J Nanopart Res (2010) 12:1307–1318
DOI 10.1007/s11051-010-9862-0
RESEARCH PAPER
Permanent magnetism in phosphine- and chlorine-capped
gold: from clusters to nanoparticles
Miguel A. Muñoz-Márquez • Estefanı́a Guerrero •
Asunción Fernández • Patricia Crespo • Antonio Hernando
Raquel Lucena • José C. Conesa
Received: 20 September 2009 / Accepted: 20 January 2010 / Published online: 6 February 2010
Ó Springer Science+Business Media B.V. 2010
Abstract Magnetometry results have shown that
gold NPs (*2 nm in size) protected with phosphine
and chlorine ligands exhibit permanent magnetism.
When the NPs size decreases down to the subnano-
metric size range, e.g. undecagold atom clusters, the
permanent magnetism disappears. The near edge
structure of the X-ray absorption spectroscopy data
points out that charge transfer between gold and the
capping system occurs in both cases. These results
strongly suggest that nearly metallic Au bonds are
also required for the induction of a magnetic
response. Electron paramagnetic resonance observa-
tions indicate that the contribution to magnetism from
eventual iron impurities can be disregarded.
M. A. Muñoz-Márquez (&)
E. Guerrero
A. Fernández
Instituto de Ciencia de Materiales de Sevilla (CSIC-US),
Av. Americo Vespucio 49, 41092 Sevilla, Spain
e-mail: miguel.angel@icmse.csic.es
P. Crespo
A. Hernando
Instituto de Magnetismo Aplicado (UCM-ADIF-CSIC),
P.O. Box 155, 28230 Las Rozas, Madrid, Spain
P. Crespo
A. Hernando
Departamento de Fı́sica de Materiales, UCM,
Av. Complutense s/n, 28040 Madrid, Spain
R. Lucena
J. C. Conesa
Instituto de Catálisis y Petroleoquı́mica (CSIC), Marie
Curie 2, Campus de Cantoblanco, 28049 Madrid, Spain
•
Keywords Gold clusters
Gold nanoparticles

EPR spectroscopy
SQUID magnetometry

Ferromagnetic behaviour
Introduction
Currently, many nanoscale applications such as
electronic devices, systems with catalytic properties,
magnetic and optical mechanisms, and biological
systems (e.g. Andres et al. 1996; Valden et al. 1998;
Sun et al. 2000; Boyen et al. 2002; Daniel and Astruc
2004; Turner et al. 2008) include transition metal
nanoparticles (NPs) as essential parts to perform their
function. Despite the widespread study and applica-
tion of these particular systems, there are still many
unknowns to be addressed regarding the solid-state
properties and structure of capped transition metal
NPs (Whetten and Price 2007), for instance, their
magnetic properties which can certainly help the
understanding of essential questions in magnetism. It
is known that due to size and surface effects that
appear when the system size is reduced down to the
nanometre range, e.g. the electronic properties of NPs
are significantly different from bulk-like systems of
the very same materials (Alivisatos 1996); however,
sometimes not only the system size is the key factor in
the physical properties and even the capping systems
play an important role in this profound change of the
123
1308
electronic structure (Zhang and Sham 2002). There-
fore, further investigations of the parameter space,
which define the physical and chemical properties of
nanoscale systems, are always welcomed.
Regarding the NPs’ magnetic properties, it is now
widely accepted that a ferromagnetic-like behaviour
is observed in one-dimensional nanoscale systems
such as gold, silver, palladium and copper which are
diamagnetic in their bulk-like form. This claim leans
on a wealth of experimental results already published
in world-leading scientific journals (e.g. Shinohara
et al. 2003; Sampedro et al. 2003; Crespo et al. 2004;
Yamamoto et al. 2004; Negishi et al. 2006; Suber
et al. 2007; Dutta et al. 2007; Garitaonandia et al.
2008; de la Venta et al. 2009), performed by wholly
independent research groups throughout the world. In
addition, this experimental work is now supported by
recent theoretical total energy calculations (Gonzalez
et al. 2006; Luo et al. 2007; Michael et al. 2007)
which yielded encouraging results on size-dependent
magnetization and spin symmetry-breaking that
would explain the experimental results found for
thiol-capped gold and silver nanoparticles. Probably,
the most striking advance at the light of the latest
magnetization studies based on element-specific
techniques (Yamamoto et al. 2004; Negishi et al.
2006; Garitaonandia et al. 2008; de la Venta et al.
2009), such as X-ray magnetic circular dichroism
(XMCD) and Mössbauer spectroscopy, is that it has
been unequivocally determined that the gold atoms
neighbouring the chemisorption site at the nanopar-
ticle surface are the carriers of the magnetic moment.
At this point, it has to be mentioned that all the
previous studies mainly concern thiol-capped transi-
tion metal nanoparticles and, as a result, the magnetic
behaviour has been partly attributed to the charge
transfer effect measured in the metal-S bond (Guer-
rero et al. 2007); including the fact that magnetism
was not observed in polymer-like gold compounds
such as Au2S. Of course, as it will be discussed later,
this is not the only effect responsible of the
ferromagnetic-like behaviour. However, the well-
known Brust method used to synthesize highly
monodisperse thiol-capped gold NPs from gold salt
(HAuCl4) reduction (Brust et al. 1994) has lead to a
widespread use and thorough study of the Au–S bond
properties in nanometre-size systems, being these
gold-thiolate species the building blocks of many
self-assembled systems. Despite the Brust method is
123
J Nanopart Res (2010) 12:1307–1318
a highly reproducible and straightforward synthesis
procedure, it is not free of experimental difficulties.
In fact, it has been proved almost impossible to obtain
subnanometric gold clusters using this synthesis
method which would justify the importance of the
synthesis methods developed by Weare et al. (2000)
for small phosphine-stabilized Au NPs (diameter *2
nm) and, the procedures established by Bartlett et al.
(1978) to obtain phosphine-capped undecagold clus-
ters (subnanometric particles with a diameter around
0.8 nm) that have played a very important role in the
research presented in this article. These synthesis
methods allowed the fabrication of smaller particles
than the ones obtained by the Brust method and, in
addition, they have the potential of exchanging the
phosphine shell with a protecting thiolated ligand
layer (Song et al. 2003; Woehrle and Hutchison
2005); so the final result is a set of very small thiol-
capped gold NPs or clusters.
