Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation

Superior dielectric breakdown strength of graphene and

carbon nanotube infused nano–oils
Purbarun Dhar, Ajay Katiyar*, Lakshmi Sirisha Maganti,
Arvind Pattamatta and Sarit K Das#
Department of Mechanical Engineering, Indian Institute of Technology Madras,
Chennai – 600036, India
*Research and Innovation Centre (DRDO), IIT Madras Research Park,
Chennai – 600113, India (2nd affiliation of author)
#School of Mechanical, Materials and Energy Engineering (SMMEE),
Indian Institute of Technology Ropar, Rupnagar–140001, India (Present address of author)

ABSTRACT
Nano–oils comprising stable and dilute dispersions of synthesized Graphene (Gr)
nanoflakes and carbon nanotubes (CNT) have been experimentally observed for the
first time to exhibit augmented dielectric breakdown strengths compared to the base
transformer oils. Variant nano–oils comprising different Gr and CNT samples
suspended in two different grades of transformer oils have yielded consistent and high
degrees of enhancement in the breakdown strength. The apparent counter–intuitive
phenomenon of enhancing insulating caliber of fluids utilizing nanostructures of high
electronic conductance has been shown to be physically consistent thorough theoretical
analysis. The crux mechanism has been pin pointed as efficient charge scavenging
leading to hampered streamer growth and development, thereby delaying probability of
complete ionization. The mathematical analysis presented provides a comprehensive
picture of the mechanisms and physics of the electrohydrodynamics involved in the
phenomena of enhanced breakdown strengths. Furthermore, the analysis is able to
physically explain the various breakdown characteristics observed as functions of
system parameters, viz. nanostructure type, size distribution, relative permittivity, base
fluid dielectric properties, nanomaterial concentration and nano–oil temperature. The
mathematical analyses have been extended to propose a physically and dimensionally
consistent analytical model to predict the enhanced breakdown strengths of such nano–
oils from involved constituent material properties and characteristics. The model has
been observed to accurately predict the augmented insulating property, thereby
rendering it as an extremely useful tool for efficient design and prediction of
breakdown characteristics of nanostructure infused insulating fluids. The present
study, involving experimental investigations backed by theoretical analyses and models
for an important dielectric phenomenon such as electrical breakdown can find utility in
design of safer and more efficient high operating voltage electrical drives, transformers
and machines.
Index Terms — Dielectric breakdown, dielectric liquids, nanofluid, graphene, carbon
nanotubes (CNT), transformer oil, analytical formulation

1. INTRODUCTION utilized to develop advanced systems along avenues of smart

materials [1-3], sensors [4-6], MEMS/NEMS [7, 8], modern
nanoelectronics [9-11], power systems [12, 13] and many
Carbon based meso–nanomaterials; viz. graphene (Gr), more. In recent times, variant phenomena, smart applications
carbon nanotubes (CNT), fullerenes, nano–diamonds etc. have and utilities of such carbon nanostructures in stable, colloidal
shown tremendous potential to revolutionize the modern phases have also been probed [14-17]. However, a majority of
technological prowess of mankind. Gr and CNTs have been
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
such endeavors have been limited to understanding the nanomaterials as potentials for enhancing breakdown strengths
chemistry of such nanostructures in the colloidal state or their of insulating fluids. Based on experimental observations and
ability to tune some special properties of the fluid phase, mathematical analysis of the electro–hydrodynamics involved,
which, majorly has been concentrated on the thermophysical a comprehensive picture of the underlying physics and
properties of the fluid [18-20]. The fact that nanomaterials mechanisms has been put forward. Moreover, given the high
such as Gr and CNT possess exceedingly high electron probability of direct applications of such nano–oils in high
transport characteristics in a manner points to the fact that their charge density/ high voltage electrical machines and
presence in a colloidal system could be engineered to tune or components, an analytical model has been proposed to predict
modify the material dielectric properties of the fluid phase. the augmentations possible for a particular nano–oil or
efficient design of such nano–oils as per necessity. The
Incorporation of inorganic/ organic additives to fluids in physically and dimensionally consistent model derived from
order to modify their dielectric characteristics has been first principles accurately predicts the enhanced breakdown
practiced for a reasonably long time. Among the dielectric voltages, as validated against experimental data.
characteristics, researchers have often focused their attention
towards modifying the dielectric breakdown strength of
2. MATERIALS AND METHODOLOGIES
insulating fluids, since it is one of the most important aspects
that govern safe operation of modern power systems which are
cooled by insulating oils, ranging from transformers, large 2.1. NANOMATERIALS: SYNTHESIS AND
scale super capacitors, resonator coils, high charge density CHARACTERIZATION
components etc. However, chemical additives often suffer
from inherent disadvantages, such as corrosion of components,
The nanomaterials utilized in the present study comprises
possible reaction at elevated electric field intensities and
two samples of nano–graphene synthesized in–house and two
temperatures, possibility of flammability in case of corona
samples of multi–walled carbon nanotubes (MWCNT)
discharge at high voltages etc. Consequently, in recent times,
procured from firms manufacturing and supplying
researchers have started looking for alternatives to such
nanomaterials. All involved in–house synthesis has been
additives in the form of nanomaterials [21, 22]. Nanomaterials
carried out with commercially available reactants and reagents
offer certain advantages over the conventional additives, viz.
which have been used as obtained and without further
high stability as dispersions, negligible corrosive action,
purification. The graphenes (Gr) utilized have been
expanded options over material selection and high dielectric
synthesized chemically in the form of reduced graphene oxide
and electronic properties which are of utmost importance for
(rGO), which has been obtained from graphene oxide (GO)
modifying the dielectric characteristics of the base fluid.
synthesized via the modified Hummer‘s method [19]. In a
Among the few reports available in the domain of enhancing typical synthesis procedure, 1 gm of graphite (G) powder
the dielectric breakdown (DB) strength employing stable (average flake size ~ 20–50 μm) is initially added to ~ 25 mL
dispersions of nanostructures in insulating fluids, the nature of
conc. sulphuric acid (98%) and allowed to stand at 90 oC for 2
nanomaterial utilized can be divided in to two broad
hours. Next, 2 gms each of anhydrous potassium
categories, semiconducting (titanium dioxide [23-26]) and
peroxosulphate and anhydrous phosphorus pentoxide is added
super paramagnetic (iron (II,III) oxide [27-29]). Both the
categories of particles have been resorted to since such to the mixture and the same stirred vigorously for 3 hours
materials possess high values of relative permittivity, which in while maintaining the initial temperature for pre–oxidation of
turn leads to rearrangement of the electric field within the fluid the graphite flakes. The resulting mixture is carefully
and electron scavenging in a manner so as to reduce the centrifuged to obtain the preoxidized graphite (pG) as
possibility of DB at the expected field intensity. Furthermore, sediment. The same is washed thoroughly with deionized (DI)
all such reports are completely experimental in nature and very water and further centrifuged to collect the pG, which is then
limited insight into the actual physics involved in the dried overnight in a hot air oven. The next step, comprising
augmented transport property can be extracted. Purely oxidation of the pG to GO involves dissolution of the pG in ~
theoretical analysis [30] reveals that augmentation of 50 mL conc. sulphuric acid (98%) and the mixture stirred
breakdown voltage by scavenging of free electrons is also vigorously while maintaining ice cold conditions by employing
possible with nanostructures of exceedingly high electrical an ice bath. The ensuing process is highly exothermic and an
conductivities. Hence, intuitively, Gr and CNT based ice cold atmosphere is compulsory in order to avoid explosive
insulating fluid dispersions are expected to exhibit high reaction. To the mixture, 3 gm of potassium permanganate is
enhancements in the DB voltages. The present paper added in small batches over a period of time. The reaction
investigates experimentally the augmented DB strength of Gr yields an algae–green mixture and the vigorous stirring under
and CNT based transformer nano–oils, as a function of ice cold conditions is continued for 6–7 hrs. Further, the
nanostructure size, concentration, electronic and dielectric
mixture is diluted by adding ~ 500 mL DI water slowly and
properties and nano–oil temperature. Thorough survey of
stirred till the mixture attains room temperature. The oxidation
existing literature on the subject matter reveals that the present
process is finalized by adding 10 mL hydrogen peroxide (5%)
article is the first of its kind to report carbon based
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
to the solution. The formation of a bright yellow solution samples, wherein, the micron scale G flakes has been
confirms the formation of GO. The solution is allowed to stand completely converted to nanoscale Gr, with the exception of a
overnight for the GO to sediment. The precipitate is collected small population among Gr2 in the 2–4 micron range. Gr1 and
via decantation and washed thoroughly with DI water. The Gr2 exhibit a maximum flake population at ~ 100 nm and 300
resulting mixture is centrifuged at ~ 12000 rpm to collect the nm respectively. (d) Visual appearance of the synthesized Gr
GO, which is then added to pure acetone (concentration ~ 0.1 (right, black) from the natural G (left, grey).
wt. %) and ultrasonicated to form stable suspensions. The
suspensions are then spread uniformly as thin films onto petri
The synthesized Gr has been characterized qualitatively
dishes and oven–dried overnight. The dried films are scraped
from its visual appearance and SEM image and quantitatively
off with scalpels to obtain pure, non–aggregated GO.
from analysis of the Raman spectra and DLS data. The SEM
The dried GO is dispersed in DI water (0.1 wt. %) and image reveals presence of nanoscale thin flakes, highly
ultrasonicated to form stable suspensions, which are then convoluted and mimicking folds on sheets of fabric, which
heated to 60–70 oC. To 250 mL of the solution, either 250 mg indicates the presence of graphenic structures. G being
of finely powdered sodium borohydride or 25 mL of hydrazine multilayered, do not exhibit such pristine convolutions and
hydrate and 10 mL ammonia solution is added, while stirring folds. The oxidation followed by reduction exfoliates the
for 1 hr. at the aforementioned temperature. Either of the layers, simultaneously creating folds, creases and wrinkles on
mentioned processes can be utilized in order to reduce the the individual flakes. Hence, presence of such nanoscale folds
functional groups on the GO to obtain rGO/Gr suspension. The and wrinkles (represented in Fig. 1(a) by arrows) qualitatively
suspensions are centrifuged at ~ 12000 rpm to obtain the Gr, confirms transformation of graphitic structures to Gr.
which is further washed, centrifuged and suspended in pure Quantitatively; the Raman spectrum is an effective tool to
acetone and dried to obtain non–aggregated Gr nano–flakes confirm the presence of Gr. As illustrated in Fig. 1(b), the
similar to that mentioned in the preceding paragraph. The spectrum exhibits the characteristic signature peaks of
synthesized Gr has been subjected to in–depth characterization graphene systems. The D (~1350 cm-1) and G (~1600 cm-1)
so as to comprehend the physical parameters. The detailed bands represent planar defects in the sheets and the in–plane
characterizations for the Gr samples have been illustrated in stretching of the sp2 hybridized carbon atoms in Gr
Fig. 1. respectively. Further, the existence of the 2D band (~2800 cm-
1) within the spectrum confirms the presence of Gr. The ratio
of intensity of the 2D band to that of the G band for the present
samples indicates Gr systems of less than equal to five layers
[31], i.e. the dielectric and electrical transport phenomena
unique to Gr are expected to manifest themselves during the
present experimentations. Two distinct Gr samples, viz. Gr1
and Gr2 have been synthesized. The samples are distinguished
based on their polydispersity, which has been characterized
utilizing DLS (as illustrated in Fig. 1(c)). Analysis reveals that
Gr2 has a much larger average flake size than that of Gr1.

