Controlled synthesis of titanium dioxide nanoparticles in a modified diffusion flame reactor

Article abstract of DOI:10.1016/S0025-5408(01)00552-9

Titanium dioxide (TiO₂) nanoparticles were synthesized by the oxidation of titanium tetrachloride (TiCl₄) in a modified diffusion flame reactor. The reactor utilized a multiport diffusion type burner composed of 5 concentric tubes. The flame configuration was (Ar + TiCl₄)/Ar/H₂/O₂/air producing several tens of grams of TiO₂ nanoparticles per hour. Flame characteristics of the diffusion flame at the modified burner outlet were investigated. TiCl₄ concentration and flow rates of combustion gases such as oxygen, hydrogen and air were chosen as key experimental variables for the control of the particle size and phase composition. TiO₂ nanoparticles ranged from 10 to 30 nm in average particle diameter, and the mass fraction of anatase synthesized was 40 to 80% in all experiments.

Full text of DOI:10.1016/S0025-5408(01)00552-9

Materials Research Bulletin 36 (2001) 627–637
Pergamon
Controlled synthesis of titanium dioxide nanoparticles in a
modified diffusion flame reactor
Hee Dong Janga,*, Seong-Kil Kimb
aDivision of Materials Development, Korea Institute of Geology, Mining, and Materials, Tajeon, 305-350, Korea
bR&D Center, Samhwa Paint Industry, Ansan, 425-110, Korea
(Refereed)
Received 1 July 2000; accepted 10 October 2000
Abstract
Titanium dioxide (TiO2) nanoparticles were synthesized by the oxidation of titanium tetrachloride
(TiCl4) in a modified diffusion flame reactor. The reactor utilized a multiport diffusion type burner
composed of 5 concentric tubes. The flame configuration was (Ar ϩ TiCl4)/Ar/H2/O2/air producing
several tens of grams of TiO2 nanoparticles per hour. Flame characteristics of the diffusion flame at
the modified burner outlet were investigated. TiCl4 concentration and flow rates of combustion gases
such as oxygen, hydrogen and air were chosen as key experimental variables for the control of the
particle size and phase composition. TiO2 nanoparticles ranged from 10 to 30 nm in average particle
diameter, and the mass fraction of anatase synthesized was 40 to 80% in all experiments. © 2001
Elsevier Science Ltd. All rights reserved.
Keywords: A. Ceramics; B. Chemical synthesis; D. Crystal structure
1. Introduction
Particles smaller than several tens of nanometer in primary particle diameter are consid-
ered as nanoparticles. Titanium dioxide (TiO2) nanoparticles shows high surface area per unit
volume. This high value causes an increase of activity as a catalyst and greater sensitivity as
a sensor. Titanium dioxide nanoparticles are used as a photocatalyst, in cosmetics for high
* Corresponding author. Fax: ϩ82-42-861-9727.
E-mail address: hdjang@kigam.re.kr (H.D. Jang).
0025-5408/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved.
PII: S0531-5565(01)00552-9