Moreover, the obtained NPs have been subject of
controversy regarding their magnetic properties
which some authors have ascribed to natural occur-
ring iron impurities that can be found in the reagent
grade chemicals used across the synthesis process or,
as a result of using stainless steel tools whilst
manipulating the synthesis products and reactants
(Abraham et al. 2005). However, it has been demon-
strated that superparamagnetic iron impurities cannot
be responsible of the ferromagnetic behaviour in gold
NPs (Crespo et al. 2006). The electron paramagnetic
resonance (EPR) studies carried out in this investi-
gation are key to probe the magnetic state of eventual
impurities and, most important, to check their role on
the macroscopic magnetic behaviour.
The structural and electronic properties of phos-
phine-capped clusters and nanoparticles have been
deeply studied in the past (Weare et al. 2000; Bartlett
et al. 1978; Mingos 1976; Briant et al. 1981; Teo
et al. 1992; Mingos 1996; Schmid 2008). However,
nobody before has explored the magnetic properties
of these systems in order to tackle the eventual
relationship between the magnetic behaviour and the
electronic/atomic structure. Moreover, there is a
long-standing issue regarding the physical and chem-
ical properties of these nanomaterials where the
driving factors of the thermodynamic stability are
still unknown (Walter et al. 2008). These were the
reasons that lead us to study the magnetic properties
of phosphine-capped gold nanoparticles and clusters.
J Nanopart Res (2010) 12:1307–1318
As previously said, the nanoparticle surface atom
bonding with the capping ligand is not the unique
concurring factor in the emergence of a ferromag-
netic-like behaviour in gold NPs. The size effect is
also a key point: the nanoparticle diameter is very
similar to the wavelength of the confined electrons in
the nanoparticle and this will determine many of the
physical properties of these particular systems at the
quantum level. In addition, the ferromagnetic-like
behaviour shall be strongly linked to the surface
effects originated by the dramatic increase of atoms
that lose the bulk 3D symmetry in the NP surface; the
atoms in the surface are only constrained in a 2D
fashion, hence, the NP physical properties undergo
significant changes associated to a modification in the
chemical environment and atomic structure. For
instance, considering the thiol-capped NPs, due to a
huge increase in the mobility of the thiol-Au species
(Yu et al. 2006), the bonding of the sulphur atom in
the thiol head to a gold surface atom leads to a deep
change of the outermost surface atoms structure
which necessarily must involve a dramatic modifica-
tion of the electronic structure.
Experimental details
Sample preparation
Au11-TPP synthesis
This subnanometric gold cluster is synthesized by
reduction of the AuCl(PPh3) precursor which was
synthesized according to the procedure described by
Braunstein et al. (1990). Following the synthesis
method established by Bartlett et al. (1978), 0.18 g
of Au(PPh3)Cl are dissolved in 8 ml of absolute
ethanol and then reduced with a solution of NaBH4 in
ethanol (0.0137 g of solute in 2 mL of solvent) over
a 15 min period whilst the reducing agent is slowly
added under nitrogen atmosphere. The mixture is
then stirred for 2-h. The formed solid is precipitated
with hexane and, filtered and washed with CH2Cl2/
hexane. The remaining solid is dissolved in CH2Cl2
and filtered to eliminate a colourless and insoluble
powder. Finally, the product is precipitated with
hexane.
1309
nAu-TPP synthesis
Phosphine-chlorine-capped Au NPs could be directly
obtained from HAuCl4 reduction in presence of
triphenylphosphine (PPh3 or TPP), using NaBH4 as
reducing agent. The synthesis method used was first
reported by Weare et al. (2000) which is a modifica-
tion of the well-known Brust method (1994). Au(III) is
transferred from an aqueous solution of 0.2 g of
HAuCl4 in 12 mL of milli-Q water to degassed and
dried toluene; a solution of 0.28 g of tetraoctylammo-
nium bromide in 12 mL of toluene was used as the
phase-transfer agent. The mixture is strongly stirred
for 5 min whilst 0.46 g of PPh3 are added to the
solution which is further stirred for 10 more minutes. A
fresh solution of NaBH4 (0.4 g of solute in 2 mL of
milli-Q water) is quickly added whilst stirring then, the
mixture is continuously stirred for three hours. The
organic phase is then separated from the aqueous
phase. Toluene is removed under reduced pressure by
means of a rotary evaporator. The solid was precip-
itated with hexane and, finally, filtered and washed
with hexane and MeOH/H2O.
Sample characterization
Particle characterization using elemental analysis,
ICP, TEM and SEM
Aliquots of the synthesized nanoparticles were
dropped onto the carbon film on a 200-mesh copper
grid and air dried. Nanoparticles were imaged
employing a Philips CM200 TEM microscope work-
ing at 200 kV at the Microscopy Service of the
Instituto de Ciencia de Materiales de Sevilla. The
approximate particle size distribution histograms were
measured using image analyser software that deter-
mines the cluster radii from a digital image of the
micrographs. The chemical composition by elemental
analysis in a LECO CNS-2000 and ICP-MS in a
Thermo Elemental X-7 was determined at the Centro
de Investigación, Tecnologı́a e Innovación de la
Universidad de Sevilla (CITIUS). In addition, the
chemical composition was also checked using EDX in
the TEM microscope and in a SEM Hitachi S-4800.
Samples for EDX-SEM analysis were placed in
carbon tape.
123
1310
Optical transitions
UV–Vis absorption spectra were recorded in transmis-
sion mode. In this experiment, the gold clusters and nano-
particles were dispersed in liquid solutions (1 mg/mL)
of ethanol and placed into a high transmission quartz
cuvette; so good transmission is achieved in the UV
region. The spectra were collected in the range
350–850 nm with a Shimadzu UV-2102 PC spectrom-
eter at room temperature.
SQUID and EPR measurements
The hysteresis curves were obtained using a Quan-
tum Design S6000 magnetometer at the Instituto de
Magnetismo Aplicado in Madrid. Data were col-
lected at temperatures between 5 and 300 K using a
liquid helium cooling system. The samples were
placed in adhesive kapton stuck to a quartz tube. The
diamagnetic contribution from the sample holder was
measured and subtracted from the total magnetiza-
tion. Meanwhile, the EPR measurements were
carried out at the Instituto de Catálisis y Petroleoquı́-
mica in Madrid on a Bruker ER200D instrument
operating in the X-band and interfaced to a digital
data acquisition system. Aliquots of the studied
samples were placed into a special spectroscopically
pure quartz cell. All the spectra were recorded in a
TE104-type double cavity. The frequency of the
microwave was calibrated for each experiment
using a standard of the stable-free radical Diph-
enylpicrylhydrazyl (DPPH) with g = 2.0036 located
in the second cavity.
