Bull. Mater. Sci., Vol. 31, No. 5, October 2008, pp. 759–766. © Indian Academy of Sciences.

Strain induced anomalous red shift in mesoscopic iron oxide prepared

by a novel technique

T N NARAYANAN, D SAKTHI KUMAR†, YASUHIKO YOSHIDA† and

M R ANANTHARAMAN*
Department of Physics, Cochin University of Science and Technology, Cochin 682 022, India
†
Bio-Nano Electronics Research Centre, Department of Applied Chemistry, Toyo University, Japan

MS received 29 November 2007; revised 30 July 2008

Abstract. Nano magnetic oxides are promising candidates for high density magnetic storage and other appli-
cations. Nonspherical mesoscopic iron oxide particles are also candidate materials for studying the shape, size
and strain induced modifications of various physical properties viz. optical, magnetic and structural. Spherical
and nonspherical iron oxides having an aspect ratio, ~2, are synthesized by employing starch and ethylene glycol
and starch and water, respectively by a novel technique. Their optical, structural, thermal and magnetic
properties are evaluated. A red shift of 0⋅24 eV is observed in the case of nonspherical particles when com-
pared to spherical ones. The red shift is attributed to strain induced changes in internal pressure inside the
elongated iron oxide particles. Pressure induced effects are due to the increased overlap of wave functions.
Magnetic measurements reveal that particles are superparamagnetic. The marked increase in coercivity in the
case of elongated particles is a clear evidence for shape induced anisotropy. The decreased specific saturation
magnetization of the samples is explained on the basis of weight percentage of starch, a nonmagnetic compo-
nent and is verified by TGA and FTIR studies. This technique can be modified for tailoring the aspect ratio and
these particles are promising candidates for drug delivery and contrast enhancement agents in magnetic reso-
nance imaging.

Keywords. Nano magnetic oxides; red shift; magnetic storage.

1. Introduction octahedral sites (Viswanathan 1990). The preparation of

gamma iron oxide from precursors is tricky because the
Mesoscopic magnetic iron oxides have attracted the atten- material turns easily into its nonmagnetic phase during
tion of researchers for a variety of reasons. Magnetic iron synthesis if adequate precautions are not taken.
oxides are candidate materials for audio, video and com- Nonsphericity is an essential criterion for enhancing
pact memories (Hyeon 2003). Recently, efforts are under- signal to noise ratio during recording (Bate 1991; Deng et al
way to find new applications of these materials—as drug 1992). But γ-Fe2O3 is a cubic crystal and does not easily
delivery agents, as superparamagnetic iron oxide (SPION) form into elongated ellipsoidal particles. Earlier, research-
particles for hyperthermia, contrast enhancing agents in ers attempted to synthesize nonspherical precursors using
magnetic resonance imaging (MRI) (Sabino et al 2004) complexing medium (Anantharaman et al 1997). Nuclea-
and also as biosensors (Dallas et al 2006). They are bio- tion will be greatly influenced by external conditions like
compatible and inexpensive. From a fundamental point of pH, temperature and constant stirring. Complexing agents
view, magnetic iron oxides are ideal templates for study- like starch retard the growth of the precursors in the crys-
ing ferrimagnetism at the nano level since they possess an tallographic a-direction, so that growth proceeds in the c-
ideal two sub-lattice with magnetic ions on either sites. direction, leading to needles (Arndt et al 1998). However,
Magnetic iron oxides belong to the class of inverse with the advent of nanoscience and nanotechnology, shape
spinels having their cation distributed on the octahedral and size induced modifications of various properties like
(B) and tetrahedral (A) sites. Maghemite or gamma ferric optical, magnetic, structural are a rich area of physics
oxide also crystallizes in the inverse spinel structure. The where quantum mechanical phenomenon like quantum
structure of maghemite is said to possess a vacancy ordered size effects can be investigated on candidate materials like
spinel structure with vacancies situated exclusively on the gamma iron oxide (Kachkachi et al 2000). It is also known
that size dependent magnetic properties manifest itself
into superparamagnetism and single domain characteris-
tics. Reduction of size results in large red shift in Fe3O4
*Author for correspondence ([email protected]) (Nair et al 2005).
759
760 T N Narayanan et al

