Magnetic properties of lanthanide chalcogenide semiconducting nanoparticles

Article abstract of DOI:10.1021/ja0620080

To understand the importance of the band gap to the magnetic ordering in magnetic semiconductors, we have studied the effect of particle size on the ferromagnetic Curie temperature in semiconducting EuS. We have synthesized capped ～20 nm EuS nanoparticles using a single-source precursor, [Eu(S ₂CNBu₂)₃Phen] decomposed in trioctylphosphine. The nanoparticles have been characterized by X-ray powder diffraction, TEM, and magnetic susceptibility measurements as a function of temperature and field. The Curie temperature, based on Arrott plots, is depressed by 1-2 K from the bulk value.

Full text of DOI:10.1021/ja0620080

Published on Web 08/05/2006
Magnetic Properties of Lanthanide Chalcogenide
Semiconducting Nanoparticles
Michelle D. Regulacio, Konrad Bussmann,† Brad Lewis,‡ and Sarah L. Stoll*
Contribution from the Department of Chemistry, Box 571227, Georgetown UniVersity,
Washington D.C. 20057
Received March 23, 2006; E-mail: sls55@georgetown.edu
Abstract: To understand the importance of the band gap to the magnetic ordering in magnetic
semiconductors, we have studied the effect of particle size on the ferromagnetic Curie temperature in
semiconducting EuS. We have synthesized capped ∼20 nm EuS nanoparticles using a single-source
precursor, [Eu(S2CNiBu2)3Phen] decomposed in trioctylphosphine. The nanoparticles have been character-
ized by X-ray powder diffraction, TEM, and magnetic susceptibility measurements as a function of
temperature and field. The Curie temperature, based on Arrott plots, is depressed by 1-2 K from the bulk
value.
Introduction
The europium chalcogenides were intensely studied in the
1970s8 and continue to be of both theoretical9 and experimental10
interest. Initial studies were motivated by applications that utilize
the large Faraday11 and Kerr effects.12 The potential for using
EuS as a “spin filter” has driven recent thin film studies.13 The
europium chalcogenides are part of a general class of materials
that exhibit novel coupled phenomena, i.e., pairwise combina-
tions of magnetic, electronic, and optical properties.14 Magneto-
optical, magnetoresistive, and optoelectronic materials are
targeted for the development of new electronic devices (such
as GMR heads or LEDs). Recently, there has been particular
interest in the synthesis and characterization of nanoparticles
of EuO and EuS in the search for new luminescent materials.15,16
Magnetic nanoparticles have been of interest for applications
in electronic devices1 and biological systems2 and raised
significant questions about the fundamental properties of
magnetic materials in the nanometer size regime.3 We have been
interested in studying the magnetic lanthanide chalcogenides,
LnQ (Ln ) lanthanide, Q ) O, S, Se, Te), as model systems
for probing solid-state properties that are likely to be influenced
by band gap tuning using particle size. Although magnetic
properties have been studied in metallic (Fe, FePt)4,5 and
insulating (for example Fe2O3, Fe3O4, and spinels MFe2O4, M
) transition metal)6 systems, we believe magnetic semiconduc-
tors should provide new insight into the question of how solid-
state properties evolve with size. In addition, because the
properties are governed by the extent of delocalization of f
electrons, these materials provide an independent probe of f
orbital overlap and the degree of bonding covalency in lan-
thanides, an area of current debate.7
(7) (a) Castro-Rodriguez, I.; Olsen, K.; Gantzel, P.; Meyer, K. J. Am. Chem.
Soc. 2003, 125 (15), 4565. (b) Jensen, M. P.; Bond, A. H. J. Am. Chem.
Soc. 2002, 124 (33), 9870.
(8) Wachter, P. CRC Crit. ReV. Solid State Sci. 1972, 3 (2), 189-238.
(9) (a) Horn, M.; Strange, P.; Temmerman, W. M.; Szotek, Z.; Svane, A.;
Winter, H. Los Alamos National Laboratory, Preprint ArchiVe, Condensed
Matter 2004, 1-7. (b) Dietl, T. Los Alamos National Laboratory, Preprint
ArchiVe, Condensed Matter 2004, 1-7. (c) Ariponnammal, S.; Natarajan,
S. Mod. Phys. Lett. B 2000, 14 (24), 843-848. Kunes, J.; Ku, W.; Pickett,
W. E. J. Phys. Soc. Jpn. 2005, 74 (5), 1408.
(10) (a) Sakalle, U. K.; Jha, P. K.; Sanyal, S. P. Bull. Mater. Sci. 2000, 23 (3),
233-235. (b) Kasuya, T. J. Magn. Magn. Mater. 1999, 195 (1), 141-147.
(c) Gorlitz, D.; Kotzler, J. Eur. Phys. J. B 1998, 5 (1), 37-43.