X-ray absorption spectra measurements
XAS of the Au samples were recorded in transmis-
sion mode at the BM29 beamline of the European
Synchrotron Radiation Facility (ESRF) storage ring.
The samples were measured as thin self-supported
pressed pellets diluted in boron nitride. Spectra were
recorded at the Au L2- and L3-edge, at 13,734 and
11,919 eV, respectively. Simultaneously, a gold foil
standard was measured to calibrate the X-ray energy.
To compare the X-ray absorption near-edge structure
region, a linear background was fitted in the pre-edge
region and subtracted before normalization to the
edge jump.
123
J Nanopart Res (2010) 12:1307–1318
Results and discussion
Chemical composition: elemental analysis,
ICP and EDX
Two different types of gold nanomaterials were
chemically synthesized for this study: triphenylpho-
sphine-capped undecagold clusters labelled as Au11-
TPP and gold nanoparticles which were named nAu-
TPP. As it will be discussed later, both particles have
a significant chlorine content coming from the
precursors AuCl(PPh3) and HAuCl4 used in the
fabrication. Across the entire synthesis process,
Teflon-coated stainless steel tools, as well as, brand
new laboratory glassware were used to avoid
unwanted effects on the magnetometry results com-
ing from eventual ferromagnetic contamination.
The undecagold compound was first synthesized
by Bartlett et al. (1978), although there were previous
theoretical studies on this system (Mingos 1976) that
focused on various Au11 and Au13 based phosphine-
passivated clusters. Since the synthesis method
followed here claims the production of undecagold
gold clusters, then, despite the obtained sample could
contain both phases (Au11 and Au13) we will refer to
it as undecagold cluster which seems to be the most
frequent phase. According to the literature, the result
of this synthesis should be a Au11(PPh3)7Cl3 cluster
which agrees pretty well with the chemical compo-
sition results reported in this section (cf. Table 1).
Regarding the phosphine-stabilized gold NPs, We-
are et al. (2000) slightly modified the Brust method for
thiol-capped Au NPs synthesis so phosphine-capped
gold NPs could be obtained. This method was born as
an alternative to the more complicated synthesis
method originally formulated by Schmid et al. (1981)
which required from strictly anaerobic conditions and
diborane gas as a reducing agent, and resulted in 1.4 nm
Au55(PPh3)12Cl6 clusters. The NPs obtained following
Weare et al. (2000) method should have a mean size of
1.5 nm which according to the authors corresponds to a
metal core of *101 gold atoms that leads to an
estimated formula of Au101(PPh3)21Cl5. However, as it
will be shown later and despite having reproduced the
synthesis method described by Weare et al. (2000), the
nanoparticles produced in our laboratory have a mean
diameter of *2 nm which, according to the composi-
tional analysis performed, would lead to an estimated
composition Au225(PPh3)80Cl16.
J Nanopart Res (2010) 12:1307–1318
Table 1 Summary of the chemical composition determined by ICP, elemental analysis and EDX (TEM and SEM), along with some
structural parameters determined from the TEM micrographs, of the phosphine-capped particles studied in this article
Sample Au Fe P C
(% wt)a (% wt)b
Au11-TPP 50.1 \0.03 5.2 36.0
nAu-TPP 43.5 \0.03 5.9 41.1
a
As determined by the ICP
b
As determined by elemental analysis
c
Qualitative determination by SEM-EDX
d
As determined by TEM-EDX
Determining the exact composition of the syn-
thesized samples has been proved a very difficult
task. Compositional data were obtained from induc-
tively coupled plasma-atomic emission spectrometry
analysis (ICP) for the heaviest elements (Au and Fe)
whilst the quantity of lightest elements (P, C and H)
was determined by elemental chemical analysis. The
results are summarized in Table 1. Unfortunately,
despite having performed a filtration procedure, still
some residual ligands remained free, not bonded to
any gold particle and, therefore, this led to an
overestimation of the gold-ligand ratio. One should
remember that ICP and elemental chemical analysis
provide a chemical composition quantification of the
sample as a whole. Instead, energy dispersive X-ray
analysis (EDX), measured in a transmission electron
microscope (TEM), is able to quantify the chemical
composition in a very small region mainly occupied
by nanoparticles which results in a more accurate
determination of the gold-ligand ratio: essentially,
Au:P ratio. TEM studies also allowed the size
determination of the clusters and nanoparticles.
However, some difficulties were found when mea-
suring the Cl content in the studied samples: in
addition to the low Cl content, the effect of a low
electron scattering cross section at 200 kV lead to a
non measurable emission line from the Cl K-shell. A
supplementary EDX spectrum was recorded in a
different electron microscope; in this case, a 20 kV
scanning electron microscope (SEM) where the Cl
K-shell at 2822 eV could be measured since the
electron cross section is noticeably higher at 20 kV.
This was used to qualitatively determine the pres-
ence of chlorine in the synthesized samples. The
EDX results and the TEM size determination are
included in Table 1.
1311
H Cl Au:P Dm SD r
(% at)c (at. ratio)d (nm) (nm)
2.5 3.0 1.3 1.41 0.14
2.9 3.6 3.3 2.1 0.2
Cluster and nanoparticle topological structure
According to previous results, an incomplete icosa-
hedral structure with an approximately C3v symmetry
axis (Bartlett et al. 1978; Nunokawa et al. 2006;
Barnard et al. 2009) is assumed for the phosphine-
capped undecagold clusters which have a core diam-
eter of 0.8 nm. As it has already been described by
Menard et al. (2006b), this value is slightly overes-
timated when measured by conventional bright field
(BF) TEM (cf. Fig. 1). For clusters below 1 nm, the
poor contrast between the smallest metal NPs and the
support films used for the electron microscopy studies
tends to bias microscopic measurements towards any
subpopulation of larger sized particles (Narayanasw-
amy and Marks 1993; Wilcoxon et al. 2000).
