Synthesis and pharmacological evaluation of novel arylpiperazine oxicams derivatives as potent analgesics without ulcerogenicity

Article abstract of DOI:10.1016/j.bmc.2019.03.007

Gastrotoxicity continues to be a major issue in therapy with nonsteroidal anti-inflammatory drugs (NSAIDs). Medicine is yet to develop absolutely safe analgesics. Numerous strategies are employed to discover new, safer NSAIDs, for example selective inhibi

Full text of DOI:10.1016/j.bmc.2019.03.007

Accepted Manuscript
Characterising exfoliated few-layer graphene interactions in co-processed nano-
fibrillated cellulose suspension via water retention and dispersion rheology
Katarina Dimi-Misic, Josphat Phiri, Kaarlo Nieminen, Thad Maloney, Patrick
Gane
PII:
S0968-0896(18)31601-8
https://doi.org/10.1016/j.bmc.2019.03.007
BMC 14799
DOI:
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
10 September 2018
26 February 2019
2 March 2019
Please cite this article as: Dimi-Misic, K., Phiri, J., Nieminen, K., Maloney, T., Gane, P., Characterising exfoliated
few-layer graphene interactions in co-processed nanofibrillated cellulose suspension via water retention and
dispersion rheology, Bioorganic & Medicinal Chemistry (2019), doi: https://doi.org/10.1016/j.bmc.2019.03.007
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Characterising exfoliated few-layer graphene interactions in co-processed
nanofibrillated cellulose suspension via water retention and dispersion rheology
Katarina Dimi-Misic1, Josphat Phiri1, Kaarlo Nieminen1, Thad Maloney1 and Patrick Gane1,2
1Aalto University, Department of Bioproducts and Biosystems, School of Chemical Engineering, FI-00076
Aalto, Helsinki, Finland
2Omya International AG, Baslerstrasse 42, CH-4665 Oftringen, Switzerland
ABSTRACT
Few-layer graphene has been produced by mechanical delamination of exfoliated and naturally obtained
graphite in aqueous suspension using the dispersion and suspension properties of nanofibrillated cellulose
(NFC). Various degrees of graphene platelet integrity were obtained depending upon the processing
conditions and the optional adoption of surfactant to aid dispersion of the hydrophobic agglomerates of
nanometre-thin carbon material. The presence of NFC in the suspension acts similarly to the presence of
surfactant, increasing the hydrodynamic coupling between the particles and water as a function of
processing time, regardless of the graphene-comprising source. By fitting the stress growth region in the
stress-shear rate relation to a concatenated series of single exponential functions of shear rate, the power
law exponent and suspension consistency parameters (n and k), within a shear rate-localised Herschel-
Bulkley (HB) expression, provide a straightforward characteristic for monitoring the desired suspension
coupling response, and hence a measure of product constancy.
Keywords: few-layer graphene, micro/nanofibrillar cellulose, coprocessing graphite delamination in
cellulose, graphene-cellulose-water interaction under flow, quality parameterisation, power law shear
response
1. INTRODUCTION
Renewable and biodegradable cellulose-based materials have shown promise for various potential
applications including replacement of fossil oil-based materials due to their excellent functional properties
[1, 2]. Cellulose fibres, which contain the most abundant natural polymer, can be processed to form
microfibrillated cellulose (MFC) or nanofibrillated cellulose (NFC) by mechanical, chemical or enzymatic
processes [3, 4]. NFC has been demonstrated in many applications related to its functional properties such

as high surface area, high strength, biodegradability, low density etc. [1, 4, 5]. The nanofibrils in NFC are
typically 4-20 nm nanometres in diameter and a 500 – 2 000 nm in length [6]. When dispersed in water,
these highly surface charged fibrils form a stable dispersed transparent water retaining gel-like suspension,
at concentrations as low as 0.03 % via hydrogen bonding between nanofibrils and water molecules [7, 8].
Due to the high aspect ratio and easily modifiable surface chemistry, micro nanofibrillated cellulose (MNFC)
has high potential to act as a building block for various applications including reinforcement in polymer
matrices, aerogels, foams, nanopaper, transparent and flexible films [9]. Chemical pre-treatments of pulp
significantly lower the energy consumption in NFC production, making the industrial manufacture of NFC
economically feasible [1, 2, 10]. One such treatment is aqueous (2,2,6,6-tetramethylpiperidin-1-yl)oxyl
(TEMPO)-mediated oxidation, due to its high efficiency and moderate chemical costs [7]. TEMPO-mediated
oxidation that drastically reduces the energy consumption during the pulp disintegration to NFC through
enabling breakdown of the fibre structure into the nanofibrillar form with high surface charge, which, due
to the high aspect ratio of the fibrils, acts as a water binding agent and provides strong repellence between
the particles, resulting in the formation of the gel, which in the case of pure nanofibrils is transparent [10,
12].
Graphene, contrastingly, is a 2-dimensional material that has attracted a lot of attention due to its
fascinatingly unique properties [15, 16]. Single graphene sheets have shown extremely high values of
electron mobility > 200 000 cm2V- 1s-1 at room temperature, which are much greater than in metals or even
semiconductors [16, 17]. Graphene is also ~200 times stronger than steel with a Young’s modulus of 1 100
GPa and fracture strength of about 125 GPa [17, 18]. Theoretically, graphene has a specific surface area of
2 630 m2g-1 and extremely high thermal conductivity of 5 000 Wm-1K-1, which is even greater than that of
diamond [20]. Apart from generating mechanical properties, graphene’s extraordinary electrical properties
will direct its utilisation as a filler in many composites designed to act as supercapacitors [21, 22]. These
properties of graphene have led to a wide range of potential applications including printed electronics,
nanocomposites, sensors, printed solar cells, transparent electrodes and catalysts [23-31]. However, large-
scale application of graphene is still a major challenge due to lack of feasible, viable and efficient
production methods [32, 33].
Fabrication of electrically conductive and mechanically strong graphene-based composites, especially those
including biodegradable and renewable polymers, is of significant interest [35, 41]. One of the major
potential and promising applications of graphene is its use as a functional reinforcing filler in the production
of composite materials [29, 31]. One such example is graphene-NFC nanocomposites, which can have a
variety of functional properties [33, 34, 35]. Additionally, strengthening could, for example, enhance the
utilisation of NFC in many practical applications including gas barriers, and where reduced weight is

