Combustion synthesis and characterization of nanocrystalline Tin and Tin Oxide (SnOx, x = 0-2) particles

Article abstract of DOI:10.1111/j.1151-2916.2004.tb06356.x

Nanocrystalline SnO_x particles (x = 0-2) were synthesized using tetramethyltin (Sn(CH₃)₄) vapor as the particle precursor reactant in hydrogen/oxygen/argon (H₂/O₂/Ar) flames. The particle composition

Full text of DOI:10.1111/j.1151-2916.2004.tb06356.x

J. Am. Ceram. Soc., 87 [11] 2033–2041 (2004)
journal
Combustion Synthesis and Characterization of Nanocrystalline Tin and
Tin Oxide (SnOx, x ؍ 0–2) Particles
D. L. Hall, A. A. Wang, K. T. Joy, T. A. Miller, and M. S. Wooldridge†
Department of Mechanical Engineering, University of Michigan, Ann Arbor, Michigan 48109-2125
Nanocrystalline SnOx particles (x ؍ 0–2) were synthesized
using tetramethyltin (Sn(CH3)4) vapor as the particle precur-
sor reactant in hydrogen/oxygen/argon (H2/O2/Ar) flames. The
particle composition and morphology were characterized us-
ing X-ray diffractometry, transmission electron microscopy,
and nitrogen (N2) surface adsorption. By controlling the
concentration of oxygen in the reactant gases and the flame
temperatures, metallic tin (Sn), tin monoxide (romarchite
SnO), and/or tin dioxide (cassiterite SnO2) were generated.
The crystalline powders consisted of both discrete primary
particles and agglomerates, with average primary particle
sizes of 23–24 nm for SnO2 and 69 nm for Sn (based on specific
surface area measurements of bulk powders collected in the
exhaust region of the flame). The compositional results were
interpreted using equilibrium and detailed chemical kinetics
models.
Nanosized metallic tin (Sn) is also a material of considerable
interest for use in anodes of lithium-ion batteries.20,21 Tin-based
compounds have high reversible capacities; however, they can be
subject to permanent loss of capacity due to decomposition on
initial charging (e.g., SnO2 decomposition into Sn and Li2O22).
Tin-based compounds can also experience loss in capacity after
high cycling. Pure tin, tin alloys, and other tin intermetallic
compounds have been proposed as alternatives to SnO2 to elimi-
nate issues of decomposition22 and to improve cycling perfor-
mance.20 The performance of the tin particles is directly related to
the size of the particles, where smaller, nanosized Sn facilitates
rapid insertion of Li into the anode material21 and reduces
dimensional changes and potential failure in the electrodes during
cycling.20,23 In previous studies, the tin particles were typically on
the order of 100–400 nm20,22,23 in size or larger.23
Tin oxides can be generated using a variety of synthesis
techniques including sol–gel processing,15,18,19 chemical vapor
deposition,24 sputtering methods,12 flame synthesis,10,25,26 gas-
phase condensation27 and mechanochemical processing.28 Pure tin
nanopowders with particles ϳ100 nm or less in size can be
produced by reductive precipitation.23 Commercial production of
SnO2 powders typically uses a sol–gel technique. Sol–gel pro-
cessing creates agglomerated particles with irregular particle
morphologies and can lead to chlorine contamination of the
powders because of the chloride particle precursors used (e.g.,
SnCl4 and PtCl4, etc.).4 Here, we introduce a new synthesis
method based on combustion of tetramethyltin (TMT, Sn(CH3)4)
that can be used to generate crystalline nanoparticles of any of
these three important materials: Sn, SnO, or SnO2.
I. Introduction
IN DIOXIDE (SnO2) is an industrially important material which is
used in numerous applications where the specific electrical,
T
optical, and mechanical properties of SnO2 are highly desirable.
For example, SnO2 is used extensively as the active material in gas
sensors,1–7 as well as in systems where optical or electrical
coatings are required, such as in low-emissivity architectural
glass,8 solar cells,9–11 liquid crystal displays,10,11 photodetec-
tors,6,11 and video touch screens,10 to name a few. Tin monoxide
(SnO) also has attractive optical and electrical properties and can
be found in gas-sensing applications12 and as anodes in lithium-ion
cells.13
Combustion synthesis has several advantages in comparison to
other nanosized particle synthesis methods. Combustion synthesis
is a method that can be used to assemble sensor materials at the
molecular level; thus, a large range of morphologies and compo-
sitions can be produced in one step, using one technique.29 The
technique is a continuous process as opposed to a batch process,
such as sol–gel methods. The high-temperature environment is
self-purifying, leading to materials with very low levels of con-
tamination. High conversion efficiencies (i.e., reactant-to-product
conversions of Ͼ90%) are readily achievable, leading to less
process waste and lower pollutant emissions. Production rates can
be high (Ͼ10 g/h) using laboratory-scale facilities, and combustion
synthesis methods are often scalable to commercial production
rates. Energy costs associated with combustion synthesis methods
are also low due to the exothermic nature of the reactions.
The primary objective of the current study is to demonstrate that
a robust range of material compositions and microstructures can be
achieved using a single synthesis method. The approach uses the
distinctive properties of a multielement diffusion burner to control
the combustion characteristics known to be the most important in
affecting particle properties—namely, the reactant concentrations
and the combined temperature field and particle residence
time.29–31 In the following sections, the synthesis methodology is
described and the results of the materials characterization studies
are presented. In Discussion, the results are interpreted using
combustion modeling.
