Available online at www.sciencedirect.

com
Transactions of
Nonferrous Metals
Society of China
Science Trans. Nonferrous Met. SOC.China 17(2007) 1451-1455
Press ~

www.csu.edu.cn/ysxb/

Size and shape effects on Curie temperature of ferromagnetic nanoparticles

CAO Ling-fei(E@ X)’,XIE Dan(# f’l-)’, GUO Ming-~ing($R@@)~,

H.S. Park’, T. Fujita’
1. School of Engineering, University of Tokyo, Bunkyo-ku, Tokyo 113-8656, Japan;
2. Laboratory of Crystallography, Department of Materials, ETH-Zurich, CH-8093 Zurich, Switzerland;
3. School of Materials Science and Engineering, Central South University, Changsha 41 0083, China
Received 15 July 2007; accepted 10 September 2007

Abstract: A simplified model was developed to describe the Curie temperature suppression of ferromagnetic nanoparticles. Based on
a size and shape dependent model of cohesive energy, the critical temperature variations of ferromagnetic nanoparticles were
deduced. It is predicted that the Curie temperature of nanoparticles depends on both size and shape conditions, among which the
temperature suppression is strongly influenced by the particle size and the shape effect is comparably minor. The calculation values
for freestanding nanoparticles are in good agreement with other theoretical model and the experimental results. The model is also
potential for predictions for the nanoparticles embedded in different siibstrates.

Key words: nanoparticle; size effect; shape effect; ferromagnet; Curie temperature

bulk value Tc(o), and the relationship can be expressed

1 Introduction by a step function as

In recent years, ferromagnetic nanoparticles have

been extensively investigated due to their scientific and
industrial importance, and unique properties resulting
from their ultrafine sizes. It has been reported that when
the size reduces to a nanometer, ferromagnetic solids
may exhibit superparamagetism[ 11, higher coercive
force[2], giant magnetoresistance[3], phase transforma- where D is the size of nanocrystals, r denotes the
tion[4] and lower Curie temperature[5-9]. The last point thickness of a monolayer, and 1 is the corresponding
of Curie temperature Tc suppression, namely, Tc of critical exponent. Though the original model is limited in
ferromagnetic nanoparticles decreases with decreasing a thin film system, the pioneering conception strongly
diameter, is quite interesting for the development of new impacts general understandings on the Curie temperature
functional materials, such as switches functioning in a suppression. For example, NIKOLAEV and SHIPILM
designed temperature range with tunable Tc, temperature [ 1 I] have proposed a Tc(D) function for ferromagnetic
sensitive ferromagnets applied in magnetic fluids[ 101. nanoparticles as
Some models have been developed to understand
the experimental observations of Tc suppression.The first (3)
one should be mentioned is the scaling theory established
by FISHER and BARI3ER[6] for finite-size effects in where AL is the thickness of surface layer of
ferromagnetic systems. Based on the spin-spin nanoparticles, characterizing the influence of the surface
correlation length (SSLC, r ) mechanism, it is predicted layer on the Curie temperature in light of exchange bond
that if the nanosolid size is smaller than the critical SSCL, variations. However, when this model is utilized to fit
the Tc will shift to a lower temperature Tc(D) than the experimental data for Fe304nanoparticles of different