important [9, 17, 23]. Furthermore, application of graphene in NFC composite papers has shown to add
hydrophobic performance to an otherwise highly hydrophilic surface [18, 33, 36].
Most of the reported work so far describes the production of graphene oxide (GO) or reduced graphene
oxide (RGO), derivatives of graphene that are prepared by oxidation of graphite rather than by mechanical
delamination [36]. Relative to the oxidised GO, RGO has superior electrical conductivity [21, 25, 30] with
oxygen containing groups on the surface providing pseudo-capacitive storage charge (via Faradaic reaction
mechanism) in addition to its double layer storage. This promising material can be qualified for condensed-
matter and high energy physics applications, as well as in material science, cellular imaging, drug delivery
and biological material applications, together with a wide range of other technological applications such as
bioelectronics and biosensing [26, 32, 35]. However, the strong tendency for RGO to agglomerate and re-
stack can deteriorate overall electrical capacitance [18, 30].
Unfortunately, the fabrication methods for GO or RGO utilise toxic chemicals and processing conditions,
which are not conducive to environmental protection and are challenging to implement in large scale
production of graphene and its derivatives to reach industry’s increasing needs [18, 35, 36, 37, 38].
Consequently, with graphene being potentially in contact with the environment, we should consider the
environmental issue and the safety of these nanomaterials regardless of their benefits. Thus, for the
sustainable implementation of graphene, it is important to develop mechanical routes for its production,
which can deliver pristine graphene, but which can offer utility in different composites [34, 36, 39].
The reported studies of mechanical fabrication of pristine graphene generally involve liquid phase
exfoliation of graphite in synthetic surfactant systems, which, if they remain on the surface of graphene
particles, are not desirable for the onward production of composites, especially for electrical applications
[33, 36]. The amphiphilic nature of cellulose fibres, however, enables them to act as both dispersing and
homogenising agent prior to use in graphene-cellulose composite suspensions [33, 35, 37, 38]. Such a direct
exfoliation process potentially using NFC provides an environmentally friendly and more cost-effective
alternative for the fabrication of graphene, and therefore it is important to understand the mechanism of
hydrodynamic interactions between a graphene particle and thixotropic NFC suspension under transition to
high strain under high speed/high pressure processing conditions [40, 41]. Moreover, since pristine
graphene has a tendency for agglomeration due to its surface chemistry, being partially hydrophobic and
morphologically unique, the suspension exhibits a complex thixotropic nature similar in some respects to
that of MNFC suspensions [35, 36, 39]. Thus, for the successful production of graphene-NFC composites, it
is important to understand and tune the set of mutual rheologically complex flow or strain conditions [37,
39]. Cellulose nanofibrils can act as surface active dispersant for graphite and ultimately graphene particles
during refining, including the potential for entering into the two dimensional layers of stacked graphene

sheets and between individual particles within graphene agglomerates for improved hydrodynamic
coupling to promote delamination during exfoliation [37, 38, 40, 41].
Given the different pre-treatments and production routes exfoliation of graphite it is not surprising that
each combination results in a difference in morphology and relative polarity of the particles, which affect
the level of repulsion or aggregational attraction between the graphene particles in native dispersion and in
the NFC gel-like suspension matrix [38, 40]. Furthermore, NFC aqueous suspension is a non-Newtonian
fluid, highly dependent on flow conditions, and the increase of shear rate enhances both the aggregation
and fragmentation of the fibrillar colloid system, which results in flow behaviour which displays thixotropy
with a final equilibrium state being dependent on parameters characteristic of the particles and their
interactions [42, 43].
The present work applies the study of rheology as a tool to examine the influence of the pre-treatment and
refining time and the behaviour of the graphene particles in the supporting NFC suspension in which the
nanocellulose component is used as constituent dispersion and production aid, as in detail described in the
study conducted by Phiri et al. [44-46]. Graphene-NFC composite suspensions will add rheological
complexity by incorporating both the flexible surface-surface bonds of the graphene aggregates and the
fibrillar gellant-like cage-structure particles interspersed with bound water [47-49]. Probing the composite
suspension under carefully defined shear gradient is designed to reveal the dispersion properties of the
graphene transitioning non-linearly from retained hydrophobic interactions of graphene, through the
sought-after dispersion via NFC-graphene interaction enhancing wettability, followed by physical
entanglement [49, 50]. Coprocessing graphite with NFC suspension is studied also in search of synergy
arising from the material co-exfoliation process, i.e. co-processing [40, 41]. The action of graphene particles
in respect to aiding dewatering of NFC suspensions, that are known to be difficult to dewater, is also
discussed [51-55]. The rheology of complex particulate gel-like suspensions is related initially to the
mechanical aspects under the application of strain from the static state [56-60]. For complex agglomerated
suspensions shear induces extension and breaking of the particle-particle bonds and the distortion and
collapse of the cage structure entrapping water in the particulate matrix [56-63].
2. Materials and Methods
Deionised water was used throughout the experimentation.
2.1 Cellulose fibre raw material and NFC production
Never dried laboratory bleached Kraft birch pulp (kappa number 0.96) was used as the raw material for the
TEMPO-mediated (2,2,6,6-tetramethylpiperidin-1-yl)oxyl, (Sigma Aldrich (St Louis, USA)) oxidation to