The performance of tin oxides is directly related to the particle
size and compositional characteristics and consequently a strong
function of the synthesis process used to generate the materi-
als.3,12,14 For example, SnO2 sensor performance (e.g., stability,
sensitivity, and selectivity, etc.) has been improved considerably
by reducing the size of the SnO2 particles used in the sen-
sors6,14–16 and/or by adding dopants (typically noble metals or
other metal oxides) to the tin dioxide.3,5,6,14,17–19 Specifically,
several researchers have shown that pure SnO2 gas-sensor perfor-
mance is enhanced by using nanosized particles where the particle
size is less than 10 nm.14,15 Doped SnO2 also exhibits improved
sensor sensitivity as the crystallite size is decreased; however, at
comparably larger crystallite sizes to pure SnO2, e.g., ϳ15–50
nm.15,16 Similarly, tin monoxide performance in power cell anodes
is enhanced considerably by reducing the characteristic particle
size.13
Z. A. Munir—contributing editor
Manuscript No. 10285. Received June 5, 2003; approved May 25, 2004.
Based in part on the thesis submitted by D.L.H. for the M.S. Degree in mechanical
engineering, University of Michigan, Ann Arbor, MI, 2001.
†Author to whom correspondence should be addressed. e-mail: mswool@umich.
edu.
2033

2034
Journal of the American Ceramic Society—Hall et al.
II. Experimental Procedure
Vol. 87, No. 11
active area of the burner is surrounded by a shroud coflow of
nitrogen (0.635 cm wide). A square optical chimney (3.8 cm ϫ 3.8
cm ϫ 34 cm) can also be used to extend the region of controlled
reactant conditions above the burner. Particle precursor reactants
can either be introduced with the fuel for the primary flame, or the
precursor reactants can be injected into the center of the burner via
the secondary fuel tube (see Figs. 1 and 2). In the current study, the
particles were generated exclusively using the secondary fuel tube.
As indicated in Fig. 1, two flame systems are used with this
synthesis method: the primary flame and the secondary flame. The
fuel and oxidizer flows to the primary flame region (see Fig. 2) are
independent, and no mixing of the reactants occurs within the body
of the burner. Instead, mixing of the primary flame reactants
occurs rapidly near the surface of the burner, and a nominally
The materials were generated using a multielement diffusion
flame burner (MEDB), shown schematically in Figs. 1 and 2.
Details of the burner are described in Wooldridge et al.,32 and
details of the TMT delivery system are provided in Hall et al.33
Briefly, the 2.54-cm ϫ 2.54-cm square burner consists of a
hastalloy honeycomb support through which 173 stainless steel
hypodermic needles (0.51-mm i.d.) are inserted at systematic
intervals (see Fig. 2). Fuel flows through the hypodermic needles,
and oxidizer flows through the remainder of the channels of the
honeycomb (channel i.d., 0.81 mm). A single independent fuel
tube, the secondary fuel tube (0.51-mm i.d.), is located at the
center of the burner. To minimize entrainment of room air, the
Fig. 1. Experimental schematic and photograph insert of the combustion synthesis facility used to generate the SnOx particles. Note the H2/O2/Ar primary
flame does not emit visible radiation for the conditions studied, and the primary flame is therefore not apparent in the photograph insert.

November 2004
Nanocrystalline Tin and Tin Oxide Particles
2035
precursor reactant. TMT (Alfa Aesar, 98%) vapor/Ar mixtures
were generated by bubbling argon through a reservoir of liquid
TMT maintained at room temperature. Vapor pressure measure-
ments were made to confirm the argon was saturated in TMT
vapor. The results indicated saturated mixtures of 21%–23% TMT
dilute in Ar (mole basis) were delivered to the secondary fuel tube
for all conditions studied, regardless of the argon flow rate. All
reactant flow rates were monitored using calibrated rotameters or
calibrated flow meters. The burner was operated at atmospheric
pressure conditions (P Х 101 kPa) for all experiments.
Microscopic and bulk properties of the particles were analyzed
using X-ray diffraction (XRD, Rigaku rotating anode X-ray
diffractometer), transmission electron microscopy (TEM, JEOL
4000EX), and surface adsorption (BET, N2 adsorption, Micromer-
itics ASAP 2010 accelerated surface area and porosimetry system).
Bulk samples were obtained by deposition of the particles onto
sampling coupons mounted on a cold plate placed in the exhaust
region of the burner. Sampling locations for the cold plate were all
greater than 20 cm above the surface of the burner and typically
were obtained for sample heights of Hs ϭ 34 cm. Sampling times
to obtain sufficient material for bulk analysis by XRD or BET
analysis were typically 5–10 min. Samples for TEM analysis were
obtained by rapidly inserting a probe into the secondary flame and
depositing particles directly onto the TEM grids (carbon film on
copper grids, Electron Microscopy Science, CF300-C450). The
TEM sampling times were all less than 1 s. Additional details of
the TEM sampling system are provided in Wooldridge et al.32
Fig. 2. Top view schematic of the multielement diffusion burner dem-
onstrating the arrangement of the primary and secondary fuel tubes and
oxidizer channels (not to scale). Critical burner dimensions are provided in
the text.
uniform one-dimensional flame sheet, the primary flame, is
formed (except in the immediate vicinity of the secondary fuel
tube). When the particle precursor reactants are directed through
the secondary fuel tube, a lifted diffusion flame, the secondary
flame, is created. The primary flame has characteristics of a
premixed flame for heights above the mixing region (Ͼ3–5 mm).
