synthesis and growth of hematite nanodiscs through a facile hydrothermal approach

The particle characteristics e.g., shape, size, crystallization and physicochemical properties e.g., magnetic, catalytic properties of the as-prepared nanoparticles are investigated by v

R E S E A R C H P A P E R

Synthesis and growth of hematite nanodiscs through a facile

hydrothermal approach

X C JiangÆ A B Yu Æ W R Yang Æ

Y DingÆ C X Xu Æ S Lam

Received: 28 November 2008 / Accepted: 7 April 2009

Ó Springer Science+Business Media B.V 2009

Abstract This study reports a facile hydrothermal

method for the synthesis of monodispersed hematite

(a-Fe2O3) nanodiscs under mild conditions The

method has features such as no use of surfactants, no

toxic precursors, and no requirements of

high-temper-ature decomposition of iron precursors in non-polar

solvents By this method, a-Fe2O3 nanodiscs were

achieved with diameter of 50 ± 10 nm and thickness

of *6.5 nm by the hydrolysis of ferric chloride The

particle characteristics (e.g., shape, size, and

distribu-tion) and functional properties (e.g., magnetic and

catalytic properties) were investigated by various

advanced techniques, including TEM, AFM, XRD,

BET, and SQUID Such nanodiscs were proved to

show unique magnetic properties, i.e.,

superparamag-netic property at a low temperature (e.g., 20 K) but

ferromagnetic property at a room temperature (*300 K) They also exhibit low-temperature (\623 K) catalytic activity in CO oxidation because

of extremely clean surfaces due to non-involvement of surfactants, compared with those spheres and ellip-soids capped by PVP molecules

Keywords Hematite nanoparticles
Nanodiscs
Hydrothermal synthesis

Introduction

Hematite (a-Fe2O3) nanoparticles have been widely studied because of their attractive properties, includ-ing stability in air, n-type semiconductinclud-ing, non-toxicity, and corrosion-resistance These properties have driven them for potential applications in catal-ysis, gas sensing, pigment, nonlinear optic, and field-effect transistor (Shin et al 2004; Schertmann and Cornell 1991; Wang and Willey 1998, 1999) A variety of synthesis methods have been employed for shape and size control, such as ball milling, co-precipitation, sol-gel method, micelle template, thermal decomposition of precursors in non-polar solvents, and hydrothermal methods (Ozaki et al

1984; Matijevic´ 1985; Matijevic´ and Hamada 1982; Matijevic´ and Scheiner 1978; Woo et al 2003; Vayssieres et al 2005; Cao et al 2005; Yin et al

2005; Raming et al.2002) Some methods focused on

X C Jiang
A B Yu (&)

School of Materials Science and Engineering, University

of New South Wales, Sydney, NSW 2052, Australia

e-mail: a.yu@unsw.edu.au

W R Yang

Australian Key Centre for Microscopy and Microanalysis

(AKCMM), Electron Microscopy Unit, University

of Sydney, Sydney, NSW 2006, Australia

Y Ding
C X Xu

School of Chemistry and Chemical Engineering,

Shandong University, Jinan, Shandong 250100, China

S Lam

CSIRO Materials Science and Engineering, P.O Box 218,

Lindfield, NSW 2070, Australia

DOI 10.1007/s11051-009-9636-8

the thermal decomposition of organometallic

precur-sors (e.g., Fe(CO)5, Fe(acac)3, Fe(oleate)3, or their

dual source systems) in non-polar solvents for

generating monodispersed iron oxide nanoparticles,

which, however, may limit their applications in

aqueous system (Yin and Alivisatos 2005; Casula

et al.2006; Cheon et al.2004) For example, the iron

oxide nanoparticles obtained by thermal

decomposi-tion in non-polar solvents are difficult to be transferred

directly into aqueous solution because of the

surface-coated surfactants The removal of such surfactants

may lead to particle aggregation, and hence affect the

covalently binding other surfactants such as

poly(eth-ylene)glycol (PEG) spacer with hydrophilic groups for

further dextran coating that is targeted toward solid

tumor treatment (Sonvico et al.2005)