produce the NFC. In the reaction 13 % NaOCl solution (Merck (Darmstadt, Germany)) was the primary
oxidant used. NFC-TEMPO suspension was made at 2.3 wt% and diluted to 0.9 wt% by mixing into water.
2.2 Surfactant solution
Sodium cholate (SC) surfactant was purchased from Sigma Aldrich and used as received without any
modification. A surfactant solution was prepared by dissolving SC in water at a concentration of 3 mg cm-3.
2.3 Graphite raw material for graphene production
Expanded graphite (EG) was provided by Asbury Carbon; natural graphite flake (NGF) material was
purchased from Sigma Aldrich.
2.4 Preparation of graphene suspensions
°
Thermal exfoliation was applied as a pre-treatment to EG by heating for ~20 s in a muffle oven at 700 C
before being mixed with the respective chosen media [36]. NGF was used as received without any
modifications.
NGF and EG graphite suspensions were prepared at a concentration of 8 mg cm-3 in three different media:
water (W), surfactant solution (S) and NFC aqueous suspension (NFC). 500 cm3 of each prepared suspension
was subjected to shear exfoliation performed using an IKA Magic Lab (1-l Module MICRO-PLANT equipped
with a single walled open 1 dm3 vessel) high shear mixing equipment with a maximum flow rate of 80 dm3h-
1
at a rotation speed of 20 000 min-1 (rpm) for a predetermined time. 100 cm3 sample aliquots were
collected after 15, 30, 60, 90, 120 min for analysis as shown in Fig. 1.
The exfoliation refining, R, described above, is referred to in terms of the medium used and the material
being exfoliated Rmediummaterial, e.g. NGF exfoliated in the process with water, W, as the medium is given the
label RWNGF etc.
As controls, direct mixes with separately processed NFC were made of each material post processing in
water, i.e. aqueous suspensions of NGF and EG processed alone, namely RW
and RW mixed with
NGF
EG
separately processed NFC suspension. The refining in each case of mixing post processing (RW
RWmaterial 2) is referred to as RWNFC + material 2, i.e. RWNFC + NGF and RWNFC + EG, respectively.
+
material 1

Fig. 1 Schematic representation of graphene exfoliation from either natural graphite flakes (NGF) or
expanded graphite (EG) in water, surfactant and two ways of refining in combination with NFC suspension,
respectively.
Table 1 Suspensions of graphene and graphene/NFC suspensions used in this research, produced as
presented in the Fig. 1.
Exfoliated refined
suspension
Material in aqueous
suspension
Process
suspension
medium
Exfoliation time:
15/30/60/90/120 min
RW
Natural graphite flakes
(NGF)
NGF
Water

RW
Expanded graphite
(EG)
(W)
EG
RW
Nanofibrillated cellulose
(NFC )
NFC
RS
Natural graphite flakes
(NGF )
NGF
SC surfactant
solution
RS
Expanded graphite
(EG)
(S )
EG
RNFC
RNFC
RW
Natural graphite flakes
(NGF )
NGF
NFC suspension
(NFC)
Expanded graphene
(EG)
EG
Nanofibrillated cellulose +
Natural graphite flakes
(NFC + NGF )
NFC + NGF = (RWNFC + RW
NGF)
mixed after separate
processing
Water
(W)
RW
Nanofibrillated cellulose +
Expanded graphite
[NFC + EG]
NFC + EG = [RWNFC + RW
]
EG
mixed after separate
processing
2.5 Gravimetric dewatering/ water retention (ÅA-GWR)
The static gravimetric dewatering of the suspensions was measured using the Åbo Akademi Gravimetric
Water Retention device (ÅA-GWR), as described in detail in our previous publications [8, 11]. A volume of 5
cm3 of suspension was inserted into the cylindrical vessel above a polycarbonate membrane (Nucleopore
Track-Etch membrane 0.5 μm (Whatman) and a stack of absorbent blotter papers. The cylinder was closed,

and the sample held under an overpressure by applying a vacuum beneath the apparatus, ǀ-Δpǀ = 50 kPa for
90 s, during which time dewatering through the membrane occurred. The blotter paper sheet/stack was
weighed before the start of the dewatering measurement and after when the pressure was released. The
weight difference was multiplied by 15 091 m-2, which is the inverse of the cylinder cross-sectional area. An
average of five determinations was computed.
2.6 Rheology
2.6.1 Shear response
Shear flow measurements were used for the determination of dynamic viscosity (η) performed with bob in
cup geometry, where the “bob” was a four-bladed vane spindle with a diameter of 10 mm and a length of
8.8 mm, while the metal cup had a diameter of 17 mm (Anton Paar, Germany) [47]. Flow curves of the NFC
gel-like suspension were measured as a function of decreasing shear rate, spanning a wide shear rate range
( = 1 000 – 0.01 s-1 ) with a logarithmic spread of data point collection duration, ranging from 1 to 100 s [64,
65].
The Ostwald-de Waele expression for the purely viscous shear thinning [8, 11] is given by
n1

  k
(1)
and used to fit the thixotropic shear response behaviour, where k and n in Eq. 1 are the consistency and