The secondary flame is a lifted, a laminar diffusion flame (with a
lift-off distance of 5 mm from the surface of the burner), where the
oxidizing species are provided from the primary flame system. No
particles are produced from the primary flame. All particles are
produced as products of the secondary flame. The temperature
profiles of the primary flame have been measured and are
well-characterized by 1D “top-hat” profiles.32 The temperature,
velocity, and composition gradients of the secondary flame have
not been measured due to difficulties associated with the adverse
combustion conditions (high temperatures, high velocities, cata-
lytic reaction, and particle accumulation effects) and due to
difficulties with interrogating the small dimensions involved (the
luminous region of the secondary flame is typically less than 6 mm
in diameter).
III. Results
(1) Composition
Table I is a summary of the experimental conditions examined
and the results obtained for powder composition and material
production rates. The effects of three important synthesis param-
eters were considered: (1) the local level of oxygen-containing
species available for reaction with the Sn(CH3)4, (2) the temper-
ature of the primary flame, and (3) the particle residence time at
high temperatures. The oxygen species concentrations were con-
trolled by changing the equivalence ratio (␾, the quotient of the
actual fuel-to-oxygen ratio of the reactants divided by the stoichi-
ometric fuel-to-oxygen ratio) of the primary flame. The TMT of
the secondary flame was not considered in the determination of the
equivalence ratio of the primary flame. The temperature of the
primary flame was examined by changing the dilution level of the
primary flame reactants with argon, while maintaining a constant
equivalence ratio. Particle residence time/temperature effects were
examined by sampling materials at various distances from the
burner surface, by extending the regions of high temperatures via
In the current work, hydrogen (H2, Cryogenic Gases, 99.99%)
and oxygen (O2, Cryogenic Gases, 99.995%) diluted in argon (Ar,
Cryogenic Gases, 99.998%) were used to generate the primary
flame. Nitrogen (N2, Cryogenic Gases, 99.998%) was used as the
shroud gas. TMT vapor diluted in argon was used as the particle
Table I. Experimental Conditions and Results
Measured
production
rate (mg/s)
Predicted
production
rate (mg/s)
Chimney
present
Powder
color§
†
‡
†
Case
␾
H2 (L minϪ1
)
O2 (L minϪ1
)
Ar (L minϪ1
)
TMT/Ar (mL minϪ1
)
Hs (cm)
Tad (K)
1
0.75
2.94
2.48
2.94
2.96
2.94
2.94
2.94
5.33
2.96
2.98
2.98
3.49
3.50
3.58
3.00
1.95
1.38
1.48
1.50
1.48
1.48
1.48
2.61
1.44
1.26
1.26
1.47
1.47
1.48
1.18
24.44
17.25
11.70
18.82
18.49
18.49
18.49
14.85
18.97
17.67
12.72
18.33
18.33
18.49
13.91
62.6
395
34
3.2/5.2¶
34
Yes
No
0.56
–
W
1.42
8.93
1.45
6.05
1.82
1.87
1.87
8.93
1.96
0.69
1.45
1.38
1.38
1.38
1.87
1410
1590
2160
1690
1700
1700
1700
2430
1650
1560
1880
1670
1670
1660
1700
2¶ 0.90
–
3
4
5
6
7
0.99
0.99
1.00
1.00
1.00
64.3
268
Yes
No
–
G/Y
W
25
34
34
34
–
80.7
82.7
82.7
395
86.6
30.3
64.3
60.9
60.9
60.9
82.7
Yes
Yes
Yes
No
0.30
0.23
0.13
–
W
W
W
8¶ 1.02
5.2¶
–
9
10
11
12
13
14
15
1.03
1.18
1.18
1.19
1.20
1.21
1.27
21
34
34
34
34
30
21
No
0.03
0.33
0.29
–
–
0.04
0.13
W
Yes
Yes
Yes
Yes
No
G/Y
G/Y
G/Y
G/Y
W
No
G/Y
†The equivalence ratios and the adiabatic flame temperatures are based on the reactants of the primary flame and do not include fuel or argon concentrations from the secondary
flame. ‡Hs ϭ sample height. §W ϭ white; G ϭ gray; Y ϭ yellow/brown. ¶Conditions used to obtain TEM samples from the secondary flame. The sampling heights were variable;
however, Hs was less than 6 cm for all TEM samples.

2036
Journal of the American Ceramic Society—Hall et al.
Vol. 87, No. 11
the chimney, and by changing the flow rates of the TMT/Ar and
H2/O2/Ar gases.
Powder composition was determined by comparison of the
XRD patterns of the powder samples with accepted standards for
SnO2, SnO, and Sn. Figure 3 shows representative XRD patterns
obtained in the current study for each material. The SnO2 peaks all
indexed to the tetragonal crystallographic phase of ␣-SnO2 or the
cassiterite form of tin dioxide. The SnO peaks indexed to the
romarchite form of tin monoxide, and the Sn peaks indexed to
metallic tin. Throughout the study, all XRD analyses of white
powder samples indexed to SnO2, yellow/brown powder samples
indexed to SnO, and gray powder samples indexed to Sn. There-
fore, the sample color (listed Table I) was considered a good
indication of the material composition.
Fig. 4. Schematics indicating the compositional distribution on the
sample coupon as a function of equivalence ratio (for moderate tempera-
tures). Note the schematics are indications of the composition of the bulk
materials and are not meant to represent composition of individual
nanoparticles.
As indicated by the results shown in Table I, the most important
parameters in determining the composition of the product powders
were the equivalence ratio and temperature of the primary flame.
For near-stoichiometric or fuel-lean operation (␾ Յ 1.0) at
moderate temperatures (temperature effects on composition are
presented below), SnO2 was formed. For fuel-rich conditions (␾ Ͼ
1.0), both SnO and Sn were formed. Figure 4 is a schematic
representing the results of the macroscopic composition of the
powders as a function of equivalence ratio. (Note that the figure is
not meant to depict the composition of individual nanoparticles.)
