ARTICLE IN PRESS

Journal of Magnetism and Magnetic Materials 320 (2008) 2330 2334

Contents lists available at ScienceDirect

Journal of Magnetism and Magnetic Materials

journal homepage: www.elsevier.com/locate/jmmm

Site occupancy and magnetic study of Al3+ and Cr3+ co-substituted Y3Fe5O12
K. Bouziane a,, A. Yousif a, H.M. Widatallah a, J. Amighian b
a
Department of Physics, College of Science, Sultan Qaboos University, P.O. Box 36, PC 123, Muscat, Sultanate of Oman
b
Department of Physics, University of Isfahan, Isfahan, Iran

a r t i c l e in f o a b s t r a c t

Article history: Single-phased polycrystalline Y3Fe52xAlxCrxO12 garnet samples (x 0, 0.2, 0.4 and 0.6) have been
Received 25 August 2007 prepared by the conventional ceramic technique. Rietveld renement of X-ray diffraction patterns of the
Received in revised form samples shows them to crystallize in the Ia3d space group and the corresponding lattice constant to
4 March 2008
decrease with increasing Al3+ and Cr3+ contents (x). Mossbauer results indicate that Cr3+ substitutes for
Available online 10 May 2008
Fe3+ at the octahedral sites whilst Al3+ essentially replaces Fe3+ at the tetrahedral sites. This result
Keywords: indicates that co-doping of Y3Fe5O12 does not affect the preferential site occupancy for separate
Garnet individual substitution of either Cr3+ or Al3+. The magnetization measurements reveal that the Curie
Co-substitution temperature (Tc) monotonically decreases with increasing x while the magnetic moment per unit
Site occupancy
formula decreases up to x 0.4 and then slightly increases for x 0.6. This reects a progressive
Magnetism
weakening of the ferrimagnetic exchange interaction between the Fe3+ ions at octahedral and
tetrahedral sites due to co-substitution. The magnetic moment was calculated using the cations
distribution inferred from the Mossbauer data and the collinear ferrimagnetic model, and was found to
agree reasonably with the experimentally measured value. The phenomenological amplitude crossover,
characterized by the temperature T*, has also been observed in the doped YIG and briey discussed.
& 2008 Elsevier B.V. All rights reserved.

1. Introduction
Since its discovery in 1956, yttrium iron garnet (YIG) and
substituted-YIG have attracted a lot of attention because of their
application in microwaves domain [1] and magneto-optical
devices [2]. YIG has the chemical composition {Y3}[Fe3+ 3+
2 ]a(Fe3 )d
O12, where Y ions occupy the dodecahedral site with 12
O2 ligands, a represents the octahedral site with six O2 ligands,
and d the tetrahedral site with four O2 ligands. Its crystal
structure is described in the cubic space group Ia3d by Bonnet
et al. [3] and in the trigonal space group R3 by Rodic et al. [4], to
account for the lowering of symmetry due to spontaneous
magnetostriction. YIG is ferrimagnetic with two magnetic a and
d sublattices of Fe3+ that are coupled antiparallelly. It is well
established that composition modication alters the magnetic
properties of garnets by magnetic dilution associated with the
replacement of Fe3+ ions by diamagnetic ions of approximately
the same size and by magnetoelastically active ion substitutions
usually in small concentration [5]. Also, monitoring the selective
site doping can shed light on the magnetism of each sublattice
from fundamental point of view as well as enable control of
physical properties for applications purposes. Various levels and
 Corresponding author. Tel.: +968 24415489; fax: +968 24414228.
E-mail address: bouzi@squ.edu.om (K. Bouziane).
types of doping of Fe3+ sites have been performed. Among others,
substitution of magnetic Fe3+ by nonmagnetic Al3+ has shown
interesting features [6]. It was shown that Al3+ goes to the
tetrahedral 24d sites in Al-doped YIG [7,8] whereas Cr3+
preferentially occupies the unexpected octahedral site, which is
attributed to the electronic conguration of Cr3+ [7,9]. On co-
substituting nonmagnetic Al3+ and magnetic Cr3+ for Fe3+ ions
[10], it was reported that Cr3+ and 60% of Al3+ ions go to the
tetrahedral sites whereas the remaining 40% of Al go to octahedral
sites. However, a recent study [11] showed both Cr3+ and Al3+
substitute for Fe3+ in the octahedral sites only. Therefore, the site
occupancy of both Cr3+ and Al3+ in YIG is controversial. It is to be
mentioned that the cation distribution in the latter two studies
was based on measurements at room temperature of either
magnetization [10] or Mossbauer spectroscopy [11].
The canting effect induced by substitution of magnetic ions by
non-magnetic or less magnetic ones leads to a decrease on one
site of the effective exchange [12]. In fact, this phenomenon is
very sensitive to local disorder and inhomogeneities in the
substitutions. It can lead to various local canted structures with
different hyperne elds, but the thermal disorder can only make
the difference when the balance of the exchange interactions is
close to the canting condition. However, the magnetic ordering of
ferrites like YIG changes signicantly from its arrangement at
room temperature, as shown for x 0.4 and x 0.6 in this study.
In particular, the thermal disorder would reduce the net