power law index, respectively, and
is the shear rate.
2.6.2 Viscoelastic response
For the oscillatory viscoelastic measurements, plate-plate geometry with serrated surface profile on both
upper and lower plates was used to avoid apparent wall-slip due to sample depletion. The upper plate
diameter was 20 mm and gap setting of 2.3 mm. The bottom plate was connected to a Peltier block
temperature control, set to a constant temperature of 23 ˚C. All suspensions were measured five times and
for each measurement a new sample was used [45, 48, 49].
Prior to the oscillatory measurements, the samples were pre-sheared at a shear rate of = 100 s-1 for 60 s
and then left to rest for another 60 s [51-53]. Each oscillatory measurement started with an amplitude
sweep, applying strain (γ) across the range between 0.1 and 500 %, maintaining a constant angular
frequency (ω) of 10 (rad)s-1, to determine the effective linear viscoelastic region (LVE) [55, 64]. For the
following corresponding frequency sweep measurement, a constant strain of γ = 0.1 % in the LVE region
was applied whilst angular frequency (ω) spanned the range of 0.01-100 (rad)s-1. The dynamic moduli,
storage modulus (G´) and loss modulus (G´´), were followed as a function of strain amplitude (γ) and span of

angular frequency (ω). The influence of increase in angular frequency (ω) on the complex viscosity (η*) was
evaluated by measuring its response as a function of increasing angular frequency (ω = 0.1-100 (rad) s-1).
To define differences between the respective suspensions regarding their colloidal interactions, packing
effects and friction between graphene particles and/or nanofibrils during the flow, log-log plot flow curves
of the complex viscosity (η*) were also fitted to a power law according to an equivalent of the Ostwald-de
Waele empirical model, Eq. 3. This method is allowing for the comparison of different water binding
properties within the linear viscoelastic region for an effective complex viscosity, using an effective shear
rate determined as the root mean square value under oscillation, and effective dynamic viscosity beyond
the elastic region (η), respectively [53, 55] .
Fig. 2 Derivation of the rms shear rate from a sinusoidal oscillation.
For a sinusoidal oscillation with angular frequency ω, Fig. 2, the root mean square velocity is given by
2
vrms

A0
(2)
π
where A0 is the amplitude of the oscillation. We can, therefore, derive a root mean square shear rate =
vrms/A0, such that the Ostwald-de Waele shear thinning model can be approximated in the case of complex




n 1
c
2
nc 1




viscosity as*  kc (rms
)
 kc

(3)
π

where kc and nc in Eq. 3 are the flow consistency index and the power-law exponent, respectively, for the
complex viscoelastic case: nc = n = 1 indicates a Newtonian fluid, whereas n < 1 indicates a pseudo-plastic
(shear thinning) behaviour.
2.6.3 Rheological data variation
Data variation in the rheological measurement for the gel-like thixotropic systems was within 10 %. Noise
reduction in the raw data was additionally applied using Tikhonov regularisation [10, 13]. The first step in
the correction process is to remove obvious outliers (primarily negative values and values out of scale). The
actual Tikhonov regularisation step consists of minimising a linear combination of a residual term between
raw and smoothed data and a term representing the amount of remaining noise in the smoothed data.
Adjusting the ratio between the coefficients in the linear combination determines the level of smoothness
of the smoothed data. Fitting low order polynomials to the ends of the data and replacing the y-coordinates
of the ends with values calculated from the fits prevents the appearance of unnecessary winding at the
ends. If necessary, the smoothing procedure could be advanced segmentally.
2.6.4 Consecutive yield stress
Yield stress is perhaps the most important rheological property of complex suspensions, as it needs to be
exceeded in order that they can flow, which is challenging to evaluate for such wall slip/solids depletion
prone materials as NFC-containing suspensions, which display shear banding to minimise stress when
sheared [8, 12, 43]. Therefore, we use both static yield stress (τs0), obtained from the LVE region conducted
0
with plate-plate geometry, and dynamic yield stress (τd ), from flow curves conducted from vane in cup
0
geometry [14, 45]. The dynamic yield stress (τd ) is defined as the minimum stress required for maintaining
the flow, while static yield stress (τs0) is defined as the stress required for initiating flow, and the latter often
has the higher value for NFC suspensions [4, 10, 45].
0
The Herschel-Bulkley equation describes the presence of a dynamic yield stress (τd ) from the plot of the
flow curves as [50-52],
0
n

τd  τd  kγ
(4)
where the terms k and n are the consistency and flow index, respectively, as presented in Eq. 1, with, once
again, the value of n < 1 describing a material exhibiting shear thinning, and n = 1 Newtonian behaviour.
As strain (γ) is constantly increased at constant frequency (ω) from oscillatory amplitude sweep
measurements, the maximum in the elastic stress (τs) that corresponds to the static elastic yield stress (τs0)
is determined as the first point of deviation from linear elastic deformation occurring at a corresponding
critical strain value (γc) [53, 55]. The elastic stress component in oscillatory mode is given by

0

(5)
s  G c
2.6.5 Suspension structure recovery
Structure recovery measurements were performed with the three-interval thixotropy test (3ITT) using bob in cup
geometry [57-59]. In this test, the evolution of transient viscosity η+ after high shear rate is found by applying a
step-wise shear rate with three defined intervals of applied shear, namely a low shear interval/ high shear interval/
low shear interval procedure. During the first and third intervals, the sample is sheared at low shear rate, = 0.1 s-1,
and in the second interval at high shear rate, = 1 000 s-1. Recovery behaviour is defined after the dilatant
+
increase of transient viscosity ηt=3 in third interval has been overcome it its maximum value and recorded
as the reduced values of ηt=3+ /η0, where η0 is the viscosity value at the start-up of the first interval [13, 80].
2.7 Scanning electron microscopy (SEM)
To visualise the structure of the NFC-graphene composites and so provide a correlation with the observed
rheological properties, images were recorded using a Zeiss Sigma VP scanning electron microscope at 1–2
kV acceleration voltage. The samples for SEM imaging were prepared by freeze-drying the suspensions,
followed by sputtering with a ~4 nm platinum film.
2.8 Particle and agglomerate size in suspension
The particle size distributions of the water medium-processed and post-process mixed agglomerate
structures were measured with a Mastersizer (Malvern Instruments, UK) and compared with the surfactant
processed graphene product to illustrate the role of graphene dispersion in the process [9, 14]. Prior to
measuring, the samples were diluted with de-ionised water to a solid content of 0.01 %. The light scattering
volume equivalent median diameter (dsv(50)) is also reported as an average of at least five separate
measurements. In the case where the density of a material is constant throughout its size distribution, the
volume and weight determined size distribution are identical.
3. Results and Discussion
3.1 Water retention
Flocculation and/or agglomeration within the suspension particulate matrix results in particle clusters [81-
83]. The finer the constituent particles, the greater is the water retention tendency for the case of
hydrophilic surfaces [59-61]. Thus, water retention is strongly dependent on process refining time and/or
whether surfactant-acting material is present to ensure the hydrophilicity of the constituent graphene
particle surfaces [36, 37]. Exfoliation of graphite to few-layer graphene in the presence of a dispersing