(11) Thongchant, S.; Hasegawa, Y.; Tanaka, K.; Fujita, K.; Hirao, K.; Wada,
Y.; Yanagida, S. Jpn. J. Appl. Phys., Part 2 2003, 42 (7B), L876-L878.
(12) Fumagalli, P.; Schirmeisen, A. J. Appl. Phys. 1996, 79 (8), 5929.
(13) (a) Hao, X.; Moodera, J. S.; Meservey, R. Phys. ReV. B 1990, 42 (13),
8235-8243. (b) Metzke, R.; Nolting, W. Phys. ReV. B: Condens. Matter
1998, 58 (13), 8579-8589. (c) Horn, M.; Strange, P.; Temmerman, W.
M.; Szotek, Z.; Svane, A.; Winter, H. Los Alamos National Laboratory,
Preprint ArchiVe, Condensed Matter 2004, 1-7.
(14) Wachter, P. CRC Crit. ReV. Solid State Sci. 1972, 3, 189.
(15) (a) Thongchant, S.; Hasegawa, Y.; Wada, Y.; Yanagida, S. Chem. Lett.
2001, 12, 1274. (b) Hasegawa, Y.; Thongchant, S.; Wada, Y.; Tanaka, H.;
Kawai, T.; Sakata, T.; Mori, H.; Yanagida, S. Angwe. Chem., Int. Ed. 2002,
41 (12), 2073-2075. (c) Hasegawa, Y.; Thongchant, S.; Kataoka, T.; Wada,
Y.; Yatsuhashi, T.; Nakashima, N.; Yanagida, S. Chem. Lett. 2003, 32 (8),
706-709.
(16) (a) Thongchant, S.; Hasegawa, Y.; Wada, Y.; Yanagida, S. J. Phys. Chem.
B 2003, 107, 2193. (b) Regulacio, M. R.; Tomson, N.; Stoll, S. L. Chem.
Mater. 2005, 17 (12), 3114-3121. (c) Mirkovic, T.; Hines, M.; Nair, P.;
Scholes, G. Chem. Mater. 2005, 17, 3451.
† Present address: Materials and Sensors Branch (6360), Naval Research
Laboratory, 4555 Overlook Ave. SW, Washington, DC 20375.
‡ Present address: Department of Chemistry, The Pennsylvania State
University, University Park, PA 16802.
(1) Awschalom, D. D.; Flatte, M. E.; Samarth, N. Sci. Am. 2002, 286 (6), 66-
73.
(2) Gould, P. Mater. Today 2004, 7 (2), 36.
(3) (a) Majetich, S.; Jin, Y. Science, 1999, 284, 470. (b) Gambardella, P.;
Rusponi, S.; Veronese, M.; Dhesi, S. S.; Grazioli, C.; Dallmeyer, A.; Cabria,
I.; Zeller, R.; Dederichs, P. H.; Kern, K.; Carbone, C.; Brune, H. Science
2003, 300, 1130. (c) Dumestre, F.; Chaudret, B.; Amiens, C.; Renaud, P.;
Fejes, P. Science 2004, 303, 821. (d) Kodama, R. H. J. Magn. Magn. Mater.
1999, 200, 359-372.
(4) Farrell, D.; Majetich, S. A.; Wilcoxon, J. P. J. Phys. Chem. B 2003, 107,
11022.
(5) Sun, S.; Murray, C. B.; Weller, D.; Folks, L.; Moser, A. Science 2000,
287 (5460), 1989.
(6) (a) Vestal, C.; Zhang, Z. J. J. Am. Chem. Soc. 2003, 125, 9828. (b) Vestal,
C.; Song, Q.; Zhang, Z. J. J. Phys. Chem. B 2004, 108 (47), 18222. (c)
Vestal, C.; Zhang, Z. J. Chem. Mater. 2002, 14 (9), 3817. (d) Huh, Y. M.;
Jun, Y. W.; Song, H. T. J. Am. Chem. Soc. 2005, 127 (35), 12387. (e)
Wan, S. R.; Zheng, Y.; Liu, Y. Q. J. Mater. Chem. 2005, 15 (33), 3424.
(f) Gu, H. W.; Xu, K. M.; Yang, Z. M. Chem. Commun. 2005, 34, 4270.
9
10.1021/ja0620080 CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 11173-11179
11173

A R T I C L E S
Regulacio et al.