Most of our earlier studies on nonspherical particles are 2.3 Characterization

limited to the synthesis of nonspherical γ-Fe2O3 particles.
Most of the particles lie well above the nanometer dimen- The structural analysis and the identification of the phase
sion (Anantharaman et al 1997). However, size, shape were conducted with an X-ray diffractometer (Rigaku
and strain dependent optical and magnetic properties are a Dmax-C) using CuKα radiation (λ = 1⋅5418 Å). A scanning
virgin area where the effect of size and shape can be stu- rate of 5° min–1 was applied in the 2θ range of 10°–70°.
died in a single shot if nanosize particles can be synthe- The grain size was estimated by employing Debye–
sized. Such a study assumes importance because of the Scherrer’s formula
resulting magneto optical properties of the nanocrystals.
D = 0⋅9λ/β cosθ, (1)
In this work, spherical and elongated mesoscopic gamma
iron oxide particles are synthesized employing a novel where λ is the wavelength of X-ray used, β the FWHM of
technique and using starch/water/ethylene glycol as com- XRD peak with maximum intensity and θ the angle of
plexing agents. Their structural, magnetic and optical diffraction. A JOEL JEM 2200 FS electron microscope
properties are evaluated. Emphasis is laid in studying the using an accelerating voltage of 200 kV was used to de-
shape induced optical properties of gamma iron oxide duce the particle size distribution of nanoparticles. The
nanoparticles. Thermogravimetric (TG) and FTIR analysis transmission electron microscope (TEM) images, high
are carried out to probe the retention of organic residues resolution transmission electron microscope (HRTEM)
in the sample. Attempts are made to correlate the optical images and energy dispersed spectra (EDS) were taken for
and magnetic properties with size and shape. morphological and compositional studies. Planes were
identified from HRTEM. Room temperature magnetic
properties were obtained from the hysteresis loop re-
2. Experimental
corded in a vibrating sample magnetometer (VSM) model
EG&G Par 4500. A Thermo Nicolet Avatar 370 DTGS
2.1 Preparation of spherical γ-Fe2O3 nanoparticles
model spectrophotometer using KBr method was employed
to record the FTIR spectrum of the samples. A Jasco V
In order to prepare spherical γ-Fe2O3 particles, 20% of
530 UV-visible spectrophotometer was used to determine
starch solution was prepared by adding 20 g of soluble
the diffused reflectance studies. DTG studies were con-
starch into 100 cc of hot water. The resultant solution was
ducted using Perkin Elmer Thermal Analysis in nitrogen
heated with constant magnetic stirring to 50°C till it be-
atmosphere in the temperature range 30–550°C.
came transparent. 25 g of anhydrous ferrous sulphate was
dissolved in 75 cc starch solution. This solution was heated
to 70°C. To this solution 0⋅2 molar ethylene glycol was 3. Results and discussion
added. It was presumed that ethylene glycol inhibits growth
of the particles along one particular direction. Addition of 3.1 Structural and morphological studies using
ethylene glycol was done drop-wise with constant stirring XRD and TEM
in order to ensure the formation of spherical particles.
These particles were filtered off and dried in a furnace at The X-ray powder diffraction pattern for the spherical and
370°C in ambient atmosphere to ensure the formation of nonspherical maghemite particles are depicted in figure 1.
γ-Fe2O3 (Venkatesh Rao et al 1974). This led to a brown They represent a spectrum typical of an inverse spinel.
magnetic powder confirming the formation of γ-Fe2O3
nanoparticles (Khaleel 2004). These samples are hereafter
named N-1.