agent/surfactant or hydrodynamic shear-coupling NFC is, therefore, seen to promote water retention,
suggesting that the material is finer and hydrophilic [36]. In addition, the NFC presence creates a gel-like
structure which itself develops higher water retention [8, 10, 11]. However, if hydrophobic large graphene
agglomerates are present, the water retention is greatly reduced. Thus, by studying the results shown in
Table 2, the lowest quoted dewatering value (highest water retention) corresponds to the combination of
the gel-like action of NFC combined with the highly refined graphene created by processing the graphite in
a medium of NFC suspension RNFC and RNFCEG, and this despite the lower solids consistency of the NFC-
NGF
containing samples.
Table 2 Gravimetric water retention results (ÅA-GWR) for graphene suspensions and graphene-NFC
complex suspensions, compared with processing in the presence of surfactant, as a function of exfoliation
time.
Gravimetric dewatering
Refining
time
/gm-2
Graphene suspensions
Graphene-NFC composite suspensions
/min
-
consistency 4 w/w%
RS RS
-
consistency 2 w/w%
RW
RW
RW
RW
RNFC
RNFC
NGF
EG
NGF
EG
NFC + NGF
NFC + EG
NGF
EG
15
30
60
90
6 829
6 714
6 423
5 823
5 324
6 697
6 234
5 897
5 421
5 198
6 478
6 143
5 876
5 473
5 137
6 256
5 887
5 623
5 276
4 987
5 889
5 344
4 982
4 503
4 098
3 687
5 203
4 848
4 524
4 145
3 556
3 249
5 387
4 824
4 577
4 125
4 653
4 218
3 819
3 524
120
We can, therefore, conclude from the presented data that products originating from expanded graphite
(EG) comprise a finer structure than those originating from natural graphite flakes (NGF), and that the
water retention of the graphene/NFC suspension materials formed via simultaneous co-exfoliation (co-
processing) of graphite with NFC is greater than when graphite is mixed afterwards with NFC. Even at this
early stage of the analysis, there is a clear observable synergy derived in co-exfoliation by co-processing of
graphite with NFC versus addition of NFC suspension after processing [40, 41]. Furthermore, given the

greater water retentivity achieved by co-exfoliation with NFC, it is likely that there is a greater
hydrodynamic coupling between the liquid phase and the particles during the high shear exfoliation
process itself. This, then, is the main hypothesis to be tested rheologically.
3.2 Rheological experiments
3.2.1 Graphene dispersions in water – action of surfactant
3.2.1.1 Dynamic shear
The steady state shear curves of graphene suspensions, Fig. 3, produced from graphite in water, and
graphite in surfactant solution, reveal that the suspensions have shear thinning behaviour at low shear rate,
and a slight tendency to transient dilatancy at higher shear rate [43, 45, 47, 50]. The Ostwald-de Waele
expression (Eq. 1) can be fitted to the shear thinning low and medium shear region, thus providing
characteristic fitting values for consistency coefficient k and power index n. Smaller n indicates greater
shear thinning and larger k indicates higher static state viscosity, related to flocculation/agglomeration, and
driven mainly by the hydrophobic nature of the graphene particles in water. Hence, when processing
together with surfactant the consistency value k drops as the particles become at least partially stabilised.
Dilatant behaviour at the high shear rate may be considered related to shear-induced aggregation as the
weak stabilising force is overcome during forced close contact of the particles, leading to coalescence due

to
the
attractive
hydrophobic
force
[49,
50].
Fig. 3 Flow curves of graphene in water and surfactant suspension in response to refining time: (a) NGF
processed in water, (b) EG processed in water, (c) NGF processed in surfactant solution, and (d) EG
processed in surfactant solution. The effect of surfactant in reducing viscosity is clearly seen, though the
tendency to shear induced aggregation for the surfactant containing system is relatively marked due to the
mechanism of close encounter overcoming the weak stabilisation.
The values of consistency coefficient k and power index n are shown in Table 3. Both k and n decrease as a
function of refining time and further on using surfactant in the refining fluid medium.
Table 3 Dynamic viscosity (η) rheological parameters according to the Ostwald-de Waele shear thinning
power law.
Refining time
/min
k consistency coefficient
n < 1 shear thinning
/Pa.sn
RW
RW
RS
RS
RW
RW
RS
RS
NGF
EG
NGF
EG
NGF
EG
NGF
EG
15
2.63
2.24
1.94
1.61
0.21
0.16
0.12
0.09

30
60
2.17
1.85
1.43
1.28
1.94
1.71
1.33
1.18
1.64
1.31
1.22
1.14
1.44
1.16
1.11
1.09
0.16
0.14
0.11
0.09
0.14
0.12
0.10
0.08
0.09
0.08
0.07
0.06
0.08
0.06
0.06
0.03
90
120
3.2.1.2 Dynamic yield stress
Extrapolating the sheared material stress curve back to the limit of zero shear,
0

