Formation of Magnetic Fe_xO_y/Silica Core-Shell Particles in a One-Step Flame Aerosol Process

Abstract

In this study, iron silicon oxide particles were generated in a one-step flame assisted spray pyrolysis (FASP) process using H₂/air or H₂/O₂ diffusion flames. A colloidal precursor solution was used, which contained dissolved iron nitrate and stably suspended silica nanoparticles. H₂/air flames resulted in magnetic Fe_xO_y/silica core-shell particles. There was a correlation between particle size and particle structure; particles larger than 500 nm had the core-shell structure, but smaller particles had non-core-shell structures. H₂/O₂ flames only resulted in nanoparticles that had non-core-shell structures. The core-shell particles had a iron oxide core that was hermetically enclosed in a uniform silica shell with a typical thickness of approximately 100 nm; they were superparamagnetic with a room-temperature saturation magnetization greater than 24 emu/g. Temperature history of the particles may be used to explain the correlation between flame type and particle structure. The correlation between particle size and structure may be due to size-dependent thermodynamic stability of the structures, or kinetics of heat and mass transfer. The results from this study suggest that micrometer sized iron oxide silica core-shell magnetic particles could be generated from a one-step flame aerosol process, but Fe_xO_y/silica nanoparticles (< 100 nm) with the core-shell structure cannot be generated in a one-step flame aerosol process.

INTRODUCTION

Magnetic Fe_xO_y/silica (Fe _x O _y = Fe₃O₄ or γ -Fe₂O₃) core-shell particles have attracted much research interest due to the many potential applications for such particles, for example, immobilization and separation of biomolecules, magnetically controlled drug delivery, and MRI imaging enhancement (Gupta and Hung 1989; Weissleder et al. 1990). There have been numerous publications on the synthesis of core-shell magnetic particles with an iron oxide core and a silica shell. Most of the reported synthesis methods for iron-oxide/silica core-shell magnetic particles were “wet” methods (Philipse et al. 1994), involving multi-step, batch-mode processes in a solution. An alternative to the “wet” methods is aerosol synthesis. Aerosol methods can offer continuous production of particle materials, and they are flexible in scale. Therefore, aerosol processes capable of generating core-shell magnetic Fe_xO_y/silica particles may offer potential advantage over the “wet” methods. In the past, aerosol methods based on flame and furnace have been used for the synthesis of iron oxide silica composite particles. For example, McMillin et al. and Ehrman et al. used Bunsen burner type premixed flame processes (Ehrman et al. 1999; McMillin et al. 1996), Li et al. (2006) used a flame spray pyrolysis approach to synthesize iron silicon oxide particles. The products from these processes were nanoparticles that had non-core-shell structures. That is, these nanoparticles were composed of segregated iron oxide and silica regions that did not have the Fe_xO_y-core/silica-shell structure.

Recently, a furnace-based aerosol process was reported to be capable of generating magnetic core-shell iron oxide silica nanoparticles (Basak et al. 2007). Most recently, a two-step approach based on a flame reactor was reported for the synthesis of Fe₂O₃/SiO₂ core-shell nanoparticles (Teleki et al. 2009). Teleki et al. used a flame spray pyrolysis (FSP) reactor to generate Fe₂O₃ nanoparticles; by introducing hexamethyldisiloxane vapor in the post-flame region, they were able to encapsulate the Fe₂O₃ nanoparticles with a very thin silica coating.

The motivation of this study was to answer the question: Is it possible at all to use a one-step flame process to generate Fe_xO_y/silica core-shell particles? Here a one-step flame aerosol process is defined as a process in which all the ingredients of the product particles are introduced into the flame reactor at a single inlet simultaneously with the fuel gas. Due to its relative simplicity, a one-step flame process apparently can offer some advantages over the furnace-based processes or multi-step flame processes, provided each process is capable of the same function.

