Journal of

Welcome Comments Help

MATERIALS RESEARCH

Electrically controlled flame synthesis of nanophase TiO2 , SiO2 ,

and SnO2 powders
Srinivas Vemury,a) Sotiris E. Pratsinis,b) and Lowinn Kibbeyc)
Department of Chemical Engineering, University of Cincinnati, Cincinnati, Ohio 45221-0171
(Received 20 April 1996; accepted 27 October 1996)

Nanophase particles with precisely controlled characteristics are made by oxidation of

their halide vapors in electrically assisted hydrocarbon flames using needle-shaped or
plate electrodes. The particle size and crystallinity decrease with increasing field strength
across the flame. The field generated by the electrodes across the flame decreases the
particle residence time in the high temperature region of the flame. Furthermore, it
charges the newly formed particles, resulting in electrostatic repulsion and dispersion
that decreases particle growth by coagulation. Electric fields reduced the primary
particle size of TiO2 , the agglomerate size of SnO2 , and both the agglomerate and
primary size of SiO2 .

I. INTRODUCTION particles unipolarly during their synthesis can reduce the

Nanosized particles have distinctly different proper-
ties compared to bulk materials because the number of
atoms on the particle surface is comparable to that inside
the particle.1 As a result, these particles are characterized
by lower melting point, better light absorption, and struc-
tural properties. Nanosized particles are also used to form
catalysts with high specific surface area and large density
of active sites. Though a number of processes have been
developed for synthesis of nanoparticles,2 their produc-
tion cost remains high, limiting, thus, development of
their applications. Flame reactors, on the other hand,
are routinely used in industrial synthesis of submicron
powders with relatively narrow size distribution and high
purity.3 As such, the flame route has high potential
for relatively inexpensive synthesis of finer, nanosize
particles at high yields and production rates.
Charging particles during their formation can have
a profound effect on primary particle size, crystallinity,
degree of aggregation, and agglomerate size. Hardesty
and Weinberg4 showed that the silica primary particle
size can be reduced by a factor of three when an electric
field is applied across a counterflow CH4 -air diffusion
flame. They attributed it to the rapid deposition of
particles on the electrodes, thus decreasing the particle
residence time in the high temperature region of the
flame. Likewise, Katz and Hung5 showed that the size of
TiO2 , SiO2 , and GeO2 particles made in a similar reactor
was greatly influenced by the presence of electric fields.
Xiong et al.6 showed theoretically that charging titania
a)Currently at Lucent Technologies Inc., Murray Hill, New Jersey
07974.
b)
Author to whom correspondence should be addressed.
c)
Currently at Protein Technologies International, St. Louis, Missouri
63164.
particle size and narrow the particle size distribution.
Recently, Vemury and Pratsinis7 showed that the
average primary particle size of TiO2 particles made
by flame oxidation of TiCl4 was decreased from 50 to
30 nm, and its crystallinity was drastically altered by
applying a unipolar electric field with needle electrodes
across a CH4 -air diffusion flame. They attributed the
observed changes to particle charging and repulsion
and the ionic wind from the needles across the flame,
which decreased the particle residence time at high
temperatures. Though this study nicely revealed the
potential of electrical discharge on flame synthesis of
powders, it was confined to a single material and a
diffusion flame which is hard to characterize for its broad
variation of temperature and gas velocities. Furthermore,
the ionic wind can have a dramatic effect on the diffusion
flame itself since it may affect reactant gas mixing,8
which, in turn, can alter the particle diameter by as much
as a factor of 10.
Here, the effects of externally controlled electric
fields on the characteristics of flame-generated powders
are investigated in a burner-stabilized flat flame and a
nonstabilized laminar flame. These fields are created
either by needle electrodes that introduce ions in the
flame or by plate electrodes that merely attract flame-
generated ions. Three materials with distinctly different
sintering rates and applications, TiO2 , SiO2 , and SnO2 ,
are produced here. Titania is used as a pigmentary
material,9 photocatalyst,10 and as a catalyst support.11
Fumed silica particles are widely used for optical fibers,
catalyst supports, and as fillers and dispersing agents.12
Nanosized tin oxide powders are used as semiconductors
and gas sensors.13 The objective of the present study
is to explore the potential of electric fields for flame
synthesis of nanophase materials with closely controlled

J. Mater. Res., Vol. 12, No. 4, Apr 1997  1997 Materials Research Society 1031

http://journals.cambridge.org Downloaded: 14 Aug 2014 IP address: 128.120.194.194