, provides the
limd 

 0
0
value of stress required just to keep the suspension flowing, termed the dynamic yield stress (τd ) [60, 61].
This is a measure of the flocculation tendency in the particulate dispersion. It can be seen clearly, as
expected, that the use of surfactant increases the flowability of the dispersion by acting against
flocculation, and the shear energy input (time) also breaks down the graphite into graphene platelets which
tend to form looser structures in the static state.
0
Fig. 4 Comparing the dynamic yield stress (τd ) for graphene produced in water and graphene produced in
surfactant solution as a function of refining time.
3.2.2 Graphene-NFC composite suspensions
Since the dynamic shear viscosity properties of graphene-NFC composite suspensions are dominated by the
water-holding gel-like and the strongly shear thinning behaviour of NFC, and so the graphene-graphene

interactions tend to be hidden, it is decided here to present and compare the viscoelastic, and yield stress,
response to strain amplitude and frequency, and shear, applied respectively, to the product composites [4,
10, 11].
3.2.2.1 Viscoelastic analysis
The strain applied to the suspensions ranges γ = 0.01–500 % using parallel plate geometry. As strain
increases we see that the elastic response follows two structural modes, as shown by the two arrows in Fig.
5. Firstly, a weaker static loose floc/agglomeration structure, included in the NFC gel-like matrix, yields and
then the system retains a stronger particle-particle aggregation interaction until the strain is sufficient,
secondly, to break this down into a more purely viscous state [8, 44]. The decoupling of the weaker first
stage floc behaviour followed by the particle-particle aggregation structure reveals an interesting trend in
respect to refining time, in that the refining leads to big differences in respect to the weaker floc structure,
but less so in respect to the particle-particle aggregation [32, 43, 45]. Notable is the fact that the behaviour
of the viscoelastic response is similar whether the graphite has been co-exfoliated by co-processing with
NFC (RNFC and RNFCEG) or the separately processed NFC has simply been added to the graphene obtained
NGF
by exfoliation in water (RWNFC + NGF and RWNFC + EG). This tells us that the weakening of the floc structure, as we
saw earlier linked with increasing water retention, is dependent on the particle number and hydrophilicity.
The longer the refining time, the greater the number of material particles that are generated, and the
particle size is reduced [38]. Therefore, the larger scale loose floc structure associated with coarse particles
decreases but the stronger particle-particle aggregation is retained.
To aid comparison of the magnitude of the viscoelastic response, moduli values at two strain levels are
shown in Table 4. The strain levels are chosen to reflect the two-stage structure regime, i.e. at γ = 0.01 %
and γ = 1.12 %, Fig. 5, to indicate the separation of agglomerated graphene particles from the gel-like loose
floc within the NFC matrix at very low strain, and their subsequent mixing together with the NFC fibrils at
higher strain within the second stage linear viscoelastic region LVE, before reaching critical strain γc, at
which the elastic moduli drastically decrease.

Fig. 5 Viscoelastic moduli G´ (solid symbols ) and G´´ (open symbols ) obtained from amplitude sweep tests
where the strain span is from 0.01 to 500 % revealing the combined floc and gel-like graphene-NFC
suspension structure: post processing mixed samples (a) RWNFC + NGF, (b) RWNFC + EG, and coprocessed samples
(c) RNFCNGF, and (d) RNFC
.
EG
Table 4 Values of elastic (G´) and loss moduli (G´´) at two different values of strain (γ = 0.01 % and 1.12 %),
capturing the two-stage structure breakdown of loose flocs followed by stronger aggregates.
Refining
time
G´γ = 0.01 %
G´γ = 1.12 %
/Pa
/Pa
/min
RW
RW
RNFC
RNFC
RW
RW
RNFC
RNFC
NFC + NGF
NFC + EG
NGF
EG
NFC + NGF
NFC + EG
NGF
EG
15
30
60
98.23
84.78
69.56
92.12
71.34
59.45
132.23 103.23
32.94
24.91
17.25
28.69
21.44
14.57
36.44
29.41
23.57
34.32
28.52
22.13
102.53
79.12
99.12
75.23

90
58.09
37.76
51.34
32.07
66.23
42.23
62.12
39.34
13.13
10.52
11.37
9.23
19.37
14.84
18.52
12.75
120
G´´γ = 0.01 %
/Pa
G´´γ = 1.12 %
/Pa
RW
RW
RNFC
RNFC
RW
RW
RNFC
RNFC
NFC + NGF
NFC + EG
NGF
EG
NFC + NGF
NFC + EG
NGF
EG
15
30
31.96
29.77
23.36
16.56
14.72
11.62
46.23
34.36
27.56
21.73
17.73
43.12
21.77
19.57
14.55
11.00
7.42
28.36
21.18
17.95
14.00
11.17
22.67
25.43
18.28
16.76
12.96
31.41
24.28
19.56
14.96
15.88
12.74
9.86
17.52
14.49
11.41
8.57
60
90
120
7.14
5.09
Interestingly, also in Table 4, the suspected increased coupling between NFC and graphene particles when
the two are co-exfoliated, via co-processing, is supported by comparatively higher structural strength and
viscosity values, i.e. {G´, G´´ (RNFC )} > {G´, G´´ (RWNFC + NGF )} and {G´, G´´ (RNFCEG ) > {G´, G´´ (RWNFC + EG)},
NGF
respectively. The somewhat greater ease of refining EG compared with NGF is also reflected in the lower
moduli values, as was seen in the increased water retention behaviour for the co-exfoliated NFC-EG
composite arising from co-processing.
The elastic moduli (G´ and G´´) of all the NFC-graphene composite suspensions show similar response to
increasing angular frequency (ω) under oscillation, Fig. 6, displaying initially weak structure breakdown, as
was seen under low strain, Fig. 5, followed by high frequency vibration-induced elastic behaviour [40]. This
observation clearly indicates that the stronger aggregation structure in the NFC medium becomes ever
more rigid under vibration, following a hardening response [43, 51, 53]. Fascinatingly, the greater the
refining time, the weaker the initial static structure, as seen before in Fig. 5, but the greater the relative
induced elastic hardening, seen as a cross-over from the lowest modulus value at low frequency to the
highest or close to highest at high frequency [54, 61, 62]. This reveals that the effect is one of particle
number; the more particles present, the greater is the elastic hardening. Such materials make useful
vibration resistant materials, for example in buildings under earthquake conditions.

