10892 Langmuir 2003, 19, 10892-10900

Electrochemical Reduction of Iron Oxide Nanoparticles on

Mercury
E. Dubois* and J. Chevalet†
Laboratoire LI2C, UMR CNRS 7612, Université Pierre et Marie Curie, case 51,
4 Place Jussieu, 75252 Paris Cedex 05, France

Received July 17, 2003. In Final Form: September 19, 2003

The electrochemical reduction of well-defined γ-Fe2O3 nanoparticles dispersed and stabilized in aqueous
media which are appropriate for electrochemical techniques (analysis and preparation) is studied. A medium,
in which metallic iron can be formed and in which iron ions do not precipitate, is examined in detail
([citrate(NH4)3] ) 0.01 mol/L, NH4ClO4/NH3 pH ) 8.6). Our results clearly show that the differences
between nanoparticles and ionic species as starting material do not consist in a simple scaling. At intermediate
potentials (-0.4 to -1.2 V/ECS) we show that the nanoparticles can be reduced into FeII, with a fraction
of it being free in solution. At more negative potentials (-1.5 to -1.8 V/ECS), nanoparticles are reduced
into metallic iron nanoparticles that enter the Hg electrode due to the wetting properties of Hg on iron.
There exist strong interactions between the oxide nanoparticles and the electrode, interactions which are
shown to be partly of electrostatic nature, that is, depending on the concentration of the supporting electrolyte
(NH4ClO4/NH3).

I. Introduction
The final goal of the present investigation is the
preparation of a dispersion of metallic iron in mercury,
that is, a magnetic conductive liquid. We study here the
electrochemical reduction of nanometric particles (around
10000 atoms) dispersed in water, that is, a colloidal
dispersion. Such dispersions of metallic iron in mercury
have been obtained from the reduction of ionic iron on a
mercury cathod.1-3 Nevertheless, this method raises
several difficulties during the synthesis: production of
H2 and no control of the size of the iron aggregates formed.
The final material shows a low stability that limits its
applications. Consequently, the reduction of iron oxide
nanoparticles instead of ionic iron appears as an advantage
because nanoparticles could constitute nanoreactors,
allowing a better control of the material prepared. Such
a procedure corresponds to nonclassical electrochemistry,
and few works deal with big objects. Several articles by
Heyrovsky4-7 deal with nanoparticles, however, smaller
than ours, studied with an analytical purpose and never
totally reduced. Nevertheless, we shall compare our results
to these works in the Discussion and show that there are
indeed some similarities between both types of systems.
The reoxidation of nanoparticles formed in situ by
reduction has also been reported.8 Moreover, the interac-
tion colloids/mercury interface has been studied in several
cases, although in the absence of an electrochemical
* To whom correspondence should be addressed. E-mail:
emdubois@ccr.jussieu.fr.
† E-mail: chevalet@ext.jussieu.fr.
(1) Luborsky, F. E. J. Electrochem. Soc. 1961, 108 (12), 1138.
(2) Windle, P. L.; Popplewell, J.; Charles, S. W. IEEE Trans. Magn.
1975, Mag-11 (5), 1367.
(3) Alekseev, V. A.; Veprik, I. Y.; Minukov, S. G.; Fedonenko, A. I.
J. Magn. Magn. Mater. 1990, 85, 133.
(4) Heyrovsky, M.; Jirkovsky, J. Langmuir 1995, 11, 4288.
(5) Heyrovsky, M.; Jirkovsky, J.; Struplova-Bartackova, M. Langmuir
1995, 11, 4309.
(6) Heyrovsky, M.; Jirkovsky, J.; Struplova-Bartackova, M. Langmuir
1995, 11, 4300.
(7) Heyrovsky, M.; Jirkovsky, J. Langmuir 1995, 11, 4293.
(8) Zutic, V.; Nicolas, E.; Gérard, P.; Gierst, L. Eletroanal. Chem.
Interfacial Electrochem. 1973, 44, 107.
reaction: (i) with gold nanoparticles;9 (ii) with big objects
such as living cells.10
The reduction of nanoparticles in view of preparative
electrolysis raises several questions: (i) is it possible to
reduce such nanoparticles; (ii) are they totally reduced;
(iii) do they enter the mercury electrode? A first study
gave a positive answer to these three questions.11 However,
many difficulties appeared and were not taken care of:
(i) the variation of intensity versus the time during
electrolysis is erratic; (ii) the yield in current is bad; (iii)
the stability of the material produced is not good enough.
The resulting question is thus: do we really take
advantage of the nanoreactor given the used procedure?
We propose here to improve the understanding of the
synthesis step and, in particular, the parameters and
phenomena that govern the reduction of nanoparticles.
II. Experiments
1. Colloidal Solution. The colloidal solutions, which con-
stitute the electroactive species for the whole set of experiments,
are composed of nanoparticles of γ-Fe2O3, that is, maghemite,
which is a magnetic iron oxide. These nanoparticles were
chemically synthesized.12-14 Due to the acidic properties of surface
groups, the nanoparticle surface can be positively or negatively
charged depending on the pH (see Figure 1). In acidic medium,
the surface charge is positive whereas it is negative in alkaline
medium, and the point of zero charge of the surface is around
a pH value of 7.5. Consequently, there exist electrostatic repulsion
forces between particles that allow stabilizing the suspensions
in water if the conditions are appropriate. Nevertheless, such
bare particles are not stable for 3.5 < pH < 10.5 because the
density of the surface charge is too low. One solution for obtaining
a stable dispersion is the coating of the surface with small citrate
(9) Caselli, M.; Lippolis, G.; Gierst, L. J. Electroanal. Chem. 1972,
38, 451.
(10) Svetlicic, V.; Ivosevic, N.; Kovac, S.; Zutic, V. Langmuir 2000,
16 (21), 8217.
(11) Dubois, E.; Chevalet, J.; Massart, R. J. Mol. Liq. 1999, 83, 243.
(12) Massart, R. IEEE Trans. Magn. 1981, Mag-17, 1247.
(13) Massart, R.; Roger, J.; Cabuil, V. Braz. J. Phys. 1995, 25 (2),
135.
(14) Magnetic Fluids and Applications Handbook; Berkovskii, U. B.
M., Bashtovoi, V. G., Eds.; Begell House: 1996.