0304-8853/$ - see front matter & 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmmm.2008.04.163
 ARTICLE IN PRESS

K. Bouziane et al. / Journal of Magnetism and Magnetic Materials 320 (2008) 23302334 2331

magnetization, making the Neels collinear ferrimagnetic model to a cubic structure (space group Ia3d), reecting that the
not valid. For example, the cations distribution reported by compounds are single-phased. The lattice constant decreases
Lahlooh et al. [11] for Y3Fe52xAlxCrxO12 garnets would lead to with increasing concentration x of the doped ions as shown in
magnetic moment inconsistent with that obtained from magne- Fig. 2. Such a decrease in the lattice parameters is generally
tization measurements reported in this study. In this work, three attributed to the smaller ionic radii of Al3+ and Cr3+ relative to that
techniques, namely, X-ray diffraction, low temperature magneti- of high-spin Fe3+ ions at both tetrahedral and octahedral sites. It is
zation and Mossbauer spectroscopy, were employed to clarify the interesting to note that the lattice constant does not decrease
site preference of Cr3+ and Al3+ ions in Y3Fe52xAlxCrxO12 garnets. linearly, in line with the Vegrads law, as expected when both Al3+
It is shown that the preferential sites occupancy for Al and Cr in and Cr3+ substitute Fe3+ ions at either site. While Cr3+ ions are
co-doped YIG remains nearly identical to that corresponding to known to have a strong octahedral site preference [7,9], the
separate doping of either Cr3+ or Al3+ in YIG. decrease, though limited, in the lattice parameters with increasing
x value suggests that Al3+ ions do partly replace tetrahedral Fe3+
ones. The smaller tetrahedral ionic radius of Al3+ (0.53 A) relative
2. Experimental to that of tetrahedral Fe3+ (0.69 A) supports this view [13]. It
should be noted that the Rietveld analysis of the spectrum of
Samples with nominal composition Y3Fe52xAlxCrxO12 (x 0, sample x 0.4 has also revealed the presence of a very minor
0.2, 0.4 and 0.6) have been prepared by the conventional ceramic phase (67%) that was not identied but possibly magnetic
technique. The starting materials Y2O3, Fe2O3, Al2O3 and Cr2O3 (such as traces of YCrO3 or a-Fe2O3 impurity phases [9]), which
were weighted in appropriate proportions and wet-mixed for 1 h can be at the origin of the relatively larger broadening of the
in acetone to obtain homogeneous mixtures. The mixtures were Mossbauer spectrum as shown below.
dried and then calcined at 1200 1C for 10 h followed by mechanical The magnetization curves recorded at room temperature of the
milling for 1 h to get homogeneous sub-micronic powders. The four samples are shown in Fig. 3. It can be noticed that the
powders were pressed in blocks under 2 tons/cm2 using polyvinyl
alcohol (PVA) as binder and then sintered in oxygen atmosphere
at 1400 1C for 10 h. The sintered blocks were crushed into powder 12.38
and then annealed at 700 1C for 2 h.
The crystalline nature of the resulting compounds was
analyzed using a Philips PW 1700 X-ray diffractometer with a 12.36
Lattice constant, a (A)

CuKa source. The magnetization was measured with a DMS 1660

vibrating sample magnetometer (VSM) in a magnetic eld up to 12.34
13 kOe, and in the temperature range 77573 K. The VSM was
initially calibrated using pure nickel (Ms 54.9 emu/g). The
12.32
Mossbauer spectra were recorded at 78 K using 57Co/Rh source
with the spectrometer in the transmission mode. The spectra were
tted with a distribution of hyperne elds using NORMOS 12.30
program.
12.28
3. Results and discussion Y3Fe5-2xAlxCrxO12
12.26
The X-ray powder diffraction patterns recorded from the 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Y3Fe52xALxCrxO12 samples are shown in Fig. 1. All patterns rene xAl-Cr

Fig. 2. The variation of the lattice parameters of Y3Fe52xALxCrxO12 with the

concentration x.