Fig. 6 Elastic moduli (G´ and G´´) dependence on angular frequency (ω) reveals the changing influence
from weak floc interactions, which become broken down, to that of strong particle-particle interactions
delivering a vibration-induced hardening (ringed regions): post processing mixed samples (a) RWNFC + EG, (b)
RWNFC + NGF, and co-exfoliated (co-processed) samples (c) RNFCNGF, (d) RNFC
.
EG
The shear thinning behaviour of graphene-NFC suspensions within the viscoelastic region under oscillation
was traced using the complex viscosity (η*) response to increased angular frequency (ω) under constant
strain, as presented in Fig. 7. Shear thinning behaviour at low to intermediate angular frequency can be
seen to change to dilatant behaviour at higher frequencies due to the presence of the strongly interacting
aggregated graphene particles [3, 36, 43].

Fig. 7 Complex viscosity (η*) response to increase of angular frequency ω = 0.01-100 (rad)s-1, showing the
Ostwald-de Waele complex viscosity thinning analogue parameters kc and nc, and the region of oscillation
dilatancy: post processing mixed samples (a) RW
, (b) RWNFC + NGF, and co-exfoliated (co-processed)
NFC + EG
samples (c) RNFCNGF, (d) RNFC
.
EG
The thixotropic response is parameterised using the Ostwald-de Waele analogue expression under
oscillation, Eq. 3, applied of the region ω = 0.01-10 (rad)s-1, and the equivalent values of kc and nc are shown
in Table 5 [10, 40, 43, 45]. It can be concluded that the smaller graphene particles produced from the EG
result in lower viscosity in NFC suspension, which is even more pronounced when the graphite is refined
together with the NFC due to the dispersive effect of the NFC during co-exfoliation in the co-processing,
inducing also improved coupling with the applied shear.
Table 5 Complex viscosity (η*) thixotropic response to increasing angular frequency fitted using an Ostwald-
de Waele analogue power law at low-medium frequency providing a flow kc and shear thinning coefficient
nc.
-1
-1
Refining
time
kc@ω = 0.01-10 (rad) s
nc@ω = 0.01-10 (rad)s

RW
RW
RNFC
RNFC
RW
RW
RNFC
RNFC
EG
/min
NFC + NGF
NFC + EG
NGF
EG
NFC + NGF
NFC + EG
NGF
15
30
98
85
112
109
84
106
105
74
0.21
0.18
0.22
0.24
0.19
0.15
0.11
0.22
0.21
0.18
0.14
0.11
88
65
44
37
67
56
38
23
0.17
0.15
0.11
0.09
0.16
0.11
0.08
0.07
60
90
62
56
120
49
43
3.2.2.2 Static versus dynamic stress analysis
Static yield stress (τ0 ) was evaluated from the parallel plate strain amplitude response in the oscillatory
s
measurements and defined as the maximum in static stress (τs) in response to increasing strain (γ), Fig. 8.
This corresponds to the value of elastic modulus (G´), at critical strain, multiplied by the critical strain (γc),
Eq. 5. The static yield stress (τ0 ) is expected to be larger than the dynamic yield stress (τ0 ), obtained from
s
d
shearing conditions measured with vane geometry, and presented for the cases of graphene in water and
graphene in surfactant solution in Fig. 4, as the structure is not broken prior to the application of strain [45,
52, 59].
Graphene-NFC suspensions in which graphite has been exfoliated with NFC have lower yield point (both
static and dynamic), due to the less flocculation and more disperse state of the particles [45, 61]. Longer
refining time leads to a decrease in particle size, and correspondingly results in smaller clusters of graphene
aggregates for all graphene-NFC suspensions regardless of the processing route. This reduces the energy
needing to bring the system into flow [56, 57].

Fig. 8 Static yield stress under amplitude sweep test obtained as the maximum of stress dependence on
strain, revealing the floc strength of the graphene-NFC suspensions.
The dynamic stress as a function of shear rate is complex, and perhaps displays the most promising
information in respect to describing the material content in the graphene-NFC composite dispersions, Fig.
9, which can be drawn out only in the light of the rheological analyses described above [4, 8]. Dynamic
stress is plotted for the shear rate range = 0.01 - 1 000 s-1, and illustrates shear thinning followed by a
minimum rising again rapidly to either a plateau of close to Newtonian behaviour or a slower shear
thickening, and finally a strongly rheopectic/dilatant behaviour, reminiscent of filler in polymer dispersion
[59, 60, 63]. These four concatenated domains of stress behaviour were fitted using the Herschel-Bulkley
0
model, Eq. 1, delivering a dynamic yield point (τd i) at each domain concatenation point, together with a
respective consistency index (ki) and shear thinning coefficient (ni) for each domain, providing insight into
the effect of shear rate and refining time on the agglomeration of graphene in the NFC matrix.