As reviewed above, previous studies have shown that one-step flame processes in general do not produce Fe_xO_y/silica core-shell particles. However, this study was inspired by the hypothesis that a novel combination of flame aerosol methods might be capable of producing magnetic Fe_xO_y/silica core-shell particles. Flame assisted spray pyrolysis (FASP) using a colloidal solution was identified as a novel approach for synthesizing Fe_xO_y/silica particles. A colloidal solution is a solution that contains dissolved metal salt as well as stably suspended colloidal particles. Colloidal solutions had been previously used as starting materials in furnace and flame based aerosol synthesis processes, as shown in Table 1. To the best of our knowledge, prior to this work FASP with a colloid solution precursor had not been used for the synthesis of Fe_xO_y/silica particles.

TABLE 1 Previous studies using colloidal solutions in aerosol synthesis

Reference	Colloidal solution	Reactor	Products
(Moore et al. 1992)	SiO2 colloid Al(NO)3 water solution	Furnace	Mullite particles
(Lee et al. 1999)	TiO2 colloid metal nitrate water solutions	Furnace	ZnO-TiO2, ZnS-TiO2, CdS-TiO2 and Ag-TiO2 particles
(Vander Wal and Ticich 2001)	Ferro fluid	Flame (pyrolysis)	Carbon nanotubes (colloid particles as catalysts)
(Abdullah et al. 2004)	Polystyrene latex (PSL) colloid metal nitrate water solutions	Furnace (two steps)	Porous SiO2, TiO2, ZrO2, Al2O3 and Y2O3 particles
(Wang et al. 2007)	SiO2 colloid Gd(NO)3 and Eu(NO)3 solution	Furnace	Gd2O3:Eu particles

In this study, either air or O₂ was used as the oxidant stream to result in either a H₂/air flame or a H₂/O₂ flame for the FASP. Two different precursor concentrations were used while maintaining the Fe/Si ratio constant. Thermophoretic sampling was used to help understand the particle evolution in the flame. In the following sections descriptions of the experimental methods are given. The results are described and discussed, followed by the conclusions.

Experimental

The experimental methods are described below. The experimental conditions of the major tests are also summarized at the beginning of the Results section in Table 2.

TABLE 2 Experimental conditions, characterization methods, and major results (Fe/Si molar ratio = 1; MS: Mössbauer Spectroscopy; MM: Magnetization Measurement; TS: Thermophoretic Sampling)

Test	Flame	Precursor concentration	Characterization	Major results
1	H2/Air	0.65 M	TEM, XRD, MS, MM, TS	Core-shell particles (mean diameter ∼ 1 μ m), very few sub-300 nm non-core-shell particles.
2	H2/Air	0.03 M	TEM	Core-shell particles (mean diameter ∼ 0.5 μ m), many sub-300 nm non-core-shell particles, and nano-aggregates.
3	H2/O2	0.65 M	TEM, XRD, MS, MM, TS	Non-core-shell nanoparticles smaller than ∼ 50 nm.
4	H2/O2	0.03 M	TEM	Non-core-shell nanoparticles smaller than ∼ 40 nm.

Synthesis Conditions

The FASP apparatus used in this study was similar to that used in previous studies (Guo and Luo 2008; Guo et al. 2009a). A schematic illustration of the process is shown in Figure 1. The colloidal solution was prepared by dissolving Fe(NO₃)₃· 9H₂O (Alfa Aesar, Ward Hill, MA) in a silica colloid (Silicon(IV) oxide, particle size 10 nm, 30% in H₂O, colloidal dispersion, CAS# 7631-86-9, Stock # 43111, Lot # K11Q038, Ward Hill, MA). The Fe(NO₃)₃· 9H₂O was first dissolved in deionized water; then the aqueous solution was mixed with the silica colloid. The concentration of the iron nitrate solution and the mixing ratio between the iron nitrate solution and the silica colloid were controlled so that the Fe/Si molar ratio was maintained constant at 1. The colloidal precursor solution was fed into an atomization vessel at 5 mL/h and atomized continuously by a 1.7 MHz ultrasonic transducer (Guo and Luo 2008). The H₂ fuel gas, at a flow rate of 1 standard liter per minute (SLM), carried the droplets of the colloidal precursor solution from the atomization vessel through a tubular burner. At the burner mouth, the H₂ gas formed a stable, self-sustained coflow diffusion flame supported by air or O₂ at 10 SLM. Particles were formed in the flame after complex physical changes and chemical reactions. A vacuum pump drew the post-flame aerosol through a filter holder (not shown in schematic) and product particles were collected on a filter (Guo and Luo 2008). The particles collected on filter were characterized by the various methods described below. Particles were also sampled from the interior of the flame by thermophoretic sampling, and then analyzed by electron microscopy, as described below.