30 x=0

0.2
x = 0.6
Intensity (arb. units)

20 0.4

0.6

x = 0.4 10
M (emu/g)

0
x = 0.2
640

642

-10
400

422

642
420

444

800
521

532

840

864
440
331
321

431

x=0
-20

30 40 50 60 70 80 -30
2 (deg) -15 -10 -5 0 5 10 15
H (kOe)
Fig. 1. Powder X-ray diffraction patterns of Y3Fe52xAlxCrxO12 garnet samples. The
symbol (*) indicates the presence of a non-identied impurity phase for x 0.4. Fig. 3. Room temperature hysteresis loops of Y3Fe52xAlxCrxO12 garnet samples.
 ARTICLE IN PRESS

2332 K. Bouziane et al. / Journal of Magnetism and Magnetic Materials 320 (2008) 23302334

coercivity is negligible, regardless of the content x. The saturation Table 1

magnetization (Ms) is reached rapidly and Ms decreases progres- Magnetic moment at 0 K (in mB) of Y3Fe52xAlxCrxO12 garnet: m(x) is determined
from Mossbauer measurements using Eq. (1) mcal calculated using the Neels
sively with x. This result indicates that the doping has taken place
collinear model combined to the site occupancy displayed in Table 3, mmes and
with some sort of preferential site occupancy for either Cr3+ or amplitude crossover temperature (T*) determined from magnetization M(T)
Al3+ ions. measurements using T3/2 and T2 laws; respectively, and Curie temperature (Tc)
The temperature dependence of magnetization has been
recorded (Fig. 4(a)) in order to estimate in part the magnetization Content x mcal (mB) mmes (mB) Tc (K) T* (K)

M(0) at 0 K by using the appropriate power function of absolute T 3/2

law 2
T law
temperature [14]. It is well known that the T3/2 Bloch law is only
valid for temperatures below about Tc/20, which means for the 0 5 5.0170.01 4.9770.03 566 287
YIG: To30 K. Furthermore, in the more exact formula of Dyson 0.2 4.4 4.6570.07 4.5570.03 520 279
0.4 3.9 3.9370.04 3.8470.03 455 238
[15] there are terms in T5/2, T7/2, and T4 besides the rst one in T3/2;
0.6 4.1 4.0570.05 3.9270.03 428 224
so it is expectable that in some range of temperature 430 K the
mean effective exponent will be close to 2 and even greater,
depending on the exact balance of the involved exchange
interactions. We have tested both the T3/2 Bloch law and the T2 600 5.5
law. The linear extrapolation of M(T) to T 0, although not giving Y3Fe5-2xAlxCrxO12
very precisely the true absolute values of M(0), is good enough for Tc
the relative variations with the substitution rate and their

Magnetic moment (B)

550 5.0

Curie temperature (K)

correlation to the occupancy of the sites. Furthermore, Both
T3/2 and T2 give the same values of the absolute magnetization,
within an uncertainty of less than 4%. The Blochs T3/2 law
[M(T) M(0)(1BT3/2)] was successfully applied for an extended 500 4.5
range of temperature (Fig. 4(b)) whereas the T2 power law has
been tested and the result shows that it is valid only for a very
limited range of (low) temperature (Fig. 4(c)). The corresponding
450 4.0
tted values of M(0) were used to determine the magnetic
moment per formula unit (mmes) at 0 K.
The values obtained for the magnetic moment mmes at 0 K using
both T2 and T3/2 laws as well as the Curie temperature Tc are 400 3.5
reported in Table 1 and Fig. 5. The net magnetic moment per 0.0 0.1 0.2 0.3 0.4 0.5 0.6
formula unit mcal was also calculated on the basis of Neels model xAl-Cr
and the cation distribution determined from Mossbauer analysis
Fig. 5. The magnetic moment per formula unit at 0 K and Curie temperature
(see Table 1). It is interesting to note the monotonic decrease of Tc
versus x for Y3Fe52xAlxCrxO12 garnets.