628
H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
absorption of ultraviolet light, in toners and coating materials, etc [1]. TiO2 nanoparticles are
routinely produced by the gas-to-particle conversion in flame reactors because this method
provides good control of particle size, particle crystal structure and purity [2]. The diffusion
flame that exhibits strong gradients in concentration and temperature during particle forma-
tion and growth is one of the interesting tool among the various flames to synthesize TiO2
nanoparticles. Formenti et al. [3] have reported the production of TiO2 nanoparticles by the
oxidation of titanium tetrachloride (TiCl4) in (O2 ϩ TiCl4)/N2/H2/O2 diffusion flames. They
produced 6 to 145 nm sized anatase particles by varying the flow rate of TiCl4 and that of
oxygen carrying out TiCl4 vapor. Vemury and Pratsinis [4,5] investigated the effect of
dopants (SiCl4, SnCl2 and AlCl3) and electric discharge (corona) on the particle size and
phase composition of TiO2 nanoparticles at the diffusion flame configuration of (Ar ϩ
TiCl4)/air/CH4. SiCl4 decreased primary particle size and inhibited the transformation of
anantase to rutile. Meanwhile, SnCl2 and AlCl3 were found to enhance the transformation of
anatase to rutile. The corona discharge also showed same effect as the effect of SiCl4 on the
particle size and phase composition of TiO2 nanoparticles. Pratsinis et al. [6] reported the
effect of flame configurations such as (Ar ϩ TiCl4)/CH4/O2, (Ar ϩ TiCl4)/O2/CH4, etc on
the specific surface area and phase composition of the TiO2 nanoparticles in a diffusion flame
reactor. Yang et al. [7] also used (air ϩ titanium isopropoxide)/CH4/O2 diffusion flames for
processing TiO2 nanoparticles from titanium isopropoxide. TiO2 particles obtained in their
experiments were 5.2 to 11.0 nm in crystal size and 64 to 100% in anatase mass fraction.
Although those previous studies have obtained a many interesting and important results,
more studies on the characteristics of the formation, growth and phase transformation of
TiO2 nanoparticles in the diffusion flame are still required because the flame has complex
environments. There are many variables for the synthesis of nanoparticles in the flames such
as the configurations of the flame, the concentration of precursor, gas composition, gas flow
rate, etc. Those parameters affect the flame characteristics, particle formation, particle
aggregation, sintering, and phase composition of particles.
The standard diffusion flame having a configuration of precursor/fuel/oxidant has been
most frequently used to synthesize particles by the gas phase reaction in the previous studies.
But as Pratsinis et al. [6] have reported, the different flame configuration has great effect on
the particle size and phase composition of TiO2 nanoparticles. In this respect, the develop-
ment of the well-designed burner having a specific flame configuration is very important and
can be considered as know-how because it gives the best reaction condition of the precursor
in the flame such as optimum temperature profile with feasible consumption of oxidant gas,
stable flame for the optimum residence time of precursor, etc.
In the present study, the flame synthesis of TiO2 nanoparticles with a modified diffusion
flame reactor having a different flame configuration from the previous ones is presented and
discussed. The modified diffusion flame reactor was prepared using the multiport burner that
was composed of 5 concentric tubes, and the flame configuration was (Ar ϩ TiCl4)/Ar/H2/
O2/air. A modified diffusion flame reactor was designed as a bench scale to produce more
than several tens of grams of TiO2 nanoparticles from TiCl4 per hour, which was about ten
times higher than previous studies. The advantage of the proposed flame is able to introduce
large flow of air into the fifth tube of the multiport burner for minimizing the amount of pure
oxygen, keeping low concentration of TiCl4 and short residence time of TiCl4 in the flame.

H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
629
Fig. 1. A Schematic of experimental apparatus for the synthesis of TiO2 nanoparticles.
Another merit of the flame is that it can introduce inert gases such as argon into the second
tube of the burner to prevent plugging of the exit of precursor due to the deposition of the
particles.
For the controlled synthesis of TiO2 nanoparticles by the modified flame reactor TiCl4
concentration, and the maximum flame temperature that depends on gas composition and
flow rates of combustion gases in the modified diffusion flame reactor were chosen as key
experimental variables. Effects of those variables on the particle diameter and phase com-
position were investigated.
2. Experiments
The overall reaction for the vapor-phase synthesis of TiO2 from TiCl4 in the present study
is as follows;
TiCl4 ͑ g͒ ϩ 2O2 ͑ g͒ ϩ 4H2 ͑ g͒ 3 TiO2 ͑s͒ ϩ 4HCl ͑ g͒ ϩ 2H2O ͑ g͒
A schematic of the bench-scaled experimental apparatus for the generation of TiO2
nanoparticles is shown in Fig. 1. The experimental apparatus consists of an evaporator of
aerosol precursor, an aerosol generator, and a particle collector.
A modified diffusion flame burner (3.4 cm in outside diameter and 45 cm in length)
composed of five concentric stainless tubes (10, 16, 22, 28, and 34 mm in outside diameter)
was prepared since it provided a stable flame over a wide range of operating conditions.
Reactants mixing configuration of the aerosol generator is shown in Fig. 2. A quartz tube of
10 cm in diameter and 120 cm in length was installed to preserve the flame and particles.
Liquid phase TiCl4 was injected to the evaporator by using the syringe pump, and the
vaporized TiCl4 was carried into the central tube of the burner by introducing dry argon (Ar)
gas (99.999%). The burner and manifold were kept about 20°C above the evaporator
temperature to prevent condensation of TiCl4 in the lines. Hydrogen (H2) was used as a fuel
while oxygen (O2) and air were used as oxidants.