FIG. 1 A schematic of the flame synthesis process using colloidal precursor solutions.

Particle Characterization

To determine their microstructure and size, the particles were analyzed by transmission electron microscopy (TEM) and energy dispersive X-ray spectroscopy (EDS) on a JEOL 2010 microscope (JEOL Ltd., Tokyo, Japan). For TEM sample preparation, the particles were carefully scraped off the filter and suspended in isopropanol. The suspension was dropped on a copper TEM grid with carbon film support. The particles were left on the grid after evaporation of isopropanol. Using representative TEM images of a sample, particle size distribution was obtained. A total of more than 500 particles were measured to obtain the size distribution for a sample.

To obtain chemical structure information of the product particle, X-ray diffraction (XRD) was carried out on a Bruker-AXS D8 Powder X-ray diffractometer (Bruker, Madison, WI) by placing the filter samples directly on the sample stage. Also, Mössbauer spectroscopy was carried out to obtain information of iron speciation and crystallite size. Mössbauer spectra of the particles were collected on a Wissel spectrometer (Wissenschaftliche Elektronik GmbH, Starnberg, Germany) at room temperature using ⁵⁷Co source in Rh matrix. Measurements were calibrated with alpha-Fe standard absorber sample. Spectra were fitted by Lorentzian shape components providing values of isomer shift, quadrupole splitting, absorption line area, and line width for phase, iron valence state, and relative abundance analysis.

To determine the magnetic properties of the particles, magnetization measurements were carried out in a MPMS® SQUID VSM dc magnetometer (Quantum Design, San Diego, CA). The room-temperature M(H) measurement was carried out by setting temperature at 300K and scanning 7T → 0T → −7T → 0T → 7T at a step size of 50 Oe/s (10,000 Oe = 1T). For the low temperature measurement, the temperature was set at 5K (sample field cooled in 7T) and the scan was run 7T → 0T → −7T → 0T → 7T at a step size of 50 Oe/s (10,000 Oe = 1T).

Thermophoretic Sampling

To help understand the particle formation mechanism, particle samples were taken thermophoretically from the flame interior in two of the main tests. The thermophoretic sampling was carried out by quickly inserting a TEM grid into the flame center at various heights above the burner. The TEM grid typically remained in the flame for a few milliseconds to tens of milliseconds before being quickly retracted. Particles in the vicinity of the TEM grid were driven by thermophoresis towards the TEM grid due to the large temperature gradient between the flame gases and the cold TEM grid. The particles were thus deposited on the TEM grid. Particle growth was “frozen” at the moment the particles were deposited on the cold TEM grid. These particles “frozen” in a certain growth stage could be later viewed in a microscope. Therefore the particle characteristics at various growth stages could be learned. Thermophoretic sampling has been used repeatedly in understanding particle evolution in flames (Dobbins and Megaridis 1987; Guo and Kennedy 2004, 2007; Guo et al. 2009b; Mueller et al. 2004).

RESULTS

A strong correlation was found between the flame type and particle structure/size. There was also a correlation between particle structure and particle size. The experimental conditions, characterization methods and main particle characteristics of the four major tests are summarized in Table 2.

H₂/Air Flames Resulted in Core-Shell Particles

H₂/air flames resulted in core-shell particles, as shown in Figure 2. With 0.65 M precursor (Test 1), the mean particle diameter was approximately 1 μ m, and the majority of the particles had the core-shell structure. The core-shell particles had a silica shell with low electron density relative to the core, which generated clear contrast in the TEM image for identification of the core-shell structure. The silica shell thickness was uniform for each core-shell particle. From particle to particle the shell thickness varied from 80 nm to 120 nm.