2.2 Preparation of ellipsoidal gamma iron oxide

The above procedure for the synthesis of spherical iron

oxide particles was modified to prepare ellipsoidal
gamma iron oxide (hereafter named N-2) in the sense, that
instead of ethylene glycol and starch, water and starch are
employed as the complexing agents for this medium. 25 g
of anhydrous ferrous sulphate was dissolved in 75 cc
starch solution. This solution was heated to 70°C. To this
solution 0⋅2 molar solution of oxalic acid was added to
precipitate nonspherical ferrous oxalate precursors from
the medium. It was found that as the concentration of starch
increases aspect ratio also increases. Figure 1. XRD pattern of γ-Fe2O3.
 Strain induced anomalous red shift in mesoscopic iron oxide 761

Figure 2. (a) TEM, (b) HRTEM, (c) EDS and (d) size distribution histogram images of N-1.

The planes are identified and listed. The peaks (111), cles have an average length of 19 nm (figure 3a) and
(220), (311), (400), (511) and (440) correspond to that of breadth 9 nm with an average aspect ratio of 2.
γ-Fe2O3 (ICDD: 39-1346). It is evident from figure 3b that the particles are crys-
There are no indications of the existence of any impurity talline and well separated by grain boundaries. Diffusion
phase, corresponding to α-Fe2O3. However, the base line of lattice planes into the grain boundary is also noted. The
is found to be shifted due to the presence of amorphous particle size distribution for spherical particles is deter-
phase probably due to the organic inclusions in the sam- mined and a histogram depicting the size distribution is
ples. The average particle size was determined from line shown in figure 2d. The discrepancy in particle size as
broadening and the spherical particles have an average determined by XRD and TEM could be because of the
size of 10 nm while the nonspherical particles exhibited an fact that only large crystallites contribute to the Braggs
average size of 13⋅5 nm. TEM images of N-2 indicate the peak (Cohen 1990). It can be seen from figure 3b that
formation of nonspherical particles, while particles coded although the crystal lattice of particles are well defined,
N-1 are spherical in shape (figure 2). The spherical parti- the Miller indices are interlocked (buckled) and lattice
cles have an average size of 10 nm. Nonspherical parti- planes have a small bending in the grain boundary. These
762 T N Narayanan et al

Figure 3. (a) TEM, (b) HRTEM and (c) EDS images of N-2.

are evidence for the occurrence of high internal strain tent. If it is not, the same applied field will magnetize it
within the lattice (Vassiliou 1994). along a long axis. Hence the strength of the applied mag-
netic field along the short axis is such that it produces the
3.2 Magnetization studies using vibrating sample same true field inside the specimen. Thus shape alone is
magnetometer (VSM) the source of magnetic anisotropy (Cullity 1972). Thus it
can be inferred from the magnetization curves of N-1 and
The hysteresis loop for spherical (N-1) and nonspherical N-2 that increase in coercivity in the case of N-2 is
iron oxide particles (N-2) are shown in figures 4 and 5. mainly due to the elongation of grains. These ellipsoidal
Nonspherical particles exhibit a saturation magnetization particles may cause a small amount of spin freezing in
of 30 emu/g whereas spherical particles display a satura- random directions. This is one of the reasons attributable
tion magnetization of 37 emu/g and a coercivity of to the decrease in saturation magnetization value for N-2.
105 Oe. Enhancement of coercivity for sample N-2 and The magnetization curve of N-1 shows almost negligible
decrease in its saturation magnetization with respect to remanance and small coercivity showing their super-
N-1 are other evidences for the deviation from sphericity paramagnetic nature.
of sample N-2. If the particles are spherical, the same Additional evidence for superparamagnetism exhibited
applied field would have magnetized it to the same ex- by N-1 can be obtained by simulating the magnetization
 Strain induced anomalous red shift in mesoscopic iron oxide 763

