzed by an autocorrelation function decay that is connected
to the translational diffusion coefficient [38]. We diluted the
pristine CNC and the blended solutions in DI water before
Fig. 2 X-Ray diffractograms of a pristine CNCs, b pristine
PEDOT:PSS and c CNC–PEDOT:PSS nanocomposite (CNC
60 wt%). The absence of broad peaks of PEDOT:PSS in the nano-
composite sample indicates structural modification of the polymer
agglomeration

1 3
the DLS measurements. Diluted solutions are needed as the
Stokes–Einstein relation that relates the translational coef-
ficient to the hydrodynamic diameter is applicable only for
diluted solutions [38]. Fig. S3 in ESM shows the particle
size distribution with respect to intensity obtained by DLS.
We found a monomodal peak and low polydispersity index
for both the solutions. These peaks were at 68.43 nm and
101.7 nm for the pristine and blended solutions respectively.
The blended solutions gave a slightly higher polydispersity
index value (0.586) compared to the pristine CNC solution
(0.248). Both the values are less than 1 which indicates good
dispersion and homogeneity. These data revealed a higher
Z-average value (hydrodynamic diameter) for the blended
particles (~ 67 nm) compared to pristine CNCs (~ 54 nm).
Note that for a particle of nonspherical shape, the hydrody-
namic diameter is the diameter of an imaginary sphere that
yields the same translational coefficient as the actual particle
of arbitrary shape. Therefore, hydrodynamic diameter for
the high aspect-ratio nanocrystal rods is a reliable metric for
comparison purposes. The increase in hydrodynamic diam-
eter of blended nanocrystals is indicative of the formation of
a stable shell of PEDOT:PSS around the CNCs.
We performed multi-angle static light scattering
(MASLS) measurements to calculate radius of gyration,
molecular weight and second virial coefficient for pristine
CNC and CNC blended PEDOT:PSS solutions in DI water
using five different solutions of varying concentrations for
both the samples. Figs. S4 and S5 in ESM show the Zimm
plots for the pristine CNC and the blended samples. In these
plots θ and c are the incident light angle and concentra-
tion of the solution respectively, ΔRθ is the Rayleigh ratio
and K is an optical constant. The measured quantity was
the Rayleigh ratio for all the samples at various incident
angles. The plots were obtained from the calculated values
and extrapolated to zero angle at each concentration and also
to zero concentration at each angle [39]. The calculated val-
ues of radius of gyration and molecular weight for the CNC
samples were (50.2 ± 4) nm and (1.118 ± 0.038) × 107
g/
mol respectively. The corresponding values for the blended
sample were (97.9 ± 8.4) nm and (1.159 ± 0.088) × 107
g/
mol. The above values were obtained from the extrapola-
tion to zero angle at each concentration. Extrapolation to
zero concentration at each angle also yielded the values of
second virial coefficient and molecular weight. These values
for CNC samples were (7.1 ± 2.8) × 10−6
cm3
mol/g2
and
(1.12 ± 0.19) × 107
g/mol respectively. For the blended sam-
ples, the second virial coefficient and molecular weight were
(5.6 ± 1.7) × 10−5
cm3
mol/g2
and (1.16 ± 0.67) × 107
g/mol
respectively. The increase in both radius of gyration and
molecular weight for the blended samples in comparison to
pristine CNC samples further confirms the strong associa-
tion of conducting polymer chains with nanocrystal surface.
The second virial coefficient values were positive, which
indicates repulsive interactions between CNCs in both the
solutions, consistent with negatively charged CNC surfaces.