10.1021/la035294r CCC: $25.00 © 2003 American Chemical Society

Published on Web 11/18/2003
Reduction of Iron Oxide Nanoparticles on Mercury
Figure 1. Characteristics of the different kinds of stable
colloidal solutions: nature of the charge, choice of counterions,
range of stability, nature of the supporting electrolyte, and
maximum concentration of electrolyte compatible with the
colloidal stability (a range is given because cmax depends on the
size of the nanoparticles, with the smaller being stable for higher
electrolyte concentrations). Note that, in the presence of a ligand
L, the surface charge is only ensured by the ligand until pH 7
and is ensured both by the oxide sites and the ligand for higher
pH, with the proportion of both sites depending on the pH.
molecules in order to shift the point of zero charge.15 Under such
conditions, the charge is negative for pH > 3.5. However, the
value of the charge is not the only parameter controlling the
colloidal stability: the nature of the counterions is crucial. The
different kinds of surface/counterions associations are sum-
marized in Figure 1: it is clear that stabilized dispersions can
be obtained for any pH between 0.5 and 13.5. However, such
conditions are not sufficient to determine a convenient composi-
tion of the supporting electrolyte used to perform electroreduction
of the nanoparticles, and it will be shown that a specific study
is necessary to determine the appropriate conditions (see section
III.1 and the caption of Figure 1).
These colloidal dispersions are characterized by several other
parameters such as the size of the nanoparticles and the volume
fraction Φ of solid. The size of the nanoparticles is determined
from the magnetic properties of the solutions. Each particle is
indeed a monocrystal and, due to the nanometric size, a magnetic
monodomain, with a dipolar moment µ proportional to the volume
of the particle.14,16 The size distribution is described by a log-
normal distribution characterized by the mean diameter d0 and
the polydispersity σ. The volume fraction Φ of the particles is
determined from a chemical titration of iron:17 Φ (%) ) 1.577[Fe]
(mol/L). The stock solution used in the present study is
characterized by d0 ) 6.65 nm, σ ) 0.22, and Φ ) 3.4%.
2. Electrolytes and Ionic Species. All chemical used are
pure analytical grade reactants except TMACl and TMAOH,
which are 97% pure only (Aldrich). The solutions of ionic iron are
prepared from FeSO4(NH4)2SO4 for divalent ionic iron and Fe2-
(SO4)3 for trivalent ionic iron.
3. Voltammetric Experiments. Depending on the purpose
of the experiments, that is, preparative or analytic, two different
geometries for the electrode are used. For preparative experiments,
a mercury pool was used, the surface of which is around 1 cm2.
It is associated with a calomel reference electrode and with a
platinum counter electrode. The potentiostat is a Radiometer
PGP201. For analytical purposes, the use of a static mercury
drop electrode or of a dropping mercury electrode is required. A
303A system from PAR EGG is used in conjunction with an Ag/
AgCl reference electrode and a Pt counter electrode connected
to a potentiostat, model EGG 273.
The main differences between these two cells are the huge
difference of surface between both mercury electrodes and the
facility to renew the electrode. Indeed, the mercury drop electrode
allows changing the surface very easily: more electrochemical
(15) Dubois, E.; Boué, F.; Cabuil, V.; Perzynski, R. J. Chem. Phys.
1999, 111 (15), 7147.
(16) Bacri, J. C.; Boué, F.; Cabuil, V.; Perzynski, R. J. Magn. Magn.
Mater. 1986, 62, 36.
(17) Charlot, G. Les méthodes de la chimie analytique; Masson et
Cie: Paris, 1966.
Langmuir, Vol. 19, No. 26, 2003 10893
techniques can be applied as compared to the case of the mercury
pool electrode; then the influence of the aging of the surface is
easily studied. Conversely, the mercury pool cannot be changed
easily, and essentially cyclic voltammetry or simple electrolysis
is performed onto it. Figures 2-6 show results obtained with the
mercury drop electrode. Potentials on the abscissa are referred
to a Ag/AgCl/KClsat reference.
4. Magnetic Properties. The magnetization measurements
are performed in order to detect the formation of metallic iron
and to quantify this formation. An extraction method is used:
a high magnetic field (10 000 Oe ) 180 kA‚m-1) is applied on the
sample placed in a cylinder surrounded by the detection coil.
The flux variation while extracting the tube from the field is
then measured. This simple basic extraction method is needed
in our case in order to avoid erroneous results due to inhomo-
geneous samples.
III. Determination of Appropriate Experimental
Conditions
The reduction of nanoparticles stabilized as colloidal
dispersions raises several questions linked to their
nonclassical electrochemical behavior as compared to that
of pure ionic solutions. First of all, before performing an
electrochemical reaction, the conditions have to be chosen
in order to keep their colloidal stability. The second point
is linked to the possibility of a reduction up to metallic
iron formation under the previously fixed conditions, a
question which raises the problem of the possible total or
partial reduction of such big objects as nanoparticles. The
third point is related to the specific behavior of these
nanoparticles, which have to be compared with ionic iron.
This comparison implies additional constraints on the
composition of the solutions studied due to the complex
chemistry of iron. To solve these different problems, the
crucial points are the nature of the ions and their
concentrations in the solutions. Both have to be determined
in order to define appropriate conditions for studying the
electroreduction of nanoparticles.
1. Stability of the Colloidal Suspension. The first
point is the need for maintaining the stability of the
colloidal solution, that is, of the electroactive species,
despite both the nature and concentration of the sup-
porting electrolyte and the nature of the new species
produced during the electrolysis. Indeed, the salt con-
centration brought by the colloidal solution is too low to
ensure a convenient electrical conduction. Moreover, the
stability depends on several parameters, as pointed out
in section II. (i) It depends on the value of the salinity:
if it is too high, the electrostatic repulsion is screened and
particles aggregate.18 (ii) It depends on the nature of the
added ions: multivalent ions or small monovalent ions
which have a charge opposite to that of the nanoparticles
must be avoided, as well as ions that tend to strongly
adsorb on the surface.19 (iii) As indicated previously, it
depends on the pH: this parameter can shift during the
electrolysis and can be strongly affected near the electrode.
(iv) It also depends on the size of the nanoparticles: smaller
particles are more stable (see the last column in and
caption of Figure 1).
Therefore, we worked with small γ-Fe2O3 particles of
low polydispersity. The stability is checked at several
scales. Macroscopically, a stable dispersion is black
whereas a destabilized dispersion becomes brown and
readily separates in two phases. At a microscopic scale,
optical microscopy allows detecting aggregates of a few
microns. Note that light scattering cannot be used for our
(18) Israelachvili, J. N. Intermolecular and Surface Forces, 2nd ed.;
Academic Press: New York, 1991.