The MWCNT systems utilized in the present study have

been procured from Nanoshel Inc. (USA) (manufactured via
Chemical Vapor Deposition (99.5 % purity), illustrated in Fig.
1 (b)) and the characteristic dimensions of the nanostructures
have been verified utilizing Transmission Electron Microscopy
(TEM). The TEM images of the CNT samples have been
Figure 1: Characterization of the graphene samples, namely illustrated in Fig. 2. The nanotube outer diameters for the two
Gr1 and Gr2. (a) Illustrates the High Resolution Scanning samples utilized, CNT1 and CNT2, have been verified against
Electron Microscope (HRSEM) image of the synthesized Gr. the manufacturer‘s claims and illustrated in Fig. 2(c) and (d).
The fabric mimicking texture and the nanoscale folding and The average diameter for CNT1 and CNT2 are 20 nm and 10
wrinkles on the sheets, as indicated by arrows, qualitatively nm respectively whereas that specified by the manufacturer
indicates exfoliation of G to form Gr. (b) Raman spectroscopy was 20–40 nm and 5–20 nm respectively, thereby confirming
intensity vs. shift plot for Gr. The presence of the unique D, G the manufacturer‘s claims. The aspect ratio lies within the
and 2D bands at ~1350 cm-1, 1600 cm-1 and 2800 cm-1 range 100–400 and 200–500 for CNT1 and CNT2 respectively
respectively within the spectrum establishes the presence of as per the characterization report provided by the
Gr. (c) Dynamic Light Scattering (DLS) analysis for the manufacturer. The nanomaterials synthesized and/or procured
samples reveals the polydisperse size distribution for the have been utilized to prepare nano–oils utilizing two types of
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
transformer oils procured from certified dealers. Two grades compartment with HV transformer and terminals and the
of transformer oils have been utilized and have been denoted controller circuit and actuator compartment. The equipment
as Oil1 and Oil2 hereafter. The nano–oils have been utilizes single phase 220 V AC as the input and a HV
synthesized by adding the required amount of nanomaterial to transformer steps up the voltage as governed by the automated
the oil (in wt. %) followed by ultra-sonication for 1 hour. Oleic controller. The HV is fed across the electrode terminals with
acid (OA) has been utilized as the capping agent/ surfactant for the test fluid in between and a precision micro–ammeter reads
samples with nanomaterial concentrations of and above 0.05 the current passing across the terminals. At the moment of DB,
wt. % so as to induce more stability to the suspensions. The an arc of high current is discharged across the electrodes and
amount of OA utilized in each case varies from 0.5-2 mL per the controller circuit trips the HV circuit. The voltmeter
500 mL of oil depending upon the concentration of reading at that tripping point is read as the breakdown strength
nanomaterials. It is noteworthy that the presence of OA in the of the dielectric fluid. The tests have been performed in
oil in such dilute proportions has negligible effect on the accordance to the ASTM D–877 utilizing two hemispherical
breakdown strength of the base oil; as established from electrodes of 5 mm diameter (the apex points of the
experimental observations. Also, the average shelf life of the hemispheres face each other) with electrode spacing of 5 mm
dilute samples ranges above 1 month when kept undisturbed. (the spacing is fixed utilizing specific gauges). The HV across
Samples have been denoted hereafter in accordance to its the terminals is raised by an automated circuitry at an approx.
constituents, e.g. Gr2 and Oil1 based nano–oil has been named rate of 2 kVs-1. The DB strength of each set of oil is
Gr2 Oil1. Illustrations of the base transformer oil, CNT and determined 10 times and the arithmetic mean is taken as the
Gr based nano–oils have been provided in Fig. 3 (c), (d) and breakdown strength of the same. Experiments have also been
(e) respectively. performed to deduce the effect of temperature on the DB
strength of nano–oils, wherein, the oil is first heated to
elevated temperatures and the experiments performed
thereafter. The temperature of the oil in between the gaps of
the electrodes has been monitored utilizing a precision
thermometer and the temperature difference from the required
value has not been allowed to traverse beyond ± 2 oC. The
mean of 10 readings at and around (within the allowed
tolerance) a particular temperature has been considered as the
DB voltage at that temperature.

