You are on page 1of 5

Journal of Magnetism and Magnetic Materials 111 (1992)29-33

North-Holland

Magnetocaloric effect in superparamagnets


R.D. McMichael +, R.D. Shull, L.J. Swartzendruber, L.H. Bennett
National Institute of Standards and Technology, Gaithersburg, MD 20899, USA

and

R.E. Watson *
Brookhaven National Laboratory, Upton, NY 11973, USA

Received 22 August 1991; in revised form 2 December 1991

The magnetocaloric effect is calculated for superparamagnetic materials as a function of temperature, field and cluster
size. Assuming classical behavior, a universal curve is calculated from which an optimum cluster moment may be found for
maximum entropy change upon application of a given field H at a given temperature T. Quantum effects are shown to be
small for temperatures above 10 K and fields less than a few tesla. A comparison with results for a spin -7 paramagnet such
as GdaGasOx2 (GGG) is made, which indicates that superparamagnetic materials such as magnetic nanocompositeo -ffer
the possibility of extending the upper useful temperature limit of paramag~etic materials for magnetic refrigeration.

I. Introduction use cf Dy3AIsO12 [4]. For higher operating tem-


peratures, the magnetocaloric properties of ferro-
In recent years, there has been increasing in- magpeti2 materials have been studied for use
terest shown in the magnetocaloric properties of with the operating temperature at or near the
materials for use in magnetic refrigerators [1,2], Curie point of the material [5-9].
especially for cryogenic refrigerators for use-in The magnetocaloric effect is the adiabatic tem-
specialized aerospace applications, although heat perature change of a material upon application of
pumps for use near room temperature have been a magnetic field. It is Very closely related to the
constructed as well [3]. For refrigerators operat- isothermal field induced entropy change which,
ing in the temperature range between 1.5 and 20 for small temperature changes, differs from the
K, the refrigerant material of choice has been magnetocaloric effect by a factor of the tempera-
almost exclusively Gd3GasOlz (GGG), although ture divided by the specific heat of the material.
the upper temperature limit of paramagnetic re- By grouping spins together in superparamagnetic
frigerants has been increased somewhat by the clusters, the magnetic moments are more easily
aligned than in paramagnetic systems, and for
certain ranges of field, temperature and cluster
Correspondence to: Dr. R.D. McMichael NIST, Bldg. 223, size, the entropy of the spins is more easily
Rm B152, Gaithersburg, MD 20899, USA. changed by application of a field [10]. In this
+ National Research Council Postdoctor~d Research Associ-
paper, we calculate the magnetocaloric effect for
ate.
* Supported by the Division of Materials S,.iences, US DOE the entire range of parameters, demonstrating
under contract DE-AC02-76CH00016. how grouting atoms in superparamagnetic clus-

0304-8853/92/$05.00 © 1992 - Elsevier Science Publishers B.V. All rights reserved


30 R.D. McMichael et al. / Magnetocaloric effect in superparamagnets

ters can produce larger entropy changes than in where k is Boltzmann's constant. The magnetic
atomic paramagnets. An optimum cluster size is entropy S c of the sytem can then be found from
determined for a given temperature and applied the thermodynamic relation
field.
The entropy change of a magnetic system can o( :r In z)
be calculated from the thermodynamic Maxwell Sc = - i)T
equation,

(0s) -- , (1)
--Nk [ l n ( 4 r r ) + In (sinh ( __~_
)/(mH --~mH))

T "~ Ii
+l-~coth --~ . (5)
or, integrating over the field,

This expression for the entropy of the magnetic


= dH. (2) moments has a value of Nkln(4~-) at H - 0,
tt
which is analogous to tne quantum-mechanical
As long as the magnetic equation of state, expression, Nk!n(2J + 1), for N moments of an-
M(H, T), is known, the magnetic entropy change gular momentum J. The limit of high H and low
can be calculated. T is S c = Nk[ln(4rr) + 1 - ln(2mH/kT)] which
A sum rule can be derived from eq. (2) con- becomes negative for large H and low T. This
cerning the area under a plot of AS vs T for a anomaly is due to the classical nature of the
given applied field, H. system and is related to the fact that in the
classical system there is no minimum energy scale
comparable to gl.tBH in the quantum system
oO

fo AS dT= -HM o. (3)