(19) Jolivet, J. P. De la solution à l′oxyde; Actuels, S., Ed.; CNRS
Editions: Paris, 1994.
10894 Langmuir, Vol. 19, No. 26, 2003 Dubois and Chevalet

Table 1. Summary of the Results of Preparative Experiments Performed To Select the Conditions for Metallic Iron
Formation on the Mercury Cathodea
surface charge counterions supporting electrolyte Fe0 E0(Fe)/EECS, V
+ ClO4 - HClO4 pH ) 1.4 no
- TMA+ TMAOH 0.2 M yes -1.85
- Na+ Na3cit 0.15 M no
- Na+ TMACl 0.2 M yes -1.8
- TMA+ TMA3cit 0.15 M no
- TMA+ TMA2oxalate 0.2 M yes -1.8
- TMA+ TMACl 0.2 M yes -1.65
- NH4+ NH4cit/NH3 pH ) 8.6 yes -1.65
- TMA+ TMAH2PO4/(TMA)2HPO4 pH ) 7.1 yes -1.7
- TMA+ TMAHPO4/(TMA)3PO4 pH ) 11.1 no
- TMA+ TMAborate pH ) 9 no
a The three columns on the left indicate the experimental conditions used to maintain stability while performing electrolysis. The two

columns on the right give the result of nanoparticles’ reduction and the approximate reduction potential where metallic iron is obtained.

solutions due to the strong absorption of the particles in

the visible range (except for very dilute dispersions). The
presence of very small aggregates of several particles is
checked by a technique taking advantage of the specific
properties (magnetic and optical) of our magnetic oxide
nanoparticles: the measurement of the relaxation of the
birefringence induced by a low field perturbation.20,21 With
the relaxation determined being proportional to the
volume of the rotating nanobject, aggregates are very well
detected.
The pairs electrolyte/particles that fulfill the stability
condition have thus been determined and are given in
Table 1.
2. Conditions of Metallic Iron Formation. The
second point in the setting of experimental conditions is
to test whether the electrochemical reduction can be fully
achieved, that is, oxide nanoparticles converted into Figure 2. Influence of citrate concentration on the cyclic
metallic iron. This has been done using preparative voltammograms of Fe3+. The electrolyte is 0.15 mol/L NH4-
experiments for a large variety of electrolytes over the ClO4, and [citrate(NH4)3] varies from 0 to 0.1 mol/L (see legend).
whole range of pH where these different electrolytes were [iron] ) 6 × 10-4 mol/L, twaiting ) 3 s (equilibration time at the
initial potential, here always the most positive potential), and
tested both with ionic iron (when possible) and nanopar- scan rate ) 50 mV/s.
ticles. At the end of the preparative electrolysis, magne-
tization measurements are performed on the mercury
electrochemical behavior of nanoparticles. Indeed, the
electrode, to conclude if metallic iron was formed. The
electrochemical response of FeII or FeIII in the electrolyte
trials are summarized in Table 1.
tested must be at least characterized for identifying the
After these preliminary steps, four main conclusions
different steps occurring over the entire potential range.
arose: (i) nanoparticles can be reduced into metallic iron;
Such a comparison of iron ions and oxide nanoparticles
(ii) the reduction into metallic iron always occurs for
is, however, difficult due to the precipitation of iron species
potentials more negative than -1.5 V/ECS, thus close to
when their concentration is not low enough and/or when
the border of the domain of electroactivity; (iii) metallic
the pH is not sufficiently acidic. Nevertheless, at low
iron can be formed only if pH > 7; (iv) when alkali metal
enough iron ions concentrations, it is possible to study
ions are present, they are simultaneously reduced with
experimentally ionic iron for pH < 7 without precipitation
the iron. Metallic sodium, potassium, and so forth are
during a short period of time (around 1 h), as observed
dissolved into mercury, thus modifying the resulting
during the set of experiments, for example in NH4ClO4.
material. Moreover, these alkali metals provoke a strong
Under these conditions, FeIII is reduced into FeII at
pH increase: this leads in all cases to the destabilization
potentials more positive than the border of the domain of
of the colloidal suspensions because these cations are not
electroactivity, and the couple FeIII/FeII is thus not
convenient counterions for negatively charged particles
observable (see Figure 2). Only the reduction into metallic
in alkaline medium. In a first series of experiments, we
iron occurs. Furthermore, electrolytes such as NH4ClO4
tried to avoid such a modification to simplify the problem.
do not stabilize the colloidal dispersions over a long period
Consequently, to produce metallic iron from the reduc- of time, since the citrate molecules adsorbed on the surface
tion of dispersed oxide nanoparticles, these dispersions desorb with time. If the citrate concentration in the
should be prepared in the absence of alkali metal solution is zero, the surface charge, that is, the repulsion
counterions and are constituted of citrate surface coated force between particles, becomes too low, and the particles
particles. then tend to precipitate. Solving this problem is possible
3. Comparison between Ionic Iron and Nanopar- by adding citrate in the supporting electrolyte: maghemite
ticles as Starting Material. Such a comparison is nanoparticles remain stable, and iron ions are complexed
important for the understanding of the nonconventional so that they do not precipitate. Moreover, the potential of
FeIII/FeII is shifted inside the domain of electroactivity.
(20) Wilhelm, C.; Gazeau, F.; Roger, J.; Pons, J. N.; Salis, M. F.; Nevertheless, the reduction into metallic iron is also
Perzynski, R.; Bacri, J. C. Phys. Rev. E 2002, 65, 031404.
(21) Bacri, J. C.; Perzynski, R.; Salin, D.; Servais, J. J. Phys. 1987, shifted toward the border of this domain while increasing
48, 1385. citrate concentration (Figure 2). This implies choosing a
Reduction of Iron Oxide Nanoparticles on Mercury
Figure 3. Comparison of the behavior of ions and nanoparticles. Cyclic voltammograms in [NH4ClO4/NH3 pH ) 8.6] ) 0.015 mol/L,
[citrate(NH4)3] ) 0.01 mol/L. [iron] ) 6 × 10-4 mol/L in all experiments (Fe2+, Fe3+, nanoparticles), twaiting ) 3 s, and scan rate )
50 mV/s. Part b is the expansion of the nanoparticles’ signal plotted in part a.
citrate concentration as a compromise between its effect
on the colloidal stability and its influence on the value of
the potential of reduction into metallic iron. We have used
[citrate] ) 0.01 mol/L in the experiments that follow.