Meanwhile, the nAu-TPP sample with an average
diameter of *2 nm as shown in Fig. 1, and considering
a slight experimental overestimation, would correspond
to a bulk-like fcc structure with *225 gold core atoms,
which is consistent with one of the ‘magic number’
structures already reported in previous studies (Brust
et al. 1994; Whetten et al. 1996): on the basis of X-ray
diffraction (XRD) experiments recorded by Whetten at
al. (Whetten et al. 1996), a 2% expansion of the bulk
gold lattice constant is considered. In Fig. 2 there is a
perspective view diagram of the assumed models for
both, undecagold clusters and 2 nm nanoparticles. It
should be noted that the theoretical Au:P atomic
ratio shown in the nAu-TPP model nearly match the
experimental results obtained from the EDX analysis.
Despite the Au:P ratio might seem very low compared
to the same ratio as calculated in the well-
established Au55-TPP clusters, it is very similar to the
Au:P ratio obtained for the also well-characterized
[Au39(PPh3)14Cl6]Cl2 cluster (Teo et al. 1992). This
123
1312
Fig. 1 Transmission
electron micrographs of
phophine-capped Au
clusters and nanoparticles,
and their corresponding size
distribution histograms. The
solid line represents the
fitting curve assuming a
log-normal function. The
calculated mean particle
diameter (Dm) and the
standard deviation (r) are
shown in the histograms
illustrative figure provides a very good idea on how
well-packed the PPh3 molecules must be in the NP
surface, so *80 of them, along with 16 chlorine atoms,
will fit in one single particle. Incidentally, regarding
the molecular packing structure of phenyl rings on
Au(111) surfaces, it has been reported that phenyl rings
which usually lie flat on the gold surface when the
coverage is below half-monolayer, can also adsorb
along the ring edge in stand up position to improve the
surface coverage in saturated conditions (Müllegger
et al. 2006). The adsorption site of PPh3 molecules and
Cl ligands is in agreement with previous structural
studies of solvated nanoparticles (Periyasamy and
Remacle 2009) and conventional surface science
investigations (e.g. Steiner et al. 1992; Kastanas and
Koel 1993; Baker et al. 2008); the adsorption of
electronegative species are found to affect signifi-
cantly the gold surface structure, in particular, these
adsorbates can lift gold atoms from the surface. The
preferred adsorption sites are surface defects such as
adatoms, step edges, kinks and gold vacancies. Hence,
according to previous results by Periyasamy and
Remacle (2009), the PPh3 molecules have been coor-
dinated to the face edge atoms of the gold core outer
layer, i.e. to the corner atoms. Meanwhile, the chlorine
ligands are symmetrically coordinated to the face-
centred gold atoms in the sides of the nanoparticle core.
123
J Nanopart Res (2010) 12:1307–1318
The formation of a self-assembled monolayer which, as
it has been proved before (Yu et al. 2006), would have
profound implications concerning the shell-core bond-
ing that might go beyond a plain chemisorption. A
strong bonding to the nanoparticle gold atoms will
deeply change the outermost surface core atoms
topological structure and, subsequently, the electronic
structure: this point may be crucial to explain the origins
of the magnetic properties.
Unfortunately, no additional information could be
extracted from these nanoparticles and clusters by
means of high-resolution transmission electron
microscopy (HRTEM). Although valuable informa-
tion has been gathered from HRTEM studies in larger
supported nanoparticles, the application of this method
to smaller clusters (D \ 3nm) is limited by image-
contrast, momentum-transfer, and beam-induced
mobility considerations (Cleveland et al. 1997). How-
ever, the UV–Vis absorption spectra were extremely
valuable since the quality of the synthesized particles
could be checked using the absorption features. The
UV-Vis spectra are shown in Fig. 3. The dashed line
corresponds to the absorption spectrum of the Au11-
TPP cluster which shows an absorbance feature at
*420 nm typical of subnanometric phosphine-stabi-
lized undecagold (Bartlett et al. 1978). Meanwhile, the
solid line represents the absorption spectrum of the
J Nanopart Res (2010) 12:1307–1318
Fig. 2 Perspective view of the cluster and nanoparticle
structure. On the left hand side panel, the subnanometric
cluster Au11-TPP with an icosahedral structure, whereas the fcc
bulklike *2 nm nAu-TPP nanoparticle is represented in the
right hand side panel. The gold core atoms are yellow, the red
balls correspond to phosphorus and the green ones to chlorine.
For simplicity’s sake, the aromatic rings of the triphenylphos-
phine molecule are not included
Fig. 3 UV–Vis absorption spectra of nAu-TPP (solid line) and
Au11-TPP (dashed line) in ethanol
nAu-TPP nanoparticles with a broad plasmon reso-
nance feature at 520 nm, characteristic of *2 nm
nanoparticles (Alvarez et al. 1997).
Magnetic behaviour: SQUID and EPR
measurements
The magnetic behaviour of the systems presented here
was studied by means of superconducting quantum
1313
interference device (SQUID) magnetometry. The
recorded hysteresis cycles at the lowest and highest
temperature, presented in Fig. 4a and c, provided
sound information regarding the macroscopic magne-
tization of the studied systems and, clearly show the
ferromagnetic-like behaviour of the phosphine-stabi-
lized gold nanoparticles (nAu-TPP) whilst the phos-
phine-capped clusters (Au11-TPP) do not present any
magnetic feature being diamagnetic at low and high
temperature. Moreover, the magnetization versus
temperature is shown in Fig. 4b and d for samples
nAu-TPP and Au11-TPP, respectively. The depen-
dence of the magnetization with temperature of the
nAu-TPP sample is consistent with the observed
behaviour in thiol-capped gold nanoparticles (Crespo
et al. 2004): the hysteresis phenomenon is observed
up to room temperature which means that this
particular system has a blocking temperature above
300 K corresponding to an anisotropy constant of
3 9 107 J m-3. Only the coercive field decreases from
850 Oe down to 300 Oe when raising the temperature
from 5 K up to room temperature. Meanwhile, the
magnetic saturation of the nAu-TPP is 0.08 lB/NP.