However, one report by Ureña-Benavides and Kitchens [39]
found a negative value and this apparently non-intuitive phe-
nomenon was explained by a possible existence of a gel-like
layer of sulfated oligosaccharides. This layer was claimed
to surround the CNC surface and trap protons and sulfate
ions, which in turn partially screen the electrostatic repulsive
interactions [39, 40]. We used high power ultrasonication
energy for 1 h (experimental section) in order to obtain a
homogeneous high concentration CNC dispersion. There-
fore, this gel-like layer is unlikely to adhere to the CNC in
solution. However, in the blended sample, the value of the
second virial coefficient value decreased by nearly an order
of magnitude. This is not surprising since CNCs are partially
covered by the polymer as both the hydrodynamic diameter
and radius of gyration values are higher for the blended solu-
tions compared to pristine CNC solution. Therefore, CNCs
are expected to experience weaker repulsive interactions in
the blended solution.
Atomic force microscopy (AFM) is a powerful and widely
used method to investigate the size distribution of CNCs.
AFM gives valuable information regarding surface topog-
raphy [1, 41]. We used AFM to determine the 3-D shape
and dimensions of pristine CNCs and CNCs blended with
PEDOT:PSS. Figure 3a, c show the height images of a single
nanocrystal blended with conducting polymer and pristine
CNC respectively. Figure 3b, d show the corresponding line
scan analysis of Fig. 3a, c. It is evident from these images
and plots that cellulose nanoparticles in the blend have larger
physical dimensions in both the vertical (height) and lat-
eral directions. Therefore, these AFM images corroborate
the results obtained from DLS and MASLS experiments
where we found a larger hydrodynamic radius and radius
of gyration for the blended nanoparticles compared to pris-
tine CNCs. Note that in AFM, the height values are more
reliable than the width data as width values are generally
overestimated due to tip convolution [42]. The size increase
for the blended nanocrystals is consistent with the FESEM
cross-sectional images we discussed earlier in the previ-
ous section. As mentioned in the experimental section, we
have used H2SO4 hydrolyzed CNCs. Acid hydrolysis gives
a broader distribution of CNC sizes as a result of diffusion-
controlled nature of the process [1, 12, 31]. Fig. S6 in ESM
shows the AFM image of the CNCs on an SiO2/Si substrate.
The estimated length distribution of the nanocrystals based
on this image is between 50 and 400 nm and the diameter is
between 3 and 6 nm.
3.3 Electrical conductivity measurements
Electrical characterization was conducted at room tempera-
ture with a four-probe conductivity system integrated with

1 3
a Keithley source meter. Electrical conductivity (σ) was cal-
culated from the measured sheet resistance (RS) and film
thickness (t) according to the following equation [43, 44]:
Prior to the experiment, film thickness measurements
were performed using FESEM and ellipsometry for the
drop-coated and spin-coated films respectively. According
to Fig. 4, the conductivity of PEDOT:PSS is more or less
retained up to a significantly high CNC loading (60 wt%) for
both the drop-coated and spin-coated film samples despite
the fact that CNCs are insulating. The drop-coated films had
an average thickness of 25 µm and could be delaminated
from the substrate by scratching the edges of the films to
yield free-standing membranes such as shown in the Inset
of Fig. 4 and Fig. S8 in ESM.
3.4 Infrared and optical spectroscopy
We performed Fourier Transform Infrared Spectroscopy
(FTIR) on our samples in ATR (attenuated total reflec-
tion) mode in order to investigate the interaction of cel-
lulose nanocrystals with the conducting polymer in the
CNC–PEDOT:PSS nanocomposite. Figure 5 shows the
FTIR spectra between 1700 and 3600 cm−1
for pristine
CNCs, pristine PEDOT:PSS and the blend. The inset of
(1)𝜎 =
1
RS × t
Fig. 3 Atomic force microscope
(AFM) images of a single CNC
wrapped by PEDOT:PSS poly-
mer and c pristine single CNC.
AFM size measurement by line
scan of b single CNC wrapped
by PEDOT:PSS polymer and
d pristine single CNC. The
increased horizontal distances
and heights for the nanocom-
posite show the association of
the polymer chains with the
nanocrystals
Fig. 4 Four-probe conductivity measurement of CNC–PEDOT:PSS
nanocomposite (of varying CNC load) for drop casted thick solid
films (black triangle) and spin coated thin solid films (red square).