4. Electroanalytical Methods: Preliminary Stud-
ies. Using the different electrolytes that fulfill the
constraints described above, several electrochemical meth-
ods were applied at the mercury drop electrode in order
to explore the behavior of ions and nanoparticles. The
objective was to separate the specific contribution due to
one type of electrolyte or to one ion from the general trends.
Some general characteristics that are specific to the
experiments with nanoparticles and that do not depend
on the nature of the electrolyte can be extracted from our
results. Experiments using cyclic voltammetry (CV) show
that the intensities obtained depend (i) on the initial
potential of the cycle, (ii) on the region of potential explored,
(iii) on the waiting time (twaiting) at the initial potential,
and (iv) on the values of potentials when the scan is blocked
inside a cycle. These results mean that the intensities
depend on the story of the mercury drop surface. Pulsed
techniques that allow jumping above a potential region
are thus interesting for the study of the present system.
Normal pulse polarography (NPP) indeed shows that the
intensity depends (i) on the initial potential, (ii) on the
time spent at this potential, and (iii) on the story of the
surface, because the intensities are totally modified if the
drop is hanging instead of dropping (as in classical NPP).
As no reoxidation of metallic iron is observed from CV
experiments, reverse pulse polarography (RPP) was used
in order to explore shorter times. No clear reoxidation of
metallic iron was observed. Finally, chronoamperometric
(CA) experiments show that the intensity after a potential
step toward the potential of reduction into metallic iron
deviates from the classical 1/t0.5 Cottrell’s law.
As all these experiments have shown qualitatively
similar behaviors in the various electrolyte solutions
employed, further experiments were performed on the
electrolyte that appears as the best compromise given all
the constraints: NH4ClO4/NH3 pH ) 8.6, [citrate(NH4)3]
) 0.01 mol/L, with a ferrofluid coated with citrate and
with NH4+ counterions. Indeed, the anion ClO4- is a low
complexing agent, citrate ions complex iron and stabilize
particles, and the solution is a buffer. Moreover, ions and
nanoparticles can be directly compared in the same
electrochemical environment.
Langmuir, Vol. 19, No. 26, 2003 10895
Figure 4. Cyclic voltammograms of nanoparticles in 0.15 mol/L
NH4citrate/NH3 pH ) 8.6. [iron] ) 4 × 10-3 mol/L, twaiting ) 3
s, and scan rate ) 50 mV/s. Dashed lines correspond to two
cycles with nanoparticles without o-PNT. Plain lines correspond
to two cycles in the same solution after addition of o-PNT.
IV. Studies of the Electrolyte NH4ClO4/NH3 pH )
8.6, [citrate(NH4)3] ) 0.01 mol/L
In the previous paragraph, the best conditions have
been defined. It is now possible to compare iron ions’ and
iron oxide nanoparticles’ reduction in order to determine
the parameters that govern the electrochemical behaviors.
The voltammograms obtained with FeII, FeIII, and nano-
particles in this electrolyte are plotted in Figure 3. The
first obvious observation is that nanoparticles produce a
much lower intensity than ions. In all cases, metallic iron
is obtained around -1.5 V and there is no clear reoxidation
peak. Nevertheless, for nanoparticles, there appears some
current in the region -0.3 to -1.2 V, which may correspond
to their partial reduction. This last point is the first striking
difference between ions and nanoparticles. Such an
original electrochemical behavior for nanoparticles can
be expected given the complex chemistry of iron (com-
plexation, formation of hydroxides and oxides, equilibrium
between multiple species); nevertheless, it does not appear
with ionic iron.
1. Intermediate Reduction States of the Nano-
particles. Figure 4 shows two cycles in the region [0 to
-1 V]. The reoxidation peak noted A in Figure 4 (around
-0.25 V) appears only if the species reduced for potentials
in the region -0.4 to -0.8 V are formed. In the second
cycle appears a new reduction peak noted B in Figure 4
(around -0.3 V). Given the results for ions shown in Figure
10896 Langmuir, Vol. 19, No. 26, 2003
3, one may suppose that these two peaks correspond to
the couple FeIII/FeII. The potentials are not identical in
Figures 3 and 4, however, as they vary with the intensity
of the complexation (see Figure 2) and they depend on the
ratio [ionic iron]/[citrate] that probably differs in the two
experiments. This result means that the nanoparticles
can be broken into FeII ions.
This hypothesis is supported by other results. First,
when adding a specific complexing agent of FeII ions, such
as o-phenanthroline (o-PNT), the cyclic voltammogram is
modified, as shown in Figure 4: the oxidation peak A no
longer exists, and neither does the reduction peak B
observed previously in the second cycle. This is consistent
with the previous result: due to the complexation by
o-PNT, FeII cannot be oxidized into FeIII; thus, the further
reduction of FeIII (during the second cycle) is likewise not
observed. Macroscopically, this complex FeII/o-PNT can
be detected due to its very intense red color, allowing the
direct detection of FeII formed close to the mercury drop.
This is clearly observed with ionic FeIII. The same
experiment with nanoparticles is more difficult to perform,
due to the strong optical absorption of the particles around
the same wavelengths as those for the complex. Never-
theless, this colored complex can be observed during
nanoparticles’ reduction using a special cell of low thick-
ness, an optimized nanoparticles’ concentration, and a
simple optical device to magnify the mercury drop. The
puzzling point is the lower quantity of FeII formed from
nanoparticles than the amount formed from ferrous ions,
as deduced from the difference of the color intensity around
the mercury drop, although the global concentration of
iron is identical: It is obvious that much less FeII is formed
while nanoparticles are reduced. Several reasons can be
suggested: (i) The quantity of reduced material is much
lower than that for ions, as shown in Figure 3 (lower
intensity for nanoparticles). (ii) Only a part of the ions are
free ions, and some may precipitate: the local pH and
concentration values near the drop may be different from
the bulk conditions. (iii) The species produced may block
the surface of mercury. (iv) The FeII ions may interact
with negative particles and not with o-PNT.
Another evidence of the partial reduction of nanopar-
ticles can be deduced from another set of experiments: if
a preparative electrolysis is made with the same system
on a mercury pool, running a cyclic voltammetry after
several hours of preparation shows the presence of FeII
ions in the solution.
Thus, from all this set of experiments, one can conclude
that free FeII ions are obtained under the present
conditions after a partial reduction of the nanoparticles
in the region -0.4 to -1.2 V. Consequently, the particles
are partly broken: such a situation has to be avoided
during preparative experiments; otherwise, the benefit of
the nanoreactor is lost. Indeed, several species are present
in the solution: nanoparticles, ferrous iron, and/or pre-
cipitated oxides or hydroxides (depending on the pH and
on ionic strength conditions), the characteristics of which
are different from those of the initial nanoparticles and
obviously not controlled. In such a case, a nanoparticle
cannot be considered as a nanoreactor.