 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders
characteristics. The effects of material properties, electric
field intensity, electrode geometry, and position along
the flame axis are investigated.
II. EXPERIMENTAL
Figure 1 shows a schematic of the experimental
setup. A laminar premixed burner-stabilized flame re-
actor is used to make titania, tin oxide, and silica
particles in the presence of externally controlled electric
fields. The flame is stabilized on the burner mouth
using a 2 cm long mullite monolith (Corning) honey-
comb with 48 openingsycm2 . The burner [alumina tube
(Coors), 1.875 cm ID] is packed to 3y4 of its volume
with glass beads (6 mm in diameter) supported on a
screen to provide good mixing of the reactants before
they enter the honeycomb. The advantage of using a
burner-stabilized, premixed, flat flame is that most of
the particles experience similar temperatures and gas
velocities across the flame.14 The above powders are
also made in a nonstabilized, laminar flame reactor. This
reactor is also an alumina tube (Coors, 1.25 cm ID), but
without the monolith.
Clean, dry argon gas (Wright Brothers, 99.8%) is
bubbled into a gas-washing bottle containing TiCl4
(Aldrich, 99.9%), SiCl4 (Aldrich, 99%), or SnCl4
(Aldrich, 99%), premixed with particle-free nitrogen
(Wright Brothers, 99.8%), oxygen (Matheson, 99.9%),
and methane (Wright Brothers, 99.8%) and sent through
the burner. A check valve is used before the precursor is
mixed with the mixture of air and methane to prevent any
back flow of the flame. In the case of the nonstabilized
flame reactor, air is used as the oxidant. Oxide particles
are formed in the flame by oxidation/hydrolysis of the
FIG. 1. Schematic of the experimental setup.
1032 J. Mater. Res., Vol. 12, No. 4, Apr 1997
http://journals.cambridge.org Downloaded: 14 Aug 2014
precursors. The particles are collected on glass fiber
filters (Gleman Scientific) for subsequent analysis in an
open-faced filter (Graseby/Anderson) kept 10.5 cm from
the burner face, by the aid of a vacuum pump.
Stainless steel, needle-like electrodes with sharp tips
(ø100 mm) or plate electrodes [3.8 3 2.5 3 0.38 cm
(long plates) or 2.5 3 2.5 3 0.38 cm (short plates)] are
used to create the electric field across the flame.7 One
electrode is connected to the DC power supply (Gamma
High Voltage Research, Inc.; Spellman) with reversible
polarity and the second one to the ground. For the
powders made in the burner-stabilized, premixed, flat
flame, the distance between the electrode tips is 5 cm,
while in the nonstabilized flame, it is 4 cm. The plate
electrodes are arranged in such a way that the bottom
edge of the plate is in the same plane as the burner face.
The specific surface area of the powder, A, is meas-
ured by nitrogen adsorption at 77 K (Gemini 2360,
Micromeritics) using the BET equation. The average
primary particle size, dp , is obtained from dp ­ 6yrp A,
where rp is the particle density. The particle morphology
is obtained by transmission electron microscopy (TEM)
on Philips CM 20 or Philips EM 400 microscopes operat-
ing at 100 kV. The TEM primary particle size is obtained
by visually counting at least forty primary particles.15
The weight fractions of the anatase and rutile phases
in the titania samples are obtained by x-ray diffraction7
(D500, Siemens; using Cu Ka radiation). The XRD crys-
tallite size was not determined since this estimate may be
inconclusive for powders with large particle size differ-
ences.16 For flame-generated powders it has been shown
that XRD crystallite sizes are consistently smaller than
the BET determined particle sizes.15 A 0.038 cm gauge
Pt–Rh thermocouple (Omega Engineering) insulated
with a mullite sheath is used for measuring the flame
temperature and is corrected for radiation losses.17
III. RESULTS AND DISCUSSION
A. Flame synthesis of oxides in the
absence of electric field
First, titania, tin oxide, and silica powders are made
in both flames in the absence of electric fields. At the
burner-stabilized flat flame reactor, the room temperature
flow rates of CH4 , N2 , and O2 streams are 445, 3250,
and 1300 cm3ymin, respectively, while 250 cm3ymin of
argon are bubbled through TiCl4 and SnCl4 . The SiCl4
bubbler is kept in ice (at 0 ±C) to reduce its vapor
pressure, and only 200 cm3ymin of argon are bubbled
through SiCl4 . To keep the total gas flow rate the
same in all experiments, the N2 flow rate is set to
3300 cm3ymin during silica synthesis. Thus, the TiCl4 ,
SnCl4 , and SiCl4 concentrations leaving each bubbler
are 1.75 3 10–4 , 2.9 3 10–4 , and 8 3 10–4 molymin,
respectively. The specific surface areas of products TiO2 ,
IP address: 128.120.194.194
 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders

SnO2 , and SiO2 are 50, 20, and 148 m2yg, respec-
tively. The differences in the specific surface area of
these oxides synthesized under similar conditions stem
from their different sintering rates.18 Tin oxide has low
melting point (1127 ±C) and sinters by evaporation-
condensation,19 while titania sinters by grain boundary
diffusion20 and silica sinters by viscous flow.21
When a nonstabilized flame is used, the flow rates
of the CH4 and air streams are 237 and 1100 cm3ymin,
respectively, while 200 cm3ymin of argon are bubbled
through the precursor. The concentrations of TiCl4 ,
SnCl4 , and SiCl4 leaving the bubbler are 1.4 3 10–4 ,
2.3 3 10–4 , and 8 3 10–4 molymin, respectively. The
specific surface areas of product TiO2 , SnO2 , and
SiO2 powders are 16, 13, and 130 m2yg, respectively.
The lower overall flow rate through the latter burner
(1537 cm3ymin) compared to that through the flat flame
burner (5205 cm3ymin) results in less gas dilution and
longer exposure at high temperatures.8 Consequently,
particles made in the nonstabilized, laminar flame are
sintered more effectively having lower specific surface
area than those made in the flat flame. This is supported
also by XRD of titania powders that were mostly rutile
when they were made in the nonstabilized flame while
they were exclusively anatase (rutile-free) in the flat
flame. Rutile is the thermodynamically stable form of
TiO2 , while anatase and brookite are metastable at all
temperatures and transform to rutile upon heating.22
B. Flame characteristics in the
presence of electric fields
Figure 2 shows pictures of the burner-stabilized
flame during synthesis of SnO2 particles at various
positive field strengths and electrode configurations.

FIG. 2. (a – f) Burner-stabilized, premixed, laminar flames producing SnO2 particles under the influence of various electric fields created by needle
(b –d) and plate (e, f) electrodes. Gas flow rates: CH4 ­ 445, O2 ­ 1300, N2 ­ 3250, and Ar ­ 250 cm3ymin; SnCl4 ­ 2.9 3 10–4 molymin.

J. Mater. Res., Vol. 12, No. 4, Apr 1997 1033

http://journals.cambridge.org Downloaded: 14 Aug 2014 IP address: 128.120.194.194

 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders
Figure 2(a) shows this flame in the absence of any
electric field. The needle electrodes are kept at 0.1 cm
above the burner face [Figs. 2(b)–2(d)]. When an
electric field of 1.4 kVycm is applied, the flame is visibly
affected by the ensuing corona discharge [Fig. 2(b)].
At low field strengths, the flame is tilted toward the
negative electrode (cathode). The established current
across the electrodes assures equal movement of positive
and negative charges; the positive ions must equal
the negative ions and electrons. As the positive ions
are larger than the electrons, the motion of positive
ions “drags” neutral gas molecules, creating a net gas
movement toward the cathode.14,23 Further increasing
the field strength up to 2.2 kVycm drastically reduces
the flame height [Figs. 2(c) and 2(d)]. At the highest
field strength, the flame front is nearly on the burner
face, resulting in substantial particle deposition on
the monolith. This represents the upper limit of the
employed field strength. The reduction of the flame
height is attributed to the ionic wind (also known as
the Chattock wind or corona wind) flowing across the
flame.24,25 When a corona discharge is created across a
flame, ions flow from the discharging electrode toward
the ground electrode. The flow of ions increases with
increasing field strength. As these ions move, convection
is created across the flame, reducing the flame height.7
Figure 2 also shows pictures of the burner-stabilized
flame producing SnO2 particles under the electric field
created by long plate electrodes [Figs. 2(e) and 2(f)]. In
this configuration the flame ions are attracted toward the
plate electrodes. Under a field intensity of 1 kVycm, the
flame height is similar to that in the absence of fields,
with a slight inclination of the flame toward both cathode
and anode. As the field intensity is increased to 2 kVycm,
the flame is pulled toward the electrodes and its height
further decreases. Literally, flame 2(a) can be viewed as a
flower bud that opens up [Figs. 2(e) and 2(f)] as the field
intensity increases and plate electrodes attract more and
more flame ions. When plates are used as the electrodes,
the ionic wind is reduced and the flame is substantially
more stable than when needle electrodes are used.
Figure 3 now shows pictures of the nonstabilized
flame producing SnO2 particles at various positive field
intensities created across the flame by needle electrodes
at 1 cm above the burner face. Since the gas flow rates in
this flame are lower, the onset of the ionic wind appears
at a slightly lower field strength (1.2 kVycm). Again, as
the field strength increases, the flame decreases in height
and leans toward the cathode [Figs. 3(b) –3(e)]. The
ionic wind has a stronger impact on this flame since the
gas velocity through the burner is lower and nonuniform
compared to that of the burner-stabilized flat flame. As a
matter of fact, at a field strength of 2.5 kVycm, the flame
is barely holding onto the burner [Fig. 3(f)] as the flame
is vigorously oscillating, and a slight increase in the field
1034 J. Mater. Res., Vol. 12, No. 4, Apr 1997
http://journals.cambridge.org Downloaded: 14 Aug 2014
strength extinguishes the flame. The latter property is
used for extinction of pool flames.26 It should be noted
that the maximum (“breakdown”) field strength that can
be used is limited by sparking between electrodes, which
is a function of electrode tip diameter, gas composition,
and electrode distance. For the employed electrodes and
SnO2 aerosol, this limit was 2.75 kVycm, which is of
little significance here since the flame is extinguished by
the ionic wind at lower field intensities. However, this is
an important limit for burner-stabilized flames that are
not extinguished by ionic wind or flames that are not
affected by burner deposits, as it was the case with the
flames forming silica and titania.
Figure 4 shows the temperature along the axis of
the burner-stabilized, flat flame in the absence of any
precursor, but with the carrier gas argon flowing through
the burner along with CH4 , N2 , and O2 . Precursor gases
were not used to avoid particle deposition on the tip of
the thermocouple.27 In the absence of an electric field,
the maximum temperature is about 1110 ±C, slightly
above the luminous zone of the flame (open circles). The
temperature is measured 0.5 cm away from the flame
center up to 4 cm above the burner face. The measured
temperature across the burner face in a horizontal plane
is almost the same (±20 ±C) as that measured at the
center. The temperature quickly drops away from the
maximum along the flame axis. Figure 4 also shows
measured flame temperatures with 21.6 kVycm applied
across the flame using the needle electrodes (filled cir-
cles). So as not to disturb the field across the flame
with the thermocouple, temperatures are measured 5 cm
above the electrodes and higher. In the presence of
the corona discharge, the downstream temperature is
about a hundred degrees lower than without the field.
Figure 4 also shows flat flame temperatures measured in
the presence of a negative electric field (21.6 kVycm)
created by long plate electrodes (filled diamonds). The
temperature downstream of the flame is not influenced by
the electric field created by the plate electrodes, implying
that the ionic wind with this configuration does not have
as strong an impact as with the needles.
Figure 4 also shows the temperature profile of the
nonstabilized flame with a maximum temperature of
1000 ±C at the tip of the flame (open squares). Also
shown are the measured temperatures of the nonstabi-
lized flame under an electric field created by the needle
electrodes of 11.75 Vycm (filled squares). The electric
field has a profound effect on the flame temperature. The
ensuing ionic wind across the flame dissipates heat by
convection.7 This reduces the measured temperature by
as much as 200 ±C away from the burner face.
C. Synthesis of titania powders
Figure 5(a) shows the specific surface area of TiO2
particles as a function of the applied positive or negative
IP address: 128.120.194.194
 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders

FIG. 3. Nonstabilized, laminar flames producing SnO2 particles under the influence of various electric fields created by needle electrodes. Gas
flow rates CH4 ­ 237, air ­ 1100, Ar ­ 200 cm3ymin; SnCl4 : 2.3 3 10–4 molymin.

field strength created by the needle electrodes across
the burner-stabilized, flat flame. In the absence of elec-
tric field, the specific surface area of titania particles is
50 m2yg and increases to about 85 m2yg at 21.8 kVycm
and to 75 m2yg at 11.8 kVycm. The specific surface
area increases or the primary particle size decreases with
increasing applied potential across the flame, regardless
of the polarity of the electric field. When a corona dis-
charge is applied, the temperature of the flame decreases
by the ionic wind (Fig. 4) which decreases the particle
sintering rate, resulting in smaller primary particles or
powders with high specific surface area. Also, the parti-
cles are charged unipolarly, causing them to repel each
other, resulting in reduced coagulation rates. Further-
more, electrostatic dispersion of unipolarly charged parti-
cles reduces their concentration,28 slowing down further
coagulation and, subsequently, particle growth.
The negative electric field is more efficient than
the positive one in reducing particle growth since it
results in smaller particles. A similar effect was observed
in diffusion flames.7 The velocity of negative ions is
20 –50% greater29 than that of positive ions, while that
of electrons is 1000 times greater,23 resulting in more
efficient particle charging.30 Furthermore, the ionic wind
sets in at lower field strengths with negative than with
positive polarity because of the greater mobility of nega-
tive ions. As a result, particle growth by coagulation and
coalescence is retarded more effectively with negative
than with positive coronas. The small reduction in the
specific surface area at the highest applied electric field
is attributed to the proximity to the “breakdown” field
intensity (1.8 kVycm) that destabilizes this flame and
the onset of particle deposition on the burner face. The
monotonic increase of specific surface area between 0.8
and 1.7 kVycm shows that electric fields can be used to
control the TiO2 primary particle size from 30 to 18 nm
at these conditions.
Needle electrodes introduce new ions in the flame
by corona discharge. An alternative is to create an
electric field using the ions generated by the flame
itself. Thus, needle electrodes are replaced by plate
electrodes. As shown in Figs. 2 and 4, the ionic wind
is reduced substantially when plate electrodes create
the electric field. Figure 5(b) shows the specific surface

J. Mater. Res., Vol. 12, No. 4, Apr 1997 1035

http://journals.cambridge.org Downloaded: 14 Aug 2014 IP address: 128.120.194.194