Zero CNC load (0 wt%) corresponds to pristine PEDOT:PSS. Inset:
fully dried free standing flexible solid flake of CNC–PEDOT:PSS
nanocomposite (CNC 60 wt%) thick solid film held between two
fingers. Remarkably, both the spin-coated and drop-coated films of
the CNC–PEDOT:PSS blend show very little loss in conduction up
to CNC loading of 50 wt%. A sharp drop in the conductivity of the
blended film is observed only for CNC loading higher than 80 wt%.
(Color figure online)

1 3
Fig. 5 shows the normalized hydroxyl stretching region
between 3000 and 3600 cm−1
for CNCs. There are four
minor peaks at 3329, 3282, 3439 and 3486 cm−1
. The former
two peaks represent intermolecular hydrogen bonding while
the latter two are ascribed to intramolecular hydrogen bond-
ing [45]. The main hydroxyl stretching peak for the pris-
tine CNC is stronger and sharper compared to the blend
(CNC–PEDOT:PSS nanocomposite). This is an indication
of the presence of higher number of −OH groups on the
pristine CNC surface compared to the blend. The inset
shows the minor peaks shifted to higher wavenumbers in
the blended composite. Similar shifting was found in the
CNC composite with polyethylene matrix as a consequence
of hydrogen bonding [45]. This hydrogen bonding of CNC
with PEDOT is in agreement with the results of light scat-
tering experiments and AFM as we have seen an increase
in particle size in the blended nanocomposite. Figure 6 is a
schematic illustration of the plausible association of CNCs
with PEDOT:PSS, based on light scattering, AFM and FTIR
data. DLS and MASLS measurements are very clear about
a shell of polymer chains surrounding the CNCs, also con-
firmed by the topographic differences in AFM images. FTIR
data suggesting hydrogen bonding between the polymer and
CNCs implies that PSS strongly adsorbs on to the CNCs
as depicted in Fig. 6, which is also consistent with the fact
that PSS and CNCs are hydrophilic (and therefore likely to
interact) while PEDOT is hydrophobic (and therefore less
likely to adsorb on the CNCs).
Raman spectra of pristine CNC films, pristine
PEDOT:PSS films and the blended composite films are
shown in Fig. 7. In the Raman spectra of pristine CNCs
(Fig. 7a), the bands at 1098 and 1120 cm−1
arise from the
Fig. 5 FTIR spectra of pristine CNCs (black), pristine PEDOT:PSS
(red) and blended nanocomposite (CNC 60 wt%). The vertical scale
(absorption) is same for all the samples, but the spectra have verti-
cally split for improved visibility. Inset: normalized absorbance for
pristine CNC and blended nanocomposite in the hydroxyl stretching
region; The minor peaks ascribed to hydrogen bonding are shifted to
higher frequencies for the blended nanocomposite compared to the
pristine CNC. (Color figure online)
Fig. 6 Schematic illustration of the proposed association between CNCs and PEDOT:PSS chains based on the light scattering, AFM and FTIR
data

1 3
C–O–C stretching modes in cellulose [46–48]. The bands
between 250 and 600 cm−1
arise due to skeletal bending
modes [47]. In Fig. 7b for the pristine PEDOT:PSS, the
peak at 1436 cm−1
corresponds to the C=C symmetrical
stretching vibration [36, 49]. The peaks at 1256, 1368 and
1535 cm−1
are associated with the Cα–Cα inter-ring stretch-
ing modes, Cβ–Cβ stretching modes and symmetric Cα–Cα
stretching modes of PEDOT respectively [50] while the
peaks centered at 1504 and 1550 cm−1
have been associated
with the C=C asymmetric stretching vibrations that corre-
spond to thiophene rings in the middle and at the end of
the chains, respectively [36, 49]. The peaks at 445, 575 and
989 cm−1
correspond to the oxyethylene ring deformations
while the peak at 710 cm−1
was assigned to the symmet-
ric C–S–C deformation [51]. All the peaks of PEDOT:PSS
also appeared in the blend (Fig. 7c) except the peak at
2890 cm−1
while the relatively weak CNC bands are not
visible in the Raman spectra of the blended nanocomposite.