2. Interactions Particles/Mercury, celec ) 0.015 mol/
L. The preliminary experiments described in section III.4
have clearly shown from the influence of twaiting that
accumulation of electroactive material takes place on the
Hg electrode under specific conditions. It is important to
precisely analyze these phenomena in order to understand
their origin.
All experiments with nanoparticles show an influence
of the initial waiting time twaiting spent at the initial
Dubois and Chevalet
Figure 5. Influence of twaiting on chronoamperometry experi-
ments on nanoparticles. The conditions are [NH4ClO4/NH3 pH
) 8.6] ) 0.5 mol/L, [citrate(NH4)3] ) 0.01 mol/L, and [iron] )
6 × 10-4 mol/L. A potential step from Einitial ) 0 V toward Efinal
) -1.5 V is performed at t ) 0. The curve of the electrolyte is
given as a reference. The inset plots the intensity versus t-0.5
for the data obtained with nanoparticles.
potential. For instance, in Figure 3, for nanoparticles,
twaiting is 3 s (at 0 V). If this time is increased, the intensities
over the whole cycle are higher. The same behavior is
observed while performing NPP between 0 and -1.7 V:
the intensity depends on the time spent at 0 V. This can
also be clearly seen in Figure 5 using chronoamperom-
etry: the intensity during the first second after the
potential step is much higher if twaiting ) 120 s instead of
3 s, although it tends toward a limit independent of twaiting
for long times. The inset of Figure 5 plots the intensity as
a function of t-0.5: the decrease of the intensity clearly
deviates from the classical t-0.5 law, showing that the
process is not purely diffusional. These different experi-
ments prove that there is accumulation of electroactive
material on the electrode at least at 0 V (applied potential).
After this first set of experiments, and in order to
compare ions’ and nanoparticles’ behavior versus ac-
cumulation effects, another type of study has been
performed. The idea was to check whether this type of
interaction was weak or sufficiently strong to resist a
complete removal of the electrolyte. With ionic iron (6 ×
10-4 mol/L in the mentioned electrolyte), after twaiting )
120 s under Einitial ) 0 V, if the solution is carefully replaced
by pure deaerated electrolyte (the mercury drop always
remaining under Einitial), a voltammogram shows the
normal feature of a pure electrolyte response. Conversely,
performing the same replacement of electrolyte with
nanoparticles (6 × 10-4 mol/L in the indicated electrolyte),
the first cycle of the set of voltammograms shows the same
type of signal as does the solution of nanoparticles during
a first cycle. The same situation occurs if chronoamper-
ometry is applied in place of cyclic voltammetry.
After such a treatment, the second cyclic voltammogram
(or the second chronoamperogram) shows a peculiar shape.
(i) There is no detectable singularity corresponding to the
original cyclic voltammetry with nanoparticles. However,
this shape is different from that of the cyclic voltammo-
gram of the electrolyte alone; a higher intensity is observed
and the limit of the domain for negative potentials is
shifted, by approximately +100 mV. (ii) If nanoparticles
are introduced again in the cell, the next voltammogram
performed on the same drop produces a modified signal
as compared to the reference signal of Figure 3b. This
indicates a likely evolution of the surface.
The influence of the initial accumulation potential has
been then examined too, and the effects are as follows:
Reduction of Iron Oxide Nanoparticles on Mercury
Figure 6. Influence of the electrolyte concentration on the behavior of nanoparticles. Conditions: [NH4ClO4/NH3 pH ) 8.6] ) celec,
[citrate(NH4)3] ) 0.01 mol/L, [iron] ) 6 × 10-4 mol/L, and twaiting ) 3 s. (a) Cyclic voltammetry: scan rate ) 50 mV/s. celec is given
in the inset. (b) Chronoamperometry: Einitial ) 0 V and Efinal ) -1.5 V. celec is given in the inset.
(i) Under the same experimental protocol (exchange of
electrolyte), and with nanoparticles, after twaiting ) 120 s
at Einitial ) Ewaiting ) -1 V, no signal appears. (ii) When
applying the NPP method on nanoparticles, some ac-
cumulation phenomena due to the time spent at the initial
potential are observed (and are very efficient) only if the
initial potential lies between 0 and -0.5 V. (iii) If cyclic
voltammetry is performed between -1 and -1.7 V only,
no reduction into metallic iron is observed except during
the first cycle, whatever the initial waiting time at -1 V.
Given these results, one can suppose that these ac-
cumulation phenomena are essentially related to elec-
trostatic repulsion between the maghemite nanoparticles
and the mercury electrode, since no real chemical affinity
is expected between maghemite and mercury. However,
high electric local fields could have an influence. For the
nanoparticles, the structural charge is around 2 charges/
nm2, that is, 0.32 C/m2. The surface of mercury is positive
in the range 0 to -0.5 V and negative in the range -0.5
to -1.7 V. Far from the point of zero charge (around -0.5
V), the charge of the mercury is of the same order of
magnitude as that of the nanoparticles. With the con-
centration of 0.015 M in NH4ClO4/NH3, the electrolyte
used here, the Debye length (which is the typical size of
the electric double layer) is around 25 Å, so that the range
of the electrostatic repulsion is high enough to avoid the
particles approaching the electrode.
3. Influence of the Electrolyte Concentration celec.
Given the very likely dependence on electrostatic phe-
nomena deduced from the above experiments, it is
reasonable to examine the influence of the ionic strength,
that is, of the concentration of the supporting electrolyte.
Indeed, contrary to fundamental behaviors of ions in
electrolytes, which obey a diffusional regime except for
very low concentrations (migration), an influence of the
electrolyte concentration on electrochemical behavior
exists for nanoparticles (see Figure 6). Increasing the
electrolyte concentration changes the shape of the cyclic
voltammograms (Figure 6a). The intensities corresponding
to intermediate species increase and change in morphol-
ogy. For the last reduction step (metallic iron), the intensity
is much higher for high salinities. Chronoamperometry
(Figure 6b) also shows that the intensity is much higher
for short times and remains higher even at a very long
time (15 s). Note that, given the time scale, no capacitive
current influence is detectable.
Concerning the accumulation phenomena described in
the previous section, it is clear that they still exist at higher
Langmuir, Vol. 19, No. 26, 2003 10897
salinities, as suggested from cyclic voltammetry, chro-
noamperometry, and experiments where the electrolyte
is removed as described in section IV.2. The surface is
also modified after experiments, as described previously.