Experimental Section
The europium chalcogenides are small band gap semiconduc-
tors, which exhibit a variety of magnetic ordering from
ferromagnetic (EuO TC ) 66.8 K, EuS TC ) 16.6 K) to
antiferromagnetic (EuTe TN ) 9.64 K) and metamagnetic (EuSe,
TN ) 4.6 K becomes ferromagnetic at a field of 0.5 GPa).17,18
One of the advantages of using particle size to study magnetic
properties in this system is that the ordering temperature has a
clear dependence on the energy gap. Considered almost a
textbook Heisenberg ferromagnet,19,20 models of EuS typically
consider the 12 nearest (Z1) and 6 next-nearest (Z2) neighbors
of the cation fcc lattice. Using the mean field approximation,
the Curie temperature can be related to the exchange integrals,
J1 (ferromagnetic coupling) and J2 (antiferromagnetic coupling)
through the equation
The dithiocarbamate precursors were prepared according to previ-
ously published procedures.27 The EuS nanoparticles were obtained by
the dissolution of [Eu(S2CNiBu2)3Phen] (1.75 g, 1.85 mmol) in
trioctylphosphine (TOP, 25 mL) and oleylamine (15 mL), and the
solution was heated to 240 °C and held at this temperature for 7.5 h.
Initially, the solution was orange-red and began to darken at ap-
proximately 70 °C, and finally it was purple-black by the time the
temperature reached 240 °C. After 7.5 h the temperature was reduced
to 60 °C, and anhydrous methanol (40 mL) was added to the solution.
In a glovebox the solution was transferred to a centrifuge tube and
centrifuged at 3500 rpm for 40 min. The yellow-green supernatant was
discarded, and the black powder was dissolved in anhydrous heptanes
(20 mL). Fresh methanol (40 mL) was added to the dark purple solution,
and the precipitate was isolated by centrifugation, washed with
methanol, and dried in vacuo. Elemental analysis gave C (6.8%), H
(1.52%), P (1.83%), and N (<0.5%).
2
kBTC ) S(S + 1)[Z1J1 + Z2J2]
(1)
3
UV-visible spectra were recorded from 200 to 800 nm in acetonitrile
on a Perkin Elmer UV-visible spectrometer in quartz cuvettes. Infrared
spectra were measured in the range 450-4000 cm-1 as pressed pellets
in KBr on a Perkin Elmer FTIR. X-ray powder diffraction patterns
were obtained using a Rigaku RAPID Curved IP X-ray powder
diffractometer with Cu KR radiation and an image plate detector.
Magnetic measurements were made on a QD SQUID from 50 to 5 K
in fields ranging from 500 to 5000 Oe. Arrott plots were obtained by
calculating isotherms for seven temperatures. For each field, the mag-
netization was squared (emu2/g2) for T1 ) 14.0037(7), T2 ) 14.9966(6),
T3 ) 15.9972(3), T4 ) 16.9967(6), T5 ) 17.9970(23), T6 ) 18.9957(8),
T7 ) 19.9984(20) K and plotted as a function of H/M (Oe g/emu) (Fig-
ure 5). The values of M2s where H/M is zero (for each isotherm) were
then graphed as a function of temperature (Figure 6). The M2s vs T plot
where M2s goes to zero was used to determine the Curie temperature,
TC ) 15.1834(2) K based on a linear regression (squared correlation
coefficient of R2 ) 0.9982(1)). Samples were prepared for TEM
measurements by dipping carbon-coated copper TEM grids 5 to 6 times
into solutions of the nanoparticles, allowing the grids to dry briefly
before reimmersion. Images were taken on a JEOL JEM 1200 EXII
TEM operated at 80 keV using a high-resolution Tietz F224 camera.
where TC is the Curie ferromagnetic ordering temperature, kB
7
is Boltzmann’s constant, S ) /2 for Eu(II) 4f7 (8S7/2) ground
state, Z1 ) 12, and Z2 ) 6.21 The connection between the Curie
temperature and the Eg, band gap, is through J1, an exchange
parameter that has the form
Ab2
Eg2
J1 )
(2)
where A is a function of intra-atomic exchange (4f valence to
5d conduction band exchange, a measure of the extent of
delocalization of f electrons in the conduction band), b is a
measure of the orbital overlap, and Eg is the band gap.22
Experimentally the relationship between electronic structure
and magnetic properties has been investigated using pressure,23
which unfortunately affects both the orbital overlap and the band
gap.24 Doping has also been used to probe the electronic
structure (in particular the energy separation between the 4f
valence band and the conduction band) and the magnetic
properties in the EuQ materials.25 At low doping levels,
increasing the electron concentration causes an oscillation in
the Curie temperature as typically found for materials described
by the RKKY interaction.26 Here we have investigated the role
of particle size to study the inter-relationship between the band
gap and the magnetic properties. Our evidence points to the
importance of particle size on the ordering temperature, and
we discuss three mechanisms for how this might occur.
Results and Discussion
Synthesis. The synthetic route to form nanoparticles is
extremely important to elucidating properties because the size,28
shape,29 and surface properties30 all have a profound influence
over optical and magnetic measurements.31,32 Previously, EuS
nanoparticles have been made by solid-state diffusion of
powdered EuS into the pores of zeolites,33 formation in liquid
ammonia solutions,34 and decomposition from single source
precursors35 or white LED irradiation resulting in uncapped
nanoparticles.36 Unfortunately, surface conditions substantially
(17) Wernick, J. H. In Treatise on Solid State Chemistry, the Chemical Structure
of Solids; Hannay, N. B., Ed.; Plenum Press: New York, 1974; Vol. 1, pp
175-272.