We used the technique of Shi et al. [52] which exploits the
invariance of the number and structure of oxyethylene rings
with doping, in order to compare the difference in aromatic
absorbances of pristine PEDOT:PSS films and the blended
CNC–PEDOT:PSS nanocomposite films. By normalizing
the spectra to the intensity of the peak at 445 cm−1
(Fig. S7
in ESM), we sought to compare subtler features in the spec-
tra which now represented the same number of thiophene
units [52]. The full-width at half-maximum (FWHM) of the
PEDOT peak at 1436 cm−1
for the pristine PEDOT:PSS and
nanocomposite are 59.5 cm−1
and 43.3 cm−1
respectively,
which corresponds to a 28% reduction (inset of Fig. 7 and
Fig. S7 in ESM). Likewise, the PEDOT peak at 989 cm−1
narrowed to a FWHM value of 11.2 cm−1
in the blended
nanocomposite film from a FWHM value of 12.5 cm−1
in
the pristine PEDOT:PSS film. Such dramatic reductions in
the width of multiple peaks provides evidence of improved
molecular ordering and increased effective conjugation
length of PEDOT upon blending with CNCs [53, 54].
Figure 8 shows the absorption spectra of the pristine
CNC, pristine PEDOT:PSS and the blended nanocom-
posite films, which were collected in diffuse reflectance
mode using an integrating sphere to eliminate the influ-
ence of particle scattering effects. The characteristic fea-
ture in the optical spectrum of the pristine PEDOT:PSS
film (red curve in Fig. 8) is the monotonically increas-
ing absorption for visible and near-infrared wavelengths
starting from roughly 400 nm, which is due to the bipo-
laron sub-bandgap transition [55, 56]. Bipolaronic states
originate from PSS doping which induces charged defects
in the conductive polymer PEDOT [56]. This transition
involves optically excited transition within electronic
band gap and is similar to free carrier absorption in
degenerately doped inorganic semiconductors [55, 56].
The UV–Vis spectrum of the pristine CNC film (black
curve in Fig. 8) is flat due to the absence of free carriers
in the film. However, even for a CNC loading of 82 wt%
(navy blue curve in Fig. 8), the bipolaron absorption fea-
ture is present and does not resemble the flat vis–NIR
absorption of pristine CNC film. The bipolaron absorp-
tion feature is strongest for the pristine PEDOT:PSS
film and gradually decreases in intensity as increasing
amounts of CNCs are added to the blend (Fig. 8) indicat-
ing the reduction in mobile carriers. However, the two
Fig. 7 Raman spectra of a pristine CNCs, b pristine PEDOT:PSS
and c CNC–PEDOT:PSS nanocomposite (CNC 60 wt%). Inset of (c)
shows the PEDOT peak at 1436 cm−1
has narrowed in the blended
nanocomposite compared to the pristine CNC
Fig. 8 UV–Vis (absorption) spectra of pristine CNCs (black), pristine
PEDOT:PSS (red) and CNC–PEDOT:PSS nanocomposites contain-
ing different amounts of CNCs. (Color figure online)

1 3
signatures of electronic dedoping, namely a dramatic
decrease in the bipolaron absorption band and the appear-
ance of neutral or polaronic bands, are completely absent.
The absence of chemical dedoping is also confirmed in
the normalized Raman spectra (Fig. S7 in ESM) wherein
the intensity ratio of the benzoidal to the quinoidal bands
is roughly the same in the pristine CNC films and the
CNC–PEDOT:PSS blended nanocomposite films. The
conductivity is directly proportional to the product of the
number of free carriers per unit volume and the carrier
mobility. In light of the UV–Vis data (Fig. 8) showing a
gradual reduction in the free carrier density with increas-
ing CNC loading, we infer a higher carrier mobility from
the conductivity data in Fig. 4 which shows almost no
reduction in electrical conduction of the blended nano-
composite films even for CNC loading as high as 50%.