Fig. 9 Dynamic stress of graphene-NFC obtained using vane geometry, with four different intervals used for
fitting curves following the Herschel-Bulkley power law expression, Eq. 1: (a) RWNFC + EG, (b) RWNFC + NGF, and
co-exfoliated (co-processed) samples (c) RNFCNGF, and (d) RNFC
.
EG

The four shear rate domains were divided logarithmically into an ultralow shear region = 0.01 - 0.1 s-1, a
low shear region = 0.1 - 1 s-1, an intermediate region = 1 - 10 s-1 and a high shear region = 10 - 1 000 s-1.
As we can see in Fig. 9, the benefit of studying the dynamic stress is that the method provides a large
amount of data for curve fitting using the concatenated Herschel-Bulkley models. The ultralow shear region
provides confirmation of the initially sheared floc and retained gel-like structure, the low shear and
medium shear indicate the breakdown and transition minimum of the structured state, and the short
region of quasi steady state, respectively [54, 61]. Use of the stress data provides the real benefit of
studying the high shear region in detail [44, 51]. Analysing this region will now allow us to interpret the
primary aggregate structure of the various systems.
Fig. 10 The four regions of exponential change, showing the fourth characteristic behaviour of stress
growth fitted using the Herschel-Bulkley expression.
0
By observing the Herschel-Bulkley concatenate dynamic yield stress τd i for each domain together with the
respective ki and ni (i = 1 to 4) values with respect to the processing conditions for the two graphene
sources it is possible to correlate the nanofibrillar cellulose effect of dispersing seen as a decrease in
0
flocculation and agglomeration (indices ki) and increase in yield points (τd i) of co-exfoliated/co-processed
graphene-NFC suspensions when smaller particles are present in the matrix throughout the shear rate
range, as shown in Fig. 9 and summarised in Fig. 11. Refining time has a high impact on the agglomeration
of graphene particles and creation of clusters of graphene within the NFC matrix, which affects the energy
that needs to be introduced to bring the system into flow.

0
Fig. 11 Rheological parameters of (a) dynamic stress τd i = 4, and (b) ki = 4 and ni = 4, showing the effect of
increased dispersing when coprocessing with NFC and the clear trend of shear thinning followed by shear
thickening as a function of shear rate.

3.2.2.2 Structure recovery
Refining time decreases structure recovery time, which was measured as the maximum reached in
+
transient viscosity (η max), after which a drastic decrease in transient viscosity is observed for all
suspensions, Fig. 12. This combination of rheopectic behaviour, seen as transient viscosity increase at
constant shear rate and results in an overshoot peak upon eventual increase of shear rate, and then relates
to the break-up of graphene clusters at high shear rates, as can be seen with the accompanying steady
state curves, Fig. 12. Thus, we can expect a breakage of graphene aggregates within the NFC matrix after
prolonged time, and this is revealed by the dramatic transient viscosity decrease and subsequent fast
structure recovery [10, 13, 58].
Transient viscosity recovery in the third interval, is longest for the graphene derived from the NGF graphite
flakes post-process mixed with separately processed NFC and displays the slowest rheopectic behaviour,
which in turn decreases with increasing refining time, i.e. resulting from smaller agglomerates size. To trace
structure recovery in the third interval, we reduce the values after it reaches the maximum, which occurs at
different time intervals for each suspension, using the static starting state viscosity (η+max/η0). For graphene-
NFC suspensions derived from expanded graphite (EG), recovery after high shear is faster than for NGF-
derived graphene, and correspondingly for those where graphene was ground with NFC, regardless of
graphene origin.

Fig. 12 Normalised transient viscosity (η+/η0) in third interval (low shear of 0.1 s-1), measured after second
interval of high shear ( = 1 000 s-1), to trace structure recovery in 3ITT recovery tests using vane geometry
showing apparent dynamic viscosity up to the maximum overshoot values in the second, high shear, region:
(a) RW
displaying the greatest shear thinning, the most rapid rheopectic maximum and the fastest
NFC + NGF
recovery.

3.3 Structure as viewed in the SEM
SEM images obtained from graphene in water suspensions upon exfoliation at different refining times (for
duration of refining of 30, 90 and 120 min) with water as medium reveal that few-layer graphene produced
from EG suspensions have different morphological characteristics and reduced tendency to agglomerate
than graphene formed from NGF suspensions, as presented in Fig.13, and that refining time leads to a
decrease in agglomerate size [36].
Fig. 13 SEM images revealing the influence of refining time, namely 30, 90 and 120 min, on agglomerate
size and shape in two different sets of refined few layer graphene suspensions: (a), (b), (c) from NGF; (d),
(e), (f) from EG.
When graphene is produced via co-exfoliation/co-processing of graphite with NFC suspension, its particles
are much more uniformly distributed throughout the NFC suspension, which leads to more compact and
less porous graphene platelets, Fig. 14, for both EG and NGF ground together with NFC [36, 58]. Refining
time has an influence on the morphology of the platelets, as in the previous Fig. 13 with water only, but the
result of more uniform and less aggregated agglomeration of the graphene particles demonstrates the
universal dispersing power and increased hydrodynamic coupling with the fluid phase when including NFC
[36, 40, 41].

Fig. 14 SEM images showing the influence of co-exfoliation with NFC suspension by co-processing on few
layer graphene platelet morphology and size for the two graphite sources: (a), (b) NGF, and (c), (d), (e) EG,
respectively, for the different refining times, 30, 90 and 120 min.
Contrasting Fig. 15 with Fig. 14 shows clearly the advantage of co-exfoliation with NFC by co-processing
versus simple mixing after processing [40, 41]. When graphite is processed separately from the NFC
suspension, and only with the poorly coupled hydrodynamic effect of water, it retains much bigger
agglomerates than when co-exfoliated simultaneously in the co-process, and with further flocs amongst the
post-added NFC fibrils creating the appearance of a more broken platelet structure [38]. Once again, as In
Fig. 3, we see the greater uniformity of the product when using EG versus NGF, Fig. 15.