Figure 2: Characterization of the MWCNT samples,

namely CNT1 and CNT2. (a) TEM image of the MWCNT
samples confirming tube structure (b) Image of the procured
sample (c) TEM image of CNT1 (20 nm average outer
diameter) (d) TEM image of CNT2 (10 nm average outer
Figure 3: Components of the dielectric breakdown setup (a)
diameter).employment history and special fields of interest.
Front view of the test setup. The indicators exhibit the status of
the high voltage (HV) terminals across the sample (red lamp
2.2. INSTRUMENTATION indicated active HV circuit whereas green indicates
dormancy), which can be viewed through the safety window.
Determining the DB strength of the transformer oils and the The system is automated to increase or decrease the voltage
nano–oils requires a voltage measurement instrument operated across the terminals (displayed by the HV voltmeter; rated 100
in accordance to specified ASTM/IEC standards and codes. kV with least count of 200 V) at the desired rate. (b) The
For the present study, a custom manufactured automated high sample holding basin with the nano–oil and the electrodes
voltage (HV) test compartment has been utilized and assembled. The protruding ends of the electrodes are rested on
components of the same have been illustrated in Fig. 3 (a). The the HV terminals. (c) Transformer oil (T.O.) (d) CNT based
setup consists of two compartments, viz. sample basin holding nano–oil (e) Graphene based nano–oil.
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation

3. RESULTS AND DISCUSSIONS Figure 4: Dielectric breakdown characteristics of graphene

based nano–oils (a) Gr1 based nano–oils (b) Gr2 based nano–
3.1. Augmented dielectric breakdown strength oils.

Experimental observations and analysis reveals that the

presence of Gr and CNT in transformer oils, when in the form
of a stable suspension with appreciably long shelf life, leads to
augmented DB strength of the oil when compared to its virgin
form. Oil1 and Oil2 have been observed to possess DB
strength of ~ 30 kV and 40 kV respectively when tested with
the 5 mm electrode gap configuration (i.e. 6 kV/mm and 8
kV/mm strength respectively). However, since the electrode
gap has been maintained constant for all the present
experiments, the BD strengths have been expressed in their
potential form rather than the field intensity form. The BD
characteristics of Gr1 and Gr2 based nano–oils has been
illustrated in Fig. 4(a) and (b) respectively and the respective
enhancements and critical concentration at which maximum Figure 5: Magnitudes of augmentation of the DB strength of
enhancement is obtained has been illustrated in Fig. 5. As graphene based nano–oils, comparison between various
evident, the presence of Gr nanoflakes in the oil leads to samples and critical concentrations for maximum obtainable
enhanced insulting properties, with a maximum augmentation enhancements.
of ~ 45% observed for Gr2 Oil2 nano–oil at very dilute
dispersion concentration of 0.075 wt. %. Likewise, Fig 6(a)
and (b) illustrate the breakdown performance of CNT based The phenomenon of enhanced DB strength of dielectric oils
nano–oils with the respective enhancements and concentration upon addition of nanostructures possessing very high electrical
corresponding to peak position illustrated in Fig. 7. In case of conductivities such as Gr and MWCNT, though
CNTs, the observed maxima in enhancement of breakdown counterintuitive at first glance, is physically consistent and can
voltage exceeds the capability of Gr, with ~ 50% enhancement be explained qualitatively based on the electrodynamic
observed for CNT2 Oil1 nano–oil at a more dilute processes within the nano–oil when exposed to HV stresses.
concentration of 0.025 wt. %. The superior DB voltage exhibit by the nano–oils is a direct
consequence of the morphed electrodynamics within the oil
caused by the presence of the nanostructures, thereby making
the electrical response of the nano–oil grossly different from
that of the base oil. Among variant parameters, the extent of
morphing of the dielectric response of the system can be
deduced qualitatively from the charge relaxation parameter for
the system. In brief, DB within a liquid occurs due to the
establishment of a narrow, highly conductive ‗tunnel/ channel‘
which propagates outwards from the electrodes, the merger of
which leads to the formation of a highly ionized pathway for
the charge to propagate across the dielectric medium [32, 33].
The transient development of the zones of ionized molecules
that propagate towards each other (known as streamers) is
governed by the DB strength of the liquid and the growth
period is directly proportional to the DB strength and the
applied electric field intensity. Thereby, if the growth and/or
response period of the ionic constituents responsible for
development of the streamers can be reduced or delayed, the
probability of DB for the liquid at a given electric stress is
reduced accordingly. This is the possible mechanism by which
the presence of nanostructures enhances the DB strength of the
oil. In the event that the charge relaxation time constant of the
nanostructures is markedly lesser the characteristic time of
streamer development and propagation, the free electrons
released due to near–field ionization at the electrodes is
effectively captured or scavenged by the nanostructures,
thereby appreciably preventing the growth of streamers to the
requisite strength. A qualitative illustration of the mechanism
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
has been provided in Fig. 8 (a). Au contraire, the converse absolute permittivity (ε) and electrical conductivity (σ) of the
leads to negligible or even adverse effects on the DB strength nanostructure (subscripted np) in conjunction with the
of the liquid and the possible mechanism has been discussed in surrounding fluid (subscripted f) [30]. Initially, a nanostructure
subsequent portions. (of characteristic length ‗dnp‘) situated at a spatial location
describable by (r,θ,z) coordinates at a frozen instant of time (t)
is considered, such that the origin of the describing polar
coordinate system coincides with the geometrical center of the
nanostructure. An electric field (E0), such that the field lines
are directed collinear to the z–coordinate axis, with the nano–
oil as the participating medium, is applied between the
electrodes. In absence of nanostructures, the field is expected
to distribute uniformly within the oil and ionize the oil in the
region between the electrodes homogeneously. However, the
presence of stably suspended nanostructures leads to
disruption of the uniformity of the field in the vicinity of the
nanostructure–oil interface, leading to marked different
electrical repose of the oil in between the electrodes. The
electric field, when solved in accordance to the Laplace
equation establishing null divergence of the field potential,
yields the radial and polar components of the spatio–temporal
distribution of the electric field within the nanostructure as
[30]

   d np    np   f
3
 t     d np    np   f
3
 t 

Er (r , )  E0 cos  1   3    e     3    (1  e  ) 
     2
   4 r   np f  
   4 r    np  2 f  

(1)

Figure 6: Dielectric breakdown characteristics of graphene  

 d np    np   f
3
 t     d np    np   f
3
 t 

E (r , )  E0 sin   1   3    e     3    (1  e  ) 
based nano–oils (a) CNT1 based nano–oils (b) CNT2 based      2
 
 8r   np f  
   8r    np  2 f  

nano–oils. (2)