Such an improvement in carrier mobility in the blended
film is consistent with the electron micrographs showing
stacked self-organization and the Raman spectra indicat-
ing improved molecular ordering.
Commercial PEDOT:PSS formulations (including the
one used in this work) contain an excess of the hydro-
philic PSS component to achieve good dispersibility in
aqueous suspensions, which are typically characterized by
colloidal micelles containing hydrophobic PEDOT with
some associated PSS, surrounded by a shell of “free PSS”
[57]. Casting of PEDOT:PSS films from such micellar sus-
pensions results in a phase-segregated film morphology
with sub-optimal electrical conduction wherein PEDOT-
rich domains are interconnected by PSS-rich domains
[58]. Consequently, a variety of treatments and additives
reported in the scientific literature that dissolve away or
otherwise remove the free PSS, result in improved con-
ductivity in the PEDOT:PSS films due to a reduction in
the insulating PSS component [59, 60]. In this work, the
excess (free) PSS is not removed. Instead, the hydrophilic
PSS preferentially adsorbs on to the surfaces of the simi-
larly hydrophilic CNCs, which in turn enables the forma-
tion of percolating PEDOT-rich networks that are also
characterized by improved molecular ordering of PEDOT
chains. Furthermore, cast-evaporation is a slow process
which allows CNCs as well as the conjugated polymer
chains to be interconnected and form percolation net-
works [1]. The formation of such percolation networks is
of paramount importance in the fabrication of reinforced
nanocomposites. However, no dedoping occurs since the
charge donating dopant (PSS) is still present in the blend.
Therefore, this percolation network retained the electrical
conductivity of PEDOT:PSS even after blending with high
concentrations of CNCs. These results bode well for the
deployment of CNC-based composites as substrates in thin
film organic optoelectronic devices, and for the synthesis
of conductive paper based on cellulosic nanomaterials.
4 Conclusion
Cellulose nanocrystals (CNCs) exhibit a strong interaction
with PEDOT:PSS which has multiple consequences for
the structure, morphology and optoelectronic properties of
the resulting blended films. Dynamic light scattering and
multi-angle light scattering indicate a significant increase
in the hydrodynamic radius/radius of gyration of the CNCs
due polymer chains of PEDOT:PSS wrapping around them
in blends, a result also confirmed by atomic force micros-
copy. The blended films spontaneously self-organize into a
layered stacked structure with a pitch corresponding to the
length of 1–2 CNCs that are vertically oriented in the layers
constituting the stack. Hydrogen bonds form between the
CNCs and PEDOT:PSS which is evident from the infrared
spectroscopic data and is indicative of PSS adsorption on the
surface of the CNCs. While X-ray diffraction results did not
show long range order in the stacking of PEDOT chains, the
dramatic narrowing of the Raman bands corresponding to
both the quinoidal and benzoidal conformations of PEDOT
chains in the blended nanocomposite indicated an improve-
ment in short range molecular ordering. UV–Vis spectra
showed a gradual reduction in the density of mobile carriers
through the weakening of the bipolaron absorption band as
the CNC loading was increased in the blended nanocom-
posite film. However, the conductivity of the films remained
nearly constant up to a CNC loading of 50 wt%. We infer a
higher carrier mobility in the blended nanocomposite films
from these results, which is a conclusion consistent with the
observation of self-organized stacks and enhanced molecular
ordering.
Acknowledgements This work was made possible by funding support
from NSERC, Alberta Innovates, FPInnovations, CFI and NRC-NINT.
We thank Dr. Wadood Hamad and his team at FPInnovations for con-
structive discussions and assistance with sample characterization. P.