(18) Fujiwara, H.; Kadomatsu, H.; Kurisu, M.; Hihara, T.; Kojima, K.;
Kamigaichi, T. Solid State Commun. 1982, 42, 509.
(19) Nolting, W. J. Phys. C: Solid State Phys. 1982, 15, 733-745.
(20) Rhyne, J.; McGuire, T. P. IEEE Trans. Magn. 1972, 8 (1), 105.
(21) (a) Wolf, W. P.; McGuire, T. R.; Shafer, M. W. J. Appl. Phys. 1964, 35
(3), 984. (b) Zinn, W. J. Magn. Magn. Mater. 1976, 3, 23.
(22) Goodenough, J. B. Magnetism and the Chemical Bond; John Wiley and
Sons: New York, 1963; pp 144 and 167.
(27) (a) Su, C.; Tan, M.; Tang, N.; Gan, X.; Liu, W. J. Coord. Chem. 1996, 38,
207. (b) Regulacio, M. D.; Tomson, N.; Stoll, S. L. Chem. Mater. 2005,
17, 3114.
(28) Alivisatos, A. P. Science 1996, 271, 933.
(29) Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates, B.; Yin, Y.; Kim,
F.; Yan, H. AdV. Mater. 2003, 15 (5), 353-389.
(30) Kalyuzhny, G.; Murray, R. J. Phys. Chem. B 2005, 109, 7012-7021.
(31) (a) Cheon, J.; Kang, N.-J.; Lee, S. M.; Lee, J.-H.; Yoon, J.-H.; Oh, S. J. J.
Am. Chem. Soc. 2004, 126, 1950-1951. (b) Seo, W. S.; Jo, H. H.; Lee,
K.; Kim, B.; Oh, S. J.; Park, J. T. Angew. Chem., Int. Ed. 2004, 43, 1115.
(c) Song, Q.; Zhang, J. Z. J. Am. Chem. Soc. 2004, 126, 6164.
(32) Zhao, F.; Sun, H.-L.; Gao, S.; Su, G. J. Mater. Chem. 2005, 15, 4209.
(33) Chen, W.; Zhang, X.; Huang, Y. Appl. Phys. Lett. 2000, 76 (17), 2328-
2330.
(34) Thongchant, S.; Hasegawa, Y.; Wada, Y.; Yanagida, S. J. Phys. Chem. B
2003, 107, 2193-2196.
(35) (a) Regulacio, M. R.; Tomson, N.; Stoll, S. L. Chem. Mater. 2005, 17
(12), 3114-3121. (b) Mirkovic, T.; Hines, M.; Nair, P.; Scholes, G. Chem.
Mater. 2005, 17, 3451
(23) (a) Goncharenko, I. N.; Mirebeau, I. Phys. ReV. Lett. 1998, 80 (5), 1082-
1085. (b) Schwob, P.; Vogt, O. Phys. Lett. 1967, 24A, 242. (c) Fujiwara,
H.; Kadomatsu, H.; Kurisu, M.; Hirara, T.; Kojima, K.; Kamagaichi, T.
Solid State Commun. 1982, 42, 509.
(24) Goni, A. R.; Strossner, K.; Syassen, K.; Cardona, M. Phys. ReV. B 1987,
36 (6), 1581.
(25) (a) Holtzberg, F.; McGuire, T. R.; Methfessel, S.; Suits, J. C. Phys. ReV.
Lett. 1964, 13 (1), 18-21. (b) Maletta, H. J. Appl. Phys. 1982, 53 (3),
2185. (c) von Molnar, S.; Methfessel, S. J. Appl. Phys. 1967, 38 (3), 959-
964. (d) Maletta, H. J. Appl. Phys. 1982, 3 (3), 2185.
(26) Holtzberg, F.; McGuire, T. R.; Methfessel, S.; Suits, J. C. Phys. ReV. Lett.
(36) Hasegawa, Y.; Afzaal, M.; O’Brien, P.; Wada, Y.; Yanagida, S. Chem.
Commun. 2005, 242-243.
1964, 13 (1), 18.