In the above equations, the charge relaxation time of the

nanostructure with respect to the fluidic phase is expressible as
2  
 1 2
2 1   2 (3)
Considering Gr based nano–oils, an estimate of the relaxation
time can be performed. The relative permittivities of the oils
have been determined utilizing a custom built parallel plate
capacitor system and a multi–frequency LCR meter (HP,
USA). The capacitor is first calibrated against air and a graded
mineral oil with standard set of data available. Experiments
reveal that at frequency ranges of 10–100 Hz (prioritized in
accordance to the generation and distribution frequencies
utilized worldwide) Oil1 and Oil2 possess relative
permittivities of ~2.25 and ~2.48 respectively. The electrical
Figure 7: Magnitudes of augmentation of the DB strength of conductivity of the oils has been considered to be of the order
graphene based nano–oils, comparison between various of 10-12 S/m, as obtained from published data [30]. In
samples and critical concentrations for maximum obtainable similitude, the relative permittivity of the Gr samples have
enhancements. been considered equivalent to that of graphitic systems
(ranging from 2–2.5) and the electrical conductivity has been
determined based on the room temperature electronic mobility
in impure (to keep the predictions inclined towards the worst
A qualitative assessment of the efficacy of a given case scenario side) graphene sheets, which is of the order of
nanostructure in augmenting the breakdown strength of a 108 S/m. Based on the above discussions, the magnitude of ‗τ‘
liquid can be made based on the charge relaxation time (τ), for the Gr based nano–oils has been predicted to be ~ 10-19 s.
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
However, depending on composition and dielectric properties, Based on the fact that Gr/CNT has tens of orders of
the characteristic time scale for streamer growth within magnitudes larger electrical conductivity compared to
electrically stressed oils ranges from the order of 10 -10 s to the insulating oils, Eqn. 1 can be restructured based on an infinite
order of 10-6 s [30]. Evidently, the charge relaxation time for conductivity particle assumption. Since the relative
Gr in oil is at least 10 orders of magnitude smaller than the permittivity of graphitic systems is very similar in magnitude
streamer development time scale for the oil. Consequently, Gr to that of insulating oils and the electrical conductivity largely
flakes tend to capture or scavenge the ionic entities released as overshadows that of insulating oil, the restructured Eqn. 1 for
a consequence of ionization of the oil much earlier than the the electric field at the nanostructure–oil interface at
charged entities can propagate and lead to the growth of the timescales much larger than charge relaxation time is
streamer (Fig. 8 (a)). As a result, the probability of DB at the expressible as
expected value is drastically reduced, providing the oil with
Er (r  dnp 2, )  3E0 cos 
increased electrical stress bearing capacity. (4)
As evident, the electric field intensity at the interface is thrice
as strong as the incident field intensity within the system,
leading to high concentration of converging field lines onto the
nanostructures. As far as particles of lesser electrical
conductivities are concerned, the interfacial field intensity is
much weaker (compared to Eqn. 3) and therefore, presence of
Gr/CNT leads to gross changes in the local electric field,
leading to highly effective particle scavenging since the
liberated electrons traverse along field lines, which in turn
strongly converge onto such nanostructures. This explains the
reason as to why even very minute concentrations of such
nanostructures can lead to highly augmented insulation
properties. For nanostructures with high dielectric constants,
such as titanium dioxide and iron (II, III) oxide, the
enhancement in breakdown voltage obtained is large, however,
similar magnitudes as Gr/CNT systems is only realized at high
concentrations (> 0.5 wt. %) [24, 27]. Since the electrical
conductivities of such systems are magnitudes lower than
GR/CNT, a much larger population of particles is required to
achieve similar enhancements. However, these materials also
possess high dielectric constants which lead to modification of
the local permittivity of the nano–oil to aid the scavenging
process, and thereby similar enhancements can be obtained at
moderately high concentrations.

As the nanostructure population scavenges liberated electrons

from the streamers, the average charge density per
nanostructure approaches the saturation density. When the
average charge density per nanostructure reaches the saturation
Figure 8: Qualitative and scale exaggerated illustration of the density, the population is unable to scavenge further, and
mechanism of scavenging charge released due to ionization of breakdown occurs. Consequently, with increasing population
the oil molecules in the vicinity of the electrodes (red arrows of nanostructures, i.e. with increased concentration, the
indicate direction of traverse) (a) Dilute suspensions amount of charge scavenged by the population before the
effectively capture the charges (indicated by green arrows) and average charge per nanostructure exceeds the saturation
delay the merging of the two streamers, thereby enhancing magnitude is higher than a low concentration case. As a result,
breakdown strength. (b) Above a critical concentration, the increased concentration leads to enhanced magnitude of
inter–flake distance between Gr is reduced to the point augmentation in breakdown voltage. However, although the
wherein the liberated charges can traverse through the flakes enhancement is intuitively expected to be a linear function of
(indicated by blue arrows), ‗hop‘ on the next flake (yellow growing concentration, in reality the enhancement follows a
arrows) due to the proximity and/or direct contact and reach decaying growth rate, as evident from Figs. 5 and 7. Moreover,
the opposite streamer zone. The percolation chains thereby act unlike titanium dioxide and iron (II, III) oxide based nano–oils
as agents that induce ‗shorting‘ of the steamers, often leading [24, 27, 28] (although the exhibit enhancement is higher for
to numerous intermittent arc discharge and eventually leading Gr/CNT) the growth of enhancement saturates within dilute
to reduced breakdown strength. regimes. The charging dynamics of the nanostructure
population can be considered analogous to the charging of a
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
capacitive system, i.e. follows an exponentially decaying the fluidic domain, leading to reduced interparticle collision
growth pattern. Ionization of fluid molecules leads to mean free path and increased particle interactions. The
formation of both mobile free electrons as well as positively increased interaction and randomness increases the degree of
charged molecules. However, since the massive nature of the chaotic motion within the fluid, which supplemented with the
molecules lead to very low mobility compared to the electrons, increased instability of the charged entities, leads to further
the nanostructures come in contact with the electrons more deterioration of the streamer stability. Consequently, the stable
frequently than positive ions, leading to predominant growth of the streamers is hampered and their merger delayed,
scavenging of electrons. Consequently, the growths of negative leading to further augmented DB strength.
charge on the nanostructures grow with time. Gr and CNT,
both nanomaterials with high free electron densities, thereby
are soon overcrowded with electrons, resulting in a high
negative charge density on the nanostructures. Consequently,
they repel the incoming electrons which results in decaying
charge scavenging efficacy in addition to the reduced potential
of the nanostructure approaching its saturation point. With
increasing population, the effective repulsion increases and
thus the growth of enhancement decays out like a charging
capacitor to a concentration where the charge acceptance
capability of the system reaches saturation. Since the electron
density in titanium oxide or iron oxide is much low compared
to that of Gr/CNT, the resultant repulsion of inbound electrons
are much smaller in magnitude at the same level of charging.
As such, these systems can acquire charges up to greater
concentrations and the growth of enhancement in such systems
is often linear or bell shaped. However, if Gr/CNT systems
cannot lead to enhanced breakdown beyond a certain threshold
concentration, one would expect the DB voltage to attain a
constant value for all concentrations beyond the threshold.
However, sharp decrease in enhancement is observed and the
same has been discussed in the next section.