FIG. 2 Representative TEM images of particles from H₂/air flames: (A) as-synthesized core-shell particles from Test 1 (some appear “hairy” due to presence of nanoparticle aggregates); (B) a higher magnification image showing nanoparticle aggregates; (C) particles from Test 1 after washing in 0.75 M sulfuric acid for 24 h to remove the nanoparticle aggregates (white arrow points to a sub-300 nm particle with non-core-shell structure), and (D) as-synthesized particles from Test 2 (black arrows point to particles with non-core-shell structures).

Nanoparticle aggregates were found among the core-shell micrometer-sized core-shell particles. In lower magnification images, the core-shell particles appeared “hairy” due to the presence of these nanoparticle aggregates. The primary particle size of these aggregates was about 10 nm, as shown in Figure 2b. These aggregated nanoparticles did not have the core-shell structure. EDS analysis suggested that these nanoparticles were mostly iron oxide. To verify the chemical nature of these nanoparticles, some of the sample from Test 1 was placed in 0.75 M sulfuric acid at room temperature for 24 h. After the acid treatment, the particles were washed with deionized water and their TEM images were taken, as shown in Figure 2c. Aggregated nanoparticles were no longer visible in the acid-treated sample. The fact that these nanoparticles dissolved in acid further suggested that they were iron oxide (Lide 2001). Similar nanoparticle aggregates were found among larger particles in a previous study on formation of Fe₂O₃ particles in a flame aerosol process (Guo and Kennedy 2007).

There was a correlation between size and structure of particles from the H₂/air flames. Larger particles (> 500 nm) had the core-shell structure. However, particles smaller than about 300 nm all had non-core-shell structures. This was true with either precursor concentration level. With 0.65 M precursor (Test 1), most of the particles were larger than 500 nm and had the core-shell structure, but the very few sub-300 nm particles had non-core-shell structures, an example of which is seen in Figure 2c. With 0.03 M precursor concentration (Test 2) the mean particle diameter was approximately 0.5 μ m; many particles were smaller than 300 nm and had non-core-shell structures, as seen in Figure 2d.

H₂/O₂ Flames Resulted in Non-Core-Shell Nanoparticles

H₂/O₂ flames with both precursor concentration levels (Test 3 and Test 4) resulted in nanometer sized non-core-shell particles. As shown in Figure 3, in TEM images the particles were nearly spherical, with a high electron density (darker) region that was distinct from a low electron density region. The particles were all smaller than 50 nm in diameter. The particle size with the 0.03-M precursor was smaller than that with the 0.65-M precursor.

FIG. 3 Bright field TEM images of non-core-shell nanoparticles from H₂/air flames with (A) 0.65 M and (B) 0.03 M precursor concentrations.

Thermophoretic Sampling Results

Particles were taken from the flame interior by thermophoretic sampling and viewed by TEM, in order to obtain “snapshots” of particles at various stages of growth. This was done in for Tests 1 and 3. The most notable finding was that micrometer-sized core-shell particles formed at intermediate stages of both the H₂/air flame and the H₂/O₂ flame. However, in the H₂/O₂ flame (Test 3) the core-shell particle formed in early stages of the flame, and then disappeared in the maximum temperature region. In contrast, in the H₂/air flame the core-shell particles formed near the maximum temperature region and remained through the flame. The TEM images of particles thermophoretically sampled from the flame interior are shown in Figure 4.

FIG. 4 Representative TEM images of thermophoretic particle samples and flame photos of (A) an H₂/air flame in Test 1 and (B) an H₂/O₂ flame in Test 3. (Figure provided in color online.)

At 5 ∼ 10 mm above the burner exit (ABE) of the H₂/air flame, micrometer sized spherical homogeneous particles were found; at this height of the H₂/O₂ flame, micrometer sized core-shell particles were found, along with some homogeneous particles. At 15 ∼ 25 mm ABE of the H₂/air flame, micrometer-sized core-shell particles were found; at the same height of the H₂/O₂ flame, only nanoparticles (mostly < 20 nm) were found, and there were no micrometer-sized particles any more. At 35 ∼ 45 mm ABE of the H₂/air flame, micrometer-sized core-shell particles were found; at the same height of the H₂/O₂ flame, only nanoparticles were found. All the nanoparticles from the thermophoretic samples had non-core-shell structures. Major findings of the thermophoretic samples are summarized in Table 3. The significance of the findings is discussed in the Discussion section.