Foundation item: Project(FY2006) supported by the JSPS Postdoctoral Fellowship For Foreign Researches; Project supported by the 21 st Century COE
Program, “Mechanical Systems Innovation,” by the Ministry of Education, Culture, Sports, Science and Technology, Japan
Corresponding author: CAO Ling-fei; Tel: +8 1-3-5841-7079; Fax: +81-3-5841-7075; E-mail: caolingfei@gmail.com
1452 CAO Ling-fei, et al/Trans. Nonferrous Met. Soc. China 17(2007)
sizes, a constant AL is unsuccessfd in the full size range even disappear. At that time, the spin-spin interaction is
of nanometer. This may be caused by the variation of AL disordered by thermal stimulus, and ferromagnet
with the solid size, and the relationship between AL and becomes paramagnet of magnetic disorder. To destroy
D is expected to be established. the magnetic exchange, sufficient energy has to be
Another approach has been reported by SUN et a1 provided to break all atomic bonds and promote the
[7].In their work, the bond order-length-strength(B0LS) atoms for thermal vibration. During this process, the total
correlation mechanism is incorporated into the king energy, or exchange interaction energy Eex,(T) can be
premise to advance a unified expression for TC regarded to equal the sum of the cohesive energy E and
predictions of ferromagnetic, ferroelectric and the thermal vibration energy Ev(r) needed to disorder the
superconductive nanosolids. The lower Curie spin-spin interaction, namely,
temperature is related to the decrease of atomic cohesive
Eexd r)=E+Ev(T) (4)
energy that caused by the coordination number(CN)
imperfection of the lower coordinated atoms near the Based on a mean field approximation, the thermal
surface edge. In other terms, JIANG et a1[8,12] have vibration energy has a proportional relationship with
discussed the size and interface effects on the critical temperature, i.e., E,(T)=kBT, where kB is the Boltzmann
temperatures mentioned above. Based on a constant. At Curie temperature, the thermal vibration
size-dependent cohesive energy model, a unified energy required to disorder the exchange interaction is a
function is modeled to predict the Tc suppression with portion of the atomic cohesive energy at T=O[7-8,
decreasing nanocrystal size. The same prediction is also 17-18]. From this point of view, the Curie temperature
obtained in their discussion on the Curie transition can be regarded directly proportional to the cohesive
temperature of ferromagnetic low-dimensional metals energy ( T C x E ) as a first order approximation, since the
(particles, wires and thin films) with different nature of any phase transition is related with the
morphologies in full size range. potentials of the two related phases. As an empirical
All these developed models are significant to result,
understand the nature of Tc suppression of ferromagnetic
nanosolids from different perspectives. However, another
important aspect of particles, the shape of nanoparticles
has not been paid enough attention to in some work.
where E and Tc denote cohesive energy and Curie
Currently, the shape of nanoparticles is generally
temperature, and the subscripts n, b denote nanoparticle,
regarded as spherical type in thermal calculation, while
corresponding bulk respectively. According to Eqn.(5),
depending on different production conditions, particles
the Curie temperature Tc should have the same size and
may also exhibit cubic, columnar, polyhedral and other
shapes. For a nanoparticle with the same size, different shape dependence as cohesive energy E. In this term, to
shapes lead to different specific surfaces, which can figure out the size and shape effects on cohesive energy
remarkably influence its physicochemical properties becomes critical to develop a size and shape dependent
[13-161. Therefore, variations on the shape of model for the Curie temperature of ferromagnetic
nanoparticles should not be ignored. For a better nanoparticles. This concept is helphl to understand the
understanding of both size and shape effects on the Curie Curie temperature variations of ferromagnetic
temperature variation of ferromagnetic nanoparticles, a nanoparticles, especially for the comparison within the
simplified model was developed in this work. Based on same series of sample system.
the cohesive energy model and the relationship between
Curie temperature and cohesive energy, the temperature 3 Results and discussion
variations for freestanding ferromagnetic nanoparticles
were described. The Curie temperature of nanoparticles 3.1Calculation of cohesive energy
embedded in substrate films was discussed as well. The cohesive energy equals the energy needed to
destroy all bonds when dividing the crystal into isolated
2 Model atoms, namely, the cohesive energy is directly
determined by the product of the bond amount and its
The Curie temperature is a critical parameter for unit energy[ 15- 16, 19-20]. Supposing the bond energy
ferromagnetic nanoparticles, which is determined by the equals each other for the same atom, the cohesive energy
spin-spin exchange interaction. Upon a temperature will depend on its bond amount. For nanoparticles, due
higher than the Curie point of ferromagnets, the to the well-known surface effects, the values of cohesive
exchange interaction caused by the spin magnetic energy are not all the same for the superficial part and
moment of neighboring 3d electrons will be very weak or interior part, caused by their thermal behavior difference.
 CAO Ling-fei, et al/Trans. Nonferrous Met. SOC. China 17(2007) 1453
One case in point is their atomic relaxation caused by as
surface coordination-number imperfection[2 11. Such a
difference for finite particles cannot be ignored, so the
cohesive energy of nanoparticles should be the sum of
the superficial part and the interior part. Accordingly, we where n denotes the atomic number of nanocrystals, C
obtain the surface-to-volume atomic ratio dependent is the atomic number of one structure cell, and k is the
expression for cohesive energy of nanoparticles as ratio between equivalent atomic radius and lattice
follows: parameter. For FCC, BCC and HCP structures, C are 4, 2
and 2, and k are & / 4 , & I 4 and 1/2, respectively
En=(1-a)Eb+aEs (6)
~41.
where the subscript 's' represents the surface, a means Substituting Eqn.( 10) into Eqn.(7), the cohesive
the surface-to-volume atomic ratio. Supposing the bond energy for freestanding nanoparticles can be described as
amount of interior atom is m, the bond amount of a ~ 3 1
surface atom with interior atoms is (1/4) m for simplicity.
Such opinion is supported by Ref.[22], where the surface
relaxation is regarded as only about one-fourth of the
area of each surface atom embedded in the lattice. Then As shape variation is significant to the thermal
we obtain Eqn.(7) for freestanding nanoparticles and properties of nanoparticles, a shape factor ,u can be
Eqn.(8) for embedded nanoparticles[ 15-1 61: defined as the ratio of two surface areas to describe the
3 difference between non-cubic and cubic particles, i.e.,
(7)
Eb