 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders
FIG. 4. Axial temperature profiles of the burner-stabilized (circles)
and nonstabilized (squares) premixed flames in the absence (open
symbols) and presence (filled symbols) of electric fields.
area of titania particles as a function of applied field
strength across the burner-stabilized, flat flame, for pos-
itive (circles) and negative (squares) electric fields using
short (open symbols) and long (filled symbols) plates.
At a positive field strength of 1.6 kVycm, the specific
surface area of titania particles is about 85 m2 gyg and
70 m2yg using the long and short plates, respectively.
The fact that the flame height and temperature do not
decrease substantially downstream of the flame (Figs. 2
and 4) indicates that the observed increase in the particle
specific surface area of titania particles can be attributed
largely to electrostatics rather than convection.
At a given field strength, the increase in specific
surface area of titania particles is larger with long
electrodes (115 m2yg) than with short ones (85 m2yg)
at 21.6 kVycm. By exposing the particles to an electric
field for longer residence times (long plates), particle
collisions and growth are further inhibited. However, fur-
ther downstream, the temperatures decrease drastically
(Fig. 4), and no further coalescence and sintering of the
particles takes place.
Figure 5(c) shows the specific surface area of the
TiO2 particles made in the nonstabilized, laminar flame
as a function of the applied positive field strength for
two needle-electrode positions along the flame axis.
In the absence of any electric field, TiO2 particles
made at these conditions have a specific surface area
of 16 m2yg, corresponding to primary particle size of
90 nm. The specific surface area is about 18 m2yg at
11.5 kVycm with the needles positioned at 0.1 cm from
the burner face and further increases to 27 m2yg at
12.75 kVycm, corresponding to an average primary
1036 J. Mater. Res., Vol. 12, No. 4, Apr 1997
http://journals.cambridge.org Downloaded: 14 Aug 2014
particle size of 55 nm. At lower field strengths, the
position of the needle electrodes along the axis of the
flame does not influence the particle specific surface area.
However, at an applied potential of 12.75 kVycm, TiO2
particles have a specific surface area of 21 m2yg when
the electrodes are at 1 cm from the burner face compared
to 27 m2yg when the electrodes are at 0.1 cm. Creating
the electric field close to the particle formation region
(near the burner face) results in convection and particle
charging at earlier stages of their inception, slowing
down, thus, particle growth by coagulation.
Figure 6(a) shows a micrograph of titania particles
synthesized in the nonstabilized, laminar flame in the
absence of an electric field corresponding to the data
in Fig. 5(c). The primary particles are polyhedral, with
very well-defined edges. The primary particle size varies
from about 20 to 100 nm, with an average primary
particle size of about 75 nm. At a field strength of
12 kVycm, with electrodes places at 0.1 cm, above the
burner face, the primary particle size decreases to about
60 nm [Fig. 6(b)] and to about 45 nm at a field strength
of 12.25 kVycm [Fig. 6(c)]. The primary particles still
retain their polyhedral shape. Increasing the field strength
to 12.75 kVycm decreases the primary particle size
to about 40 nm, and the variation of the particle size
decreases [Fig. 6(d)]. The decreasing TEM primary par-
ticle size with field intensity is consistent with the BET
data [Fig. 5(c)]. The necking between primary particles
results in larger average primary particles by BET than
TEM or XRD as is commonly observed with flame-made
particles.3,15,26
The titania particles made in the burner-stablized,
premixed, flat flame are .99.9 wt. % anatase, and the
phase composition of these powders does not change
as the field strength across the flame is increased. On
the other hand, TiO2 powders made in the nonstabilized,
laminar flame in the absence of an electric field are
82 wt. % rutile and 18 wt. % anatase. Figure 7 shows the
rutile content of TiO2 powders made in the nonstabilized
flame as a function of the applied positive field strength
across the flame. The rutile content of TiO2 decreases
down to 65 wt. % with increasing field strength across
the flame, as observed in our earlier diffusion flame
studies,7 though, here, powders with much higher rutile
content are obtained: 82 vs 20 wt. %. Of course, this is
attributed to the different flame structures: coflow diffu-
sion flame versus premixed flame. As the field strength
across the flame is increased, the flame height and the
particle residence time in the high temperature region
of the flame are reduced, inhibiting the anatase to rutile
phase transformation.31
D. Synthesis of silica powders
Figure 8(a) shows the specific surface area of SiO2 par-
ticles made in the burner-stabilized, premixed flat flame
IP address: 128.120.194.194
 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders

(a) (b)

(c)
FIG. 5. Specific surface area of TiO2 particles as a function of the applied field strength across the (a) burner-stabilized, premixed flame (needle
electrodes), (b) burner-stabilized, premixed flame (long and short plate electrodes), and (c) nonstabilized, laminar flame (needle electrodes).