TABLE 3 Summarized results of the thermophoretic samples

	Distance above burner exit
	5 ∼ 10 mm	15 ∼ 25 mm	35 ∼ 45 mm
H2/Air flame	Homogeneous, spherical, micrometer-sized particles	Core-shell, micrometer-sized particles	Core-shell, micrometer-sized particles
H2/O2 flame	Homogeneous particles and core-shell particles	Nanoparticles	Nanoparticles

Combining the experimental findings presented above, an insight may be gained into the correlation between the flame type and the particle structure. This is described in detailed in the Discussion section.

Characterization of Iron Species

XRD and Mössbauer spectroscopy were used to characterize the iron species in the particles. Representative XRD patterns of the core-shell particles from H₂/air flame (Test 1) and the non-core-shell particles from H₂/O₂ flame (Test 3) are shown in Figure 5. Maghemite (γ -Fe₂O₃) and Fe₂(Fe_0.565Si_0.435)O₄were found to be two most probable matches, the standard patterns of which are also shown in Figure 5. It is possible that the core-shell particles and the non-core-shell particles both contained γ -Fe₂O₃ and Fe₂(Fe_0.565Si_0.435)O₄. However, the peak at 2-θ of approximately 43° was more prominent with the core-shell particles. Both phases have a peak near this angle, but that with the iron silicon oxide Fe₂(Fe_0.565Si_0.435)O₄ has higher relative intensity (see the standard XRD patterns). Therefore it appears that the core-shell particles had a higher fraction of Fe₂(Fe_0.565Si_0.435)O₄ than did the non-core-shell particles.

FIG. 5 Representative XRD patterns of the core-shell and the non-core-shell particles; vertical lines show the peak location and intensity of the standard XRD diffraction patterns of γ -Fe₂O₃ and Fe₂(Fe_0.565Si_0.435)O_4.

The Mössbauer results (Figure 6) suggested that the crystallite size, and possibly the iron valence state, were different in the core-shell particles (Test 1, H₂/air flame) than in the non-core-shell particles (Test 3, H₂/O₂ flame). Both spectra exhibited relaxation of the magnetic order, which resulted in the collapse of sextet components into a doublet in the middle of the spectrum. However, this effect was especially pronounced in the spectrum of the core-shell particles, as seen in Figure 6a. The non-core-shell spectrum showed magnetically split sextet, as seen in Figure 6b, indicating the presence of maghemite (γ -Fe₂O₃) (Lin et al. 1996). The relaxation of the magnetic order could be attributed to very small crystallite size (Lin et al. 1996). The core-shell spectrum showed no sextet pattern, suggesting that the crystallite size was smaller in the core-shell particles than in the non-core-shell particles. Also, the core-shell spectrum showed substantial bulge on the right side shoulder of the central absorption peak. This suggested that the core-shell particles contained Fe^{2 +} (Murad and Cashion 2004). Thus the Mössbauer results suggested that, in the core-shell particles, a fraction of the iron could be Fe^{2 +}; while in the non-core-shell particles, apparently the iron was mostly Fe^{3 +}.

FIG. 6 Mössbauer spectra of (A) core-shell particles from H₂/air flame (Test 1) and (B) of non-core-shell nanoparticles from H₂/O₂ flame (Test 3).