This value is similar to the magnetization values
associated to the Au 5d orbitals as measured with
XMCD experiments by Garitaonandia et al. (2008)
and, more recently, by de la Venta et al. (2009). The
fractional magnetization values (\1 lB) confirm that
the electron is partially shared between the surface
gold atom and the organic ligand. Since, the blocking
has its origins in the strong spin-orbit field (Hernando
et al. 2006b) then, the blocking behaviour will exist
for magnetization values well below 1 lB.
At this point an important question arises: is the
magnetic behaviour due to the existence of ferro-
magnetic and/or paramagnetic iron impurities? At the
light of the compositional results which show an iron
content below 0.01% wt., the aforementioned possi-
bility sounds more than plausible. However, since the
measured Fe content is just below the detection limit
of the available chemical analysis device, then a more
sensitive technique was needed. Here is where the
electron paramagnetic resonance spectroscopy comes
into play. This technique is able to detect very low
quantities of chemical species that have one or more
unpaired electrons. Therefore, the EPR results will
provide information on the presence of magnetic
impurities and also on their magnetic state. These
results are shown in Fig. 5. The spectra were
123
1314
(a)
(b)
Fig. 4 SQUID magnetometry results. Panels in the left-hand
side correspond to the nAu-TPP sample and panels in the right-
hand side are the results obtained from the subnanometric
cluster Au11-TPP. Figures a and c contain the magnetization
curves from nAu-TPP and Au11-TPP, respectively, measured at
recorded at 77 and 298 K. The almost featureless
EPR signals suggest that there are not magnetic iron
impurities in any of the studied samples; in particular,
according to Jiang et al. (2002) if Fe3? impurities
were present an intense peak around 1,600 gauss (g =
4.28 at crossover point) should appear in the EPR
spectrum. The only remarkable feature can be found
in the spectra recorded from the nAu-TPP sample,
which presents a broad peak at low g values (g * 2.1
or B * 3000 Oe). This feature that is observed at
both high and low temperatures increases its line-
width and shifts towards higher g values at low
temperature which is a characteristic behaviour of
ferromagnetic centres (Tronconi et al. 1993; Fittip-
aldi et al. 2009) such as our phosphine-capped gold
nanoparticles. Finally, at 77 K, between 5000 and
123
J Nanopart Res (2010) 12:1307–1318
(c)
(d)
5 and 300 K. Meanwhile, bottom panels show the thermal
dependence of the magnetization under an applied magnetic
field of 500 Oe: figures b and d correspond to samples nAu-
TPP and Au11-TPP, respectively. The inset in figure (a) shows
a magnification of the hysteresis curve between -1 to 1 k Oe
8000 gauss in both spectra, there are spiky peaks
which correspond to paramagnetic O2 molecules that
were in the analysis chamber due to pressure
conditions, slightly higher than the ones set at
298 K. The EPR results are very useful to rule out
any potential contribution of magnetic iron impurities
to the measured hysteresis cycles. Thus, the ferro-
magnetic-like behaviour of the nAu-TPP sample must
have its origins somewhere else.
Electronic structure
Another point to be explored is an eventual polari-
zation of the Au–P and Au–Cl bonds. Here, the X-ray
absorption spectroscopy (XAS) experiments have
been very valuable. In particular, the X-ray
J Nanopart Res (2010) 12:1307–1318
Fig. 5 Electron paramagnetic resonance spectra from the
studied samples: nAu-TPP upper panel and Au11-TPP bottom
panel. Data were collected at T = 298 and 77 K. In both
samples, spiky features appear between 5,500 and 8,000 gauss
magnetic fields in the EPR spectra taken at 77 K; these
correspond to molecular oxygen present in the analysis
chamber due to vacuum conditions slightly higher than the
typical base pressure
absorption near-edge structure (XANES) data of the
Au-L3 edge, which have a first resonance around
5 eV above the threshold energy (whiteline) associ-
ated with a 2p3=2 ! 5d5=2;3=2 dipole transition that
is actually probing the density of unoccupied d states
at the Fermi level (Zhang and Sham 2002). Despite
gold should have a nominally full 5d band, due to s-
p-d hybridization, a faint whiteline is still detected
for bulk gold (cf. Fig. 6). The area under the XANES
curve will provide an accurate estimation on the
charge redistribution, i.e. a charge transfer phenom-
enon taking place in the gold-ligand bond. As it has
been previously discussed, the charge transfer effect
that occurs in the Au–S bond of the thiol-capped gold
NPs is an essential point to explain the origins of the
magnetic behaviour. However, considering that
1315
according to the XANES results reported in this
article where subnanometric phosphine-protected
gold clusters present a noticeably large charge
transfer in the Au-ligand bond and, according to the
SQUID experiments on the very same clusters where
a diamagnetic behaviour is observed, then the charge
transfer effect can be ruled out as the only responsible
effect of the ferromagnetic-like character. This claim
is reinforced if the charge transfer of the nAu-TPP is
considered. In this case, the measured resonance of
the XANES edge is significantly less intense than in
the Au11-TPP sample and, in spite of this, the
ferromagnetic-like character appears in the nanopar-
ticles whilst it is not observed in the subnanometric
clusters. Recently, Walter et al. (2008) have per-
formed density functional theory calculations of
structurally characterized ligand-protected gold clus-
ters and nanoparticles. As a result, their determina-
tion of the electronic structure, i.e. d-hole generation,
agrees with the magnetic behaviour previously
observed in thiol-capped gold NPs (Crespo et al.
2006). Walter et al. (2008) also claim that, in the
Au11(PPh3)7Cl3 clusters, the phosphine ligands are
weak surfactants that cannot significantly modify the
electron shell structure of the gold cluster core and, it
would be the chlorine ligand the one that plays the
role of the thiol ligand in the aforementioned
magnetic nanoparticles (Crespo et al. 2006). In our
case, the strong charge transfer observed from the
XANES spectrum of the undecagold clusters might
be solely attributed to the Au–Cl bond. Unfortu-
nately, the data presented in this research work are
not able to discern whether the major charge transfer
effect is due either to phosphine or chlorine ligands.
The presence of possible individual Au?-species has
not been detected by EPR. However, regardless of
what is the exact relative number of Au–P and Au–Cl
bonds, it seems to be very clear that a balance
between Au-ligand charge transfer and Au–Au bond
is indeed necessary in order to originate a net
magnetic moment. This balance as observed from
the XANES whiteline represents a slight charge
transfer large enough to originate a strong electric
dipole along the bond but, at the same time, weak
enough to keep the metallic character of the inner
gold atoms in the nanoparticle core. The optimum
balance has been found for the thiol-capped gold NPs
synthesized following the Brust method and also for
the nAu-TPP NPs.