630
H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
Fig. 2. Reactants mixing configuration of the aerosol generator (Remark: Shaded area at the top view of the burner
is the reducer of outlet diameter of 1st tube).
A Pt-Rh R type thermocouple (Omega Engineering) was used to measure the flame
temperature of the burner. The measurements were performed three times for each condition,
and mean values were chosen as the flame temperatures in the axial direction of the flame
in the absence of the precursor.
TiO2 nanoparticles generated from TiCl4 by the gas phase reaction were collected with a bag
filter that was made of teflon fiber (Tefaire, Dupont Co.). The particle morphology and size were
characterized by transmission electron microscopy (TEM, Philips Model CM12). The average
particle diameter was determined by counting more than 200 particles from TEM pictures. The
specific surface area of the powders was measured by nitrogen adsorption at Ϫ196°C using the
BET equation (Micrometrics Model ASAP 2400). Assuming spherical particles, the average
particle size dp was calculated from measured specific surface area A and particle density ␳p by
dp ϭ 6/(␳ ⅐A), and calculated particle sizes and observed ones were compared. X-ray diffrac-
p
tometer (XRD, Rigaku Co. Model RTP 300 RC) was used to obtain X-ray diffraction patterns of
powders. The phase composition of TiO2 particles was calculated from the relative intensities of
the strongest peaks corresponding to anatase and rutile [8].
3. Results and discussion
3.1. Flame characteristics
Characteristics of the modified diffusion flame were investigated without introducing the
TiCl4 vapor into the burner. Flow rate of each tube was as follows; 1st: 2 1/min of Ar, 2nd:

H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
631
Fig. 3. Temperature distribution along axial direction of the diffusion flame.
5 1/min of Ar, 3rd: 6 1/min of H2, 4th: 15 1/min of O2, and 5th: 60 1/min of air. Fig. 3 shows
the temperature distribution of the flame along the axial distance at different radial position.
The temperature profiles having maximum at the different axial distance are found and the
highest temperature in the flame is 1700°C, while the radial distance of the flame is 6 mm
and axial distance of that is 10 mm. The adiabatic flame temperature from the thermody-
namic calculation is about 2,000°C at same condition. The maximum temperature of the
flame rapidly decreased from 1,700 to 1,300°C as the distance from the burner outlet
increased from 5 to 25 mm. The height of the flame was about 6cm.
3.2. Effect of precursor concentration
The effect of precursor (TiCl4) concentration in the flame synthesis on the primary particle
size was investigated. TiCl4 concentration in the flame was varied from 1.13 ϫ 10Ϫ5 to
4.54 ϫ 10Ϫ5 mol/l [9] by changing the feed rate of TiCl4 at the maximum flame temperature
of 1700°C, holding the gas flow rate as shown in Fig. 3. Fig. 4 shows change of the specific
surface area of TiO2 nanoparticles from BET analysis. As TiCl4 concentration increased, the
specific surface area decreased from 79 to 57 m2/g. The average diameter from the BET
analysis was 19nm at 1.13 ϫ 10Ϫ5 mol/l and increased to 28 nm at 4.54 ϫ 10Ϫ5 mol/l. Fig.
5 shows the TEM pictures of TiO2 nanoparticles made in the diffusion flame at the condition
of Fig. 3. The average diameter from the TEM pictures was about 20 nm at the concentration
of 1.13 ϫ 10Ϫ5 mol/l and increased to about 30 nm by increasing the concentration to 4.54 ϫ
10Ϫ5 mol/l. But the size distribution of TiO2 nanoparticles became polydisperse as the
concentration increased. The average particle sizes calculated from BET-adsorption areas

632
H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
Fig. 4. Effect of TiCl4 concentration on BET surface area (O2: 15 1/min, Air: 60 1/min, H2: 6 1/min, and Ar: 5
1/min, Carrier gas (Ar): 2 1/min).
were nearly the same as obtained from TEM study. The amount of collected particles at
4.54 ϫ 10Ϫ5 mol/l of the TiCl4 concentration was about 20 grams per hour. Plugging at the
exit of the burner was not observed through the experiments. These results suggest that the
synthesis of TiO2 nanoparticles with the production rate of several tens of grams per hour
was successful with the modified flame configuration in the present study.
As the concentration of aerosol precursor increases at the fixed flame temperature and gas
glow rate, the growth rate of nuclei by the coagulation and surface reaction becomes faster
than the nucleation rate during the gas phase reaction. Thus, larger primary particles are
expected to be generated at the higher precursor concentrations [10]. Nanometer-sized
primary particles form aggregates in the gas phase because of interactive forces between
particles. Such aggregates change into large particles when the temperature in the gas phase
is high enough to sinter the aggregates [11,12]. As can be seen in Fig. 5, the increase in the
number of larger particles with TiCl4 concentration appears to indicate that they are produced
by the sintering. However, numerous smaller particles that seemed to be grown by the
coagulation and surface reaction are observed at the low TiCl4 concentration. If the sintering
is dominant in the particle growth, it is expected that there should be many large particles at
the low TiCl4 concentration because the primary particle size is comparatively small at the
low TiCl4 concentration. Therefore, it is suggested that particle growth by the sintering is not
dominant, and is only effective at the higher TiCl4 concentrations in the present experimental
condition.
From the analysis of X-ray diffraction patterns the crystalline phases of the TiO2 nano-
particles generated from different TiCl4 concentrations were found nearly same; it was about
45% anatase in mass fraction. The residence time at the condition of Fig. 3 is 0.033 sec that