Fig. 4 also shows the variation in composition of the bulk materials
as a function of location on the sample coupon. Approximate
dimensions are provided for reference on the figure. Due to the
geometry of the secondary flame (i.e., the radial symmetry), the
powder depositions were nonuniform in composition for the
fuel-rich conditions, with lower levels of tin oxidation found near
the center of the samples. Because it was difficult to isolate the Sn
from the SnO samples and vice versa (due to the lack of a distinct
transition from Sn to SnO on the sample coupons (see Fig. 4) and
due to aerosolization of the powders on removal from the sample
coupon), some Sn and SnO XRD patterns indicated small peaks
attributable to the alternative material. No compositional variation
was observed with the ␾ Յ 1.0 samples obtained at moderate
temperatures, and no Sn or SnO peaks were identified in the SnO2
XRD patterns.
To further examine temperature/residence time effects indepen-
dent of equivalence ratio effects, experiments were performed
where ␾ of the primary flame was kept constant and the argon
dilution of the primary flame was altered. Reducing the argon
dilution of the primary flame increases the temperature of the
primary flame, while potentially simultaneously decreasing the
residence time of the SnO2 particles formed in the secondary flame
(due to mixing of the primary and secondary flame gases near the
surface of the burner). Increasing the temperature of the primary
flame leads to added decomposition of the products of the primary
flame.
Based on previous characterization studies of the MEDB
including in situ temperature measurements using laser absorption
spectroscopy and thermocouple measurements,32,34 the maximum
temperature of the primary flame, Tad, can be estimated using
adiabatic equilibrium calculations and the known equivalence ratio
and level of argon dilution of the primary flame (for calculation
Fig. 3. Typical X-ray diffraction patterns of as-produced SnO2, SnO, and Sn powders. The experimental conditions for each sample were as follow: SnO2,
␾ ϭ 1.03, TMT/Ar ϭ 86.6 mL minϪ1, no chimney, Hs ϭ 21 cm; SnO, ␾ ϭ 1.19, TMT/Ar ϭ 60.9 mL minϪ1, chimney, Hs ϭ 34 cm; Sn, ␾ ϭ 1.27, TMT/Ar ϭ
82.7 mL minϪ1, no chimney, Hs ϭ 21 cm.

November 2004
Nanocrystalline Tin and Tin Oxide Particles
2037
details, please see Donovan et al.34). The calculated values for Tad
are provided in Table I as an indication of the magnitude of the
changes in the primary flame temperatures examined in the study.
For reference, a 37% change in the argon flow rate to the primary
flame leads to a 27% change in the predicted adiabatic flame
temperature.
Similar to the equivalence ratio, the temperature of the primary
flame was found to be a significant factor affecting the material
composition. Higher temperatures (lower argon concentrations)
led to formation of Sn and SnO at ␾ ϭ 0.99, a condition where
previously at lower primary flame temperatures only SnO2 was
produced. The change in the powder composition as the temper-
ature of the primary flame is increased is consistent with model
predictions, which are discussed below.
Changing the TMT/Ar flow rate, varying the sampling location,
and/or using the chimney to affect particle residence times at high
temperatures, for all cases where ␾ Յ 1.0 and the dilution of the
primary flame was moderate (i.e., primary flame temperatures
were moderate), had no effect on the material composition. SnO2
was produced in all cases (see Table I). For conditions where ␾ Ͼ
1.0 or conditions with low dilution of the primary flame reactants
(i.e., primary flame temperatures were high), if the Sn or SnO
samples were not obtained before significant mixing with room air
occurred, the Sn and SnO would further oxidize rapidly to form
SnO2. For example, comparing cases 14 and 15 listed in Table I
(which consider approximately the same equivalence ratio, ␾ Х
1.2–1.3, and TMT/Ar flow rate, both without the use of the
chimney) showed that sampling 9 cm higher in the exhaust region
provided the particles with sufficient time at high temperatures to
lead to complete oxidation of the tin. Collectively, the study of
particle residence time effects indicated oxidation of the tin species
occurred rapidly provided sufficient oxygen was available locally
for reaction.
nature of the particles is clearly evident (see Fig. 6), and the
different lattice fringe orientations indicate sintering was relatively
slow between particles.
Particle size distributions were obtained for selected TEM
samples. Results for a typical fractional population for synthesis of
SnO2 nanoparticles are shown in Fig. 7. The particle size statistics
were obtained using the methods described in Wooldridge et al.;32
however, a larger number of particles were counted in the current
work (N Ͼ 800), to ensure more accurate particle statistics
៮
(geometric mean particle diameters, dp, were changing by less than
3% with the incorporation of additional particle count data). In
general, the geometric mean particle diameters of primary particles
sampled from the secondary flame were less than 12 nm, and
particle aggregates were less than 100 nm in size (maximum chord
length). The particle size distributions were relatively well repre-
sented by lognormal population distributions (R2 ϭ 0.9998; see
Fig. 7), and there was little variation in the particle size statistics
as a function of height above the burner for the sample locations
examined via TEM imaging (Hs ϭ 32, 52 mm).