123
1316
Fig. 6 Au L3-edge XANES spectra of the the synthesized
samples: nAu-TPP and Au11-TPP, compared with bulk Au and
the polymeric precursor Au(PPh3)Cl used in the cluster
synthesis
Beyond this argument, there is a point to be
considered yet: the electronic structure of the system
as a whole. As it has been previously investigated
(Bartlett et al. 1978; Yang and Chen. 2003; Menard
et al. 2006a), the phosphine-capped gold clusters
have a molecule-like electronic structure which
presents almost discrete electron energy levels below
500 nm in the UV–visible absorption range. Instead,
the electronic structure of the phosphine-capped gold
NPs presents a broad plasmon feature around
520 nm. This particular behaviour is due to the
presence or not of metallic Au–Au bonds in the
system; the surface plasmon resonance feature is
strongly marked for gold particles above 10 nm
which have an almost metallic character. The pres-
ence of adsorbates in the nanoparticle surface,
namely PPh3 molecules, which reduce the surface
electrons mobility would significantly damp the
surface plasmon resonance as it has already been
proved (Guerrero et al. 2008). However, if the
adsorbate molecules, in this case phosphine and
chlorine ligands, form a well-ordered and close-
packed structure then, small domains of ordered PPh3
molecules and Cl atoms could self-assemble in the
nanoparticle core facets. This would finally lead to an
123
J Nanopart Res (2010) 12:1307–1318
arrangement of the charge dipoles formed at the Au-
ligand bond and, consequently, the nanoparticle
surface electrons would collectively move along the
domain boundaries. The coexistence of a collective
electron movement that leads to a weak surface
plasmon resonance and some degree of mobility is a
key point in the appearance of a permanent magnetic
moment (Hernando et al. 2006a). The well-ordered
structure of the capping system, far from ideal, is a
more than possible situation. In fact, a close-packed
layer is the only way to explain such a high number
of ligands: *80 PPh3 molecules and 16 Cl atoms for
a given nanoparticle, as it has been determined by the
chemical composition study.
Conclusions
In summary, the existence of a permanent magnetic
behaviour in phosphine-chlorine-capped gold nano-
particles has been shown for the first time by means
of SQUID magnetometry. The EPR experiments
confirm that Fe3? species are not present in the
samples; therefore, the origins of the observed
ferromagnetic behaviour must be somewhere else.
These results agree with previous investigations on
thiol-capped gold nanoparticles carried out using
XMCD and Mössbauer spectroscopy. The peculiar
magnetic behaviour does not have its origins in one
single factor; instead, there are several factors that
contribute to the appearance of a permanent magnetic
moment. One factor is a charge transfer effect
occurring at the Au-ligand bond which must be
linked to the presence of Au–Au bonds that provide a
metallic character: this is the reason that explains the
diamagnetic behaviour of phosphine-chlorine-capped
clusters, where there is very probable that the Cl
anions are dominating the charge transfer pulling the
electrons towards the ligand and, therefore limiting
the mobility of the nearby electrons that finally
vanishes any chance of generating an orbital mag-
netic moment. Additionally, the EPR spectra also
show a faint peak at low g values in the nAu-TPP
sample which could be attributed to the ferromagne-
tism observed in these nanoparticles. Finally, the self-
assembling of an ordered capping layer is also a key
factor since the formation of charge dipole domains
will drive the movement of the surface electrons that
finally lead to the appearance of a magnetic moment.
J Nanopart Res (2010) 12:1307–1318
Acknowledgements The authors would like to acknowledge
the support of the European Synchrotron Radiation Facility and
BM29 beamline staff. This research has been supported by the
Spanish Ministry of Science ‘Ministerio de Ciencia e Innovación
(MICINN)’ (Strategic Action NAN2004-09125-C07) and the
Andalusian Government ‘Junta de Andalucı́a’ (Excellence
Project P06-FQM-02254 and P09-FQM-4554, group TEP127).
M.A. Muñoz-Márquez thanks the Spanish Research Council
‘Consejo Superior de Investigaciones Cientı́ficas (CSIC)’ I3P
programme, E. Guerrero acknowledges the MICINN for
financial support and R. Lucena thanks CSIC for a PhD grant.
References
Abraham DW, Frank MM, Guha S (2005) Absence of mag-
netism in hafnium oxide films. Appl Phys Lett 87(25):252,
502
Alivisatos AP (1996) Semiconductor clusters, nanocrystals,
and quantum dots. Science 271(5251):933–937
Alvarez MM, Khoury JT, Schaaff TG, Shafigullin MN, Vez-
mar I, Whetten RL (1997) Optical absorption spectra of
nanocrystal gold molecules. J Phys Chem B 101(19):
3706–3712
Andres RP, Bielefeld JD, Henderson JI, Janes DB, Kolagunta
VR, Kubiak CP, Mahoney WJ, Osifchin RG (1996) Self-
assembly of a two-dimensional superlattice of molecularly
linked metal clusters. Science 273:1690–1693
Baker TA, Friend CM, Kaxiras E (2008) Chlorine interaction
with defects on the Au(111) surface: a first-principles
theoretical investigation. J Chem Phys 129(10):104, 702
Barnard AS, Young NP, Kirkland AI, van Huis MA, Xu H
(2009) Nanogold: a quantitative phase map. ACS Nano
3(6):1431–1436
Bartlett PA, Bauer B, Singer SJ (1978) Synthesis of water-
soluble undecagold cluster compounds of potential
importance in electron microscopic and other studies of
biological systems. J Am Chem Soc 100(16):5085–5089
Boyen HG, Kästle G, Weigl F, Koslowski B, Dietrich C,
Ziemann P, Spatz JP, Riethmüller S, Hartmann C, Möller
M, Schmid G, Garnier MG, Oelhafen P (2002) Oxidation-
resistant gold-55 clusters. Science 297(5586):1533–1536
Braunstein P, Lehner H, Matt D, Burgess K, Ohlmeyer MJ
(1990) Inorganic Syntheses, vol 27, chap 5: transition
metal cluster complexes. Wiley InterScience, Malden.