Magnetization Properties of Core-Shell Particles and Non-Core-Shell Nanoparticles

Magnetization measurements were carried out for core-shell particles from H₂/air flame (Test 1) and non-core-shell nanoparticles from H₂/O₂ flame (Test 3). The hysteresis loops are shown in Figure 7a and Figure 7b. Both the core-shell particles and the non-core-shell nanoparticle exhibited superparamagnetic behavior at room temperature (300 K). At 5 K, the particles became blocked, resulting in significant increases in coercivity and remanence. The core-shell particles and the non-core-shell particles apparently had different magnetic domain sizes, and hence different temperature-dependence of the magnetic properties (Tartaj et al. 2002). The main magnetic properties of the core-shell particles and the non-core-shell nanoparticles are shown in Table 4, in comparison with magnetic properties of iron oxide silica core-shell nanoparticles generated in other aerosol processes (Li et al. 2006; Teleki et al. 2009). The core-shell particles in this study exhibited very soft room-temperature magnetization behaviors (low coercivity and remanence) in comparison to particles from the other studies. This difference in magnetization properties may be attributed to the difference in crystallinity and domain size. The difference in iron speciation (e.g., γ -Fe₂O₃ and Fe₂(Fe_0.565Si_0.435)O₄) may also be responsible for the difference in magnetization properties.

TABLE 4 Comparison of particle magnetic properties with other studies

			This study
	(23 wt % SiO2-coated Fe2O3) (Teleki et al. 2009)	(γ -Fe2O3/SiO2 nanoparticles) (Li et al. 2006)	Core-shell microparticles (Test 1)		Non-core-shell nanoparticles (Test 3)
Measurement temperature	300 K	300 K	300 K	5 K	300 K	5 K
Saturation magnetization (emu/g)	32	30	24	39	25	34
Magnetic moment at 8 kOe (emu/g)	32	28	8	21	17	24
Coercivity H c (kOe)	0.1	0.5 (at 5 K)	0.008	4.8	0.04	0.8
Remanence M r (emu/g)	6.6		0.07	11	0.5	11

FIG. 7 Hysteresis loops for (A) core-shell particles from H₂/air flame (Test 1) and (B) non-core-shell nanoparticles from H₂/O₂ flame (Test 3).

DISCUSSION

The findings of this study suggest that the FASP method using colloidal solution precursor is capable of generating micrometer-sized iron oxide particles that are hermetically encapsulated with a silica shell. However, this capability is dependent on the type of flame and affected by the correlation between particle structure and particle size. Herein we give an explanation for the effect of flame type. In addition, particle size-structure correlation and the relevance of the micrometer-sized magnetic core-shell particles are also discussed.

Correlation of Flame Type and Particle Structure

The different results from the two different flame types (H₂/air and H₂/O₂) may be explained with the temperature history of the particles. Based on the experimental observations, a phenomenological particle formation mechanism is proposed and shown schematically in Figure 8. The figure illustrates the events that take place from when the precursor droplet enters the flame aerosol reactor, to when the product particles leave the flame aerosol reactor. Along the axis of the burner and in the flow direction, the gas temperature first rises to reach a maximum due to the combustion reaction, then decreases as the coflowing room-temperature gas (air or O₂) cools the combustion products. In general, from precursor droplets entering the flame to product particle leaving the flame, the possible physical and chemical events are: solvent evaporation, metal salt decomposition and metal oxide formation, melting and gasification of metal oxides, re-solidification of molten metal oxide and/or condensation of gaseous species. This depiction of particle growth does not include particle coagulation and coalescence. Nevertheless, this approach has been found useful for understanding metal oxide particle formation mechanisms in co-flow diffusion flames (Guo and Kennedy 2007; Guo et al. 2009b).

FIG. 8 Schematic illustrations of the different temperature history in a H₂/air flame and in a H₂/O₂ flame. (Figure provided in color online.)

The magnitude of the above-mentioned events and their locality within the flame are dependent on the temperature history of the particles. In the H₂/air flame, the highest temperature is about 2200 K; in the H₂/O₂ flame, the highest temperature can reach approximately 2700 K (Guo and Kennedy 2004; Guo and Luo 2008). The maximum temperature of the flames used in this study was approximately 20 ∼ 25 mm above the burner exit.