The interaction between oxide nanoparticles and mercury
is thus not purely electrostatic.
However, for high electrolyte concentrations, the par-
ticles approach the electrode for potentials in the range
-1 to -1.7 V, contrary to what was observed for celec )
0.015 mol/L (experiments in section IV.2). These results
mean that, due to the high screening of the electrostatic
repulsion if celec is high (the Debye length is around 4 Å
at 0.5 mol/L), the oxide nanoparticles can approach the
electrode by diffusion whatever the potential. The current
measured for E ) -1.5 V is thus due both to adsorption
and to diffusion. However, for low celec values, only adsorbed
nanoparticles contribute to the cathodic current at -1.5
V. The change of behavior when celec increases is in
agreement with an interaction partly of electrostatic
nature between the maghemite nanoparticles and the
mercury interface (repulsion effect at negative potential).
V. Discussion
The study of the electrochemical behavior of nanopar-
ticles formed by about 10 000 atoms raises the question
of the comparison to usual ionic solutions. In the present
case, this comparison is difficult due to the nature of the
element used, iron. The behavior of iron as an ionic species
is complicated from the electrochemical point of view due
to the number of different species that can exist, given the
chemical properties of ferrous and ferric ions.22,23 Com-
plexes are formed with many ions, several aquo/hydroxo
complexes are formed depending on the pH, precipitation
as a solid easily occurs, and intermediate species (poly-
nuclear species or colloidal particles) can also be obtained.
These species may have different electrochemical behav-
iors, and the local conditions near the electrode resulting
from reactions inside the reaction layer can be very
different from the bulk solution conditions, leading to
locally different species and, therefore, increasing the
complexity of the phenomena.
Using nanoparticles of iron oxide such as the species
studied here raises other difficulties. The first difficulty
(22) Baes, C. F.; Mesmer, R. E. The hydrolysis of cations; Wiley-
Interscience: 1976.
(23) Königsberger, L.-C.; Königsberger, E.; May, P. M.; Hefter, G. T.
J. Inorg. Biochem. 2000, 78, 175.
10898 Langmuir, Vol. 19, No. 26, 2003
is the precise characterization of the nanoparticle, which
is the electroactive species. It is necessary to know its
exact composition, the repartition of the components in
the particles if several components are involved, the nature
of the material (crystalline or not), the surface state
(interface oxide/solution), as well as the size distribution
of the particles. In the present case, the nanoparticles
have been studied for many years using numerous
techniques (see, for example, refs 14-16, 20, 21, 24, and
25). The particles are constituted of maghemite γ-Fe2O3,
each particle is a monocrystal, and the size distribution
can be measured by several techniques.14 The surface is
covered with negatively charged citrate molecules that
ensure a density of charge of 2 charges/nm2, and this charge
is compensated with ammonium counterions.15 The second
difficulty is the colloidal stability of such dispersions: they
can be destabilized by the supporting electrolyte before
any reaction. After electrochemical reactions occurred,
the dispersions could be destabilized by the intermediate
species formed, by the local conditions near the electrode
surface, or as a result of the modification of the nano-
particles’ surface (partial dissolution, modification of the
surface charges, and so forth). Such an evolution of the
solution by formation of aggregates can be useful or not,
depending on the aim of the experiment; however, it
complicates the studies, modifying the nature of the
electroactive species during the evolution of the experi-
ments. In the present experiments, the colloidal suspen-
sions are stable before electrochemical reaction and remain
macroscopically stable after electrochemical reactions
(checked as explained in section III.1). Moreover, no
aggregation of the nanoparticles can occur on the electrode
when no reaction occurs (due to interparticle repulsion);
however, one cannot assert that there is no local aggrega-
tion on the electrode after reactions occurred. Note that
if particles are partly dissolved into FeII, for example, we
observe the behavior of ionic species, complicated by the
presence of nanoparticles. Both behaviors are not inde-
pendent: indeed, adsorption of iron ions can occur on the
oxide nanoparticles.
In a first very simple approximation, the difference
between small species such as ions or molecules and big
species such as nanoparticles might be seen as a simple
change of scale. For ions, the intensity of the reduction
current is supposed to be proportional to D[FeIII], with D
being the diffusion coefficient. This can be applied to
nanoparticles, considered as big electroactive species with
a diffusion coefficient D. For FeIII, D ∼ 0.7 × 10-9 m2/s,
although, for the nanoparticles, D is much smaller. A value
of 2.7 × 10-11 m2/s has been measured for nanoparticles
slightly bigger than the particles used in the present
study.25 Here D is estimated as 6 × 10-11 m2/s. However,
one nanoparticle of 6 nm contains around 4300 iron atoms.
Comparing two solutions of the same concentration in
iron, the current should be multiplied by 4300/12 ∼ 370.
This is in contradiction with experiments that show a
smaller current for nanoparticles as compared to ionic
iron. The difference between true and colloidal solutions
as well as the particularities of colloidal solutions have to
be considered.
1. Comparison to Other Particles/Electrode
Systems. Few electrochemistry experiments has been
performed with colloidal particles; however, there exist
very interesting works which are relevant to the present
subject. Some studies deal with electrochemistry in the
(24) Gazeau, F.; Dubois, E.; Bacri, J. C.; Boué, F.; Cebers, A. R. P.
Phys. Rev. E 2002, 65 (3), 031403.
(25) Bacri, J. C.; Cebers, A.; Bourdon, A.; Demouchy, G.; Heegaard,
B. M.; Kashevsky, B.; Perzynski, R. Phys. Rev. E 1995, 52, 3936.
Dubois and Chevalet
sense that there is an electrode and an applied potential
but not always linked to an electronic transfer.9,26,27 Other
studies deal with electron transfer on nanoparticles,
however, not always through an electrode reaction.4-7,28-37
The results are sometimes difficult to interpret within a
unified framework. Nevertheless, a survey of some of these
studies will be made with comparisons with the present
results wherever possible.
First, several studies deal with reactions occurring at
an electrode with particles, that is, with electron transfers.