Section 42, A platinum-gold cluster: chloro-1 j Cl-
bis(triethylphosphine-1jP)bis(triphenyl-phosphine)- 2jP,
3jP-triangulo-digold-platinum(1?) trifluoromethane-
sulfonate, pp 218–221
Briant CE, Theobald BRC, White JW, Bell LK, Mingos DMP
(1981) Synthesis and X-ray structural characterization of
the centered icosahedral gold cluster compound [Au13(P-
Me2Ph)10Cl2](PF6)3: the realization of a theoretical pre-
diction. J Chem Soc Chem Commun (5):201–202
Brust M, Walker M, Bethell D, Schiffrin DJ, Whyman R
(1994) Synthesis of thiol-derivatised gold nanoparticles in
a two-phase liquid–liquid system. J Chem Soc Chem
Commun, p 801
Cleveland CL, Landman U, Schaaff TG, Shafigullin MN,
Stephens PW, Whetten RL (1997) Structural evolution of
1317
smaller gold nanocrystals: the truncated decahedral motif.
Phys Rev Lett 79(10):1873–1876
Crespo P, Litrán R, Rojas TC, Multigner M, de la Fuente JM,
Sánchez-López JC, Garcı́a MA, Hernando A, Penadés S,
Fernández A (2004) Permanent magnetism, magnetic
anisotropy, and hysteresis of thiol-capped gold nanopar-
ticles. Phys Rev Lett 93(8):87, 204
Crespo P, Garcı́a MA, Fernández-Pinel E, Multigner M, Al-
cántara D, de la Fuente JM, Penadés S, Hernando A
(2006) Fe impurities weaken the ferromagnetic behavior
in Au nanoparticles. Phys Rev Lett 97(17):177, 203
Daniel MC, Astruc D (2004) Gold nanoparticles: assembly,
supramolecular chemistry, quantum-size-related proper-
ties, and applications toward biology, catalysis and
nanotechnology. Chem Rev 104(1):293–346
de la Venta J, Bouzas V, Pucci A, Laguna-Marco MA, Haskel
D, te Velthuis SGE, Hoffmann A, Lal J, Bleuel M, Rug-
geri G, de Julián Fernández C, Garcı́a MA (2009) X-ray
magnetic circular dichroism and small angle neutron
scattering studies of thiol capped gold nanoparticles.
J Nanosci Nanotech 9:6434–6438
Dutta P, Dal S, Seehra S, Anand M, Roberts CB (2007)
Magnetism in dodecanethiol-capped gold nanoparticles:
role of size and capping agent. Appl Phys Lett 90(21):102,
213
Fittipaldi M, Sorace L, Barra AL, Sangregorio C, Sessoli R,
Gatteschi D (2009) Molecular nanomagnets and magnetic
nanoparticles: the EMR contribution to a common
approach. Phys Chem Chem Phys 11(31):6555–6568
Garitaonandia JS, Insausti M, Goikolea E, Suzuki M, Cashion
JD, Kawamura N, Ohsawa H, Gil de Muro I, Suzuki K,
Plazaola F, Rojo T (2008) Chemically induced permanent
magnetism in Au, Ag and Cu nanoparticles: localization
of the magnetism by element selective techniques. Nano
Lett 8(2):661–667
Gonzalez C, Simón-Manso Y, Marquez M, Mujica V (2006)
Chemisorption-induced spin symmetry breaking in gold
clusters and the onset of paramagnetism in capped gold
nanoparticles. J Phys Chem B 110(2):687–691
Guerrero E, Rojas TC, Multigner M, Crespo P, Muñoz-Márquez
MA, Garcı́a MA, Hernando A, Fernández A (2007) Evo-
lution of the microstructure, chemical composition and
magnetic behaviour during the synthesis of alkanethiol-
capped gold nanoparticles. Acta Mater 55(5):1723–1730
Guerrero E, Muñoz-Márquez MA, Garcı́a MA, Crespo P,
Fernández-Pinel E, Hernando A, Fernández A (2008)
Surface plasmon resonance and magnetism of thiol-cap-
ped gold nanoparticles. Nanotechnology 19(17):175, 701
Hernando A, Crespo P, Garcı́a MA (2006a) Origin of orbital
ferromagnetism and giant magnetic anisotropy at the
nanoscale. Phys Rev Lett 96(5):57, 206
Hernando A, Crespo P, Garcı́a MA, Fernández-Pinel E, de la
Venta J, Fernández A, Penadés S (2006b) Giant magnetic
anisotropy at the nanoscale: overcoming the superpara-
magnetic limit. Phys Rev B 74(5):52, 403
Jiang M, Terra J, Rossi AM, Morales MA, Saitovitch EMB,
Ellis DE (2002) Fe2?/Fe3? substitution in hydroxyapatite:
theory and experiment. Phys Rev B 66(22):107, 224
Kastanas GN, Koel BE (1993) Interaction of Cl2 with the
Au(111) surface in the temperature range of 120 to
1000 K. Appl Surf Sci 64:235–249
123
1318
Luo W, Pennycook SJ, Pantelides ST (2007) s-Electron fer-
romagnetism in gold and silver nanoclusters. Nano Lett
7(10):3134–3137
Menard LD, Gao SP, Xu H, Twesten RD, Harper AS, Song Y,
Wang G, Douglas AD, Yang JC, Frenkel AI, Nuzzo RG,
Murray RW (2006a) Sub-nanometer Au monolayer-pro-
tected clusters exhibiting molecule-like electronic
behavior: quantitative high-angle annular dark-field
scanning transmission electron microscopy and electro-
chemical characterization of clusters with precise atomic
stoichiometry. J Phys Chem B 110(26):12874–12883
Menard LD, Xu H, Gao SP, Twesten RD, Harper AS, Song Y,
Wang G, Douglas AD, Yang JC, Frenkel AI, Murray RW,
Nuzzo RG (2006b) Metal core bonding motifs of mono-
disperse icosahedral Au13 and larger Au monolayer-
protected clusters as revealed by X-ray absorption spec-
troscopy and transmission electron microscopy. J Phys
Chem B 110(30):14564–14573
Michael F, Gonzalez C, Mujica V, Marquez M, Ratner MA
(2007) Size dependence of ferromagnetism in gold nano-
particles: mean field results. Phys Rev B 76(22):224, 409