where S is the surface area of the particle in random

where the subscript M denotes the embedding matrix, q shape, and Scubic is the surface area of the cubic particle,
represents the misfit between nanoparticles and matrix. which is supposed to have the same volume as the former.
In a case of incoherent interface between nanoparticles The conception of shape factor is especially practical
and matrix, q=O; and for a completely coherent interface, benefit when the experimental observation of particle
q=l. One point that should be noted is that behaviors shape is combined with theoretical prediction. For
between freestanding nanoparticles and embedded ones example, the particle shape may be determined by the
with incoherent interface are not quite the same. So if set high resolution transmission electron microscopy
q=O in Eqn.(8), En/Eh=1-(3/8)a will be obtained for an (HRTEM), then the shape factor can be figured out and
incoherent interface between particles and matrix, which used in an exact prediction, such as thermal stability of
is not consistent with Eqn.(7) for freestanding nanomaterials[ 161.
nanoparticles. A possible reason for such phenomenon Combining Eqns.( 11) with (12), the cohesive energy
may be that the thermal vibration amplitudes of surface of freestanding nanoparticles with random shape can be
atoms are less than those of interface atoms, which leads described as
to the above different behaviors between freestanding En -1-
-- w
1 I3 c2 13 .Ilk2
and embedded nanoparticles. Eb
In Eqns.(7) and (8), a denotes surface-to-volume
Based on the same conception, the cohesive energy
atomic ratio, where the surface details should be noted.
of nanoparticles under embedded condition is
Usually, the nanoparticle is simplified as spherical shape
for calculation. Then according to the definition, the E"
--I- - 3p (1-p-) E M
surface-to-volume atomic ratio can be described as Eb
1i 3 c2I 3 .Ilk2 Eb