where g is the gyromagnetic ratio and /.t B is the
Bohr magneton.
This sum rule implies that for materials with the By increasing the field from 0 to H, the en-
same saturation magnetic moment, M0, those with tropy changes by
high entropy change at a given temperature will
have low entropy changes at other temperatures, ASc = S ( H ) -S(0)
and those materials which do not have a large
entropy change at any particular temperature can = Nk 1 k r coth ~
undergo a moderate entropy change over a
broader temperature range.
+ In (sinh[ mH mH (6)
t t
2,7 Classical superparamagnetic case or, in terms of the unitless variable x = mH/kT,
the entropy change can be written as
In order to avoid the integration specified in
eq. (2), the magnetic entropy of N independent T sc 1[ =- 1-xcoth x+ln
(sinhxtl
- . (7)
magnetic moments m, in a field H, at tempera- M 0H x L x ,a
ture T, is calculated directly through the partition
fucntion Z, assuming classical behavior of the A plot of this function gives the universal curve
moments: and is shown in fig. 1. This curve has a maximum
at

mH
t mH -k--T- ' (4) x= kT = 3.5, (S)
R.D. McMichael et al. / Magnetocaloric effect in superparamagnets 31

0.28 saturation effects wi!i occur at higher tempera-


tures and lower field, eq. (9) shows that AS in the
0.24
optimized superparamagnetic material is propor-
0.20' tional to H / T .
: ~ 0.16

~ 0.12 3. Quantum superparamagnetic cas,


0.08
Because of the anomalous behavior of the
0.04 entropy of classical magnetic moments in large
fields at low temperatures, it is useful to compare
0
0 ' a ' ,i ' 6 ' a ' 10 the classical results with quantum-mechanical re-
mH/kT sults. The entropy of a collection of spinal parti-
Fig. 1. Entropy change AS of a superparamagnet with satura-
cles is given by
tion moment M 0 as a function of particle moment m, induced
by removal of a field H at a temperature T. For a given value I (J+ ½)gl.LBH (J+½)gl~B H
of H and T, the maximum entropy change occurs at m H / k T coth
= 3.5 where TAS = 0.272MoH.
S O = Nk - kT kT
gi~BH gP~BH
+ coth
where the entropy change is given by
Moll
AS c --- - 0 . 2 7 2 - - - - ~ (9)
2kT

+ In sinh
(i 2kT
(J+2)g#a H )
kT
for constant values of M 0, H and T.
The broadness of the maximum in eq. (7) - In sinh ( g/.t2kTB
H ) (ll)

shows that the entropy change for superparamag-


netic particles drops off slowly as particle sizes At H = 0, this expression, reduces to the familiar
deviate from optimum. This has importance in result S(H = O) = Nk ln(2J + l) and upon appli-
that, in any real system, there will be a distribu- cation of a small field, the entropy changes by
tion of particle sizes, but the performance of the
material will not be drastically degraded by a Ng21J,2BJ(J+ 1 ) H 2
particle size distribuAon. The decrease in AS c is ASo= - 6kT2 (12)
due to a decreasing number of clusters as the
cluster size grows. To compare the entropy changes described by the
In the low field limit, m H / k T << 1, eq. (6) classical and quantum forms, the quantity
gE/.LEJ(J + 1) is identified as m 2, and in the case
reduces to
of large J, m = gl~B(J + ½). Taking the difference
-Nm2H 2 -mMoH 2 between the classical and quantum entropy
= (10)
AS C = 6kT 2 6kT 2 changes, it is found that

Note that in this limit, AS c increases linearly g/zaH gp.aH


AS o - AS c = Nk 1 -, ,------~c o t h . , ,-----~
with m when the saturation magnetization M 0 is L /-,~1 ,~ ~.. 1
held constant. For paramagnets with magnetic
moments consisting of single atoms, eq. (10) is a
good approximation except for very low tempera-
tures and very high fields, where saturation ef-
fects can be seen. In the low field limit, the
entropy change is proportional to H Z/T 2. In which is the clas~tcal expression for the eatropy
contrast, in a superparamagnetic material where clmnge of a spin-5i paramagnet, which will be
32 R.D. McMichael et al. / Magnetocaloric ef,t%:t in superparamagnets