Pioneering works have been performed on the electro-
chemical reduction behavior of many insoluble substances
(oxides, sulfurs, and so forth).28-32 The polarograms
obtained usually show one or two reduction peaks. A
transfer of electrons from the particles to the mercury
electrode has been observed. The size of the particles seems
to be an important parameter, as well as the interaction
between the particles and the mercury electrode. However,
one of the limitations of these studies is the lack of control
and measurement of the size of the particles and of the
colloidal stability of the solutions studied, parameters that
appear to be crucial. Heyrovsky tried to clarify this field
in a series of four papers introduced by an article in which
he attempted to summarize the questions. The behaviors
of nanoparticles of several semiconductive materials (TiO2,
SnO2, TiO2/Fe2O3) have been studied on electrodes,4-7 in
order to obtain information on their properties, especially
on the reduction processes. The polarographic currents
measured were diffusion-controlled in most cases. The
current corresponds to the exchange of n electrons with
the nanoparticle. This number n depends on the reaction
involved between the electrode and the particle: it can
take place in the bulk of the particle or on its surface. The
number n can also depend on the value of the potential.
Due to the size polydispersity, parameters such as n and
D are also “dispersed”, leading to a spreading of the
signals.
For both the SnO2 and TiO2 nanoparticles studied, the
electroreduction of the surface H+ ions on the particle
surface is first observed. At more negative potentials, a
direct reduction of metal surface ions may occur. For mixed
TiO2/Fe2O3 colloids, the results obtained with pure TiO2
are superimposed with the reduction of FeIII to FeII, which
is an electron transfer with the particle bulk.
The present experiments are consistent with these
results. Indeed, it is observed that an electronic transfer
can occur between the electrode and the nanoparticle. With
γ-Fe2O3, the reduction process reaches the bulk of the
particle and not only the surface, in agreement with the
results on TiO2/Fe2O3 colloids.5
(26) Andrade, E. M.; Molina, F. V.; Gordillo, G. J.; Posadas, D. J.
Colloid Interface Sci. 1994, 165, 459.
(27) Andrade, E. M.; Molina, F. V.; Posadas, D. J. Colloid Interface
Sci. 1994, 165, 450.
(28) Micka, K. Collect. Czech. Chem. Commun. 1956, 21, 647.
(29) Micka, K. Collect. Czech. Chem. Commun. 1957, 22, 1400.
(30) Micka, K. Depolarisation of the dropping mercury electrode by
suspensions of insoluble substances. In Advances in Polarography;
Second International Polarographic Congress, Cambridge 1959; Per-
gamon Press: 1960.
(31) Micka, K. Collect. Czech. Chem. Commun. 1965, 30, 235.
(32) Micka, K.; Kadlec, O. Collect. Czech. Chem. Commun. 1966, 31,
3837.
(33) Heyrovsky, M.; Jirkovsky, J. NATO ASI Ser., Ser. 3: High
Technology (Nanoparticles in Solids and Solutions) 1996, 18, 161.
(34) Heyrovsky, M.; Jirkovsky, J. Scientific Papers of the University
of Pardubice, Series A: Faculty of Chemical Technology 1998, 3, 227.
(35) Mulvaney, P.; Cooper, R.; Grieser, F.; Meisel, D. Langmuir 1988,
4 (5), 1206.
(36) Mulvaney, P.; Swayambunathan, V.; Grieser, F.; Meisel, D. J.
Phys. Chem. 1988, 92, 6732.
(37) Mulvaney, P.; Swayambunathan, V.; Grieser, F.; Meisel, D.
Langmuir 1990, 6, 555.
Reduction of Iron Oxide Nanoparticles on Mercury
Second, several works deal with the behavior of particles
toward electrodes in the absence of electronic transfers. It
is shown that interactions between the particles and the
surface exist and depend on the potential.9,26,27 With the
aim of studying heterocoagulation, that is, the adsorption
of colloidal particles on a planar surface, the behavior of
gold nanoparticles (13-22 nm) on a mercury cathode has
been studied in a large potential range.9 No reaction and
no electron exchange occurs, but the nanoparticles are
incorporated into the mercury, forming an amalgam and
allowing us to study the rate of adsorption on the electrode.
The rate of adsorption is shown to be dependent on the
global interactions between the mercury surface and the
gold nanoparticles. The latter are negatively charged
whereas the charge of mercury depends on the potential.
If the two metals bear charges of the same sign, no
adsorption occurs.
The adsorption of particles on a metallic surface was
also studied with hematite (R-Fe2O3) on silver and
mercury, however, with much bigger particles (1 µm) than
the present ones.26,27 The amount of adsorption is shown
to depend on the pH of the solution and on the electrode
potential. The origin of this adsorption is not purely
electrostatic because adsorption also occurs if the particles
and the electrode have opposite charges, that is, if the
interaction between the particles and the electrode is
repulsive. This is due to the fact that all forces between
both materials contribute to the interaction, and the details
are not fully understood yet.
These results on heterocoagulation are fully consistent
with the present observations: nanoparticles remain on
the surface of the mercury after a complete removal of the
electrolyte (see section IV.2). This means that there is
adsorption of the nanoparticles on the mercury surface,
with the magnitude of the phenomenon depending on ionic
strength and potential and, thus, on the surface charge
and on the structure of the solid/electrolyte interface. The
influence of the potential applied to the mercury electrode
on adsorption indicates that an electrostatic interaction
is probably involved in this adsorption. However, the
existence of adsorption whatever the ionic strength
indicates that the interactions are not purely electrostatic.
In our experiments, the number of electrons exchanged
can be calculated from the first chronoamperometric curve
measured after this first electrolyte removal. If it is
assumed that all these electrons are used to reduce FeIII
into Fe° and that each particle is totally reduced, a
maximum degree of coverage of the surface can be
estimated. With the following typical conditions, [iron] )
1.2 0.10-3 M, [NH4cit/NH3] ) 0.015 M, pH ) 8.6, and twaiting
∼120 s, a coverage around 30% is determined. If a simple
calculation of coverage is assumed considering a random
sequential adsorption, one expects for the fractional
surface coverage θ: θ ) 2cparticulesS(D/π)1/2t0.5, where S is
the particle section, t is the deposition time, and D is the
diffusion coefficient.38 For particles of 7 nm, D ∼ 6, 10-11
m2/s, and [iron] ) 6 × 10-4 mol/L, this formula leads to
a coverage of the surface of 14% for the experiments. This
value is a good order of magnitude as compared to the
experimental value of 30%.
2. Results of the Electron Transfer for Maghemite
Nanoparticles on Mercury. When appropriate condi-
tions are realized to observe an electron transfer (i.e. if
the particles come sufficiently close or even in contact
with the electrode), the main question is then what is the
immediate future of the nanoparticle, that is, how are its
(38) Johnson, C. A.; Lenhoff, A. M. J. Colloid Interface Sci. 1996,
179, 587.
Langmuir, Vol. 19, No. 26, 2003 10899
structure and its chemical composition modified by this
overall reduction step? The present experiments indicate
that metallic iron is formed at potentials negative enough.