3 . 2 Calculation of Curie temperature

Based on the size-and-shape dependent model of
cohesive energy, the Curie temperature of ferromagnetic
where d is the atom diameter, R is the nanoparticle nanoparticles can be further deduced. Considering that
radius, and the surface is assumed to be shell structure the Curie temperature is proportional to the cohesive
consisting of one layer of atoms. Similarly, supposing the energy, it is reliable to express the Curie temperature of
shape of the nanoparticle is ideal cubic,a can be deduced nanoparticles with both size and shape dependence as
1454 CAO Ling-fei, et al/Trans. Nonferrous Met. SOC.China 17(2007)
For freestanding nanoparticles: numerically shown in Fig. 1 with the reported
TC" -
--I- 6Y experimental values. From Fig. 1, it is clear that the Curie
n 113 c 2 13 ,
&2 temperatures of ferromagnetic nanoparticles decrease
TCh
with the decreasing size (described by the amount
For embedded nanoparticles: variation of atoms), which is in good accordance with the
experimental observations mentioned before. The curves
in Fig.1 are also compatible with the prediction by other
theoretical models[7-8,121. It is also interesting to find
From the above expressions of Curie temperature in that the Curie temperature difference caused by shape
Eqns.( 15) and (1 6), it is obvious that when the atoms that effect becomes obvious only within a quite ultrafine size
constitute a nanoparticle are determined, the parameters range, i.e. below 10 nm. Therefore, in a general situation,
C and k are fixed. Then the Curie temperature of the more emphases should be put on the size effect.
nanoparticle is strongly dependent on its shape factor and Furthermore, compared the calculation curves with the
atomic number. Here the atomic number can be regarded available experimental data for Ni nanoparticles, it seems
as one index of particle size, as it is natural that more that the experiment data are well located in or between
atoms lead to larger size of nanoparticles. The values of our prediction lines representing nanoparticles in cubic
shape factor for nanoparticles in different shape can be shape and spherical shape. Considering that the
calculated according to their geometric characters and measured particles are approximating sphere together
the definition in Eqn.( 12). Calculation results for some with other shapes, our prediction is in good accordance
typical shapes are listed in Table 1. Accordingly, it's safe with the experimental data.
to say that the Curie temperature of a nanoparticle
depends on its size and shape effects, as well as its I400
cohesive energy.
1200
Table 1 Values of shape factor ,u for nanoparticles
Shape Shape factor ,u 1 000
Cube 1 x
Sphere 0.806 0
2 800.

