1
32
Y. Zhu et al. / Materials Research Bulletin 44 (2009) 130–133
travel towards the centre of the furnace (maximum temperature
550 8C), that is further intensified because of the strong
1
exothermic effect of the oxidation–combustion reaction of Si.
The superficially molten particles spontaneously obtain spherical
shape (stage ii) due to the effect of surface tension under a micro-
gravity field (the fluidized bed conditions [10] balance the effect of
gravity). Bigger particles (stage ii) are expected to be less spherical
than smaller ones (stage v) because the shorter the radius the
stronger the cohesive force (
place at the molten surface layer of Si-particles (stage iii). The
inevitably developed stresses at the SiO /Si interface, due to the
Dp = 2g/r) [11]. Oxidation easily takes
2
lattice misfit and the mismatch of thermal expansion coefficients
of the contacting phases, result in crack formation and propagation
in the surface but also in the bulk of the particles (Fig. 4a).
Therefore, the particles collapse in such a way as shown in the
stage (iv) of Fig. 4b.
Accordingly, the collapse of the particles is a stage of high
importance. The interface between the oxide layer and the core of
silicon at three different temperature ranges is sketched in Fig. 4c.
Qualitative earlier studies have suggested that a thin film of
crystalline SiO
2
forms at the interface between Si and vitreous SiO
2
[
12–14]. There is pronounced difference among the coefficients of
ꢂ6 ꢂ1
thermal expansion (CTE) of crystalline SiO
vitreous SiO
2
(15.6 ꢁ 10
K
K
) [15],
) [17].
ꢂ6 ꢂ1
ꢂ6 ꢂ1
2
(0.58 ꢁ 10
K
) [16], and Si (3.5 ꢁ 10
However, it has been postulated [18] that this particular interface
is stress-free between 975 8C and 1000 8C. This means that the
stress due to CTE mismatch balances the stress due to lattice misfit
within that narrow temperature range (i.e. 975–1000 8C). In other
words, that stress-free regime suggests a transition stage, with
regard to the type of the stresses dominating at the interface (i.e.
due to lattice misfit or CTE mismatch). In the light of the big
difference among the CTEs of the three phases (mentioned above),
it would be assumed that the stress at the interface at
temperatures higher than 1000 8C is due to CTE mismatch. The
lower value of the CTE of vitreous silica (than the other two phases
mentioned above) implies development of tensile stress at
temperatures higher than 1000 8C. The maximum misfit strain,
e, between 1000 8C (i.e. the upper limit of the stress-free regime)
and 1500 8C (i.e. the maximum temperature of the furnace) is
calculated as
Fig. 4. (a) SEM image of a superficially melted Si-sphere and a similar sphere locally
cracked at the surface (inset). (b) Schematic representation of the stages of the
process. (c) Regimes of stresses at the Si/SiO
2
interface for different temperature
2
is also anticipated to form at the
ranges (note that a thin film of crystalline SiO
interface).
non-reacted Si in the produced powder, suggesting imperfect
oxidation. However, it is worthy to note that the investigated
processing completely suppresses formation of crystallized
silica. The HRTEM image of Fig. 3c confirms the amorphous nature
of the produced silica since the observed spots, presumably
attributed to atomic level units, are randomly arranged. Moreover,
the SAED pattern (inset of Fig. 3c) indicates an amorphous nature
¼ ð15:6 ꢁ 10ꢂ6
ꢂ1
ꢂ 0:58 ꢁ 10ꢂ6
ꢂ1Þ500 K ¼ 0:751%
e
K
K
Assuming the stress-free regime (975–1000 8C) at the interface
(
since the patterns of crystallized nano-particles are broad but
of Fig. 4c as a transition stage and the development of tensile stress
at temperatures higher than 1000 8C, compressive stresses must be
developed at temperatures lower than 975 8C, predominantly
resulted from lattice misfit at the interface. Experimental results of
thermal oxidation of silicon substrates, used in MEMS and VLSI
processes, where stoichiometric dry oxidation took place at
temperatures above 900 8C [19], support fairly well our hypothesis
(the silica developed was also non-crystalline, similar to the silica
produced in the present study). Compressive stresses of ꢀ300 MPa
were measured at room temperature [19].
distinct rings).
To shed light in the stages of the process, we collected
intermediate products after short reaction (oxidation) time, i.e.
ꢀ
30 s. Fig. 4a shows a typical superficially oxidized Si micro-
sphere with size of about 10 m. We also observed many
m
spheres whose surface was extensively but locally cracked (inset
of Fig. 4a).
4
. Discussion
The stages (ii)–(iv) of Fig. 4b are repeated (i.e. stages v and vi)
until particle size is reduced down to nano-level. In our plant and
with the given (in this study) materials’ initial stage and
pretreatment as well as processing conditions, this process lasts
10 min. Then, rapid cooling (which is feasible because of the whole
experimental set up, based on fluidized bed concept) maintains the
vitreous nature of the resultant silica powder (stage vii).
According to the experimental results, the stages of the process
can be depicted in Fig. 4b. When the silicon particles (Fig. 2b) enter
the furnace, the organic binder immediately burns out (i.e. at
9
00 8C) and hence, the morphology of Si-particles must be close to
Fig. 2a (stage i of Fig. 4b). Bulk Si melts at 1412 8C. However, earlier
independent experiments (unpublished data, whose detail
description is beyond the scope of the present paper) have shown
that mechano-chemical activation causes reduction of superficial
melting temperature of Si-particles at 900 8C. Therefore, the
temperature at the entrance of the furnace was set at 900 8C. The
melting at particles’ surface layer expands because the particles
5. Conclusions
Floating combustion synthesis was successfully employed to
produce vitreous silica powder with spherical nano-particles. The