Among the achieved hematite nanocrystals

obtained, non-spherical particles have become

attrac-tive because of their anisotropic properties To date,

one-dimensional (1D) iron oxide nanoparticles (e.g.,

rods and wires) have been widely studied, but only a

few studies were reported on two-dimensional (2D)

ones (e.g., plates and discs) Their growth

mecha-nisms are not well understood A representative

example in this area was reported by Casula et al

(2006) who demonstrated the preparation of iron and/

or iron oxide nanodiscs through a thermal

decompo-sition (at *293°C) of iron pentacarbonyl in the

presence of an oxidizer and surfactants (tridecanoic

acid or 3-Chloro peroxybenzoic acid) Niederberger

et al (2002) reported the fabrication of hematite

disclike particles with outer diameter of *1 lm and

the thickness of *250 nm through a hydrolysis and

subsequent hydrothermal approach Chen and Gao

(2004) prepared the crystalline a-Fe2O3nanodiscs by

a hydrothermal method at 150°C and through aging

for 24 h in the presence of surfactants The above

synthesis methods suggested that the utilization of

surfactants is necessary and important for shape

control This is also evidenced by other

wet-chem-istry methods in the past However, drawbacks

resulting from surface-adsorbed surfactants have the

unpredictable influence on the surface functionality

of nanoparticles and the diminished accessibility to

particle surface A small amount of residual may

significantly reduce the functionalities in catalysis or

gas sensing For example, the pyrolysis of oleic acid

could introduce some reducing agents such as carbon

(C), carbon monoxide (CO) and hydrogen (H2),

which can cause a negative effect on the particle performances (Kim et al 2007) Therefore, to develop facile synthesis methods to produce mon-odispersed iron oxide nanodiscs with extremely clean surfaces is still a challenging task

In this study, we demonstrate a facile hydrother-mal approach to generate monodispersed a-Fe2O3 nanodiscs in the absence of surfactants under mild conditions The particle characteristics (e.g., shape, size, crystallization) and physicochemical properties (e.g., magnetic, catalytic properties) of the as-prepared nanoparticles are investigated by various advanced techniques The influence of a few exper-imental parameters (e.g., pH, temperature, time, and concentration of Fe3?) on the particle growth in the surfactant-free system is then investigated The catalytic CO oxidation, as one of typical functional-ities of hematite nanoparticles, is also examined

Experimental work

Synthesis of iron oxide nanoparticles

The iron oxide nanoparticles could be prepared by the hydrolysis of FeCl3salt in an acid solution under mild conditions This approach is similar to the previous studies (Matijevic´ and Scheiner1978; Raming et al

2002), and the modification of experimental para-meters has been adopted to prepare plate-like nano-particles In a typical procedure, three steps were involved First, 0.5 g FeCl3
6H2O (Sigma-aldrich, 99.9%) was put in 10 ml of water, followed by vigorous stirring to ensure that all the powders got dissolved completely Second, the transparent yellow-brown solution was quickly injected into a conical flask containing 90 ml of hot water (*90°C) and 0.75 ml of dilute HCl (1.0 M), followed by vigorous stirring to ensure that the reaction system was homogeneous Finally, the mixed solution was refluxed heating at 90°C for around 5 min before being transferred into an oven for heating at 90°C In order to avoid water evaporation, the flask was sealed

by aluminum foil and a glass lid After heating for

48 h, the solution turned deep red color The particles were found homogeneously dispersed in this solution

In the surfactant-assisted synthesis, poly(vinyl pyr-rolidone) (PVP, Mw= 55,000, Sigma-aldrich, 99.9%) was used to control particle shape and size, but other

parameters were kept constant Ultra-pure water was

used in all the synthesis processes All the glasswares

were cleaned with aqua regia, thoroughly rinsed with

ultra-pure water and alcohol prior to use

Characterization

Various techniques were used to characterize the

particle characteristics and properties in this study, as

described below:

(i) Particle characteristics such as shape, size, and

size distribution were checked using Philips

CM200 field emission gun transmission electron

microscope (TEM) operated at an accelerated

voltage of 200 kV The specimen was prepared

by dropping the solution onto a Formvar-coated

copper grid and dried in air naturally The data

for particle-size distribution were collected

based on TEM analysis, and also assisted by

Image Processing and Analysis Program (ImageJ

1.37v, 2006);

(ii) The composition of the as-synthesized sample

was identified by powder X-ray diffraction

(XRD), and recorded using Siemens D5000 at

a scanning rate of 0.5°/min in the 2h range of

20–80°;

(iii) The atomic force microscope (AFM) image

was obtained by a Molecular Imaging Picoscan

II instrument in tapping mode The sample was

prepared by depositing a few drops of a dilute

solution of the nanoparticles onto a mica disc

and then dried in air naturally Analysis of the

AFM image was performed using the WSxM

software (version 3, Nanotec Electronica S.L.,

Spain);

(iv) The Brunauer–Emmett–Teller (BET) surface

area of the as-prepared particles was measured

at 77 K (liquid nitrogen) on a Quantachrome

Autosorb-6B Surface Area & Pore Size

Ana-lyzer Before BET measurements, the sample

was degassed at 150°C for 3–4 h to ensure that

no gas molecules adsorbed on the particle

surfaces;

(v) The magnetic properties were investigated on

a Quantum Design MPMS XL-5 (SQUID)

magnetometer The sample was put in a

low-susceptibility plastic sample holder for

mea-surements The magnetic moment from the

sample holder was found to be at least three orders of magnitude smaller than the signal from the sample and thus can be ignored; (vi) The catalytic oxidation of CO gas was per-formed in a home-built fixed bed microreactor

A reactant gas containing CO (6,000 ppm) in

O2 atmosphere (CO/O2= 1/10) was buffered with N2gas and with a total flow rate of 60 ml/ min An appropriate amount of hematite pow-ders (*50 mg) was used in the measurement The heating and cooling cycles were monitored

in a temperature range of 30–450°C

Results and discussion

Microstructure of iron oxide nanoparticles

The microstructure of the as-prepared nanoparticles obtained by the proposed synthesis strategy was checked by TEM technique Figure 1a shows the TEM image of one representative sample The particles were found nearly monodispersed with diameters of 50 ± 10 nm based on their size distri-bution (Fig.1b) In order to confirm the crystalliza-tion, the selected area electron diffraction (SAED) was carried out under TEM operations Several clear diffraction rings doped with spots were recorded and shown in a pattern (inset of Fig.1a), suggesting that the nanoparticles are of crystalline structure The further confirmation on single crystalline or poly-crystalline particles needs other techniques like XRD and HRTEM These diffraction rings could be assigned to (104), (110), (113), (024), (116), and (300) crystallographic planes of rhombohedral phase a-Fe2O3, respectively, based on the standard JCPDS card (No 02-915) (Cullity and Stock2001)

A close inspection on the particle shape and crystallization was conducted by various techniques Figure2a shows a magnification TEM image reveal-ing that the as-prepared particles are spherical parti-cles, and some of them overlapped, as pointed by arrows Further evidence could be directly obtained from the AFM image shown in Fig.2d The curve plotted in Fig.2d reveals that the particle is of plate-like structure with a thickness of *6.5 nm Combin-ing the structural analysis of TEM and AFM, the as-prepared hematite particles are of nanodiscs in