Octahedron 0.953 2 600

Pentagonal decahedron 1.011 0
Regular tetrahedron 1.049 1 400
Ellipsoid, Cylinder, Six-edge column 30.806 0
Tetragonal, Rhombohedra1 31 1.0 1.5 2.0 2.5 3.0 3.5 4.0
tg(n113)
Meanwhile, comparison between Eqns.( 15) and (16) Fig.1 Tc function for freestanding Fe, C, Ni nanoparticles in
suggests that the size and shape effects for freestanding terms of Eqn.( 15) shown as solid lines for spherical shape and
nanoparticles will be more remarkable than those for dash lines for cubic shape
embedded ones, as the latter depends on more variables
in Eqn.( 16). This phenomenon can be partly explained in For the systems of embedded nanoparticles, Curie
the way that the freestanding materials have a larger temperature also depends on the interaction between the
surface-to-volume atomic ratio, so that their size and particle and substrate, together with the parameters
shape dependence is stronger than the embedded mentioned above. When there is no or a weak interaction
particles. For a certain nanoparticle, for example, at the interface, the embedded nanoparticles may behave
freestanding Fe nanoparticles in cubic shape, the similarly to freestanding nanoparticles, namely the Curie
parameters p, C, k and Tcb in Eqn.( 13) are 1,2, & / 4 and temperature will decrease with the decreasing size. In
1 043 K, respectively. Then the relationship between the case of a strong interaction or higher value of Curie
Curie temperature of nanoparticles and their atomic temperature of the matrix (TCM in Eqn.( 16)), the Curie
number can be plotted. Under different conditions, the temperature of the embedded nanoparticle maybe change
parameters may change and the corresponding values of in an opposite way. Such phenomenon can also be
Curie temperature can be obtained. explained in a view of energy variation. It has been
reported that there exist different degrees of spin-spin
3.3 Model validity and discussion interactions between inner and surface atoms, caused by
To validate the reliability of our model, the Curie the number reduction of spin interactions on the surface.
temperatures of ultrafine Fe, Co and Ni particles are This will lead to a variation of lattice vibration.
 CAO Ling-fei, et aliTrans. Nonferrous Met. SOC. China 17(2007) 1455
Therefore, when the size is decreased, the total energy, or [5] ZHONG W H. SUN C Q. LI S, BAI H L, JIANG E Y. Impact of
bond-order loss on surface and nanosolid magnetism [J]. Acta Mater,
the thermal vibration energy to disorder the 2005, 53(11): 3207-3214.
ferromagnetic ordering state may change with the [6] FISHER M E, BARBER M N. Scaling theory for finite-size effects
interaction conditions[26]. Until now, magnetic particles in the critical region [J]. Phys Rev Lett, 1972, 28: 1516-1519.
[7] SUN C Q, ZHONG W H, LI S, TAY B K, BAI H L, JIANG E Y.
embedded in Ag or Cu matrix are two major types of
Coordination imperfection suppressed phase stability of
particle-film giant magnetoresistance(GMR) materials in . ferromagnetic, ferroelectric, and superconductive nanosolids [J]. J
industry and research work. However, their Curie Phys Chem B, 2004, l08(3): 1080-1084.
temperature variations for the embedded nanoparticles YANG C C, JIANG Q. Size and interface effects on critical
temperatures of ferromagnetic, ferroelectric and superconductive
are not easily available at present. Eqn.(l6) with nanocrystals [J]. Acta Mater, 2005, 53(11): 3305-331 1.
parameters properly chosen will be helphl to be referred EVANS R, NOWAK U, DORFBANER F, SHREFL T,MRYASOV 0,
in theoretical investigation and practical production. CHANTRELL R W, GROCHOLA G. The influence of shape and
structure on the Curie temperature of Fe and Co nanoparticles [J]. J
Furthermore, based on the definition of Appl Phys, 2006,99: 080703-1-3.
surface-to-volume atomic ratio a, there should be a direct BERKOVSKI B, BASHTOVOV V. Magnetic fluids and applications
relationship among the values of a for nanoparticles, handbook [MI. Wallingford: Begell House, 1996.
NIKOLAEV V I, SHIPILIN A M. The influence of breaking of
nanowires and nanofilms with the same size R. For
exchange bonds on the Curie temperature [J]. Phys Solid State, 2003,
nanoparticles, R means the normal radius; for nanowires, 45: 1079-1080.