Studies have also been performed to understand the nature in

which the presence of nanostructures affects the insulating
performance of the oil at elevated temperatures. Understanding
such a phenomenon is of utmost importance technologically
since most HV electrical devices generate large quantities of Figure 9: Dielectric breakdown characteristics of nano–oils
heat, thereby causing the cooling oil to perpetually operate at with respect to temperature exhibits interesting phenomena
elevated temperatures compared to the ambient. Experiments governed by electro–thermal diffusion of nanostructures within
reveal increases insulating performance of the base oil with the oil (a) Response of DB strength of Gr and CNT based
increasing temperature, as illustrated in Fig. 9 and a thermal nano–oils (0.01 wt. %) compared to that of the base oil as a
fluctuation based explanation can be conceived for the same. function of temperature (b) Enhancement of DB strength of
With increasing temperature, the viscosity of the oil reduces various nano–oils as a function of temperature, with respect to
exponentially whereas the thermal energy of the liberated the base oil at a constant temperature (referred as ‗with temp‘)
electrons and charged positive ions increases linearly. The as well as the base oil at ambient temperature (referred as
enhanced thermal energy leads to enhanced amplitude of ‗w.r.t base‘).
thermal fluctuations and thermal velocity experienced by the
charged entities, resulting in disruption of the formative zones
of the streamers. Owing to increased thermal randomness of 3.2. Analytical Formulation
the charges, the stability of the growth and development of the
streamers is delayed, leading to delayed merger and Although the mathematical analysis and associated
augmented insulating capabilities. However, it has been discussions presented in the preceding section provides a
observed (Fig. 9(a) and (b)) that addition of nanostructures comprehensive physics of the mechanisms by which the
leads to additional enhancement in breakdown voltage with presence of highly conductive nanostructures such as Gr and
increasing temperature, a phenomenon which can also be CNT can lead to efficient augmentation of the DB voltage of
explained based on increased thermal energies of the liquids even at low nanostructure concentrations, when viewed
nanostructures. As the temperature rises, the nanostructures at from an application or engineering point of view, an
experience enhanced degrees of Brownian randomness within simplistic analytical formulation to derive the magnitude of
augmentation is of utmost importance, since such a
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
methodology would enable efficient and economic design of condition yields an expression for the polar window available
nano–oils for application fronts just based on theoretical (θ) for electron scavenging at a given field intensity and
formulation. The present section describes a dimensionally nanostructure size as
consistent and physically coherent analytical model which can
Q(t )
be used at ease to evaluate the enhanced value of DB voltage
 cos 
for a given mineral oil or similar fluid in the event that the DB 3 f d np2 E0
voltage of the base oil, size distribution and concentration of (7)
Gr and CNT structures, dielectric and electrical properties of The delayed breakdown is promoted by the capture of the
the base oil and values of similar material properties are streamer forming charges by the nanostructures, thereby
known. The model provides further insight into the mechanics delaying the merger of the streamers beyond the usual
of the phenomena from a mathematical point of view and has breakdown voltage. The instant at which the nanostructure
been derived based on simplifications of the field distribution population in the vicinity of the streamer development regions
equation pertaining to perfectly conducting nanostructures. is saturated and unable to scavenge further, the streamers
The same is justified since the electrical conductivities of Gr develop au natural and leads to DB. Determining the
and CNT at room temperature are approximately 20 orders of enhancement requires an estimate of the scavenging capacity
magnitude higher than that of the transformer oils and hence of the nanostructures utilized. At the point of saturation for a
they behave similar to nanostructures with infinite conductivity nanostructure, the interface is unable to capture any more
when dispersed in such insulating media. external charge, i.e. mathematically, the polar window
The anchoring philosophy behind determining the available for charge scavenging decays down to zero (θ=0). It
augmentation lies in comprehending the magnitude of is at this instant that the charge gathered by a single
streaming charged entities scavenged by the nanostructure nanostructure can be termed as the saturation charge (QS,np)
population in the suspension. As the nano–oil is subjected to and can be accordingly deduced from Eqn. (6) as
increasing electrical stresses, the oil molecules in the vicinity
of the electrodes are ionized and the stripped off electrons lead
QS ,np  3 f dnp
2
E0
(8)
to formation of streamers that try to propagate towards the For mono–disperse systems involving spherical
opposite electrode. However, the Gr/CNT nanostructures in nanostructures with the diameters confined to a tight Gaussian
the vicinity of the electrodes begin scavenging the free distribution, the usage of a simple mean diameter suffices for
electrons, thereby delaying the merging of the streamers and the present endeavor. However, for highly poly–disperse and
eventually breakdown. As the charge scavenging by the high aspect ratio such as Gr and CNT respectively, the
nanostructure initiates, the net charge on it grows with time, effective size characterizing the system requires being
which in turn reduces its electron scavenging efficacy due to determined. For a low or near unity aspect ratio but high poly–
repulsive forces exerted by the scavenged electrons. The dispersity index system such as Gr (assuming the flakes to be
transient growth of charge (Q(t)) on the infinitely conducting near squares), the characteristic size is obtained from the DLS
nanostructure consequentially affects the localized electric data (Fig. 1(c)). Based on the approximate intensity of a
field at the oil–nanomaterial interface and its radial and polar particular peak (I), the flake size corresponding to the peak
components can be mathematically expressed as [30] position (d) and the number of peaks (m), a weighted mean
representative flake size (deq) can be deduced as
 d np
3
 Q(t ) 
m

Er  E0 1  3  cos  
  
 f d np2 
 (d I j j )
 4r  (5) d eq ,Gr 
j 1
m

 d 3
 I j
E  E0  1   sin 
np j 1
(9)
 8r 3
  (6) Since the Gr based system is a flake based population, the
area exposed to electron scavenging is taken care of once the
Scavenging of electrons by a given Gr/ CNT is possible
representative flake size has been determined. However, for a
only under the circumstance that there exists regions on the
low poly–dispersity index and high aspect ratio (AR) system
nanostructure surface wherein the net charge is predominantly
such as CNT (low poly–dispersity with reference to the outer
positive in comparison to the incoming ‗volley‘ of external
diameter), the representative diameter can be effectively
electrons, such that the region of surface can absorb the
modeled based on equivalence of the surface area of a
inbound electrons without repulsion hampering the scavenging
nanotube to that of a hypothetical spherical particle. The
process. With increasing surface charge accumulation due to
surface area based equivalence is of prime importance since it
scavenging, the polar window available for absorbance of
is the only method to ensure uniform scavenging by the
charges gradually diminishes. Mathematically, charge capture
nanotubes for the analysis, which otherwise scavenge in
is possible at those regions of the nanostructure–oil interface
accordance to its orientation with respect to the streamers. The
(r=dnp/2) where the net electric field is finitely positive (Er≥0).
equivalent diameter for a CNT population can be thus
Evaluating Eqn. (4) in accordance to the described Dirichlet
determined as
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation

deq ,CNT  2dCNT AR For quantification of the mobility, an estimate of the charge
(10) scavenged by each particle is necessary. Since the present
formulation is analytical in nature, the charging dynamics is
Furthermore, the expression in Eqn. (7) is devoid of the irrelevant as the final outcome desired is to determine the
dielectric property of the nanostructure and unless modified to degree of enhancement, a time by which the nanostructure
cater to the same, provides accurate predictions only for cases population has acquired saturation charge. Thereby, the
wherein the relative permittivities of the fluid and the magnitude of saturation charge can be utilized in the present
nanomaterial are of similar magnitudes. However, for formulation. Furthermore, since the characteristic sizes are in
situations not pertaining to the above condition (which is the micro–nanoscales, the nanostructures can be assumed to
expected to be a majority since solids often possess higher behave similar to spherical particles whose hydrodynamics in
dielectric constants than oils), a simplistic formulation known the fluid media can be represented accurately by Stokesian
as Matijevic equation [34] expressed in Eqn. (10), can be dynamics. Thereby, the Stokes–Einstein formulation for
utilized to obtain the effective relative permittivity of the Brownian diffusivity [18] can be utilized to obtain the
nano–oil from the relative permittivities of its constituents and expression for the electro–thermal mobility of a representative
nanomaterial concentration (φ, wt./wt.), and the same maybe nanostructure as
utilized in lieu of the dielectric constant of the fluid phase in
Qs ,np
Eqn. (7) to obtain a modified expression for saturation charge. e t 
3 f ( np   f ) 3 d np
(14)
 eff   f  
2 f   np In the above equation, the diameter of nanoparticle should
(11) be replaced by the equivalent diameter for Gr and CNT, as
obtained from earlier discussions. As evident, the viscosity of
In order to estimate the number density of the net charge the fluid decreases with increasing temperature, thereby
scavenged the spatial distribution of Gr/ CNT in the oil should leading to augmented mobility of the nanostructures, which in
be determined. Assuming a uniform distribution of turn leads to enhanced probability of scavenging, leading to
nanostructures in the oil phase, the number density of enhanced DB strength at elevated temperatures (Fig. 8).
nanostructures (nnp) can be evaluated from the nanomaterial However, the efficiency of the diffusive form of hampering
concentration, densities (ρ) of the fluid and nanomaterial positive streamers can be as high as the magnitude by which
phases and approximate volume of a single representative the nanoparticle mobility is greater than the positive charge
nanostructure (vsnp) as mobility (μC) (in general of the order of 1x10-9 m2V-1s-1).
Thereby, the ratio of the two diffusivities governs the
f mechanism. Finally, the maximum possible number density of
nnp 
np vsnp charge which can be scavenged (χnp) from the streamers by the
(12) involved nanostructure population before attaining the state of
Moreover, scavenging of free electrons is not the sole saturation is expressible as
mechanism by which the DB strength is enhanced in such
 e t 
systems. As the electrons are stripped off the oil molecules, the  np  nnp QS ,np  
molecules attain positive charges, which lead to formation of
independent positive streamers and can also lead to DB [33].
 c  (15)
As the nanostructures scavenge electrons, the net negative In insulating fluids such as oils, the mechanism of
charge on them grows and their propensity of attraction breakdown is governed by the ionization and subsequent
towards the positive streamer grows, leading to disruption of release of free electrons which leads to the formation of a
formation of the positive streamers, which further reduces the conducting ambience beyond a certain intensity of electrical
possibility of DB. Also, random thermal fluctuations of the stressing. Thus, the DB strength of any oil can be expected to
nanostructures within the fluid medium enhances the window hold an inverse relationship with its maximum free charge
of scavenging since a larger traversal region signifies exposure density, i.e. the magnitude of charge per unit volume which
to a greater number of free electrons. Consequently, the ability can be stripped off from its bound form when subjected to
of the nanostructure to diffuse across the fluid space due to the electrical stresses and it can be intuitively argued that
acquired electro–thermal fluctuations need to be incorporated breakdown of the matter occurs when the free charge density
within the formulation. Considering a single Gr flake or CNT, attains a fraction of its maximum limit; the fraction being a
the effective electro–thermal mobility (μe-t) is expressible as a constant for a particular material. The nanomaterials present in
function of the acquired charge (q), intrinsic thermal energy (in the nano–oils effectively scavenge and absorbs a portion of the
terms of the Boltzmann constant kB and absolute temperature charge population required to bring about DB, thereby
T) and Brownian diffusivity (DB) as [34] delaying the process. The enhancement achieved thereby
qDB depends on the magnitude of charge scavenged per unit
e t  volume by the nanomaterials up to the point of electrical
k BT (13) stressing where the base oil is expected to yield and undergo
breakdown. Substituting the electric field intensity (the ratio of
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
the breakdown voltage to the electrode spacing) at which the
base oil undergoes DB (Vf) in Eqn. (14), the saturation charge
density of the nanostructure population at that level of
electrical stress is determinable. Since a portion of the required
charge density is scavenged, breakdown does not occur at the
DB voltage of the base oil. Therefore, enhanced electrical
stress is required to ionize the oil further so as to cause DB.
Based on the maximum free charge density of the fluid (χf) (~
550 Cm-3 and 450 Cm-3 for Oil1 and Oil2 respectively, as
determined from linear extrapolation from known data [30]),
the probability (p(b)) by which breakdown at the expected
point has been delayed can be expressed as
 np
p(b) 
f
(16)
Thereby, an estimate of the probability by which the oil
does not undergo DB at the expected voltage due to
scavenging by the dispersed Gr/CNT can be made from Eqn.
(15). Hence, it may be argued that the probability by which the
breakdown is averted at the expected field intensity is the
probability by which the magnitude of DB voltage is enhanced
over and above the same. Although DB is a complex non–
linear dynamical phenomenon; the linearized argument put
forward can be utilized for ease of mathematical formulation
without suffering appreciable deviation, as observed later from
the validations. In accordance to the mechanism proposed, the
DB voltage of the nano–oil (Vn) is expressible as
Vn  V f (1  p(b)) Figure 10: Validation of the predictions obtained from the
(17)
theoretical model against experimental data for (a) Gr based
Based on the proposed mathematical formulation, the nano–oils (b) CNT based nano–oils. Owing to stability issues
predicted DB voltages for the various samples of nano–oils of the colloidal phase beyond certain critical concentrations as
have been validated against experimental findings and accurate well as additional breakdown favoring phenomena such as
predictability has been observed. The validation plots for Gr percolation chains creeping in, the predictability of the model
and CNT based nano–oils have been illustrated in Fig. 10(a) is restricted to the ascending regions of the augmented
and (b) respectively. It is observed that the prediction curves voltages. The dotted vertical lines provide an estimate of the
are nearly linear functions of nanostructure concentration; critical concentrations, beyond which the model fails to
however, experimental data holds the quasi–linear growth up capture the phenomena, neither qualitatively nor
to certain concentration and the enhancement decays beyond quantitatively.
that. In reality, the decreasing stability of the suspension in
conjunction with the propensity of hastened corona discharge
between the electrodes caused by percolation chain structures Since the electronic mobility within Gr/CNT is very high,