as a function of applied field strength. The specific sur- in the presence of 1.6 kVycm negative and positive elec-
face area of SiO2 particles increases from 148 m2yg in tric fields, respectively, across the flat flame. As with
the absence of an electric field to about 240 and 220 m2yg titania, the negative electric field is more influential at

J. Mater. Res., Vol. 12, No. 4, Apr 1997 1037

http://journals.cambridge.org Downloaded: 14 Aug 2014 IP address: 128.120.194.194

 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders

FIG. 6. TEM micrographs of TiO2 particles made in a nonstabilized, laminar flame as a function of the applied positive field strength, with the
needle electrodes kept at 0.1 cm from the burner face: (a) No electric field, (b) 12 kVycm, (c) 12.25 kVycm, and (d) 12.75 kVycm.

increasing the specific surface area. Figure 8(a) typi-

FIG. 7. Rutile content of TiO2 particles made in a nonstabilized,
laminar flame as a function of the applied positive field strength for
two locations of the needle electrodes along the flame axis.
fies the capacity of electric fields to facilitate precise
control of the average primary particle size from 20
to 12 nm.
Figure 8(b) shows the specific surface area of SiO2
particles made in the burner-stabilized, flat flame as
a function of applied field strength for positive and
negative polarities, using both short (open symbols) and
long (filled symbols) plate electrodes. As with titania,
the specific surface area of silica particles increases
with increasing field strength, with the negative electric
field again being most effective in arresting the particle
growth. Creating the electric field using longer plate elec-
trodes results in a higher specific surface area at the same
applied field strength. As for TiO2 , extending the length
of the electric field downstream of the flame further
influences the particle growth. TEM analysis of silica
particles reveals that the agglomerate size decreases with
increasing field strength across the flame.32
Figure 8(c) shows the specific surface area of SiO2
particles made in the nonstabilized, laminar flame as a
function of the applied positive field strength for two
different locations of the electrodes along the axis of the
flame. Silica particles made in the absence of an electric

1038 J. Mater. Res., Vol. 12, No. 4, Apr 1997

http://journals.cambridge.org Downloaded: 14 Aug 2014 IP address: 128.120.194.194

 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders

(a) (b)

(c)
FIG. 8. Specific surface area of SiO2 particles as a function of the applied positive and negative field strength across the (a) burner-stabilized,
premixed flame (needle electrodes), (b) burner-stabilized, premixed flame (long and short plate electrodes), and (c) nonstabilized, laminar
flame (needle electrodes).

field have a specific surface area of 130 m2yg. With an 0.1 and 1 cm, respectively, away from the burner face.
applied field strength of 2 kVycm, this increases to about Again, the position of the electrodes affects the product
230 and to 200 m2yg when the electrodes are placed at particle size as with TiO2 particles.

J. Mater. Res., Vol. 12, No. 4, Apr 1997 1039

http://journals.cambridge.org Downloaded: 14 Aug 2014 IP address: 128.120.194.194

 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders

(a)

FIG. 9. Specific surface area of SnO2 particles as a function of the

applied field strength across the nonstabilized, laminar flame. The
electric field is created by the needle electrodes.

E. Synthesis of tin oxide powders

In contrast to the results of TiO2 and SiO2 , in the
case of the burner-stabilized, premixed, flat flame, the
specific surface area of SnO2 particles did not increase
and remained at about 20 m2yg at all field strengths.
This is observed when the electric field is created by
both needle and plate electrodes. As mentioned earlier,
tin oxide sinters much faster than titania and silica. The
fact that the specific surface area of SnO2 particles did
not change within experimental error indicates that the
sintering of tin oxide particle is completed at a very
early stage in the flame. As the grain size increases, the
sintering rates decrease and grain growth is very slow.33
In the presence of electric fields particles are charged.
However, the high sintering rates of tin oxide dominate
the electrostatic forces, and hence, no change in particle
characteristics is observed.
On the other hand, in the case of the nonstabilized,
laminar flame, the specific surface area of SnO2 particles
increased from 13 to 24 m2yg at a positive field strength
of 12.5 kVycm with the needles kept at 0.1 cm from
the burner face (Fig. 9). Particles made in this flame
experience a broader spectrum of temperatures than
in the burner-stabilized flame. As a result, the ionic
wind more drastically reduces the particle residence time
at high temperatures in the nonstabilized than in the
burner-stabilized flame (Fig. 4). The high temperature
encountered by the particles during the early stages of
particle growth dominates the influence of the ionic wind
(b)
FIG. 10. TEM micrographs of SnO2 particles made in a nonstabilized,
laminar flame as a function of the applied field strength, with the
needle electrodes kept at 0.1 cm from the burner face: (a) No electric
field and (b) 12.25 kVycm.
across the flame, and the specific surface area of SnO2
particles does not change when they are synthesized in
a burner-stabilized premixed flame. As the electrodes
are moved further up to 1 cm from the face of the
nonstabilized, laminar flame, the effect of the electric
field diminishes. Figure 9 shows that the SnO2 particle
specific surface area increases from 13 to 17 m2yg only
at 12.5 kVycm field strength when the electrodes are
placed at 1 cm above the burner face, while it increases
to 24 m2yg when they are just 0.1 cm above the burner.
Figure 10(a) shows the TEM micrograph of SnO2
particles made in the nonstabilized, laminar flame in the
absence of any electric field across the flame. The pri-
mary particles are polyhedral, with well-defined edges.
The primary particle size ranges from 50 to 200 nm,