9
11174 J. AM. CHEM. SOC. VOL. 128, NO. 34, 2006

Magnetic Properties of Semiconducting Nanoparticles
A R T I C L E S
Figure 1. X-ray powder diffraction pattern of ∼20 nm EuS nanoparticles.
influence the luminescence and magneto-optical properties of
the europium chalcogenides which limits the extent to which
these examples may be compared.37 Important sample charac-
teristics include capping ligand coverage, nanoparticle size, and
dispersity. The nanoparticles used for the magnetic studies
reported here were prepared with particular attention to optimiz-
ing these characteristics, although more work remains toward
these goals. The elemental analysis is consistent with a
monolayer of TOP for 18 nm nanoparticles.38 Although the
nitrogen elemental analysis was <0.5%, the UV-visible
spectroscopy suggests that some phenanthroline is still present
on the surface (as evidenced by a peak at ∼284 nm). The FTIR
of the nanoparticles indicate the presence of some oxidized
trioctylphosphine (TOPO), with a small P-O stretching peak
is important to controlling particle size, and we are currently
investigating whether the amine/phosphine ratio may play a role
in determining the particle size. Amines have previously been
found to mediate nanoparticle growth; for example, in the
synthesis of Ag2Se nanoparticles the ratio of [HDA]/[Precursor]
was important for size control.40
Structure and Particle Size. Based on the X-ray powder
diffraction pattern of the prepared nanoparticles, EuS was the
only crystalline phase present (see Figure 1) and was indexed
to the known cubic material.41 Using the (111), (200), and (220)
peaks, the fwhm was used to calculate the particle size using
the Scherrer equation.42 Based on this, the average particle was
found to be 19.1 ((0.2) nm. A representative TEM image is
shown in Figure 2 (with histogram inset), which was found to
have an average particle size of 15 ((3) nm for measurements
of ∼130 individual crystallites. Although the nanoparticles
exhibit solubility properties associated with capped nano-
particles, there appears to be some association of particles in
the TEM images, leading to error in the size determination.
Magnetic Properties. The magnetic properties of EuS have
been studied as a function of temperature (50-5 K) and field
(5000-500 Oe). The magnetization as a function of temperature
shows a sharp increase at low temperatures, near the TC, similar
to that observed in bulk EuS as shown in Figure 3.43 As the
field is reduced, the magnetization begins to show some
evidence for peak formation (indications of a blocking temper-
ature forming) as has been seen in other ferromagnetic nano-
particles44 and well studied in systems such as iron.45 This
suggests that the particle size is approaching the critical
at 1466 cm-1
.
We have explored the effect of temperature and synthetic
time on the size of the nanoparticle and found that shorter times
have resulted in distinguishably smaller average particle sizes
(based on X-ray diffraction 14.6 ( 0.1 nm compared with 19.1
( 0.2), but compared with the greater size control now possible
in other semiconducting systems (e.g., 1-15 nm with 1 nm
increments in size developed for CdSe),39 these have relatively
broad distributions. Unlike the CdSe system where temperature
appears to have a clear effect on particle size, we find that as
we lower the temperature the crystallinity is so reduced that it
is difficult to determine the effect of this variable on nanoparticle
size. By contrast, we believe that the nature of the capping ligand
(37) Hasegawa, Y.; Thongchant, S.; Wada, Y.; Tanaka, H.; Kawai, T.;
Sakata, T.; Mori, H.; Yanagida, S. Angew. Chem., Int. Ed. 2002, 41 (12),
2073.
(38) The percent mass due to TOP was estimated by using the radius of the
inorganic core of the nanoparticle and the density of bulk EuS (5.75 g/mL)
and a shell volume estimated for a molecule layer thickness of ∼10 Å and
using the density of TOP (0.831 g/mL). This gave ∼6% by mass of TOP
compared with ∼10% observed. Smaller nanoparticles have been found to
have a higher ligand coverage (see: Bowen Katari, J. E.; Colvin, V. L.;
Alivisatos, A. P. J. Phys. Chem. 1994, 98, 4109). Given the range of particle
sizes based on TEM, this difference is not great.
(40) Ng, M. T.; Boothroyd, C.; Vittal, J. Chem. Commun. 2005, 30, 3820-
3822.
(41) Nowacki, W. Z. Kristallogr. 1938, 99, 339.
(42) Bawendi, M. G.; Kortan, A. R.; Steigerwald, M. L.; Brus, L. W. J. Chem.
Phys. 1989, 91 (11), 7282.
(43) McGuire, T. R.; Argyle, B. E.; Shafer, M. W.; Smart, J. S. Appl. Phys.
Lett. 1962, 1 (1), 17.
(44) Leslie-Pelecky, D. L.; Rieke, R. D. Chem. Mater. 1996, 8, 1770.
(45) Farrell, D.; Cheng, Y.; McCallum, W. R.; Sachan, M.; Majeich, S. A. J.
Phys. Chem. B 2005, 109, 13409.
(39) Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc. 1993,
115, 8706.
9
J. AM. CHEM. SOC. VOL. 128, NO. 34, 2006 11175

A R T I C L E S
Regulacio et al.
Figure 2. TEM image of EuS nanoparticles.