at higher nanostructure concentrations leads to decrement of the small separation helps in formation of a conducting chain
the magnitude of enhancement of the DB voltage. With within the streamer development zones. These chains are
increasing concentration, the problems associated with the theorized to be formed by the closely arranged nanostructures
stability of the suspensions creep in and minimal and linked by the free electrons within the streamer zones. At
electrodynamic stressing can often lead to hastened low concentrations, the inter–nanostructure spacing is too
flocculation, thereby leading to inefficient enhancement of DB large for the free electron population in the vicinity to link
voltage. Furthermore, another mechanism initiates hampering them electrically. Qualitative evidence of such a phenomenon
the augmentation. At higher concentrations, the flake/tube has been observed, such that, the number of intermittent
density is high and thereby the separation between two discharge between the two electrodes, even at low electric
nanostructures decreases drastically. field intensities, is high for higher concentrations, leading to
hastened breakdown. The reason behind such large number of
weak, intermittent discharges before final breakdown can be
attributed to the percolation chain theory. Thereby; due to the
dual phenomena of decreased stability and ‗short–circuit‘
initiating chained structures, only a finite degree of
enhancement of the dielectric strength is possible. Further, the
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
probability of percolating networks formed by Gr flakes in References
direct contact increases the probability of the ‗shorting‘
behavior. An increased contact based percolation network is
[1] [1] T. Mueller, F. Xia, and P. Avouris, ―Graphene photodetectors for
possible at higher concentrations of flakes/tubes exactly above high-speed optical communications,‖ Nature Photonics, vol. 4, pp. 297
a certain population density. As evident from Figs. 5 and 7 the - 301, 2010.
enhancement in breakdown strength suffers a massive drop [2] [2] Y. G. Semenov, J. M. Zavada, and K. W. Kim, ―Electrically
immediately beyond the critical concentrations, thereby adding controlled magnetic switching based on graphene-magnet composite
structures ‖ Journal of Applied Physics, vol. 107, pp. 064507, 2010.
credibility to the proposed hypothesis. As the analytical model [3] [3] H. Pu, and F. Jiang, ―Towards high sedimentation stability:
does not encompass all such mechanisms, it continually magnetorheological fluids based on CNT/Fe3O4 nanocomposites ‖
predicts increasing DB voltage as a function of concentration. Nanotechnology, vol. 16 pp. 1486, 2005.
However, it accurately predicts the voltage up to the critical [4] [4] G. Lu, L. E. Ocola, and J. Chen, ―Reduced graphene oxide for room-
temperature gas sensors ‖ Nanotechnology, vol. 20 no. 445502, pp.
concentration (indicated by vertical dotted lines in Fig. 9), i.e. 445502, 2009
the point beyond which the enhancement decays, and can [5] [5] C. C. Su, T. Liu, N. K. Chang, B. R. Wang, and S. H. Chang, ―Two
therefore be used as an efficient tool to design Gr/CNT based dimensional carbon nanotube based strain sensor,‖ Sensors and
nano–oils for high charge density electrical machines and Actuators A: Physical, vol. 176, pp. 124-129, 2012.
[6] [6] M. Zhou, Y.-H. Lu, Y.-Q. Cai, C. Zhang, and Y.-P. Feng,
drives. ―Adsorption of gas molecules on transition metal embedded graphene: a
search for high-performance graphene-based catalysts and gas sensors ‖
Nanotechnology, vol. 22, no. 385502, pp. 385502, 2011
4. Conclusions [7] [7] S. M. Kim, E. B. Song, S. Lee, S. Seo, D. H. Seo, Y. Hwang, R.
Candler, and K. L. Wang, ―Suspended few-layer graphene beam
electromechanical switch with abrupt on-off characteristics and minimal
To infer, the present article reports for the very first time, the leakage current ‖ Applied Physics Letters, vol. 99, pp. 023103, 2011.
utility of Gr nanoflakes and CNT as agents that render highly [8] [8] S.-E. Zhu, M. K. Ghatkesar, C. Zhang, and G. C. A. M. Janssen,
augmented dielectric breakdown strengths to insulating fluids, ―Graphene based piezoresistive pressure sensor ‖ Applied Physics
such as transformer oils. Gr and CNT based dilute nano–oils, Letters, vol. 102, pp. 161904, 2013.
[9] [9] F. Schwierz, ―Electronics: Industry-compatible graphene
synthesized by dispersing minimal quantities of the transistors,‖ Nature, vol. 472, pp. 41-42, 2011.
nanostructures in graded transformer oils have been [10] [10] S. Rumyantsev, G. Liu, W. Stillman, M. Shur, and A. A.
synthesized and their breakdown characteristics experimentally Balandin, ―Electrical and noise characteristics of graphene field-effect
studied as per standard codes and protocols, utilizing a high transistors: ambient effects, noise sources and physical mechanisms,‖
Journal of Physics: Condensed Matter, vol. 22 pp. 395302, 2010.
voltage fluid testing device. High degrees of enhancement in [11] [11] J. Y.Q., Z. Q., and L. L., ―Planar MEMS Supercapacitor using
the DB voltage have been obtained for all the tested categories Carbon Nanotube Forests,‖ IEEE 22nd International Conference on
of nano–oils, thereby confirming the claims put forward. A Micro Electro Mechanical Systems, 2009. MEMS 2009. , pp. 587-590
mathematical analysis based on the electro–hydrodynamics of 2009.
[12] [12] S.-W. Kim, D.-H. Seo, H. Gwon, J. Kim, and K. Kang,
nanostructures in fluidic phases has been utilized to explain the ―Fabrication of FeF3 Nanoflowers on CNT Branches and Their
mechanisms and physics behind the process. A detailed and Application to High Power Lithium Rechargeable Batteries,‖ Advanced
comprehensive explanation for the various phenomena Materials, vol. 22(46), pp. 5260–5264 2010.
observed as functions of system and material properties has [13] [13] E. Antolini, ―Graphene as a new carbon support for low-
temperature fuel cell catalysts,‖ Applied Catalysis B: Environmental,
been presented. The mathematical analysis has also been vol. 123–124, pp. 52–68, 2012.
evolved into an analytical formulation which can accurately [14] [14] Z. Ming, A. Jagota, E. D. Semke, B. A. Diner, R. S. McLean, S.
predict the breakdown strengths of such nano–oils when the R. Lustig, R. E. Richardson, and N. G. Tassi, ―DNA-assisted dispersion
various system properties are known inputs. The present work and separation of carbon nanotubes,‖ Nature Materials, vol. 2(5), pp.
338-342, 2003.
thus concretely establishes the potential of Gr and CNTs in
[15] [15] A. Schierz, and H. Zänker, ―Aqueous suspensions of carbon
elevating the operating capabilities and mitigating damage and nanotubes: Surface oxidation, colloidal stability and uranium sorption,‖
safety concerns of high voltage electrical systems cooled by Environmental Pollution, vol. 157, no. 4, pp. 1088-1094, 4//, 2009.
insulating fluids. [16] [16] Y. Chen, L. Chen, H. Bai, and L. Li, ―Graphene oxide-chitosan
composite hydrogels as broad-spectrum adsorbents for water
purification,‖ Journal of Materials Chemistry A, vol. 1, no. 6, pp. 1992-
Acknowledgments 2001, 2013.
[17] [17] P. Dhar, S. Bhattacharya, S. Nayar, and S. K. Das, ―Anomalously
augmented charge transport capabilities of biomimetically transformed
The authors thank the Defense Research and Development collagen intercalated nanographene-based biocolloids,‖ Langmuir, vol.
Organization (DRDO) for funding the present work. They also 31, no. 12, pp. 3696-706, Mar 31, 2015.
thank the Sophisticated Analytical Instruments Facility (SAIF), [18] [18] P. Dhar, S. S. Gupta, S. Chakraborty, A. Pattamatta, and S. K.
Das, ―The role of percolation and sheet dynamics during heat
IIT Madras and the Central Electronics Centre (CEC), IIT conduction in poly-dispersed graphene nanofluids,‖ Applied Physics
Madras for material characterizations. PD also thanks the Letters, vol. 102, no. 16, pp. 163114, 2013.
Ministry of Human Resource and Development (MHRD), [19] [19] P. Dhar, M. H. D. Ansari, S. S. Gupta, V. M. Siva, T. Pradeep,
Govt. of India, for the doctoral research scholarship. A. Pattamatta, and S. K. Das, ―Percolation network dynamicity and
sheet dynamics governed viscous behavior of polydispersed graphene
nanosheet suspensions,‖ Journal of nanoparticle research, vol. 15, no.
12, pp. 1-12, 2013.
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
[20] [20] N. V. Sastry, A. Bhunia, T. Sundararajan, and S. K. Das, Purbarun Dhar was born in Calcutta, India in