Specific surface area (SSA) measurements of bulk samples
obtained in the exhaust region of the burner were used to estimate
the average particle size of the final product materials. Values for
SSA for ␾ ϭ 0.75, ␾ ϭ 1.00 (TMT/Ar ϭ 80.7 mL minϪ1), and
␾ ϭ 1.19 were 37.6, 36.2, and 15.0 m2/g, respectively. Although
the particles are not spherical (as seen in Fig. 6), to a first
approximation, the average particle diameter via SSA can be
determined using
6
៮
d
p,SSA ϭ
(1)
SSA␳
͑SnO or Sn͒
2
Based on the compositional analysis described above, for ␾ Յ 1.0
with moderate primary flame temperatures, the density for SnO2
(6.85 g/cm3)35 is used in Eq. (1). For high-temperature conditions
and/or ␾ Ͼ 1.0, the density for Sn (5.77 g/cm3)35 is used. For the
(2) Microstructure
៮
SSA values provided above, the values for dp,SSA are 23.3 nm
The effects of burner operating conditions on particle size and
morphology characteristics were also investigated. For all condi-
tions examined in the study, the powders consisted of both discrete
nanosized primary particles and particle aggregates. Figures 5 and
6 are low- and high-magnification micrograph images, respec-
tively, of the typical particle structures observed. The crystalline
(␾ ϭ 0.75), 24.2 nm (␾ ϭ 1.00), and 69.4 nm (␾ ϭ 1.19). The
larger dimension for the Sn particles compared with the SnO2
particles is consistent with the lower melting point of Sn compared
with SnO2 (Tmp,Sn ϭ 231.9°C,35 Tmp,SnO ϭ 1630°C 35). The Sn
2
particles will sinter more rapidly both in the secondary flame and
on the collection plate, leading to more coarse particles.
Fig. 5. Low-magnification (10K) transmission electron micrograph of
particles sampled from the H2/O2/Sn(CH3)4/Ar flame system: ␾ ϭ 0.90;
TMT/Ar ϭ 395 mL minϪ1; Hs ϭ 5.2 cm; no chimney. Size bar ϭ 500 nm.
Fig. 6. High-magnification (500K) transmission electron micrograph of
particles sampled from the H2/O2/Sn(CH3)4/Ar flame system: ␾ ϭ 1.02;
TMT/Ar ϭ 395 mL minϪ1; Hs ϭ 5.2 cm; no chimney. Size bar ϭ 10 nm.

2038
Journal of the American Ceramic Society—Hall et al.
Vol. 87, No. 11
the TEM results represent an approximate lower limit (assuming
the particles could be rapidly quenched without the occurrence of
significant particle restructuring).
(3) Mass Production
Mass production rates were measured for several samples to
provide an indication of the potential manufacturing rates achievable
(see Table I). The corresponding maximum predicted production rates
based on the known TMT/Ar flow rates and saturation levels
(assuming 100% conversion efficiency) were also determined (see
Table I). All the experimental values were significantly below the
predicted values. However, the differences were not likely due to low
reactant-to-product conversion efficiencies, as much as due to the low
collection efficiency of the bulk sampling method used. Material
could bypass the cold plate and/or deposit on the cold plate outside the
sample coupon region. Although the particle production rates were
low compared with commercial powder production methods, the
MEDB synthesis technique was not optimized for maximum produc-
tion rates. In addition, the synthesis approach can be scaled to increase
production rates by increasing the x–y dimension of the burner and
incorporating additional secondary fuel tubes into the honeycomb
matrix of the primary flame. Although the optimal spacing of the
additional secondary fuel tubes would require additional study (to
identify potential interactions between the particle-producing flames),
preliminary results indicate a spacing of 1.25 cm between secondary
fuel tubes would be sufficient.
Fig. 7. Fractional size distribution of particles obtained from TEM
images of samples taken at ␾ ϭ 0.90, TMT/Ar ϭ 395 mL minϪ1, and no
chimney.
The particle sizes determined using BET are larger than those
determined using TEM imaging; which is consistent with the
respective sampling conditions. The BET and TEM particle sizes
were obtained at dramatically different particle sampling times and
sampling locations. The BET samples, which yielded larger
IV. Discussion
៮
particle sizes of dp,SSA ϭ 23–24 nm for SnO2, were obtained from
The results for composition and particle morphology are inter-
preted using chemical kinetic and equilibrium models of the synthesis
environment. Previously, we had obtained good agreement between
one-dimensional premixed modeling of the primary flame (not in-
cluding the effects of the secondary flame) and measured radical
profiles by incorporating detailed chemical kinetics in the model
simulations.34 While the 1D premixed model does not accurately
describe the mixing conditions in the first few millimeters above the
burner surface, the results can be used to infer trends about the
conditions of the primary flame system above (i.e., downstream of)
the initial mixing region. Figure 8 shows the model results for the
minor and major species of the primary flame for a typical stoichio-
metric condition. The temperature profile used in the model is a
polynomial fit to a measured temperature profile of the primary flame
system when the secondary fuel tube was not used for particle
the exhaust region of the burner (i.e., longer residence times) using
comparatively long sampling times (minutes). The TEM samples,
៮
which yielded smaller particle sizes of dp,TEM ϭ 10–12 nm for
SnO2, were obtained directly from the secondary flame using very
short sampling times (Ͻ1 s).
Although particle sizes determined by SSA are inherently larger
than those determined by TEM imaging when necking and
bridging occurs between primary particles,36 a comparison of the
TEM and BET results for SnO2 indicates that either significant
particle growth is occurring in the secondary diffusion flame or
significant particle restructuring is occurring on the sampling
coupon on the cold plate. Therefore, the BET results for particle
size represent an upper limit to the dimensions of the particles
produced using the operating conditions presented in Table I, and
Fig. 8. Calculated species profiles for the primary flame system for typical stoichiometric (␾ ϭ 1.0) H2/O2/Ar combustion conditions. A sketch of the
secondary flame is superimposed on the computational results for reference.