1040 J. Mater. Res., Vol. 12, No. 4, Apr 1997

http://journals.cambridge.org Downloaded: 14 Aug 2014 IP address: 128.120.194.194

 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders

FIG. 11. TEM micrographs of SnO2 agglomerates made in a nonstabilized, laminar flame as a function of the applied field strength, with the
needle electrodes kept at 0.1 cm from the burner face: (a) No electric field, (b) 12 kVycm, (c) 12.25 kVycm, and (d) 12.75 kVycm.

with an average of about 100 nm. When a positive
electric field of 12.25 kVycm is applied across the flame
by electrodes kept at 0.1 cm above the burner face,
the average primary particle size decreases to about
65 nm, but with a wide size variation from 10 to 200 nm
[Fig. 10(b)]. This is in agreement with the BET results
(Fig. 9). The broad size distributions are attributed to
the various temperature histories of particles made in
the nonstabilized, laminar flame.
Figure 11(a) shows typical agglomerates of SnO2
particles made in the nonstabilized, laminar flame in
the absence of any electric field. The particles are
highly agglomerated. The agglomerates are chain-like
structures, again with a wide range of primary particles.
When an electric field of strength 12.0 kVycm is applied
across the flame using needle electrodes kept at 0.1 cm
above the burner face, the extent of agglomeration of
SnO2 particles decreases [Fig. 11(b)]. The extent of
agglomeration further decreases as the field strength
across the flame is increased to 12.25 and 12.75 kVycm
[Figs. 11(c) and 11(d)].
The ionic wind across the flame decreases the parti-
cle residence time and temperature of the flame, resulting
in decreased coagulation rates. Furthermore, electrostatic
repulsion and dispersion contribute to a further de-
crease in the coagulation rates of the particles, resulting
in smaller agglomerates with increasing field strength
across the flame. Even though the primary particle size
of SnO2 agglomerates does not change substantially with
increasing field strength across the flame, the agglom-
erate size decreases. The high sintering rates of SnO2
makes the length scales at which electrostatics become
dominant comparable to the size range of agglomerates.
Similar trends of decreasing agglomerate size are ob-
served for SnO2 particles made in the burner-stabilized,
premixed flame.32
IV. CONCLUSIONS
The effect of electric fields on the flame synthesis
of nanosized titania, silica, and tin oxide particles was
investigated in two flames: a burner-stabilized premixed,
flat flame and a nonstabilized, laminar flame. Negative
electric fields are most effective in affecting the product
powder characteristics. When the electric field is created
by needle electrodes, ionic wind is generated across
the flame, decreasing the flame height and resulting in
shorter particle residence times at high temperatures.
Furthermore, charging of the particles results in elec-
trostatic repulsion and dispersion. These two effects
result in reduced coagulation rates. The position of
electrodes along the flame axis has a strong influence

J. Mater. Res., Vol. 12, No. 4, Apr 1997 1041

http://journals.cambridge.org Downloaded: 14 Aug 2014 IP address: 128.120.194.194