shape The lattice fringes of the individual nanodisc

could be clearly seen in the high-resolution TEM

(HRTEM) image (Fig.2b), indicating that the

nano-discs are well crystallized under the reported

condi-tions Measuring the distance between two adjacent

planes gives a value of *0.411 nm, corresponding to

the lattice spacing of {110} facets of rhombohedral

a-Fe2O3 The electron diffraction (ED) pattern could

be indexed to the [001] zone of rhombohedral hematite

(inset of Fig.2b) The crystallization of the nanodiscs

was also evidenced by the well-resolved peaks in

XRD pattern (Fig.2c), in which all the diffraction

peaks could be assigned to rhombohedral phase

a-Fe2O3 (a = b = 5.028 A˚ and c = 13.728 A˚ ,

JCPDS 02-915) (Cullity and Stock2001) This is also

supportive to the indexed diffraction rings in the

SAED pattern (inset of Fig.1a) These results revealed

that the as-prepared nanodiscs are pure rhombohedral

a-Fe2O3with single crystalline structure

Particle nucleation and growth

The particle nucleation and growth occurred while the ferric ions solution was mixed with hot water (90°C), accompanied by a rapid color change from yellow to red In order to trace the growth, the colloids were isolated from the heated suspension at different times for statistic analysis Owing to the limitations in in-site observing the nucleation of colloids, several representative samples were chosen here to illustrate the growth process After heating for

*1 min, small colloids formed, but they are difficult

to be clearly distinguished in shape and size as shown

in Fig.3a After heating for *5 min, the solution turned a bit dark red The colloids isolated from this dark-red solution were checked by TEM technique Figure3b shows the TEM image that the shapes of colloids are still difficult to distinguish, but the particle size becomes larger (20 ± 10 nm) than those obtained at the reaction time of *1 min This result suggested that the nucleation is fast, as the so-called

‘‘burst-nucleation’’ happens in this reaction system Although the nucleation and the growth may be overlapping each other at the initial stage, the particle size increasing with time could be clearly observed after 5 min (Fig.3b–f) The fast nucleation and the subsequent slow growth obtaining well-crystallized nanodiscs could also be confirmed by the relationship between reaction time and particle sizes as described

in Fig.4, the corresponding data for which were collected and compared on the basis of TEM images obtained at different times This is different from the nucleation-delayed mechanism that occurred in the formation of iron oxide nanodiscs through the thermal decomposition of Fe(CO)5 precursor in non-polar solvent (Casula et al.2006)

After 5-min heating at 90°C, the reaction system was transferred into an oven with the heating continued further In order to further understand the particle growth, the particles produced at different times (e.g., 1, 6, 12, and 24 h) were separately isolated for TEM characterization Figure 3c shows the TEM image that 1-h heating merely resulted in the formation of irregular-shaped particles, but par-ticle size increased with time The 6-h heating was found to result in small particles to grow to a diameter of 50–70 nm, and some of them were still irregular in shape (Fig 3d) Again, the aggregation of small particles or the clusters could be observed in

Fig 1 a TEM image of hematite (a-Fe2O3) nanodiscs with

inset of SAED pattern; and b The size distribution of the

a-Fe 2 O3nanodiscs

this sample With continuous heating up to 12 h,

more well-shaped particles formed with diameter of

*70 nm, although the size distribution was a bit

wide In the meantime, it was found that almost the

smaller irregular particles disappeared at this stage

(Fig.3e) This process could be consistent with

Ostwald ripening, i.e., smaller particles continue to shrink, while larger particles continue to grow (Ostwald 1896) The 24-h heating could lead to the formation of nearly monodispersed particles with a mean diameter of 55 nm (Fig.3f) A close look at the particles reveals that they have a slight shrinkage in

20 30 40 50 60 70 80

2 θ (degree)

C

200 300 400 500 600 0

2 4 6 8 10

Diameter (nm)

D

Fig 2 a A high

magnification TEM image

of a-Fe2O3nanodiscs with

overlapping as pointed by

arrows; b HRTEM image

showing the lattice fringe of

{110} planes with spacing

between two adjacent

planes of 0.411 nm; c XRD

pattern of the nanodiscs

showing that the particles

are of rhombohedral phase;

d AFM image of an

individual nanodisc and the

curve showing that the

thickness is *6.5 nm

size relative to those obtained by heating for a short

time (e.g., 12 h) This was probably due to by atomic

reconstruction on particle surfaces to minimize

surface energy Moreover, the electron diffraction

rings recorded in the SAED patterns become clear,

indicating that the particles crystallized better with

longer heating duration

In order to clearly describe the time-dependent growth, the relationship between particle size and heating time was plotted and shown in Fig.4 It could be seen from Fig 4 that the nucleation is fast but the particle size increase slowly with time The particle size increases up to the maximum *70 nm around 12 h After that, more and more well-shaped