November 2004
Nanocrystalline Tin and Tin Oxide Particles
2039
Fig. 9. Equilibrium calculation results of the distribution of Sn in the combustion products of H2/O2/Sn(CH3)4/Ar flames for Tad Х 1000 K. The results
are based on constant reactant values of 1 mol of Sn(CH3)4 and 1 mol of H2.
synthesis. Additional details of the model, including the chemical
reaction mechanism used, are provided in Donovan et al.34 Again, the
chemical kinetics model does not include the effects of the secondary
flame, including any chemistry associated with the TMT. As seen in
Fig. 8, O2 is rapidly consumed within 0.5 cm of the surface of the
burner by the H2 in the primary flame.
When both the primary and the secondary flame systems are in use,
the secondary flame forms a lifted diffusion flame where the base of
the luminous visible portion of the secondary flame is located 5 mm
above the surface of the burner. For reference, a schematic of the
secondary flame is superimposed on the results in Fig. 8. On exiting
the secondary fuel tube, the TMT/Ar gases mix with the O2 from the
surrounding oxidizer channels. However, at typical operating condi-
tions, including ␾ ϭ 1.0, there is insufficient O2 in the immediate
vicinity of the secondary fuel tube (i.e., the surrounding six oxidizer
channels) to completely oxidize the Sn(CH3)4 to complete products of
combustion, i.e., SnO2, CO2, and H2O. Therefore, a potentially
significant portion of the TMT and/or Sn and SnO particles (produced
in the secondary flame) are likely oxidized by reaction with H2O—the
most abundant oxygen-containing species present in the coannular
region surrounding the secondary flame.
calculations by the increasing or decreasing the Ar concentration
in the reactants (y in Reaction 2). The reactants, major products of
combustion (CO, CO2, H2O), some minor products of combustion
(H, OH, HO2, O), and vapor and condensed phases for Sn, SnO,
and SnO2 are considered in the calculations. The thermodynamic
data are taken primarily from Cox et al.37 The thermochemical
data for TMT are from Allendorf.38
Figure 9 shows that SnO2(l) is the preferred form of tin for low
equivalence ratios, whereas at fuel-rich conditions, Sn(l) is pre-
ferred. There is a small window in ␾ where SnO(l) is present, but
never in quantities above 30% (mole basis) of the initial 1 mol of
TMT. However, as the temperature is increased, Fig. 10 shows that
SnO(g) is the preferred form of tin for a large range of equivalence
ratios. Although the calculations are not directly representative of
the synthesis system studied, the trends are in excellent qualitative
agreement with the observed dependences on T and ␾. At higher
temperatures and higher fuel concentrations, the O atoms are
consumed preferentially by carbon, yielding Sn and SnO products.
Conversely, at lower temperatures and fuel-lean conditions, 100%
of the TMT is converted to SnO2.
We are not aware of any previous flame synthesis studies of Sn
or SnO; however, the results of the current investigation can be
compared on a qualitative basis with other combustion synthesis
methods used to produce SnO2 nanoparticles. It is difficult to
compare results on a quantitative basis, due to the limited data
available on particle residence time and temperature. Lindackers et
al. examined SnO2 synthesis using low-pressure (37.5 mbar),
premixed TMT/H2/O2/Ar flat flames.10 The final product powders
indexed to the cassiterite phase of SnO2; however, an intermediate
metastable phase (a mixture of the high-temperature ␤-SnO2 phase
and the ␣-SnO phase) was also identified in the study. The authors
found the powders consisted of primary particles and agglomerates
with specific surface areas ranging from ϳ70 to 160 m2/g
(corresponding to average particle sizes ranging from ϳ5.5 to 13
nm). Higher TMT loadings led to larger particle sizes, and the
highest TMT loading presented in the work was 1348 ppm. The
SnO2 particles produced using the MEDB were larger than those
made by Lindackers et al., which is consistent with the atmo-
spheric pressures and the higher TMT loadings (21%–23% mole
basis TMT, balance Ar) used in the current work.
Equilibrium calculations which include tetramethyltin and tin
species provide additional insight into the chemistry of the
TMT/H2/O2/Ar system and aid in interpretation of the observed
temperature and equivalence ratio dependence to the SnOx com-
positions. Figures 9 and 10 present the moles of product tin species
as a function of the equivalence ratio for Tad ϭ 1000 and 2000 K,
respectively. Here, the equivalence ratio is determined using both
H2 and TMT as fuels according to the global reaction
Sn͑CH3͒4 ϩ H2 ϩ xO2 ϩ yAr 3 Products
(2)
At stoichiometric conditions, all tin is presumed to form SnO2, all
hydrogen to form H2O, and all carbon to form CO2. To separate
temperature effects from fuel-to-oxidizer ratio effects, the results
of Figs. 9 and 10 are presented for approximately constant
temperatures. For all calculations, the fuel loading is maintained at
1 mol of TMT and 1 mol of H2, the reactants are at 1 atm and 298
K, and the reaction is considered adiabatic. The equivalence ratio
is changed by increasing or decreasing the O2 concentration (x in
Reaction 2). The temperature is maintained constant in the

2040
Journal of the American Ceramic Society—Hall et al.
Vol. 87, No. 11
Fig. 10. Equilibrium calculation results of the distribution of Sn in the combustion products of H2/O2/Sn(CH3)4/Ar flames for Tad Х 2000 K. The results
are based on constant reactant values of 1 mol of Sn(CH3)4 and 1 mol of H2.
Go¨pel, T. A. Jones, M. Kleitz, I. Lundstro¨m, and T. Seiyama. VCH, Weinheim,
Germany, 1992.
spheric pressure coflow diffusion flames25 and premixed burner-
2K. Ihokura and J. Watson, The Stannic Oxide Gas Sensor—Principles and
Pratsinis and co-workers studied SnO2 synthesis using atmo-
stabilized and premixed burner-nonstabilized flames.39 For all
Applications. CRC Press, Boca Raton, FL, 1994.