curve using the Langevin function, L(x). Here it is as- The simulated Langevin function using (4) is carried out
sumed that the individual grains are single crystals with and is shown in figure 6. The fitting (figure 6) is carried
the number of crystals equal to the number of domains. out after giving due weightage for particle size distribution.
We have Langevin function represented by Thus simulated curve (linear) fits very well with the ex-
perimental curve (dotted).
1
L( x) = cot hx − . (2)
x
3.3 DTG–TG studies
Size plays a crucial role in deciding the overall magneti-
zation. The variation of particle size is then to be taken Bulk maghemite in the micron regime possesses a satura-
into account. The size distribution obtained from TEM is tion magnetization of ~74 emu/g (Cullity 1972) and coer-
used for simulation. After giving due provision for b civity of 250 Oe (Venkatesh Rao et al 1974), and they are
(width of the size distribution histogram), the Langevin generally multidomain in character. However, as particle
function can be modified as (Chikazumi 1964) size reduced they became single domain and exhibited
x (1 + b ) superparamagnetic characteristics. Although we have
1
L( x) =
2bx ∫ L( x′)dx′ (3) procured evidence for superparamagnetism by Langevin
x (1 − b ) fitting, the reduction in magnetization is to be probed by
supplementary techniques. For this TG–DTG studies are
1 ⎧ (1 − b) sin h[ x(1 + b)] ⎫ carried out. The TG–DTG curves are shown in figure 7.
= ln ⎨ ⎬. (4)
2bx ⎩ (1 + b)sin h[ x(1 − b)] ⎭

Figure 4. Magnetic hysteresis loop for N-1.

Figure 6. Langevin function fit for N-1.

Figure 5. Magnetic hysteresis loop for N-2. Figure 7. TG–DTG curves for N-1.
764 T N Narayanan et al

TG curve shows a continuous weight loss on heating. in the samples (Silverstein and Webster 1964). This –OH
A substantial weight loss corresponding to 8% at 450°K group with hydrogen bonding is coming from starch. Peaks
is observed. The net weight loss is around 14% when the between 400 and 700 cm–1 correspond to Fe–O bonds
temperature reaches 550°K. There is a possibility of a (Raksha Sharma et al 2005).
small fraction of the polymer (particularly starch which
can remain there till 1000°C) being retained in the sample
3.5 Energy bandgap calculation
which normally does not decompose at these tempera-
tures. If due weightage is provided to the retention of
Diffused reflectance spectroscopic (DRS) studies of sam-
polymeric residues, in the final gamma ferric oxide the
ples N-1 and N-2 are conducted using UV-visible NIR
observed saturation magnetization can be accounted for.
spectrophotometer. Energy band calculations are carried
out using the relation (Pankov 1971)
3.4 FT–IR studies
Ai (hν − Egi )mi
α = ∑α = ∑ , (5)
FT–IR studies are carried out on N-1 and N-2 samples in hν
order to ascertain the presence of starch. The spectrum is
where the value of Egi and mi correspond to the energy and
charted in figure 8.
the nature of the particular optical transition with absorp-
The peak at 3384⋅82 corresponds to –OH group and the
tion coefficient, αi. For allowed direct, allowed indirect,
broadening shows the presence of hydrogen bonding.
forbidden direct and forbidden indirect transitions, the
This strengthens the evidence for the presence of starch
value of mi corresponds to 1/2, 2, 3/2 and 3, respectively
(Bhattacharyya et al 1992). Since the particles are
nanocrystalline, the values of Egi and mi are determined
without presuming the nature of electronic transition
(Chakrabarti et al 2004). Equation (5) can be rewritten in
the following form

d (ln(α hν )) m
= . (6)
d (hν ) (hν − E )

The plot of d(ln(αhν))/d(hν) vs hν will produce disconti-

nuity at a particular value of hν = E, where a possible
electronic transition can occur corresponding to the bandgap
E = Egi. Figures 9 and 10 show the above plots for sam-
Figure 8. FT–IR spectrum for N-1 (N-2). ples N-1 and N-2, respectively.

Figure 9. Bandgap calculation from differential plot for N-1. Figure 10. Bandgap calculation from differential plot for N-2.
 Strain induced anomalous red shift in mesoscopic iron oxide 765

Figure 11. ln(hν–E) vs ln(α1hν) curves for (a) N-1 and (b) N-2.