However, significant reduction currents are measured in
the intermediate region (-0.3 V/ECS to -1.3 V/ECS). The
injection of electrons in similar nanoparticles (hematite,
R-Fe2O3, and goethite, R-FeOOH) has been studied, not
by the means of an electrode but using radiolytically
produced reducing free radicals.35-37 The metallic iron is
never obtained because the reductive species is not a strong
enough reductant. However, the authors show that three
products are obtained: (i) Fe2+ ions dissolved in the
aqueous phase; (ii) FeII hydroxide adsorbed at the particle
surface; and (iii) FeII trapped within the lattice.
The amount of each species is a function of the pH.
These results are fully consistent with the present
experiments. In the intermediate potential region, only
FeII can be formed and one observes free FeII, visualized
with o-PNT, but part of this FeII is not released in the
solution (see section IV.1). Note that the free FeII is a
quasi-reversible system although the reduction leading
to FeII is not reversible. As the pH is high (8.6), the
adsorption of FeII on the particles is favored but may be
different from that of the experiments of ref 36 because
of the nature of the complexing medium used (with sodium
citrate). It is, however, not possible to differentiate in the
present case surface adsorption from bulk migration.
At far enough negative potentials, metallic iron is formed.
This can be detected from the magnetic properties of iron
(i.e. iron particles in mercury) after the preparative
experiments. The voltammograms show that this reduc-
tion is totally irreversible: there is no reoxidation of iron
(neither on the mercury drop nor on the mercury pool).
After iron preparation, if the surface of the electrode is
cleaned with concentrated acid (HNO3 10% or HCl 30%),
the oxides adsorbed on the mercury surface are removed,
and the electrode looks like pure mercury. Then, although
no species adsorbed on the surface block the reactions, no
iron is reoxidized in alkaline medium. It is interesting to
note that, in the presence of HCl 30% (and without
potential applied), no calomel is formed on the electrode
if iron is present inside. The irreversibility of the reduction
into metallic iron thus does not result from blocking species
that could be adsorbed on the mercury surface. In acidic
medium, that is, in HClO4 or 0.1 M H2SO4, a reoxidation
current can be observed very close to the mercury oxidation
wall. Such an electrolysis on a mercury pool, however,
never allows extracting the whole iron formed.
These rather unexpected results can be explained by
the dependence on the potential of the wetting properties
of iron by mercury. An iron thread in contact with a
mercury pool is not wetted for E > -1 V/ECS but is wetted
for E < -1.5 V/ECS. This has been shown and studied
using iron plates,39 and has been observed with a thread
in the present study. For very negative potentials, the
traces of oxides on the metallic iron plate are removed,
leading to a perfect wetting. Dewetting is then difficult.
In the present case, once iron nanoparticles are formed,
the wetting phenomena very likely favor their dispersion
into the mercury electrode. The reoxidation would imply
dewetting or mercury oxidation at the same time as iron
oxidation; however, to date, no quantitative argument, to
our knowledge, has been proposed to explain such a
process.
From the point of view of preparative experiments,
several results can now explain the difficulties encoun-
(39) Winkler, K.; Krogulec, T.; Baranski, A.; Galus, Z. J. Electroanal.
Chem. 1990, 291, 103.
10900 Langmuir, Vol. 19, No. 26, 2003
tered. Indeed, due to accumulation of nanoparticles under
appropriate conditions, the electrode surface is modified.
Moreover, the mercury surface can be easily polluted by
adsorption of nonreduced particles or parts of particles if
a reduction mechanism implies transient dissolution. The
intermediate species could precipitate locally (the pH may
be different from the pH in the solution, far from the
electrode surface, and iron oxides seem to adsorb easily
on mercury). Depending on the time scales of the different
processes, if free FeII is formed, the benefit of the
nanoreactor can be lost.
VI. Conclusion
In the present study, the electrochemical reduction of
nanoparticles has been investigated using a well-defined
colloidal system. Indeed, both the properties of the γ-Fe2O3
nanoparticles and the properties of the dispersions,
especially their colloidal stability, are well-known. This
allows improving the understanding of their electro-
chemical behavior, essentially through the comparison of
nanoparticles (big species) and ions (small species). For
this purpose, an appropriate medium is found in order to
compare the behavior of ferric or ferrous ions with the
behavior of oxide nanoparticles. This comparison shows
that it does not correspond to a pure change of scale; that
is, the nanoparticle cannot be considered as a “macroion”.
Indeed, a nanoparticle is different from an ion because
such an association of 10 000 atoms becomes a structured
material (γ-Fe2O3 is a semiconductor); therefore, the
parameters that control the phenomena are different.
First, the interactions between the mercury electrode
and the charged γ-Fe2O3 nanoparticles can control the
approach of electroactive species toward the electrode,
depending on the conditions (nature and concentration of
ions); thus, the apparent electroactivity of the species can
be modified. The result is either the adsorption of particles
on the electrode if interactions are attractive, or the
Dubois and Chevalet
absence of reduction current if particles cannot approach
the electrode due to repulsion. These interactions are
shown to be strong and partly of electrostatic nature, so
that they depend on the electrolyte concentration.
Second, as for any material, wettability is an important
parameter: the high wettability of iron by mercury at very
negative potentials explains why here the nanoparticles
of iron formed, whatever the process followed during the
reduction step, enter the mercury electrode, while no
solubility of iron in mercury is measurable.
Finally, as a material, a nanoparticle can be broken
into smaller crystallites down to the level of an atom. We
show that this situation indeed occurs for potentials in
the range -0.4 to -1.2 V/ECS, leading to ferrous ions.
Such a reductive dissolution of the particles may also occur
during the reduction into metallic iron, but this fact is not
directly proven yet. However, depending on the media,
the dissolution at intermediate stages of the reduction
can generate problems. Ions can precipitate under the
form of species adsorbed at the electrode interface and/or
precipitate as nonreducible species. This is clearly linked
to the complex chemistry of iron species in solution and
explains the difficulties encountered in the preparative
experiments. Therefore, the nanoparticle may not always
play the expected role of a nanoreactor which would lead
to a straightforward reduction into an iron object of similar
nanometric size and shape.
The next step in these investigations is now to improve
the knowledge of the behavior of such nanobjects in order
to control their electrochemical transformation. This
implies, in particular, determining the characteristic times
of the different phenomena, that is, of the adsorption on
the electrode, of the electron transfer, and of the propaga-
tion of the reaction layer inside the material.
LA035294R