B. thanks MITACS Globalink while U.K.T. and A.M. thank Alberta
Innovates for scholarship awards. Innotech Alberta is acknowledged
for providing CNC samples.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of
interest.
References
1. Y.L. Habibi, A. Lucian, O.J. Rojas, Chem. Rev. 110, 3479 (2010)
2. T. Abitbol, A. Rivkin, Y. Cao et al., Curr. Opin. Biotechnol. 39,
76 (2016). https://doi.org/10.1016/j.copbio.2016.01.002
3. J.P.F. Lagerwall, C. Schütz, M. Salajkova et al., NPG Asia Mater.
6, e80 (2014). https://doi.org/10.1038/am.2013.69
4. N. Lin, A. Dufresne, Eur. Polym. J. 59, 302 (2014). https://doi.
org/10.1016/j.eurpolymj.2014.07.025

1 3
5. A. Sinha, E.M. Martin, K.-T. Lim et al., J. Biosyst. Eng. 40, 373
(2015). https://doi.org/10.5307/jbe.2015.40.4.373
6. X. Wang, C. Yao, F. Wang, Z. Li (2017) Small. https://doi.
org/10.1002/smll.201702240
7. R. Sabo, A. Yermakov, C.T. Law, R. Elhajjar, J. Renew. Mater. 4,
297 (2016). https://doi.org/10.7569/jrm.2016.634114
8. A.G. Dumanli, H.M. van der Kooij, G. Kamita et al., ACS Appl
Mater. Interfaces 6, 12302 (2014). https://doi.org/10.1021/am501
995e
9. X. Mu, D.G. Gray, Langmuir 30, 9256 (2014). https://doi.
org/10.1021/la501741r
10. I. Usov, G. Nystrom, J. Adamcik et al., Nat. Commun. 6, 7564
(2015). https://doi.org/10.1038/ncomms8564
11. Y. Zong, T. Zheng, P. Martins, S. Lanceros-Mendez, Z. Yue, M.J.
Higgins, Nat. Commun. 8, 38 (2017). https://doi.org/10.1038/
s41467-017-00034-4
12. M. Mariano, N.El Kissi, A. Dufresne, J. Polym. Sci., Part B:
Polym. Phys. 52, 791 (2014). https://doi.org/10.1002/polb.23490
13. J. George, S.N. Sabapathi, Nanotechnol. Sci. Appl. 8, 45 (2015).
https://doi.org/10.2147/NSA.S64386
14. Z. Lian, W. Wang, G. Li, F. Tian, K.S. Schanze, H. Li, ACS Appl.
Mater. Interfaces. 9, 16959 (2017). https://doi.org/10.1021/acsam
i.6b11494
15. W. Kang, C. Yan, C.Y. Foo, P.S. Lee, Adv. Funct. Mater. 25, 4203
(2015). https://doi.org/10.1002/adfm.201500527
16. S. Ji, B.G. Hyun, K. Kim et al., NPG Asia Mater. 8, e299 (2016).
https://doi.org/10.1038/am.2016.113
17. H. Zhu, Z. Fang, Z. Wang et al., ACS Nano 10, 1369 (2016). https
://doi.org/10.1021/acsnano.5b06781
18. Y. Zhou, C. Fuentes-Hernandez, T.M. Khan et al., Sci. Rep. 3,
1536 (2013). https://doi.org/10.1038/srep01536
19. Q. Zhang, W. Bao, A. Gong et al., Nanoscale 8, 14237 (2016).
https://doi.org/10.1039/c6nr01534d
20. Y. Fujisaki, H. Koga, Y. Nakajima et al., Adv. Funct. Mater. 24,
1657 (2014). https://doi.org/10.1002/adfm.201303024
21. D. Gaspar, S.N. Fernandes, A.G. de Oliveira et al., Nanotech-
nology 25, 094008 (2014). https://doi.org/10.1088/0957-
4484/25/9/094008
22. W. Wu, N.G. Tassi, H. Zhu, Z. Fang, L Hu, ACS Appl. Mater.