Temperature (K)
0 100 200 300 400 500 600 reecting a decrease of the exchange coupling. The magnetic
40 moment mmes decreases almost linearly with x up to x 0.4 and
then slightly increases for x 0.6. This trend was not observed in
M(H) values at room temperature (Fig. 3) and suggests a certain
30
relative partial occupancy of both doped ions at either octahedral
or tetrahedral sites for xX0.4. Since Y3+ is known to be non-
20 magnetic in YIG systems, added to the fact that there is no doping
Magnetization at 13.5 kOe (emu/g)

x=0 at the dodecahedral site, the exchange coupling in this system is

x = 0.2 essentially governed by the intra-sublattice interactions Jaa and Jdd
10
x = 0.4 as well as inter-sublattices interactions Jad. We argue that the
x = 0.6 exchange coupling is dominated by inter-sites interactions Jad as
0 reported for the parent YIG system [16]. This is based on the fact
that doping would result in signicant changes of the intra-site
40
interactions assuming preferential site occupancy of Al3+ and Cr3+
ions [710], as is also shown from the Mossbauer study below. In
30 order to deduce accurately the distribution of Al3+ and Cr3+ ions
among sublattices and to explain the magnetic behavior of
20 Y3Fe52xAlxCrxO12 garnet samples, the corresponding Mossbauer
spectra were recorded at 78 K as shown in Fig. 6. The spectra were
10
tted using two sextets corresponding to Fe3+ in the a and d
sublattices. The corresponding tting parameters are reported in
Table 2 and the nominal composition of Y3Fe52xAlxCrxO12 garnet
0
upon x is displayed in Table 3. From Table 3, it is clear that Al3+ and
0 5000 10000 0 100000 200000 300000 Cr3+ have essentially tetrahedral and octahedral site preference,
T3/2 (K) T2 (K) respectively. It should be noted that the Mossbauer spectra for
x 0.6 and particularly x 0.4 (Fig. 6) show a large broadening
Fig. 4. (a) Temperature dependence of magnetization for Y3Fe52xAlxCrxO12 garnet
samples recorded at 13.5 kOe; (b) the corresponding T3/2 Blochs law of
associated with relatively large line width G (see Table 2). This
magnetization; and (c) the corresponding T2 power law of magnetization. The effect is not clear but can be possibly assigned to the presence of
solid straight lines are linear ts. Al and Cr at both sites. The relatively larger spectrum broadening
 ARTICLE IN PRESS

K. Bouziane et al. / Journal of Magnetism and Magnetic Materials 320 (2008) 23302334 2333

Y3Fe5-2xAlxCrxO12 54 46.5
B hf -Octa
B hf -Tetra

x=0

Bhf Tetrahedral (Tesla)

Bhf Octahedral (Tesla)
46.0

4%
53
300

x = 0.2 45.5

T (K)
250
Absorption

3%
200
400 500 600
Tc (K)
52 45.0
x = 0.4
220 240 260 280 300
T (K)

2% Fig. 7. The hyperne elds at a and d sites for various x contents. Inset is the
amplitude crossover temperature T* versus Curie temperature.

x = 0.4

magnetic Cr3+ ions (ground state 4A2, L 0 and m 3 mB). The

2%
corresponding amplitude crossover temperatures T* [14] are
displayed in Table 1 too. The decrease of T* value upon co-doping
can be attributed to the exchange coupling. As the hyperne elds
-10 -8 -6 -4 -2 0 2 4 6 8 10 reect the exchange of central iron ion with its surrounding, we
Velocity (mm/s) have plotted in Fig. 7 the variation of the hyperne magnetic eld
Bhf at the a and d sites as a function of T*. The linear variation of
Fig. 6. Experimental (  ) and tted (lines) Mossbauer spectra of Y3Fe52xAlxCrxO12 Bhf versus T* establishes a direct correlation between Bhf and T*.
garnet samples recorded at 78 K. This is expected since Bhf reects the exchange of central Fe3+ ion
with its surrounding (via oxygen ions) either in octahedral or in
tetrahedral site. This is also supported by the curve inset in Fig. 7,
Table 2 which shows a direct correlation between T* and Tc indicating that
Estimated parameters from Mossbauer spectra of Y3Fe52xAlxCrxO12 garnets:
the amplitude crossover temperature T* depends closely on the
isomer shift (d) relative to a-iron, line width (G), hyperne magnetic eld (Bhf) and
the relative intensity (I)
inter-sublattice exchange coupling.