Fig 3 Time dependence of

iron oxide nanoparticles

formed in aqueous solution:

a 1 min; b 5 min; c 1 h;

d 6 h; e 12 h; f 24 h

nanodiscs formed, and the particle size gradually

decreased to *55 nm due to the possible

recon-struction of the surface atoms with extended heating

duration

In order to understand the formation and growth

mechanism of nanodiscs, a few possibilities have been

proposed previously A typical example was reported

by Casula et al (2006) who supposed a delayed

nucleation mechanism for the formation of iron oxide

nanodiscs during the high-temperature decomposition

process, which results in the occurrence of

nanopar-ticle crystallization well separated in time from the

injection of the precursors They suggested a

burst-like nucleation at a certain delayed time and

subsequent fast nanocrystal line growth at a high iron

monomer concentration that promoted the kinetically

induced formation of anisotropic discs The

retarda-tion of the nuclearetarda-tion was induced by the surfactant

(e.g., fatty acid) used as a coordinating agent, which

strongly stabilizes the monomer in solution On the

other hand, Redl et al (2004) and Hyeon et al (2001)

reported that when the reaction was carried out under

conditions that favor gradual and slower monomer

release into the solution, thermodynamically

sta-ble iron oxide nanospheres were produced For

those nanodiscs obtained in aqueous solution,

Niederberger et al (2002) supposed that the use of

an iron–polyolate complex of [N(CH3)4]2–[OFe6

(H-3thme)3(OCH3)3Cl6]
MeOH as a precursor

mate-rial can produce disclike hematite particles by a

procedure involving the hydrolysis and subsequent

hydrothermal treatment at 150°C over 24 h They

found that each of the large particles (outer diameter

*1 lm and thickness *250 nm) was made up of many small plate-like particles Chen and Gao (2004) also suggested that the presence of surfactants, such as poly(oxyethylene)(20)-sorbitan monooleate (Tween 80) and pluronic amphiphilic triblock copolymer (P123), played an important role in the formation of crystalline a-Fe2O3 nanodiscs during hydrothermal treatment at 150°C for a period of 24 h The abovementioned methods revealed that the surfactants played a key role in formation and growth of iron oxide nanodiscs

However, these proposed mechanisms seem to be unsuitable for our case On the one hand, no delayed nucleation occurred on the basis of the time-depen-dent nucleation and growth processes (Figs.3, 4) That is, the burst-like nucleation and the subsequent anisotropic growth do not significantly occur in this case under the reported conditions On the other hand,

no surfactants or complex precursors were used in our synthesis, and thus the surfactant-assisted growth by selective face adsorption could not be considered Therefore, it is believed that the particle morphology may be determined by other factor(s) in this system Let us now consider chloride (Cl-) ions first In the reaction solution, the Cl-ions are excessive due to the addition of HCl for pH adjusting below 2 (it was measured that the pH is *2 without addition of acid in this case), which is believed to cause the Cl-ions to play dual possible roles: to retard hydrolysis of the ferric ions and to reduce surface energy on a certain crystal plane to promote preferential growth Some investigators have also studied the particle preferential growth in the systems without surfactants For exam-ple, Wang et al (2008) suggested that the forced hydrolysis of ferric chloride under acidic pH could result in the direct transformation from amorphous iron oxide to crystalline hematite when aged at 100°C for a period of 48 h, and these hematite nanocrystals could assemble into large-size disclike particles via solvent evaporation Moreover, Matijevic´ and Scheiner (1978) and Matijevic´ (1985) investigated the influ-ence of inorganic ions such as chloride, nitrate (NO3-), and perchlorate (ClO4-) on the shape and size of hematite in the hydrothermal reaction carried out at 100°C They found these inorganic ions could lead to different shapes and sizes of hematite particles Raming et al (2002) reported a similar approach

by using iron chloride salt and being carried out at

0

20

40

60

80

Reaction time (h)