3W. Go¨pel and K. D. Schierbaum, “SnO2 Sensors: Current Status and Future
Prospects,” Sens. Actuators B, 26–27, 1–12 (1995).
systems, the SnO2 particles indexed to cassiterite and were larger
than those produced in the current work, with particle sizes
determined from specific surface areas ranging from ϳ52 to 164
nm for the diffusion flames and from ϳ36 to 65 nm for the
premixed flames.
4T. Brousse and D. M. Schleich, “Sprayed and Thermally Evaporated SnO2 Thin
Films for Ethanol Sensors,” Sens. Actuators B, 31, 77–79 (1996).
5E. Kanazawa, M. Kugishima, K. Shimanoe, Y. Kanmura, Y. Teraoka, N. Miura,
and N. Yamazoe, “Mixed Potential Type N2O Sensor Using Stabilized Zirconia- and
SnO2-Based Sensing Electrode,” Sens. Actuators B, 75, 121–24 (2001).
6F. E. Krius, H. Fissan, and A. Peled, “Synthesis of Nanoparticles in the Gas Phase
for Electronic, Optical and Magnetic Applications—A Review,” J. Aerosol Sci., 29
[5–6] 511–35 (1998).
V. Summary and Conclusions
7C. Nayral, T. Ould-Ely, A. Maisonnat, B. Chaudret, P. Fau, L. Lescouze`res, and
A. Peyre-Lavigne, “A Novel Mechanism for the Synthesis of Tin/Tin Oxide
Nanoparticles of Low Size Dispersion and of Nanostructured SnO2 for the Sensitive
Layers of Gas Sensors,” Adv. Mater., 11 [1] 61–63 (1999).
A multielement diffusion burner was successfully used to
produce nanocrystalline Sn, SnO, and SnO2 particles. This is the
first demonstration of combustion synthesis of pure metal Sn
nanoparticles of which we are aware. The most significant factors
controlling particle composition were the concentration of oxygen-
containing species locally available for oxidation (primarily O2
and H2O) and the temperature of the combustion system. The
equilibrium modeling results indicate that the primary flame
temperature and equivalence ratio may be used to optimize the
synthesis strategy for isolated production of Sn, SnO, or SnO2 with
high reactant-to-product conversion efficiencies (e.g., 1 mol of
TMT producing ϳ1 mol of Sn and little to no SnO or SnO2). The
average particle dimensions (based on specific surface area mea-
surements) were 23–24 nm for the SnO2 powders produced and 69
nm for the Sn powders produced. The larger Sn particle size is
consistent with the more rapid sintering rates of Sn. Note that the
Sn particle sizes produced here are substantially smaller than Sn
particles made using alternative nanoparticle synthesis methods.
8M. D. Allendorf, “On-Line Deposition of Oxides on Flat Glass,” Electrochem.
Soc. Interface, 10, 34–38 (2001).
9C. J. Giunta, D. A. Strickler, and R. G. Gordon, “Kinetic Modeling of the
Chemical Vapor Deposition of Tin Oxide from Dimethyltin Dichloride and Oxygen,”
J. Phys. Chem., 97, 2275–83 (1993).
10D. Lindackers, C. Janzen, B. Rellinghaus, E. F. Wassermann, and P. Roth, “Synthesis
of Al2O3 and SnO2 Particles by Oxidation of Metalorganic Precursors in Premixed
H2/O2/Ar Low Pressure Flames,” Nanostruct. Mater., 10 [8] 1247–70 (1998).
11K. C. Song and J. H. Kim, “Synthesis of High Surface Area Tin Oxide Powders
via Water-in-Oil Microemulsions,” Powder Technol., 107, 268–72 (2000).
12J. Calderer, P. Molina`s, J. Sueiras, E. Llobet, X. Vilanova, X. Correig, F. Masana,
and A. Rodr´ıguez, “Synthesis and Characterisation of Metal Suboxides for Gas
Sensors,” Microelectron. Reliab., 40, 807–10 (2000).
13J. S. Sakamoto, C. K. Huang, S. Surampudi, M. Smart, and J. Wolfenstine, “The
Effects of Particle Size on SnO Electrode Performance in Lithium-Ion Cells,” Mater.
Lett., 33, 327–29 (1998).
14N. Yamazoe, “New Approaches for Improving Semiconductor Gas Sensors,”
Sens. Actuators B, 5, 7–19 (1991).
15C. Xu, J. Tamaki, N. Miura, and N. Yamazoe, “Grain Size Effects on Gas
Sensitivity of Porous SnO2-Based Elements,” Sens. Actuators B, 3, 147–55 (1991).
16G. Zhang and M. Liu, “Effect of Particle Size and Dopant on Properties of
SnO2-Based Gas Sensors,” Sens. Actuators B, 69, 144–52 (2000).
17I. Sayago, J. Gutie´rrez, L. Are´s, J. I. Robla, M. C. Horrillo, J. Getino, J. Rino, and
J. A. Agapito, “The Effect of Additives in Tin Oxide on the Sensitivity and Selectivity
to NOx and CO,” Sens. Actuators B, 26–27, 19–23 (1995).
Acknowledgments
The authors thank Dr. Mark D. Allendorf for providing the thermochemical data
for Sn(CH3)4 and Mr. Michael T. Donovan for providing the chemical kinetics
modeling results of the primary flame.