―Predicting the effective thermal conductivity of carbon nanotube based May, 1990. He received a B. Tech in Mechanical
nanofluids,‖ Nanotechnology, vol. 19, no. 5, pp. 055704, 2008. Engineering in 2012 from the National Institute
[21] [21] M. Chiesa, and S. K. Das, ―Experimental investigation of the of Technology at Durgapur, India. He submitted
dielectric and cooling performance of colloidal suspensions in insulating his thesis for M.S.–Ph.D. dual degree to the
media,‖ Colloids and Surfaces A: Physicochem. Eng. Aspects vol. 335 Department of Mechanical Engineering of the
pp. 88–97, 2009. Indian Institute of Technology at Madras, India,
[22] [22] J. G. Hwang, M. Zahn, Francis M. O‘Sullivan, L. A. A. in September 2015. Since then, he is associated
Pettersson, O. Hjortstam, and R. Liu, ―Electron Scavenging by with the institute as an Institute Pre-doctoral
Conductive Nanoparticles in Oil Insulated Power Transformers,‖ 2009 Fellow. His doctoral work is focused upon
Electrostatics Joint Conference, vol. 1.1, pp. 1-12, 2009. developing novel colloidal systems involving
[23] [23] D. Yue-fan, L. Yu-zhen, W. Fo-chi, L. Xiao-xin, and L. Cheng- nanomaterials, mostly graphene and CNT, and
rong, ―Effect of TiO2 nanoparticles on the breakdown strength of probing the various augmented thermophysical and electromagnetic transport
transformer oil,‖ Conference Record of the 2010 IEEE International properties of such colloidal systems. His other research interest lies in
Symposium on Electrical Insulation (ISEI),, vol. ISEI 2010, pp. 1-3, biomedical nanotechnology for therapeutics, micro-nanoscale transport
2010. processes, engineering multiphysics and biomimetic systems and has
[24] [24] Y. Du, Y. Lv, C. Li, M. Chen, Y. Zhong, J. Zhou, X. Li, and Y. published research articles in all the fields of interest in journals of AIP, RSC,
Zhou, ―Effect of Semiconductive Nanoparticles on Insulating ACS, IEEE, Springer Verlag etc.
Performances of Transformer Oil,‖ IEEE Transactions on Dielectrics
and Electrical Insulation, vol. 19(3), pp. 770-776, 2012. Ajay Katiyar was born in Farrukhabaad,
[25] [25] Z. Jian-quan, D. Yue-fan, C. Mu-tian, L. Cheng-rong, L. Xiao- India in July 1976. He received a B. Engg in
xin, and L. Yu-zhen, ―AC and Lightning Breakdown Strength of Mechanical Engineering from the Pt. Ravi
Transformer Oil Modified by Semiconducting Nanoparticles,‖ 2011 Shankar Shukla University, Raipur, India in
Annual Report Conference on Electrical Insulation and Dielectric 2001 and M.Tech in Manufacturing
Phenomena (CEIDP), vol. CEIDP 2011, pp. 652 - 654 2011. Engineering from the UP Technical
[26] [26] D.-E. A. Mansour, E. G. Atiya, R. M. Khattab, and A. M. Azmy, University, Lucknow, India in 2005. He was
―Effect of Titania Nanoparticles on the Dielectric Properties of employed as an engineer from Bharatpur
Transformer Oil-Based Nanofluids,‖ 2012 Annual Report Conference Agriculture Implement Works,
on Electrical Insulation and Dielectric Phenomena (CEIDP), vol. Bharatpur, India from 2001 to
CEDIP 2012, pp. 295-298, 2012. 2003. He was a Senior Lecturer, at the Maharana Pratap
[27] [27] J.-C. Lee, H.-S. Seo, and Y.-J. Kim, ―The increased dielectric Engineering College, Kanpur, India from 2005 to 2007 and
breakdown voltage of transformer oil-based nanofluids by an external Senior Lecturer, CSJM University Kanpur, India from 2007 to
magnetic field,‖ International Journal of Thermal Sciences, vol. 62, pp. 2008. Since May 2008, he is a Scientist with the Defence
29-33, 2012. Research and Development Organization (DRDO), India. Since
[28] [28] V. Segal, A. Rabinovich, D. Nattrass, K. Raj, and A. Nunes, 2014, he is also pursuing a Ph.D. at the Indian Institute of
―Experimental study of magnetic colloidal fluids behavior in power Technology Madras, India. His areas of research interest are
transformers,‖ Journal of Magnetism and Magnetic Materials vol. 215- magneto and electrorheology in nanofluids, nano-finishing,
216, pp. 513-515, 2000. industrial and automotive applications of nanofluids etc.
[29] [29] J. Li, Z. Zhang, P. Zou, S. Grzybowski, and M. Zahn,
―Preparation of a Vegetable Oil-Based Nanofluid and Investigation of Lakshmi Sirisha Maganti was born in
Its Breakdown and Dielectric Properties,‖ IEEE Electrical Insulation Sattenapalli, India in August 1987. She
Magazine, vol. 28(5), pp. 43-50, 2012. received a B.Tech in Mechanical Engineering
[30] [30] J. G. Hwang, M. Zahn, F. M. O‘Sullivan, L. A. A. Pettersson, O. from Nagarjuna University, Guntur, India and a
Hjortstam, and R. Liu, ―Effects of nanoparticle charging on streamer M. Engg (with University Gold Medal) in
development in transformer oil-based nanofluids,‖ Journal of Applied Refrigeration and Air Conditioning from the
Physics, vol. 107, no. 014310, pp. 014310, 2010. College of Engineering, Anna University,
[31] [31] N. Kurra, V. Bhadram, C. Narayana, and G. Kulkarni, ―Few layer Chennai, India. She was employed as an
graphene to graphitic films: infrared photoconductive versus bolometric Assistant Professor of Mechanical Engineering
response,‖ Nanoscale, vol. 5, pp. 381–389, 2013. at the SKR Engineering College, Chennai, India
[32] [32] P. K. Watson, W. G. Chadband, and M. Sadeghzadeh-Araghi, from 2012 to 2013. Since 2013, she is pursuing a Ph.D. at the Department of
―The role of electrostatic and hydrodynamic forces in the negative-point Mechanical Engineering of the Indian Institute of Technology Madras, India.
breakdown of liquid dielectrics,‖ IEEE Transactions on Electrical Her research interests include Computational Fluid Dynamics and Heat
Insulation, vol. 26(4), pp. 543-550, 1991. Transfer, nanofluids, microfluidic systems, etc.
[33] [33] A. Beroual, M. Zahn, A. Badent, K. Kist, A. J. Schwabe, H.
Yamashita, K. Kamazawa, M. Danikas, W. G. Chadbrand, and Y.
Torshin, ―Propagation and Structure of Streamers in liquid Dielectrics,‖ Arvind Pattamatta is as an Assistant professor
IEEE Electrical Insulation Magazine, vol. 14(2), pp. 6-17, 1998. in the Department of Mechanical Engineering at
[34] [34] P. Dhar, A. Pattamatta, and S. K. Das, ―Trimodal charge Indian Institute of Technology Madras since
transport in polar liquid-based dilute nanoparticulate colloidal 2010. He received his Bachelor, Master‘s and
dispersions,‖ Journal of Nanoparticle Research, vol. 16, no. 10, pp. 1- Doctoral degrees in Aerospace Engineering from
20, 2014. the University of Madras in 2001, the Indian
Institute of Science in 2003, and the State
University of New York at Buffalo in 2009,
respectively. From 2003 till 2005, he was
employed as a Design Engineer in the
Combustion Center of Excellence at GE India
Technology Center in Bangalore, India where he was using Computational
Fluid Dynamics based tools for the analysis of fluid flow and heat transfer in
GE Aircraft engine Diffusers and Combustion chambers. During the years
2009-2010 he was a Principal Scientist in the R.D. Aga Research,
Technology and Innovation Centre at Thermax Limited, Pune. He is the
recipient of Alexander von Humboldt fellowship for the year 2013 to conduct
 Manuscript communicated to IEEE Transactions on Dielectrics and Electrical Insulation
research studies at TU Darmstadt, Germany. His research interests are in the cancer cells, heat exchangers, boiling in mini/microchannels, fuel cells, jet
areas of Computational Nanoscale energy Transport, Computational Fluid instabilities, heat transfer in porous media, and computational fluid dynamics.
dynamics, microfluidics, Turbulence Modeling, and High performance He is a recipient of the DAAD and Alexander von Humboldt Fellowship of
computing. Germany. He is a fellow of the Indian National Academy of Engineering and
the National Academy of Sciences, India. He has been awarded the Peabody
Visiting Professorship at the Mechanical Engineering Department,
Sarit Kumar Das is a Professor with the Mechanical Engineering Massachusetts Institute of Technology, Cambridge, USA, in 2011. He is the
Department of Indian Institute of Technology Madras, India. Since July 2015, Editor-in-Chief of the International Journal of Micro/Nanoscale Transport and
he is also a Professor at the School of Mechanical, Materials and Energy an Associate Editor of the journal Heat Transfer Engineering.
Engineering (SMMEE) and the Director of Indian Institute of Technology
Ropar, India. He has published four books and more than 200 research
papers. His current research interests include heat transfer in nanofluids,
microfluidics, biological heat transfer, nanoparticle mediated drug delivery in