In the beginning, solvent evaporation and nitrate decomposition take place. The gas temperature at the burner exit (base of flame) is similar to 600 K, which reaches over 2200 K within 20 mm above the burner exit (Guo and Kennedy 2004; Guo and Luo 2008). With a typical precursor droplet diameter of approximately 4 μ m in diameter (Guo et al. 2009a), the solvent evaporation may complete in less than a millisecond after the droplet enters the flame, during which time the droplet would travel less than a millimeter (See Appendix). As the solvent evaporates, the iron nitrate starts to nucleate and precipitate on the surface of the silica colloidal nanoparticles (Wang et al. 2007). Then at a temperature below 400°C the iron nitrate decomposes to form iron oxide (Wieczorek-Ciurowa and Kozak 1999). Gases that are generated from nitrate decomposition escape; iron oxide, the solid product of the nitrate decomposition, coats the surface of the silica nanoparticles. The iron-oxide-coated silica nanoparticles form a close-packed spherical structure. The particles that are formed immediately after nitrate decomposition do not yet have the desired core-shell structure. The thermophoretic sampling results from 5 ∼ 10 mm above burner support this description of events. The presence of core-shell particles at this height of the H₂/O₂ flame suggests that particle melting occurs in this region of the H₂/O₂ flame, as discussed below.

As the particle approaches the maximum temperature region, melting occurs. At temperatures about 1950 K the particles would melt; the molten iron oxide and silicon oxide would form two immiscible liquid phases (Ehrman et al. 1999). The two immiscible liquid phases form a core-shell structure, with the silica phase being the shell. In the H₂/air flame, this takes place near the maximum temperature region. In the H₂/O₂ flame, this takes place at a location closer to the burner because of the higher heating rate; the particle becomes molten well before reaching the maximum temperature region. The experimental results support this description: core-shell particles were found at 15 ∼ 25 mm of the H₂/air, and 5 ∼ 10 mm of the H₂/O₂ flame.

In the maximum temperature zone, vaporization of iron oxide takes place. Chemical equilibrium calculations have shown that Fe and FeO vapors, rather than solid iron oxide, are thermodynamically in the maximum temperature region (Guo and Kennedy 2007). In the H₂/air flame, the flame temperature is relatively low, and the vaporization does not significantly change the size or structure of the core-shell particles. However, in the H₂/O₂ flame, the extremely high temperature (∼ 2700 K) leads to rapid vaporization of iron oxide, so that the core-shell particle disintegrates to form gaseous iron species and silica nanoparticles. The thermophoretic sampling results support this description: in the H₂/air flame the core-shell particles maintained their size and structure; while no micrometer-sized particles were found in and above the maximum temperature region of the H₂/O₂ flame. Additional tests were done with 0.65 M silica colloid (no iron nitrate dissolved therein) in a H₂/O₂ flame. The post-flame filter samples revealed silica particles with a mean diameter of approximately 1 μ m. This result suggests that the vaporization of silica is not significant in comparison to that of iron oxide.

In the cooling zone of the flame, the gaseous iron species nucleate and condense to form nano-aggregates in the H₂/air flame, which are found around the dominant, micrometer-sized core-shell particles. In the H₂/O₂ flame, the gaseous iron species would condense onto the silica nanoparticles and form nanoparticles with the non-core-shell structure.

Correlation of Particle Size and Particle Structure

There are two possible explanations for the observed correlation between particle size and particle structure. It is possible that the core-shell structure is thermodynamically stable in larger particles, but unstable in smaller particles. A possible analogous phenomenon may be the correlation between particle size and crystal structure that had been found in flame-synthesized particles (Guo and Luo 2008). However, the correlation between particle size and particle structure may also be a kinetic effect. In this study, the particle diameter ranged from tens of nanomaterials to over one micrometer. The heat transfer time constant and the solid phase diffusion time constant could vary by several orders of magnitude. Therefore the correlation between particle size and particle structure might be kinetic in nature. It is not yet possible to give a definitive answer for the size-structure correlation with this study. Nevertheless, the results suggest that it would be impossible to generate Fe_xO_y/silica core-shell nanoparticles in a one-step flame process, “one-step” as being defined earlier in the Introduction. This is consistent with the findings from previous studies (Ehrman et al. 1999; Li et al. 2006; McMillin et al. 1996).