Mingos DMP (1976) Molecular-orbital calculations on cluster
compounds of gold. J Chem Soc Dalton Trans (13):1163–
1169
Mingos DMP (1996) Gold—a flexible friend in cluster chem-
istry. J Chem Soc Dalton Trans (5):561–566
Müllegger S, Hänel K, Strunskus T, Wöll C, Winkler A (2006)
Organic molecular beam deposition of oligophenyls on
Au(111): a study by X-ray absorption spectroscopy. Chem
Phys Chem 7:2552–2558
Narayanaswamy D, Marks LD (1993) Transformation in quasi-
melting. Z Phys D 26:S70–S72
Negishi Y, Tsunoyama H, Suzuki M, Kawamura N, Matsushita
MM, Maruyama K, Sugawara T, Yokoyama T, Tsukuda T
(2006) X-ray magnetic circular dichroism of size-selected,
thiolated gold clusters. J Am Chem Soc 128(37):12034–
12035
Nunokawa K, Onaka S, Ito M, Horibe M, Yonezawa T,
Nishihara H, Ozeki T, Chiba H, Watase S, Nakamoto M
(2006) Synthesis, single crystal X-ray analysis, and TEM
for a single-size Au11 cluster stabilized by SR ligands: the
interface between molecules and particles. J Organomet
Chem 691:638–642
Periyasamy G, Remacle F (2009) Ligand and solvation effects
on the electronic properties of Au55 clusters: a density
functional theory study. Nano Lett 9(8):3007–3011
Sampedro B, Crespo P, Hernando A, Litrán R, Sánchez-López
JC, López-Cartes C, Fernández A, Ramı́rez J, Calbet JG,
Vallet M (2003) Ferromagnetism in fcc twinned 2.4 nm
size Pd nanoparticles. Phys Rev Lett 91(23):203, 237
Schmid G (2008) The relevance of shape and size of Au55
clusters. Chem Soc Rev 37:1909–1930
Schmid G, Pfeil R, Boese R, Bandermann F, Meyer S, Calis
GHM, Vandervelden WA (1981) Au55[P(C6H5)3]12Cl6—a
gold cluster of an exceptional size. Chem Ber 114:3634–
3642
Shinohara T, Sato T, Taniyama T (2003) Surface ferromag-
netism of Pd fine particles. Phys Rev Lett 91(19):197, 201
Song Y, Huang T, Murray RW (2003) Heterophase ligand
exchange and metal transfer between monolayer protected
clusters. J Am Chem Soc 125(38):11694–11701
123
J Nanopart Res (2010) 12:1307–1318
Steiner UB, Neuenschwander P, Caseri WR, Suter UW, Stucki
F (1992) Adsorption of NPh3, PPh3, AsPh3, SbPh3 and
BiPh3 on gold and copper. Langmuir 8(1):90–94
Suber L, Fiorani D, Scavia G, Imperatori P, Plunkett WR
(2007) Permanent magnetism in dithiol-capped silver
nanoparticles. Chem Mater 19(6):1509–1517
Sun S, Murray CB, Weller D, Folks L, Moser A (2000)
Monodisperse FePt nanoparticles and ferromagnetic FePt
nanocrystal superlattices. Science 287:1989–1992
Teo BK, Shi XB, Zhang H (1992) Pure gold cluster of 1:9:9:1:9:9:1
layered structure: a novel 39-metal-atom cluster [(Ph3P)14
Au39Cl6]Cl2 with an interstitial gold atom in a hexagonal
antiprismatic cage. J Am Chem Soc 114(7):2743–2745
Tronconi AL, Morais PC, Pelegrini F, Tourinho FA (1993)
Electron paramagnetic resonance study of ionic water-
based manganese ferrite ferrofluids. J Magn Magn Mater
122:90–92
Turner M, Golovko VB, Vaughan OPH, Abdulkin P, Beren-
guer-Murcia A, Tikhov MS, Johnson BFG, Lambert RM
(2008) Selective oxidation with dioxygen by gold nano-
particles catalysts derived from 55-atom clusters. Nature
454:981–U31
Valden M, Lai X, Goodman DW (1998) Onset of catalytic
activity of gold clusters on titania with the appearance of
nonmetallic properties. Science 281(5383):1647–1650
Walter M, Akola J, Lopez-Acevedo O, Jadzinsky PD, Calero
G, Ackerson CJ, Whetten RL, Grönbeck H, Häkkinen H
(2008) A unified view of ligand-protected gold clusters as
superatom complexes. Proc Natl Acad Sci USA 105(51):
9157–9162
Weare WW, Reed SM, Warner MG, Hutchison JE (2000)
Improved synthesis of small (dcore & 1.5 nm) phosphine-
stabilized gold nanoparticles. J Am Chem Soc 122(51):
12890–12891
Whetten RL, Price RC (2007) Nano-golden order. Science
318:407–408
Whetten RL, Khoury JT, Alvarez MM, Murthy S, Vezmar I,
Wang ZL, Stephens PW, Cleveland CL, Luedtke WD,
Landman U (1996) Nanocrystal gold molecules. Adv Mat
8(5):428
Wilcoxon JP, Martin JE, Provencio P (2000) Size distributions
of gold nanoclusters studied by liquid chromatography.
Langmuir 16(25):9912–9920
Woehrle GH, Hutchison JE (2005) Thiol-functionalized un-
decagold clusters by ligand exchange: synthesis, mecha-
nism, and properties. Inorg Chem 44(18):6149–6158
Yamamoto Y, Miura T, Suzuki M, Kawamura N, Miyagawa H,
Nakamura T, Kobayashi K, Teranishi T, Hori H (2004)
Direct observation of ferromagnetic spin polarization in
gold nanoparticles. Phys Rev Lett 93(11):116, 801
Yang Y, Chen S (2003) Surface manipulation of the electronic
energy of subnanometer-sized gold clusters: an electro-
chemical and spectroscopic investigation. Nano Lett
3(1):75–79
Yu M, Bovet N, Satterley CJ, Bengió S, Lovelock KRJ, Milligan
PK, Jones RG, Woodruff DP, Dhanak V (2006) True nature
of an archetypal self-assembly system: mobile Au-thiolate
species on Au(111). Phys Rev Lett 97(16):102, 166
Zhang P, Sham TK (2002) Tuning the electronic behavior of
Au nanoparticles with capping molecules. Appl Phys Lett
81(4):736–738