Fig 4 The plotted curve showing the time-dependent

nucle-ation and growth of the colloids

90–100°C for different times (e.g., 1–6 days) to

prepare hematite colloids; however no disclike

parti-cles were formed In other systems, the effect of

inorganic ions on particle growth was also studied

Both Livage et al (1988) and Reeves and Mann

(1991) groups have demonstrated the influence of

inorganic ions such as chloride, phosphate (PO43-),

sulphate (SO42-), and perchlorate on the shape and

size of hematite and other transition metal oxides

They reported that the presence of Cl- ions could

result in rhombohedral hematite crystals comprising

1014 faces that exhibited relatively high energy The

formation of such high-energy 1014 faces indicates

that the Cl- anion has a profound influence on the

stability of these faces because the 1014 face has an

open structure that may be able to accommodate Cl

-(ionic radius 1.8 A˚ ) and thereby stabilize the bonding

within the surface plane The selective surface effect

of Cl-anion could finally affect the morphology and

size of particles Similar effects have been observed

for nanocrystals grown in the presence of fluoride

(F-), Cl-, and bromide (Br-) for copper nanorods

(Filankembo et al 2003; Filankembo and Pileni

2000), hydroxide (OH-) for silver nanowires (Caswell

et al.2003), as well as F-and Cl-anions for titania

nanosheets (Yang et al 2008; Penn and Banfield

1999), through selective surface adsorption

As a further confirmation, the replacement of HCl

by dilute HNO3, HClO4, and H2SO4was carried out

to adjust solution pH in this study Figure5shows the

corresponding TEM images of nanoparticles obtained

by addition of various acids as mentioned above

When NO3- ions were added, the irregular-shaped

particles were obtained (Fig.5a) The precise size of

particles was difficult to measure The addition of

SO42- ions could result in flocculation rapidly (B3 min) during heating at 90°C The particles obtained under such conditions were unshaped, and the size is difficult to estimate (Fig.5b) While ClO4 -ions were used, cube-like particles formed with edge lengths of 30–70 nm (Fig.5c), consistent with the previous literature Unfortunately, the addition of these acids could not produce disclike hematite particles under the reported conditions This sug-gested that the Cl-ions indeed played a crucial role

in the formation of hematite nanodiscs under the conditions considered, which is consistent with the observations reported by Matijevic´ and Scheiner (1978) that the inorganic ions could lead to shape and size change of hematite particles

Effects of experimental parameters

In order to better understand, the effects of other experimental parameters were also investigated including pH, reaction temperature, and concentra-tion of Fe3? ions, as discussed in the following context

pH

The hydrolysis of Fe3? ions is closely related to the

pH of the solution The pH is normally adjusted by addition of acid or base Here the effect of pH on particle growth was further investigated Figure6

shows the TEM images of the particles obtained at different pH values At pH = 1.5, the particles were found to be well crystallized with an average

Fig 5 Effect of the inorganic anions on the shape and size of nanoparticles: a NO3- ; b SO42- ; c ClO4

-diameter of 77 nm (Fig.6a), confirmed by their size

distribution shown in Fig.6d When the pH value

was increased to *2.5 by addition of an appropriate

amount of NaOH solution (5 M), the average size of

nanoparticles decreased a bit to *70 nm in diameter

(Fig.6, panels b and e) On further increasing pH to

*3.0, the particles continuously reduced in average

size to *63 nm (Fig.6, panels c and f) However,

when pH was further altered by adding HCl or

NaOH, it was found that a low pH (\1.0) could not

produce any particle but a clear solution, whereas a

high pH ([3.5) could lead to some precipitates (e.g.,

Fe(OH)3) prior to any reflux heating Further

inspec-tion of the particles (Fig.6, panels a–c) suggested

that the slight shrinkage in size with pH increasing

was probably caused by the fast hydrolysis rate at a

high pH (*3.0) The diffraction rings in the SAED

patterns (inset of Fig.6, panels a–c) revealed that the as-prepared nanoparticles are of crystalline structure