Relevance of Micrometer-Sized Magnetic Core-Shell Particles

In comparison to the large number of studies on core-shell nanoparticles, there are relatively few studies on the synthesis of micrometer-sized magnetic core-shell particles. However, micrometer-sized magnetic core-shell particles are useful for separation of biological molecules (Boom et al. 1990), and even in vivo magnetic resonance imaging (Hamoudeh and Fessi 2006; McAteer et al. 2007). Therefore, it appears that micrometer-sized magnetic core-shell particles are relevant to existing and potential applications in their own right.

In particular, the magnetic core-shell particles generated in this study had relatively thick and dense silica shells, which afforded the iron oxide core good protection against corrosive environment. Some of the core-shell particles from this study remained intact (based on TEM observation) after being placed in 0.75 M sulfuric acid at room temperature for over a month (data not shown). This suggests that these core-shell particles generated with this method could be used in corrosive (non-hydrofluoric) environment where magnetic particles are useful. The chemical robustness of these core-shell particles also means that they can withstand harsh preparation and cleaning processes, which makes it possible for them to be reusable.

CONCLUSION

Iron silicon oxide particles were generated by modified FASP in H₂/air and H₂/O₂ diffusion flames, using a precursor solution that contained dissolved iron nitrate and colloidal silica nanoparticles at two different concentration levels. Magnetic particles with a full, uniform silica shells were obtained from the H₂/air flames. There was a correlation between particles size and structure; particles larger than 500 nm had the core-shell structure, but those smaller had non-core-shell structures. H₂/O₂ flames only resulted in nanoparticles smaller than 50 nm, which all had non-core-shell structures. Formation of nanoparticles in the H₂/O₂ flames was attributed to the very high maximum temperature in these flames. The correlation between size and structure might be a result of size-dependent stability of the structures, or the size-dependent time constants of mass and heat transfer. Combining the results from this study and other studies, it may be concluded that it is possible to form micrometer-sized, but not nanometer-sized (< 100 nm), Fe_xO_y/silica core-shell particles, in a one-step flame process as defined in the Introduction of this article.

APPENDIX: CALCULATION OF PRECURSOR DROPLET DRYING TIME

The drying time of a pure water droplet was calculated to approximate the drying time of an actual precursor droplet. The gas temperature at the exit of the burner was estimated to be 600 K according to previous measurements (Guo and Kennedy 2004). The objective of the calculation was to determine the drying time of a 4 μ m water droplet suspended in air with an ambient temperature of 600 K and ambient water vapor partial pressure of 5000 Pa. The drying time thus calculated was considered a conservative (over) estimate of the actual drying time, because in the flame the gas temperature would increase rapidly above 600 K. The assumed ambient water vapor pressure of 5000 Pa corresponded to 50% of the water vapor that would result from 5 mL/h of liquid water evaporating into a gas flow of 1 standard liter per minute (SLM).

The drying time was calculated with the following equation:

where R = 8.314 J/molK is the universal gas constant, ρ is the density of liquid water, d is the droplet diameter, D _v is the coefficient of diffusion of water vapor, M is the molecular weight of water, p _d and p _∞ are the water vapor pressure at the droplet surface and that in the ambient air, T _d and T _∞ are the temperature of the droplet and that of the ambient gas.

The temperature of the droplet was found by solving the following equation through iteration:

where H _vap is the enthalpy of vaporization of water; k _v is the thermal conductivity of the gas surrounding the droplet. In the iteration process, empirical equations of H _vap and p _d were used to relate with the droplet temperature T _d (Hinds 1999; Meyra et al. 2004). The value of D _v at 373 K was found to be 3.99 × 10^{− 5} m²/s (Lide 2001). Thermal conductivity of water vapor k _v at 373 K was 0.2509 W/m.K (Turns 2006). After iteration using Equation (2) the water droplet temperature was found to be 325 K. Substituting the droplet temperature and the water vapor pressure into Equation (1), the drying time was found to be 0.7 ms. During this time, the particle would travel about 0.5 mm with the gas flow (1 SLM through a diameter of 7.5 mm).

Financial support for this work was provided by Texas A&M University and Texas Engineering Experiment Station. Technical assistance from Dr. Ibrahim Karaman, Mr. Steven Eli Rios, and Mr. Wonjoong Hwang in magnetization measurement is gratefully acknowledged.