Figure 3. Magnetization as a function of temperature for 20 nm EuS nanoparticles.
dimension (Dc). At this point single domain particles are formed,
where the coercive field drops to zero and the anisotropy energy
is close to kBT (i.e., superparamagnetism).46 However, our EuS
nanoparticles do not meet the two criteria for superpara-
magnetism, which are as follows: (1) Magnetization curves at
different temperatures superimpose when M is plotted as a
function of H/T and (2) Absence of hysteresis.47 For example,
in Figure 4, a small hysteresis in the magnetization as a func-
tion of field can be observed. Superparamagnetism occurs when
the formation of domain walls becomes energetically unfavor-
able and can be estimated for materials where the crystal
anisotropy is large (thus the wall energy γ is large, and the
wall thickness δ is small). With moderate anisotropy such
estimates become inaccurate, but the minimum Dc should be
greater than δ, the domain wall thickness, which we estimated
to be ∼0.8 nm.48
(46) Spaldin, N. Magnetic Materials: Fundamentals and DeVice Applications;
Cambridge University Press: 2003; p 141.
(47) Cullity, B. C. Introduction to Magnetic Materials; Addison-Wesley
Publishing Co.: Philippines, 1972; p 410.
9
11176 J. AM. CHEM. SOC. VOL. 128, NO. 34, 2006

Magnetic Properties of Semiconducting Nanoparticles
A R T I C L E S
15 and 16 K. The Curie point is determined by first finding the
zero field limit as the intercept on the M2 axis when extrapolated
to H/M ) 0. This value of M2s for each temperature is plotted
in Figure 6. The Curie temperature is defined as the temperature
where M2s is zero (crosses the x-axis). For 19.1 nm particles,
we determine a TC ) 15.2 K, which is depressed from that of
bulk EuS (16.6 K), as shown in Figure 6.
In our initial studies of the size dependence of EuS nano-
particles, we have found for a second sample, with a smaller
average particle size of 14.6 ((0.1) nm based on X-ray powder
diffraction, an enhanced decrease in the Curie temperature.
Using a similar Arrott analysis we determined a TC of 14.6 K
(shown in Figure 6). Our first observation is that TEM
measurements suggest a relatively broad particle size range (10
( 2 nm), and based on these data the two samples would appear
to be indistinguishable. However, the X-ray powder diffraction
is consistent with the more sensitive magnetic measurements
that suggest that these are statistically different in their average
size. Second, these data support the hypothesis that the Curie
temperature decreases as the particle size decreases.
There are three mechanisms we have considered to explain
the decrease in ordering temperature for a decrease in nano-
particle size: strain, surface effects, and band gap changes.
Recent work on EuS thin films and multilayers have shown
that, depending on the substrate, the TC can vary between +1
and -3 K as a result of substrate induced lattice strain.59 This
is likely reflected in changes to the orbital overlap term b, in
eq 2, due to expansion or compression of atomic distances.
Although this is well within the range of TC changes we have
observed, the question is whether the surface strain in nano-
particles is comparable. Studies of CdSe and CdS using EXAFs
suggest there is only very weak static strain in semiconductor
nanoparticles, with atomic distances (mean-square relative
displacements) close to that of the bulk.60 We do not see shifts
in the diffraction peaks, which would suggest uniform strain,
and it is difficult to separate line broadening due to nonuniform
strain and that due to particle size. Characterization of micro-
structure from line-broadening analysis can be carried out using
Rietveld analysis, Williamson-Hall plots, or Fourier methods
such as that of Warren-Averbach.61 We are currently investi-
gating which method is most appropriate for nanoparticles. We
also anticipate that the capping ligand should prevent surface
reconstruction, although this is by analogy to thin films and we
have no formal evidence for this.
Figure 4. Magnetization as a function of field for 20 nm EuS nanoparticles.
Ferromagnetic materials exhibit spontaneous magnetization
and are described by the temperature below which long range
ordering occurs, the Curie temperature TC. This transition from
the magnetized to unmagnetized state is a second-order phase
transition accompanied by a peak in the specific heat49 and a
sharp rise in the magnetization as the temperature is cooled
below TC.50 Estimating the Curie temperature by extrapolating
from the magnetization vs temperature curves does not provide
an unambiguous determination. This type of data could be used
to compare a series of samples to estimate relatiVe changes in
TC; however at this time we have yet to characterize a full series
of samples with appropriate steps in size. Therefore, we have
used magnetic isotherms, or Arrott Plots, to provide a more
sensitive and accurate measure of the Curie temperature.51 This
interpolation technique can be derived from either a Landau
description of the magnetization52 or a power series expansion
of the Brillouin function53 and is valid only in the vicinity of
the Curie temperature (the “critical region”). The spontaneous
magnetization (Ms) can be described by a linear equation,54 one
form55 of which is (H/M) ) a(T - TC) + bTM2, which is
typically plotted as M2 as a function of H/M at constant
temperature (as we have done in Figure 5). Often isotherms
will bend at higher and lower temperatures (if the field is low
enough) due to magnetic anisotropy.56 Because spontaneous
magnetization is a slowly varying function except within the
limits Hf 0 and T - TC f 0,57 the bending is most pronounced
at low fields and low temperatures. Thus, typically the high
field data are extrapolated for temperatures below TC.58 The
intercept of these isotherms with the y-axis (i.e., where M2s )
0) should be positive if T > TC and negative if T < TC. This
can be seen qualitatively in Figure 5, where clearly the high
field data for T ) 14 and 15 K have a negative intercept and
those for T ) 16 K have a positive intercept placing TC between
A second mechanism can be derived based on the fact that
nanoparticles have increased numbers of surface atoms. Studies
of ferromagnetic thin films have found competing forces due
to reduced coordination of atoms at the surface. The reduced
coordination at the surface can result in the loss of near-neighbor
coupling (which can lower TC) and enhanced the surface
magnetic moment (which can increase TC).62 However, studies
of the critical behavior of EuS thin films using spin-polarized
low-energy electron diffraction found no change in TC of the
(48) Domain wall thickness can be estimated as π ) (0.3kBTCπ2/3Ka)1/2 where
TC is the Curie temperature, kB is Boltzman’s constant, K is the anisotropy
constant, and a is the lattice constant (from Cullity, B. ibid, p 291).