 S. Vemury et al.: Electrically controlled flame synthesis of nanophase TiO2 , SiO2 , and SnO2 powders
on the particle characteristics. Placing the electrodes
closer to the particle inception region (near the burner
face) results in reduced particle growth. Replacing the
needle with plate electrodes substantially reduced the
ionic wind making electrostatics the dominant reason
for reduced particle growth. Increasing the plate length
reduced particle size since the particles were exposed to
electric fields for increased residence times.
The specific surface area of TiO2 and SiO2 increased
with increasing field strength in both flames. On the other
hand, the specific surface area of SnO2 powders was not
affected when they were made in the burner-stabilized,
premixed, flat flame, while it increased with increasing
field strength when they were made in the nonstabilized,
laminar flame. The difference in the temperature history
of the particles in these two flames is the reason for
this differing behavior. On the other hand, the sintering
rates of TiO2 and SiO2 are slow enough for charging
to influence the primary particle size. The agglomerate
size of SnO2 particles is decreased in the presence of the
electric field, even though the effect of electrostatics on
the primary particle size is not substantial. The length
scale at which electrostatics influence SnO2 particles
is on the agglomerate size rather than the primary
particle size.
In conclusion, external electric fields can greatly
facilitate synthesis of ceramic and possibly other par-
ticles with precisely controlled specific surface area,
crystallinity, and agglomerate structure. Given the cur-
rent low cost of flame aerosol technology, this tech-
nique has strong potential for large scale manufacture of
nanoparticles.
ACKNOWLEDGMENTS
We acknowledge support from the National Sci-
ence Foundation, Presidential Young Investigator Award
(S.E.P.), Grant CTS-8957042, and recently, Grant CTS-
9612107.
REFERENCES
1. R. P. Andres, R. S. Averback, W. L. Brown, L. E. Brus, W. A.
Goddard III, A. Kaldor, S. G. Louie, M. Moscovits, P. S. Peercy,
S. J. Riely, R. W. Siegel, F. Spaepen, and Y. Wang, J. Mater.
Res. 4, 704 (1989).
2. N. Ichinose, Y. Ozaki, and S. Kashu, Superfine Particle Technol-
ogy (Springer-Verlag, London, 1992).
1042 J. Mater. Res., Vol. 12, No. 4, Apr 1997
http://journals.cambridge.org Downloaded: 14 Aug 2014
3. G. D. Ulrich, C&EN, 62(8), 22 (1984).
4. D. R. Hardesty and F. J. Weinberg, Fourteenth Symposium (Inter-
national) on Combustion (The Combustion Institute, Pittsburgh,
PA, 1973), p. 1365.
5. J. K. Katz and C-H. Hung, Twenty-Third Symposium (Interna-
tional) on Combustion (The Combustion Institute, Pittsburgh, PA,
1990), p. 1733.
6. Y. Xiong, S. E. Pratsinis, and S. V. R. Mastrangelo, J. Colloid
Interface Sci. 153, 106 (1992).
7. S. Vemury and S. E. Pratsinis, Appl. Phys. Lett. 66, 3275 (1995).
8. S. E. Pratsnis, W. Zhu, and S. Vemury, Powder Technol. 86, 87
(1996).
9. E. J. Mezey, in Vapor Deposition, edited by C. F. Powell, J. H.
Oxley, and J. M. Blocher, Jr. (John Wiley & Sons, New York,
1966), p. 423.
10. D. F. Ollis, E. Pelizzetti, and N. Serpone, Environ. Sci. Technol.
25, 1523 (1991).
11. M. R. Bankmann, R. Brand, B. H. Engler, and J. Ohmer, Catal.
Today 14, 225 (1992).
12. J. R. Bautista and R. M. Atkins, J. Aerosol Sci. 22, 667 (1991).
13. E. U-K. Kim and I. Yasui, J. Mater. Sci. 23, 637 (1988).
14. R. Vijayakumar and K. T. Whitby, Aerosol Sci. Technol. 3, 17
(1984).
15. S. Vemury and S. E. Pratsinis, J. Am. Ceram. Soc. 78, 2984 – 2992
(1995).
16. A. Kobata, K. Kusakabe, and S. Morooka, AlChE J. 37, 347 –359
(1991).
17. S. Vemury, Flame Synthesis of Nanoparticles: Effect of Charging,
Ph.D. Thesis, University of Cincinnati (1996).
18. T. Matsoukas and S. K. Friedlander, J. Colloid Interface Sci. 146,
495 (1991).
19. P. G. Harrison and M. J. Willett, J. Chem. Soc., Faraday Trans.
1, 85 1921 – 1932 (1989).
20. M. Astier and P. Vergnon, J. Solid State Chem. 19, 67 (1976).
21. W. D. Kingery, H. K. Bowen, and D. R. Uhlmann, Introduction to
Ceramics (Wiley-Interscience, New York, 1976).
22. R. D. Shannon and J. A. Pask, J. Am. Ceram. Soc. 48, 391 (1965).
23. K. G. Payne and F. J. Weinberg, Proc. R. Soc. A250, 316 (1959).
24. D. Bradley and M. A. M. Jamel, Comb. Flame 58, 115 (1984).
25. D. Bradley, in Advanced Combustion Processes, edited by F. J.
Weinberg (Academic Press, Orlando, FL, 1986), p. 331.
26. E. Sher, G. Pinhasi, A. Pokryvailo, and R. Bar-on, Combustion
Flame 94, 244 (1993).
27. G. P. Fotou, S. E. Pratsinis, and P. A. Baron, Chem. Eng. Sci. 49,
1651 (1994).
28. G. Kasper, J. Colloid Interface Sci. 81, 32 (1981).
29. M. Adachi, Y. Kousaka, and K. Okuyama, J. Aerosol Sci. 16,
109 (1985).
30. A. Wiedensohler, J. Aerosol Sci. 19, 387 (1988).
31. M. K. Akhtar, Y. Xiong, and S. E. Pratsinis, AlChE J. 37, 1561 –
1570 (1991).
32. S. Vemury and S. E. Pratsinis, J. Aerosol Sci. 27, 951 (1996).
33. W. Koch and S. K. Friedlander, J. Colloid Interface Sci. 140, 419
(1990).
IP address: 128.120.194.194