Content x Site d (mm/s) G (mm/s) Bhf (T) I (%)

4. Conclusion
0 a 0.49 0.53 53.7 41.3
d 0.26 0.47 46.2 58.7
0.2 a 0.49 0.47 53.2 39.2 The Y3Fe52xAlxCrxO12 garnets (x 0, 0.2, 0.4 and 0.6) have
d 0.26 0.47 46.2 60.8 been prepared and their structure were found to crystallize in the
0.4 a 0.49 0.56 52.81 37.9 Ia3d space group. The co-doping of Fe3+ by Al3+ and Cr3+ at both
d 0.26 0.66 45.52 62.1
octahedral and tetrahedral sites has been studied. It is shown that
0.6 a 0.49 0.49 52.46 34.4
d 0.26 0.54 45.25 65.3
Al3+ ions prefer essentially tetrahedral site substitution while Cr3+
occupy octahedral site only. The Al3+ and Cr3+ co-doping does not
affect the initial occupancy site preference of separate doping of
either Al3+ or Cr3+. This substitution feature can be used to
Table 3 systematically monitor the magnetic properties such as the net
Sites occupancy of Al3+ and Cr3+ ions inferred from Mossbauer spectra tting at magnetic moment per unit formula and the exchange coupling
78 K assuming the same f-factor combined to magnetization measurements
within each sublattice and inter-sublattice.
Al/Cr content (x) Octahedral site (a-site) Tetrahedral site (d-site)

0 Fe2 Fe3 Acknowledgments

0.2 Fe1.8Cr0.2 Fe2.8Al0.2
0.4 Fe1.59Al0.01Cr0.4 Fe2.61Al0.39
0.6 Fe1.31Al0.09Cr0.6 Fe2.49Al0.51
The authors would like to thank K. Melghit for Rietveld
analysis.

References
for x 0.4 can be partially attributed to the presence of a
secondary magnetic phase as shown from XRD results.
[1] H.L. Glass, IEEE Trans. 76 (1988) 86.
It is worthwhile to point out that the use of T2 law can however [2] P. Vaqueiro, M.A. Lopez-Quintela, J. Rivas, J. Magn. Magn. Mater. 169
enlighten the change of the phenomenological amplitude cross- (1997) 56.
over upon co-doping. This is particularly interesting taking into [3] M. Bonnet, A. Delapalme, H. Fuess, M. Thomas, Acta Crystallogr. 31B (1975)
2233.
account that the tetrahedral sites are doped with diamagnetic [4] D. Rodic, M. Mitrie, R. Tellgren, H. Rundlof, A. Kremenovic, J. Magn. Magn.
ions Al3+ whereas the octahedral sites are principally occupied by Mater. 191 (1999) 137.
 ARTICLE IN PRESS
2334 K. Bouziane et al. / Journal of Magnetism and Magnetic Materials 320 (2008) 23302334
[5] G. Dionne, J. Appl. Phys. 81 (1997) 5064.
[6] S. Thongmee, P. Winotai, I.M. Tang, Solid State Commun. 109 (1999) 471.
[7] M.A. Gilleo, S. Geller, Phys. Rev. 110 (1958) 73.
[8] C.S. kim, B.K. Min, S.J. Kim, S.R. Yoon, Y.R. Uhm, J. Magn. Magn. Mater.
254255 (2003) 553.
[9] M.V. Kuznetsov, Q.A. Pankhurst, I.P. Parkin, L. Afeck, Y.G. Morozov, J. Mater.
Chem. 10 (2000) 755.
[10] V.D. Murumkar, K.B. Modi, K.M. Jadhav, G.K. Bichile, R.G. Kulkarni, Mater. Lett.
32 (1997) 281.
[11] A.F. Lahlooh, S. Mahmood, M. Mozaffari, J. Amighian, Hyp. Interact. 156157
(2004) 181.
[12] G. Dionne, J. Appl. Phys. 41 (1970) 4874.
[13] M.J. Winter, /www.webelements.comS, &, University of Shefeld, UK,
19952006.
[14] U. Kobler, A. Hoser, Solid State Commun. 142 (2007) 350.
[15] F.J. Dyson, Phys. Rev. 102 (1956) 1230.
[16] P.U. Sharma, K.B. Modi, V.K. Lakhani, K.B. Zankat, H.H. Joshi, Ceram. Int. 33
(2007) 1543.