(49) Kraftmakher, Y. Eur. J. Phys. 1997, 18, 448.
(50) West, A. Solid State Chemistry and Its Applications; Wiley and Sons: New
York, 1984; p 553.
(51) Arrott, A. Phys. ReV. 1957, 108 (6), 1394.
(52) Neumann, K. U.; Ziebeck, K. R. A. J. Magn. Magn. Mater. 1995, 140-
144, 967.
(53) Aharoni, A. Introduction to the Theory of Ferromagnetism; Oxford Science
Publications: 2000, U.K.; p 80.
(59) (a) Swikowicz, R.; Story, T. J. Phys.: Condens. Matter 2000, 12, 8511.
(b) Stachow-Wojcik, A.; Story, T.; Dobrowolski, W.; Arciszewska, M.;
Galazka, R. R.; Kreijveld, M. W.; Swuste, C. H. W.; Swagten, H. J. M.;
de Jong, W. J. M.; Twardowski, A.; Sipatov, A. Y. Phys. ReV. B 1999, 60
(22), 15220.
(60) Marcus, M.; Flood, W.; Stiegerwald, M.; Brus, L.; Bawendi, M. J. Phys.
Chem. 1991, 95, 1572.
(61) (a) Warren, B. E. X-ray Diffraction; Addison-Wesley: Reading, MA 1969.
(b) Langford, J. I.; Louer, D. Rep. Prog. Phys. 1996, 59, 131.
(62) Jensen, P. J.; Bennemann, K. H. Langmuir, 1996, 12, 45.
(54) Chikazumi, S. Physics of Ferromagnetism; Oxford Science: Oxford, 2000;
p 118.
(55) Aharoni, A. Introduction to the Theory of Ferromagnetism; Oxford Science
Publications: 2000, U.K.; p 80.
(56) Noakes, J. E.; Arrott, A. J. Appl. Phys. 1967, 38, 973.
(57) Arrott, A.; Noakes, J. E. Phys. ReV. Lett. 1967, 19 (14), 786.
(58) Saito, H.; Zayets, V.; Yamagata, S.; Ando, K. Phys. ReV. Lett. 2003, 90
(20), 207202-1.
9
J. AM. CHEM. SOC. VOL. 128, NO. 34, 2006 11177

A R T I C L E S
Regulacio et al.
Figure 5. Arrott plot of 20 nm EuS nanoparticles (M2 vs (H/M) isotherms).
Figure 6. Comparison of TC obtained for 14.4 nm EuS nanoparticles (red), 19.1 nm EuS nanoparticles (blue), and bulk EuS (green).
surface compared with the bulk.63 Nonetheless, a second
explanation for the reduced TC is the loss of near neigh-
bors, lowering the average Zh1 and Zh2. For a 20 nm spherical
particle, approximately 10% of the atoms are at the surface.
Assuming the coordination number of surface atoms is half that
of the bulk, one can estimate an average of neighbors for surface
and bulk atoms as Zh1 ≈ 11.4, and Zh2 ≈ 5.7. Using a
modification64 of eq 1:
TC ) TbCulk
jz1 J1 + jz2 J2
12 J1 + 6 J2
|
|
we calculate a TC ≈ 15.8 K, which is close to that observed.
Finally, the third model (and our initial hypothesis) is that J1
is altered by changes in the band gap. Knowing the band gap
and using eq 2, one can estimate changes in J1 assuming that
(63) Dauth, B. H.; Alvarado, S. F.; Campagna, M. Phys. ReV. Lett. 1987, 58
(20), 2118.