In this system, the H? ions may have two functions: to slow down the hydrolysis rate of Fe3? ions and to stabilize the oxygen-terminated crystal planes such as a-Fe2O3{0001} (Cotton and Wilkinson

1988) At a low pH (1–3), a particle prefers to grow along other planes such as a-Fe2O3{01ı¯1}, beneficial for the anisotropic growth of particles (Schertmann and Cornell 1991; De Leeuw and Cooper 2007; Goldschmidt1913/1923) After many tests, we found that the most suitable pH value for disc formation is 1–3 Lower pH (\1.0) could not produce any particle due to their rapid dissolution, whilst higher pH ([3.5) could lead to precipitates (e.g., Fe(OH)3) directly under the reported conditions This is in agreement with those previously reported that a-Fe2O3 was

Fig 6 Effect of solution

pH on the shape and size of

iron oxide nanoparticles and

the corresponding size

distributions: a, d

pH = 1.5; b, e pH = 2.5;

c, f pH = 3.0

merely obtained at pH \4 (Weiser and Milligan

1935; Mackenzie and Meldau 1959) Due to the

complicated processes involving hydrolysis (Fe3?

)-nucleation and (Fe(OH)3)-phase transformation (from

b-FeOOH to a-Fe2O3), it is believed that a further

study needs to be performed to understand the

particle growth

Reaction temperature

The particle growth is also affected by the reaction

temperature Figure7shows the TEM images of the

prepared nanoparticles at different temperatures (e.g.,

70, 80, and 100°C) Other experimental parameters

were maintained the same as those for the temperature

of 90°C When heated at 70°C for 48 h, small particles

formed with irregular shape (Fig.7a) Particle size

was in the range of 5–10 nm, and most of the particles

aggregated together The weak diffraction rings in the

SAED pattern suggested that these particles were not

well crystallized (inset of Fig.7a) When heated at

80°C, some bigger particles were formed (diameter of

40–80 nm), along with some smaller ones (Fig 7b)

The diffraction rings (Fig.7b) became clear,

indicat-ing that the particles crystallized further with

increas-ing temperature While at 100°C, the as-produced

particles were spindle or multi-armed nanostructures

with diameters of 20–50 nm and length up to several

hundred nanometers (Fig.7c) Similar scenarios were

observed as reported by Raming et al (2002)

con-firming that the same particles were present after

heating at 100°C for 1 day and for 1 week if the ferric

chloride salt was added directly into the preheated

hydrochloric acid solution (method 1) They also

described that a mixture of two particle types (i.e., spindle and oval shapes) was produced if the ferric chloride was not added directly to the preheated hydrochloric acid solution, which, however, first dissolved in cold water before heating to 100°C (method 2) In particular, the XRD analysis from Raming et al (2002) showed the presence of two phases, hematite and akagane´ite, if the reaction was carried out by method 2

Moreover, such spindle or multi-armed nanostruc-tures show rough surfaces and no well-defined crystalline faces, similar to those prepared by addi-tion of phosphate ions during the hydrolysis and growth processes Reeves and Mann (1991) sug-gested that the interaction of phosphate with hematite crystals was not specific to a single set of symmetry-related faces in forming spindle-shaped iron oxide particles Our observations are also consistent with the previous studies that were performed at a high temperature (e.g., 100°C), although the shape and the size distribution of particles are slightly different The precursor of [Fe(OH)2(OH2)5]? does not form a polycation but nucleates directly into a-Fe2O3 parti-cles, which, however, may result in various morphol-ogies (Matijevic´ and Scheiner1978)

Concentration of Fe3?ions

In order to investigate the effect of concentration of ferric salt ([Fe3?]) on particle shape and growth, the concentrations tuned from 0.038 to 5.55 mM were tested During all the tests, the reaction temperature and heating time were kept constant The pH value of the solution was adjusted carefully and kept around 2

Fig 7 Temperature dependence of iron oxide nanoparticles formed in aqueous solution: a 70°C; b 80°C; c 100°C