Interfaces 7, 26860 (2015). https://doi.org/10.1021/acsami.5b092
49
23. S. Purandare, E.F. Gomez, A.J. Steckl, Nanotechnology 25,
094012 (2014). https://doi.org/10.1088/0957-4484/25/9/094012
24. S. Yun, J. Kim, Z. Ounaies (2006) Smart. Mater. Struct. 15, N61.
https://doi.org/10.1088/0964-1726/15/3/n02
25. S. Yun, J. Kim, Smart Mater. Struct. 16, 1471 (2007). https://doi.
org/10.1088/0964-1726/16/4/062
26. H. Wang, H. Pang, D. Wei et al., J. Mater. Sci.: Mater. Electron.
29, 9829 (2018). https://doi.org/10.1007/s10854-018-9023-2
27. D. Hao, B. Xu, Z. Cai, J. Mater. Sci.: Mater. Electron. 29, 9218
(2018). https://doi.org/10.1007/s10854-018-8950-2
28. K. Karimi, E. Jabari, E. Toyserkani, P. Lee-Sullivan, J. Mater. Sci.:
Mater. Electron. 29, 2537 (2018). https://doi.org/10.1007/s1085
4-017-8176-8
29. K. Sun, S. Zhang, P. Li et al., J. Mater. Sci.: Mater. Electron. 26,
4438 (2015). https://doi.org/10.1007/s10854-015-2895-5
30. E. Tkalya, M. Ghislandi, W. Thielemans, P. van der Schoot, G.
de With, C. Koning, ACS Macro Lett. 2, 157 (2013). https://doi.
org/10.1021/mz300597j
31. A. Dufresne, Mater. Today 16, 220 (2013). https://doi.
org/10.1016/j.mattod.2013.06.004
32. M.-C. Li, Q. Wu, K. Song, S. Lee, Y. Qing, Y. Wu, ACS Sustain.
Chem. Eng. 3, 821 (2015). https://doi.org/10.1021/acssuschem
eng.5b00144
33. M.L. Hassan, C.M. Moorefield, H.S. Elbatal, G.R. Newkome,
D.A. Modarelli, N.C. Romano, Mater. Sci. Eng. B 177, 350
(2012). https://doi.org/10.1016/j.mseb.2011.12.043
34. L. Du, J. Wang, Y. Zhang, C. Qi, M.P. Wolcott, Z. Yu (2017)
Nanomaterials (Basel). https://doi.org/10.3390/nano7030051
35. F. Zabihi, Y. Xie, S. Gao, M. Eslamian, Appl. Surf. Sci. 338, 163
(2015). https://doi.org/10.1016/j.apsusc.2015.02.128
36. D. Yoo, J. Kim, J.H. Kim, Nano Res. 7, 717 (2014). https://doi.
org/10.1007/s12274-014-0433-z
37. D. Some, Biophys. Rev. 5, 147 (2013). https://doi.org/10.1007/
s12551-013-0107-1
38. Y. Boluk, C. Danumah (2013) J. Nanopart. Res. https://doi.
org/10.1007/s11051-013-2174-4
39. E.E. Ureña-Benavides, C.L. Kitchens, (2012) RSC Adv. 2: 1096.
https://doi.org/10.1039/c1ra00391g
40. S. Beck, J. Bouchard, R. Berry, Biomacromol 12, 167 (2011)
41. V. Khoshkava, M.R. Kamal, Biomacromol 14, 3155 (2013). https
://doi.org/10.1021/bm400784j
42. Z. Abas, H.S. Kim, J. Kim, J.-H. Kim (2014) Front. Mater. https
://doi.org/10.3389/fmats.2014.00017
43. W. Meng, R. Ge, Z. Li et al., ACS Appl. Mater. Interfaces 7,
14089 (2015). https://doi.org/10.1021/acsami.5b03309
44. J. Luo, D. Billep, T. Waechtler et al., (2013) J. Mater. Chem. A.
https://doi.org/10.1039/c3ta11209h
45. X. Xu, F. Liu, L. Jiang, J.Y. Zhu, D. Haagenson, D.P. Wiesen-
born, ACS Appl. Mater Interfaces 5, 2999 (2013). https://doi.