H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
633
Fig. 5. TEM micrographs of TiO2 nanoparticles collected at four different TiCl4 concentrations (a: 1.13 ϫ 10Ϫ5
b: 2.27 ϫ 10Ϫ5, c: 3.45 ϫ 10Ϫ5, d: 4.54 ϫ 10Ϫ5 mol/l).
,
is a little longer than those reported by Yang et al. [7]. The increase in residence time of
reactant in the flame may increase the conversion rate of TiO2 particles from anatase to rutile
phase.
3.3. Effect of the maximum flame temperature.
The maximum flame temperature is also one of the very important variable for the
determination of the particle size and crystal structure in the flame synthesis of TiO2
nanoparticles. For the control of the maximum flame temperature, the flow rates of com-
bustion gases such as oxygen, hydrogen and air were varied in the present study.

634
H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
Fig. 6. Effect of pure oxygen flow rate of 4th tube on the product particle size.
The maximum temperature of the diffusion flame decreases as the oxygen flow rate
decreases. Thus, the growth rate of particles due to low coagulation and sintering rates that
are proportional to the temperature decreases resulting in smaller particles. Oxygen flow rate
at the 4th tube of the burner decreased from 15 to 5 1/min with holding the gas flow rates of
Fig. 3, and 2.27 ϫ 10Ϫ5 mol/l of TiCl4 concentration. As the oxygen flow rate decreased
from 15 to 5 1/min, the air flow rate at the 5th tube increased from 60 to 75 1/min to keep
the constant total gas flow rate; it means that TiCl4 concentrations are the same in the flame.
When the maximum flame temperature was lowered from 1700 to 1400°C along with the
decrease of the oxygen flow rate, the average particle diameter calculated from the BET data
also decreased from 23 to 14 nm (Fig. 6). This decrease of particle diameter in the present
study is considered to be due to low coagulation and sintering rate of particles caused by the
decrease in the temperature of diffusion flame.
The anatase mass fraction of TiO2 nanoparticles increased from 41 to 80% as the oxygen
flow rate decreased from 15 to 5 1/min (Fig. 7). The increase of the anatase mass fraction of
TiO2 nanoparticles is considered to be due to low transformation rate from anatase to rutile
at the lower flame temperatures.
The flame temperature is also dependent on the hydrogen flow rate. As the hydrogen flow
rate decreases, the flame temperature decreases rapidly and smaller particles are generated
because of low coagulation and sintering rate in the flame. Higher mass fraction of anatase
is also expected at lower flame temperature. Effect of hydrogen gas flow rate on the particle
size and phase composition was investigated in the present study. Hydrogen flow rate at the
3th tube of the burner decreased from 8 to 4 1/min with holding the Ar flow rates (2nd tube)
at 5 1/min, air flow rate (4th tube) at 10 1/min, air flow rate (5th tube) at 65 1/min, and TiCl4
concentration at 2.27 ϫ 10Ϫ5 mol/l. The air at the 4th tube was introduced instead of oxygen

H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
635
Fig. 7. X-ray diffraction patterns of TiO2 particles showing the increase in the mass fraction of anatase with
oxygen flow rate (a: 15, b: 10, c: 5 1/min).
to decrease the flame temperature and to minimize the production cost without using pure
oxygen gas. With the decrease in hydrogen flow rate the maximum flame temperature was
decreased from 1300 to 1000°C, and then the average particle diameter also decreased from
29 to 12 nm (Fig. 8). The anatase mass fraction increased from 27% at 8 1/min to 75% at
Fig. 8. Effect of hydrogen flow rate on the product particle size and mass fraction of anatase.