R denotes the radius of the cross section; and for CUI X F, ZHAO M, JIANG Q. Curie transition temperature of
nanofilms, R represents the half value of the thickness. ferromagnetic low-dimensional metals [J]. Thin Solid Films, 2005,
472(1/2): 328-333.
Accordingly, it can be deduced that for different XIE D, QI W H, WANG M P. Size and shape dependent
low-dimensional systems, the surface-to-volume mole melting-thermodynamic properties of metallic nanoparticles [J].
ratio is in the ratio of ( ~ , ~ ~ ~ ~ ~ : a ~ ~ ~1,: awhere
, , , = asphere,
3:2: Acta Metall Sin, 2004,40(10): 1041-1044. (in Chinese)
WAUTELET M, DAUCHOT J P, HECQ M. On the phase diagram of
afilm and a,,, are the surface-to-volume atomic ratio for non-spherical nanoparticles [J]. J Phys Condens Matt, 2003, 15:
nanoparticles, nanowires and nanofilms, respectively 3651-3655.
[16]. Then the above cohesive energy formulae and XIE D, WANG M P, QI W H. A simplified model to calculate the
surface-to-volume atomic ratio dependent cohesive energy of
Curie temperature for particles can be extended to
nanocrystals [J]. J Phys: Condense Mater, 2004, 16: L401-L405.
describe the properties of corresponding wires or films. CAO L F, XU G Y, XIE D, GUO M X, LUO L, LI Z, WANG M P.
Thermal stability of Fe, Co, Ni metal nanoparticles [J]. Phys Status
4 Conclusions Solid (B), 2006,243: 2745-2755.
ZHANG Ren-jun, WILLlS R F. Thickness-dependent Curie
temperatures of ultrathin magnetic films: Effect of the range of
1) A simplified model is developed for the size-and- spin-spin interactions [J]. Phys Rev Lett, 2001, 86: 2665-2668.
SUN C Q, WANG Y, TAY B K, LI S, HUANG H, ZHANG Y B.
shape dependent Curie temperature of ferromagnetic Correlation between the melting point of a nanosolid and the
nanoparticles, based on the relationship between the cohesive energy of a surface atom [J]. J Phys Chem B, 2002, 106(41):
Curie temperature and cohesive energy. 10701-10705.
SUN C Q, TAY B K, ZENG X T, LI S, CHEN T P, ZHOU J, BAI H
2 ) The model predicts a decrease of Curie
L, JIANG E Y. Bond-order-bond-length-bond-strength (bond-OLS)
temperature with decreasing size of ferromagnetic correlation mechanism for the shape-and-size dependence of a
nanoparticles. The shape of nanoparticles also plays an nanosolid [J]. J Phys: Condense Mater, 2002, 14: 7781-7795.
important role on the critical temperature variation XIE Dan, QI Wei-hong, WANG Ming-pu. Equivalent model on
melting and superheating of metallic nanocrystals [J]. Acta Metall
within a quite ultrafine size range. Sin, 2004,41(5): 458-462. (in Chinese)
3) The simple model developed in this work agrees KIM H K, HUH S H, PARK J W, JEONG J W, LEE G H. The cluster
well with the available experimental data and other size dependence of thermal stabilities of both molybdenum and
tungsten nanoclusters [J]. Chemical Physics Letters, 2002, 354(112):
theoretical predictions. 165-172.
ADAMSON A W. Physical chemistry of surface [MI. New York:
References , John-Wiley & Sons Inc, 1990.
XIE Dan, WANG Ming-pu, CAO Ling-fei. A simplified model to
calculate the higher surface energy of free-standing nanocrystals [J].
TANG N J, ZHONG W, JIANG H Y, WU X L, LIU W, DU Y W. Phys Status Solid (B), 2005, 242(8): R76-R78.
Nanostmctured magnetite (Fe304) thin films prepared by sol-gel ASKELAND D R, PHULE P P. The science and engineering of
method [J]. J Magn Magn Mater, 2004, 282: 92-95. materials [MI. New York: Thomson Learning Inc, 2003: 84.
QIU Z Q, DU Y W, TANG H, WALKER J C. SYNK J E, VENDULA DU You-wei, XU Ming-xiang, WU Jian, SHI Ying-bing, LU
K. A Miissbauer study of fine iron particles [J]. J Appl Phys Rev, Huai-xian, XUE Rong-hua. Magnetic properties of ultrafine nickel
1988,63(8): 4100-4104. particles [J]. J Appl Phys, 1991, 70: 5903-5905.
MILNER A, GERBER A, GROISMAN, KARPOVSKY M, LANG X Y, ZHENG W T, JIANG Q. Size and interface effects on
GLADKIKH A. Spin-dependent electronic transport in granular ferromagnetic and anti ferromagnetic transition temperatures [J], Phys
ferromagnets [J]. Phys Rev Lett, 1996, 76: 475-478. Review B, 2006.73: 224444-1-8.
MENG Q, ZHOU N, RONG Y, CHEN S, HSU T Y. Size effect on (Edited by LI Xiang-qun)
the Fe nanocrystalline phase transformation [J]. Acta Mater, 2002,
50( 18): 4563-4570.