org/10.1021/am302624t
46. N. SM Gierlinger, A. Reinecke, I. Burgert (2006) Biomacromol-
ecules 7, 2077
47. A.E. Lewandowska, S.J. Eichhorn, J. Raman Spectrosc. 47, 1337
(2016). https://doi.org/10.1002/jrs.4966
48. U.P. Agarwal, R. Sabo, R.S. Reiner, C.M. Clemons, A.W. Rudie,
Appl. Spectrosc. 66, 750 (2012). https://doi.org/10.1366/11-06563
49. P. GR, S.R.V.A. Kanwat, J. Jang, Mater. Res. Bull. 74, 346 (2016).
https://doi.org/10.1016/j.materresbull.2015.10.044
50. T.P. Nguyen, S.A. de Vos, Appl. Surf. Sci. 221, 330 (2004). https
://doi.org/10.1016/S0169-4332(03)00952-8
51. A.A. Farah, S.A. Rutledge, A. Schaarschmidt, R. Lai, J.P.
Freedman, A.S. Helmy (2012) J. Appl. Phys. 112. https://doi.
org/10.1063/1.4768265
52. W. Shi, Q. Yao, S. Qu, H. Chen, T. Zhang, L. Chen (2017) NPG
Asia Mater. 9: e405. https://doi.org/10.1038/am.2017.107 https://
www.nature.com/articles/am2017107#supplementary-informatio
n
53. P. Veerender, V. Saxena, A.K. Chauhan et al., Sol. Energy Mater.
Sol. Cells 120, 526 (2014). https://doi.org/10.1016/j.solma
t.2013.09.034
54. J. Ouyang, Q. Xu, C.-W. Chu, Y. Yang, G. Li, J. Shinar, Polymer
45, 8443 (2004). https://doi.org/10.1016/j.polymer.2004.10.001
55. C.Sheng Hsiung, C. Chien-Hung, K. Feng-Sheng, T. Chuen-
Lin, W. Chun-Guey, IEEE Photonics J. 6, 1 (2014). https://doi.
org/10.1109/jphot.2014.2331254
56. J. Hwang, I. Schwendeman, B.C. Ihas et al., (2011) Phys. Rev. B.
https://doi.org/10.1103/PhysRevB.83.195121
57. J. Saghaei, A. Fallahzadeh, T. Saghaei, Org. Electron. 24, 188
(2015). https://doi.org/10.1016/j.orgel.2015.06.002
58. X. Crispin, F.L.E. Jakobsson, A. Crispin et al., (2006) Chem.
Mater. 18, 4354. https://doi.org/10.1021/cm061032%23x002B;
59. P.-W. Sze, K.-W. Lee, P.-C. Huang, D.-W. Chou, B.-S. Kao, C.-J.
Huang, Energies 10, 716 (2017)
60. J.E. McCarthy, C.A. Hanley, L.J. Brennan, V.G. Lambertini, Y.K.
Gun’ko, J. Mater. Chem. C 2, 764 (2014). https://doi.org/10.1039/
C3TC31951B

Remarkable self-organization and unusual conductivity behavior in cellulose nanocrystal-PEDOT: PSS nanocomposites

More Related Content

What's hot (15)

Similar to Remarkable self-organization and unusual conductivity behavior in cellulose nanocrystal-PEDOT: PSS nanocomposites (20)

More from Pawan Kumar (20)

Recently uploaded (20)

Remarkable self-organization and unusual conductivity behavior in cellulose nanocrystal-PEDOT: PSS nanocomposites