1.0 the peak in the heat capacity of superparamag-


netic particles [11] located near kBT/KV=0.5,
! • spin 7/2 atoms
0.8
',, ...... 10 atom clusters
where K is the anisotropy constant and V is the
[... ", ............30 atom clusters volume of the particle. The superparamagnetic
o 0.6 :,...",-..... 100 atom clusters blocking temperature is lower, at k BT/KV = 0.04.
,E Assuming an anisotropy field of 0.1 T, kBT/KV
~,.~--0.4 = 1 at 50 K for the 100 Gd atom clusters, sug-
.-._ ~i -'~'~'':' gesting that the effects of anisotropy on zero-field
o.2 "''~'v-o~,~ " ' ..... entropy will only be important below ~ 30 K,
given the assumed cluster size and anisotropy
i , , , , field.
o 2'0 4'o 6'o 8'o oo Dilution of the magnetic species by a non-mag-
i"ernp (K)
netic matrix material will, of course, reduce the
Fig. 2. Entropy change of a system of spin -7 Gd atoms
induced by removal of a 1 T field for individual spins (GGG) entropy change per unit mass or per unit volume.
and clusters of 10, 30 and 100 spins. The 10 atom clusters The degree to which this dilution effect can be
have the optimum size at 15.3 K, and the 30 atom clusters minimized by using low volume fractions of host
have the optimum size at 46 K. material depends strongly on the materials se-
lected and on the processing, but in certain cases
superparamagnetic behavior has been observed
small for temperatures above 10 K and fields less with less than 10% host material [12]. By keeping
than a few Tesla. the volume fraction of the host material iow,
A comparison of entropy changes for different however, interactions between the particles can
arrangements of Gd atoms under a field of 1 T become important, and this area is currently be-
calculated using the quantum mechanical expres- ing pursued.
sion is shown in fig. 2. At low temperatures, the It may be possible to overcome the potential
larger cluster moments are nearly aligned and the problems mentioned above because the entropy
entropy change is larger for the more numerous change for an optimized superparamagnet is pro-
and smaller atomic moments. At higher tempera- portienal to H/T rather than H Z / T 2. This may
tures, the larger clusters show the higher entropy make it possible to produce large entropy changes
change, with decreased entropy changes at low at higher temperatures, or to produce equivalent
temperatures in accordance with the sum rule entropy changes with reduced fields. In addition,
given in eq. (3). by regulating the particle size, the entropy change
at a given temperature for a given applied field
can be maximized, with performance falling off
4, Conclusions only slowly as particle sizes deviate from the
optimum. This capability for tuning the material
The main result of these calculations is that could prove to be very useful for refrigeration
nanocomposite superparamagnetic m,terials have applications.
the potential to exceed the performm~ce of GGG,
the most commonly used refrigerant in cryogenic
magnetic refrigeration, especially when using low
JL"; ~ 1 ..,,! _ _,tA. _ ? I_1
t~c~u:s attmnaole with permanent magnets. Poten- lilt1.. It II..It I,,.IUl I ~ If..¢.'J

tial problems facing technical ~pplication of this


class of materials include a reduction in the [1] J.A. Barclay and W.A. Steyert, Cryogenics 22 (1982) 73.
zero-field entropy resulting from anisotropy of [2] T. Hashimoto, Adv. Cryog. Eng. Mat. 32 (1986) 261.
[3] For a review of the field, see John A. Barclay, Adv.
the superparamagnetic particles at low T. The
Cryog. Eng. Mat. 33 (1988) 719.
temperature at which such zero-field entropy re- [4] T. Hashimoto, T. Yazawa, R. Li, T. Kuzuhara, K. Mat-
duction becomes important can be estimated from sumoto, H. Nakagome, M. Takahashi, M. Sahashi, K.
R.D. McMichael et al. / Magnetocaioric effect in superparamagnets 33

lnomata, A. Tomokiyo and H. Yayama, Adv. Cryog. Eng. T. Hashimoto, M. Sahashi and K. lnomata, Adv. Cryog.
Mat. 33 (1988) 733. Eng. 32 (1986) 295.
[5] T. Hashimoto, T. Numasawa, M. Shino and T. Okada, [10] R.D. Shull, L.J. Swartzendruber and L.H. Bennett, Proc.
Cryogenics 21 (1981) 647. 6th Intern. Crs;ocoolers Conf., eds. G. Green and M.
[6] H. Oesterreicher and F.T. Parker, J. Appi. Phys. 55 Knox, David Taylor Research Center Publ. no. DTRC-
(1984) 4334. 91/002, Annapolis, MD (1991) p. 231.
[7] G.V. Brown, J. Appl. Phys. 47 (1976) 3673. [11] J.D. Livingston and C.P. Bean, J. Appl. Phys. 32 (1961)
[8] T. Hashimoto, K. Matsumoto, T. Kurihara, T. Nu- 1964.
mazawa, A. Tomokiyo, H. Yayama, T. Go~o, S. Todo and [12] R.D. Shuli, U. Atzmony, A.J. Shapiro. L.J. Swartzendru-
M. Sahashi, Adv. Cryog. Eng. 32 (1986) 279. ber, L.H. Bennett, W.J. Green and K. Moorjani, J. Appl.
[9] A. Tomokiyo, H. Yayama, H. Wakabayashi, T. Kuzuhara, Phys. 63 (1988) 4261.

You might also like