18P. Siciliano, “Preparation, Characterisation and Applications of Thin Films for
Gas Sensors Prepared by Cheap Chemical Method,” Sens. Actuators B, 70, 153–64
(2000).
19A. Cabot, J. Arbiol, J. R. Morante, U. Weimar, N. Ba៮rsan, and W. Go¨pel,
“Analysis of the Noble Metal Catalytic Additives Introduced by Impregnation of
As-Obtained SnO2 Sol–Gel Nanocrystals for Gas Sensors,” Sens. Actuators B, 70,
87–100 (2000).
References
1W. Go¨pel, “Sensors: A Comprehensive Survey”; Edited by W. Go¨pel, J. Hesse,
and J. N. Zemel. Chemical and Biochemical Sensors I, II; Vols. 2–3. Edited by W.

November 2004
Nanocrystalline Tin and Tin Oxide Particles
2041
20J. O. Besenhard, J. Yang, and M. Winter, “Will Advanced Lithium-Alloy Anodes
Have a Chance in Lithium-Ion Batteries?,” J. Power Sources, 68, 87–90 (1997).
21J. Santos-Pena, T. Brousse, and D.M. Schleich, “Search for Suitable Matrix for
the Use of Tin-Based Anodes in Lithium Ion Batteries,” Solid State Ionics, 135, 87–93
(2000).
30S. E. Pratsinis, “Flame Aerosol Synthesis of Ceramic Powders,” Prog. Energy
Combust. Sci., 24 [3] 197–219 (1998).
31M. S. Wooldridge, “Gas Phase Combustion Synthesis of Particles,” Prog. Energy
Combust. Sci., 24 [1] 63–87 (1998).
32M. S. Wooldridge, P. V. Torek, M. T. Donovan, D. L. Hall, T. A. Miller, T. R.
Palmer, and C. R. Schrock, “An Experimental Investigation of Gas-Phase Combus-
tion Synthesis of SiO2 Nanoparticles Using a Multi-Element Diffusion Flame
Burner,” Combust. Flame, 131 [5] 98–109 (2002).
22D. L. Foster, J. Wolfenstine, J. R. Read, and W. K. Behl, “Nanocomposites of Sn
and Li2O Formed from the Chemical Reduction of SnO as Negative Electrode
Material for Lithium-Ion Batteries,” Electrochem. Solid State Lett., 3 [5] 203–204
(2000).
33D. L. Hall, P. V. Torek, C. R. Schrock, T. R. Palmer, and M. S. Wooldridge,
“Gas-Phase Combustion Synthesis of Tin Oxide Nanoparticles,” Mater. Sci. Forum,
386–388, 347–52 (2002).
23J. Yang, M. Wachtler, M. Winter, and J. O. Besenhard, “Sub-Microcrystalline Sn
and Sn–SnSb Powders as Lithium Storage Materials for Lithium-Ion Batteries,”
Electrochem. Solid State Lett., 2 [4] 161–163 (1999).
34M. T. Donovan, D. L. Hall, P. V. Torek, C. R. Schrock, and M. S. Wooldridge,
“Demonstration of Temperature and OH Mole Fraction Diagnostic in SiH4/H2/O2/Ar
Flames Using Narrow Line UV OH Absorption Spectroscopy,” Proc. Combust. Inst.,
29, 2635–43 (2003).
24C. G. Borman and R. G. Gordon, “Reactive Pathways in the Chemical Vapor
Deposition of Tin Oxide Films by Tetramethyltin Oxidation,” J. Electrochem. Soc.,
136 [12] 3820–28 (1989).
25W. Zhu and S. E. Pratsinis, “Synthesis of SiO2 and SnO2 Particles in Diffusion
Flame Reactors,” Ceram. Process., 43 [11A] 2657–64 (1997).
26G. Skandan, N. Glumac, Y.-J. Chen, F. Cosandey, E. Heims, and B. H. Kear,
“Low-Pressure Flame Deposition of Nanostructured Oxide Films,” J. Am. Ceram.
Soc., 81 [10] 2753–56 (1998).
35D. R. Lide (Ed.), CRC Handbook of Chemistry and Physics. CRC Press, Boca
Raton, FL, 1997.
36O. I. Arabi-Katbi, S. E. Pratsinis, P. W. Morrison Jr., and C. M. Megaridis,
“Monitoring the Flame Synthesis of TiO2 Particles by In-Situ FTIR Spectroscopy and
Thermophoretic Sampling,” Combust. Flame, 124 [4] 560–72 (2001).
37J. D. Cox, D. D. Wagman, and V. A. Medvedev, Eds., CODATA Key Values for
Thermodynamics. Hemisphere Publishing Co., New York, 1989.
38M. D. Allendorf, personal communication, 2002.
27V. M. Jime´nez, A. R. Gonza´lez-Elipe, J. P. Espino´s, A. Justo, and A. Ferna´ndez,
“Synthesis of SnO and SnO2 Nanocrystalline Powders by the Gas Phase Condensa-
tion Method,” Sens. Actuators B, 31, 29–32 (1996).
28L. M. Cukrov, T. Tsuzuki, and P. G. McCormick, “SnO2 Nanoparticles Prepared
by Mechanochemical Processing,” Scr. Mater., 44 [8–9] 1787–90 (2001).
29K. Brezinsky, “Gas-Phase Combustion Synthesis of Materials,” Proc. Combust.
Inst., 26, 1805–16 (1996).
39S. Vemury, S. E. Pratsinis, and L. Kibbey, “Electrically Controlled Flame
Synthesis of Nanophase TiO2, SiO2, and SnO2 Powders,” J. Mater. Res., 12 [4]
1031–42 (1997).
Ⅺ