636
H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
Fig. 9. Effect of total gas flow rate on the product particle size and mass fraction of anatase.
4 1/min (Fig. 8). The decrease in the particle size and increase in the anatase mass fraction
of TiO2 nanoparticles were thus obtained as expected.
Total gas flow rate was changed from 65 to 35 1/min by only decreasing the air flow rate
at 5th tube of the burner. TiCl4 feed rate at 2.0 ϫ 10Ϫ3 mol/min, Ar at 5 1/min, H2 at 6 1/min,
air at 10 1/min and air at 65 1/min was the initial operating condition. As the air flow rate
decreased, the adiabatic flame temperature from the thermodynamic calculation increased
from 1,650 to 2,500°C. The TiCl4 concentration in the flame increased from 1.38 ϫ 10Ϫ4 to
1.91 ϫ 10Ϫ4 mol/l with the decrease of air flow rate. Such increases of the flame temperature
and TiCl4 concentration were expected to result in the increase of particle size and decrease
of anatase mass fraction of TiO2 nanoparticles due to high coagulation rate of small particles.
Average particle diameter increased from 15 to 26 nm (Fig. 9). As the flame temperature
increased, large spherical particles produced by high sintering rate between small particles
were also observed frequently in TEM analysis. Fig. 9 also showed the variation of the
crystalline phase of TiO2 nanoparticles generated at different total gas flow rates in the flame;
it was 61% in anatase mass fraction at 65 1/min, and 43% at 35 1/min. Such increases of
particle size and decrease of anatase mass fraction of TiO2 nanoparticles are considered to
be due to increase of the flame temperature, the TiCl4 concentration and the residence time
as expected.
4. Conclusions
Formation of TiO2 nanoparticles from TiCl4 was investigated in the modified diffusion
flame reactor. Temperature distribution having maximum along the axial distance at the

H.D. Jang, S.-K. Kim / Materials Research Bulletin 36 (2001) 627–637
637
different radial position of the flame was found. The highest production rate of TiO2
nanoparticles using the modified flame reactor was about 20 grams per hour. As the TiCl4
concentration increased from 1.13 ϫ 10Ϫ5 to 4.54 ϫ 10Ϫ5 mol/l, average particle diameter
increased from 19 to 28 nm. When the maximum flame temperature was varied from 1700
to 1400°C by reducing oxygen flow rate from 15 to 5 1/min, the average particle diameter
decreased from 23 to 14 nm, and the anatase mass fraction of TiO2 nanoparticles was
increased from 41 to 80%. The decrease in the hydrogen flow rate resulted in the lowering
of the maximum flame temperature from 1300 to 1000°C, and subsequent reduction of the
average particle diameter from 29 to 12 nm. The anatase mass fraction decreased with
hydrogen flow rate from 75% at 4 1/min to 27% at 8 1/min. As the total gas flow rate was
reduced from 65 to 35 1/min, the average particle diameter increased from 15 to 26 nm, and
the crystalline phase of TiO2 nanoparticles decreased from 61 to 43% in anatase mass
fraction.
Acknowledgements
This study was supported by the Korea Institute of Industrial Technology Evaluation &
Planning.
References
[1] R.W. Siegel, Nanophase Materials: Synthesis, Structure, and Properties, F.F. Fujita (Ed.), Springer Series
in Material Science, Springer-Verlag, Vol. 27, 1994, pp. 65.
[2] A. Gurav, T. Kodas, T. Pluym, Y. Xiong, Aerosol Sci. Technol. 19 (1993) 411.
[3] M. Formenti, F. Juillet, P. Meriaudeau, S.J. Teichner, P. Vergnon, J. Colloid and Interface Science 39 (1972)
79.
[4] S. Vemury, S.E. Pratsinis, J. Am. Ceram. Soc. 78 (1995) 2984.
[5] S. Vemury, S.E. Pratsinis, Appl. Phys. Lett. 66 (1995) 3275.
[6] S.E. Pratsinis, W. Zhu, S. Vemury, Powder Technol. 86 (1996) 87.
[7] G. Yang, H. Zhuang, P. Biswas, Nanostructured Materials 7 (1996) 675.
[8] R.A. Spurr, H. Myers, Anal. Chem. 29 (1957) 760.
[9] H.D. Jang, J. Jeong, Aerosol Sci. Technol. 23 (1995) 553.
[10] S.K. Friedlander, Smoke, Dust and Haze, Wiley, New York, 1977, pp. 234.
[11] T. Seto, M. Shimata, K. Okuyama, Aerosol Sci. Technol. 23 (1995) 183.
[12] H.D. Jang, S.K. Friedlander, Aerosol Sci. Technol. 29 (1998) 81.