9
11178 J. AM. CHEM. SOC. VOL. 128, NO. 34, 2006

Magnetic Properties of Semiconducting Nanoparticles
A R T I C L E S
the terms A (4f valence to 5d conduction band exchange) and
b (orbital overlap) are unaltered by size. Generally, increasing
the Eg will reduce J1 and lower the ordering temperature, which
is consistent with the observed trend that TC decreases as particle
size decreases. The primary weakness in this analysis is
determining an accurate measure of the Eg. In addition to the
similar ambiguities of determining an absorption edge, there is
some question of assigning the transitions in these materials.
The optical properties of bulk EuS are complicated by changes
that occur with temperature and magnetic field.65 The absorption
spectra have historically been interpreted as having a small
absorption centered around the 4f-5dt2g transition with a more
intense broad absorption due to the transition from the 4f to 6s
band. There are inconsistencies over peak assignments in both
thin films and nanoparticles,66 which may be related to purity
or surface defects. These nanoparticles show an absorption at
520 nm, which if we assign this to the 4f-5dt2g transition, gives
a band gap of 2.39 eV. This value is surprisingly large, given
our estimate of the exciton Bohr radius of approximately 5.8
nm, smaller than the nanoparticles reported here.67 Using this
value, nonetheless, to estimate J1, we obtain 0.109, which would
give a calculated approximation of TC ≈ 7 K (using eq 1), which
is far lower than what we observe.
particles. If the peak at 520 nm is the optical band gap, it ought
to vary with nanoparticle radius, according to the Brus equation.
We have observed changes in intensity of this peak but not
position for different synthetic conditions. This would be the
case if we are not yet close to the Bohr radius or if this is not
due to the band gap. If it is a result of surface defects, or
oxidation at the surface, it should be possible to determine this
by varying the capping ligand or by forming core/shell type
nanoparticles. For example, a shell layer of PbS, which has a
close lattice match for EuS (PbS a ≈ 5.94 Å and EuS a ≈ 5.97
Å),68 should identify whether surface states are responsible for
this absorption.69 We are currently using room-temperature
europium Mossbauer spectroscopy to determine the coordination
environment of the europium in these nanoparticles.
Conclusions
In this work we have studied the effect of particle size on
the magnetic ordering temperature in lanthanide magnetic
semiconductors. Our evidence suggests there is a lowering in
the ferromagnetic Curie temperature with decreasing particle
size. Further experiments with a series of controlled particle
sizes (with a more narrow size distribution within a sample)
should distinguish whether strain, surface effects, or band gap
changes are responsible for the changes in strength of the
magnetic coupling.
Several questions remain from this work. The most critical
question, which may help distinguish between the possibilities
outlined above, is what is the functional relationship between
the decrease in TC and particle size? If it depends on the number
of surface atoms, it should vary as 1/R (R being the radius of
the particle), whereas if the Eg has an important role, the Tc
should vary as a function of 1/E2g. Also important is the
question of assigning the electronic transitions in these nano-
Acknowledgment. We thank the National Science Foundation
for funding this work (NER: 0304273, CAREER: 0449829),
and Brad Lewis would like to thank Tom Mallouk for support.
Supporting Information Available: The optical spectroscopy
of the nanoparticles, bulk materials, and theoretical assignment
of transitions are compared in S1-S3. This material is available
free of charge via the Internet at http://pubs.acs.org.
(64) Stachow-Wojcik, A.; Story, T.; Dobrowolski, W.; Arciszewska, M.;
Galazka, R. R.; Kreijveld, M. W.; Swuste, C. H. W.; Swagten, H. J. M.;
de Jong, W. J. M.; Twardowski, A.; Sipatov, A. Y. Phys. ReV. B 1999, 60
(22), 15220.
JA0620080
(65) Busch, G.; Wachter, P. Phys. Kondens. Mater. 1966, 5, 232.
(66) Thongchant, S.; Hasegawa, Y.; Wada, Y.; Yanagida, S. J. Phys. Chem. B
2003, 107, 2193-2196.
(68) Stachow-Wojcik, A.; Story, T.; Dobrowolski, W.; Arciszewska, M.;
Galazka, R. R.; Kreijveld, M. W.; Swuste, C. H. W.; Swagten, H. J. M.;
de Jong, W. J. M.; Twardowski, A.; Sipatov, A. Y. Phys. ReV. B 1999, 60
(22), 15220.
(69) Kuno, M.; Lee, J. K.; Dabbousi, B. O.; Mikulec, F. V.; Bawendi, M. G. J.
Chem. Phys. 1997, 106 (23), 9869.
(67) Using the approach of L. E. Brus (J. Chem. Phys. 1984, 80 (9), 4403),
where the m/h is neglected (assuming is it much larger than me/), m/e is
0.456, and the dielectric constant ꢀ0 is 10 (from Umehara, M. Phys. ReV.
B 2002, 65, 205208).
9
J. AM. CHEM. SOC. VOL. 128, NO. 34, 2006 11179