The nature of electronic transition is allowed direct and logical explanation was provided to account for the red
it is verified from figures 11a and b (m= 0⋅5). shift found in the optical bandgap of nonspherical
The red shifted electronic band (~ 0⋅24 eV) is due to the maghemite particles.
presence of ellipsoidal particles which are under very
high stress. This is further supported by HRTEM (figure
Acknowledgements
3b). This shows that, although the crystal lattice of parti-
cles is well defined, the Miller indices are buckled (figure
One of the authors (TNN) thanks the Kerala State Council
3b). From the lattice constants for spherical and non-
for Science, Technology and Environment (D.O. No. 004/
spherical γ-Fe2O3 particles unit cell compression of
FSHIP/05/KSCSTE), Kerala, for financial support in the
(ΔV/V) – 10% was observed. Stress increases as the size
form of a fellowship and (MRA) thanks the All India
reduced. The reduction in size is governed by the relation
Council for Technical Education, Govt. of India (File No.
2γ/r, where γ is pressure inside the grain and r the size of
8023/RID/RPS-73/2004-05).
the grain. So strain induced modification of bandgap pro-
vides a phenomenological explanation for the anomalous
red shift observed in mesoscopic iron oxide particles. The References
effect of matrix or any other impurities in the red shift
has been removed, since the entire measured bandgaps Anantharaman M R, Joseph K V and Keer H V 1997 Bull. Mater.
are well above the red shifted value. Sci. 20 975
Arndt V, Basf A G and Ludwigshafen 1998 Magn. IEEE Trans.
24 1796
4. Conclusions Bate G 1991 J. Magn. & Magn. Mater. 100 413
Bhattacharyya D, Chaudhuri S and Pal A K 1992 Vacuum 43
The employment of a complexing medium like starch for 313
the synthesis of nonspherical iron oxide particles can be Chakrabarti S, Ganguli D and Chaudhuri S 2004 Physica E24 333
modified to tailor the aspect ratio of nonspherical nanoparti- Chikazumi S 1964 Physics of magnetism (New York: John
cles of gamma ferric oxide, since shape controlled parti- Wiley & Sons)
cles are of interest both from a fundamental and application Cohen J B 1990 Ultramicroscopy 34 41
point of view. Cullity B D 1972 Introduction to magnetic materials (Phili-
ppines: Addison-Wesley Publishing Company)
The observed reduction in saturation magnetization
Dallas Panagiotis, Moutis Nicolas, Devlin Eamonn, Niarchos
with respect to the saturation magnetization of bulk Dimitrios and Pentridis Dimitros 2006 Nanotechnology 17
maghemite is due to the presence of superparamagnetism 5019
and residual organic impurities in the final product. Deng M C, Hsu S L and Chin T S 1992 IEEE Trans. Magn. 28
Strain induced modification of bandgap is possible by 2385
carefully controlling the acicularity of these particles. Hyeon Taeghwan 2003 Chem. Commun. 927
Shape control can also induce anisotropy and thus modify Kachkachi H, Ezzir A, Nogues M and Tronc E 2000 Eur. Phys.
the coercivity of these particles. A simple phenomeno- J. B14 681
766 T N Narayanan et al

Khaleel Abbas A 2004 Chem. Eur. J. 10 925 Silverstein Robert M and Webster Francis X 1964 Spectro-
Nair Swapna S, Mathews Mercy and Anantharaman M R 2005 scopic identification of organic compounds (New York: John
Chem. Phys. Letts 406 398 Wiley & Sons)
Pankov J J 1971 Optical processes in semiconductors (New Vassiliou John K, Mehrotra Vivek, Russell Michal W, McMichael
Jersey: Prentice-Hall) R D, Shull R D and Ronald F Ziolo 1993 J. Appl. Phys. 73 10
Raksha Sharma, Subhalakshmi Lamba and Annapoorni S 2005 Venkatesh Rao, Shashimohan A L and Biswas A B 1974 J.
J. Phys. D: Appl. Phys. 38 3354 Mater. Sci. 9 430
Sabino Veintemillas-Verdaguer et al 2004 J. Phys. D: Appl. Viswanathan B 1990 Ferrite materials: Science and Technology
Phys. 37 2054 (New Delhi: Narosa Publishing House)