In search of the most active MN4 catalyst for the oxygen reduction reaction. The case of perfluorinated Fe phthalocyanine

Article abstract of DOI:10.1039/c9ta09125d

Iron macrocyclic complexes (MN4) are promising catalysts for replacing platinum (the industrial standard) in electrocatalysis. In particular, FeN4 complexes have shown lower overpotential than Pt for the oxygen reduction reaction (ORR) in alkaline media. To predict the electrochemical activity of metal electrodes and molecular catalysts towards the ORR, reactivity descriptors with typical volcano correlation have been demonstrated. The most important are M-O₂ binding energy and M⁽ⁿ⁾⁺/M^(n-1)+ redox potentials for the complexes. We studied a new Fe complex, which possesses powerful electron-withdrawing fluorine residues at the periphery of the phthalocyanine ring. Fe hexadecafluorophthalocyanine (16(F)FePc) was characterized by electron paramagnetic resonance (EPR), and X-ray photoelectron spectroscopy (XPS) in the presence and in absence of O₂. Experimental and calculated O₂-Fe binding energies, as well as electrochemical characterization confirms the excellent activity of this complex for the ORR placing this complex at the top of the MN4 volcano correlation.

Full text of DOI:10.1039/c9ta09125d

Journal of
Materials Chemistry A
View Article Online
View Journal | View Issue
COMMUNICATION
In search of the most active MN4 catalyst for the
oxygen reduction reaction. The case of
perfluorinated Fe phthalocyanine†
Cite this: J. Mater. Chem. A, 2019, 7,
24776
a
bc
Received 20th August 2019
Accepted 15th October 2019
Gabriel Abarca,
Walter Orellana,
Marco Viera,b Carolina Aliaga,
Jose F. Marco,d
´
e
b
b
*
*
Jose H. Zagal and Federico Tasca
´
DOI: 10.1039/c9ta09125d
rsc.li/materials-a
Iron macrocyclic complexes (MN4) are promising catalysts for efforts MN4 catalysts can compete with Pt in activity and
replacing platinum (the industrial standard) in electrocatalysis. In stability only in alkaline media,6 whereas there are few of this
particular, FeN4 complexes have shown lower overpotential than Pt class of catalyst that can compete in acidic conditions.5 Activity
for the oxygen reduction reaction (ORR) in alkaline media. To predict and stability seems to be correlated.7 Indeed when increasing
the electrochemical activity of metal electrodes and molecular cata- the activity of the catalyst lower amounts of H2O2 are produced8
lysts towards the ORR, reactivity descriptors with typical volcano and therefore stability also increases. The activity of MN4
correlation have been demonstrated. The most important are M–O2 complexes can be improved by heat-treatment or by preparing
binding energy and M(n)+/M(nꢀ1)+ redox potentials for the complexes. catalysts with high Fe(II)/Fe(III) redox potential.9–14 It is impera-
We studied a new Fe complex, which possesses powerful electron- tive therefore to develop reactivity descriptors for these catalysts
withdrawing fluorine residues at the periphery of the phthalocyanine in order to rationally prepare materials.1,8,9,12,15,16 In previous
ring. Fe hexadecafluorophthalocyanine (16(F)FePc) was characterized work, the kinetic data of OPG electrodes modied with a great
by electron paramagnetic resonance (EPR), and X-ray photoelectron variety of substituted and non-substituted phthalocyanines and
spectroscopy (XPS) in the presence and in absence of O2. Experimental porphyrins of different central metals has been reported in
and calculated O2–Fe binding energies, as well as electrochemical order to establish reactivity descriptors for these catalysts for
characterization confirms the excellent activity of this complex for the the ORR.1,17–19 Several reactivity descriptors have been identied
ORR placing this complex at the top of the MN4 volcano correlation. but the most important ones are the M–O2 binding energy and
the M(n)+/M(nꢀ1)+ redox potentials of the complexes.11,16 It has
been shown that these two descriptors linearly correlate with
each other.16 When comparing activities of different molecular
catalysts (log i)E at the constant driving force of the electrode
1. Introduction
versus the M–O2 binding energy of each particular catalyst
Metal macrocyclic complexes (MN4) are extremely versatile
a non-linear, volcano-shaped curve is obtained. These types of
compounds that act as catalysts for a wide variety of electro-
correlations are well-known for metals, alloys and metal oxides
chemical reactions.1–3 However they have been investigated
extensively in the literature as potential catalysts for the oxygen
reduction reaction (ORR) since 1964.4 The goal of these studies
has been to develop substitutes for expensive Pt-containing
catalysts for the O2 cathode in fuel cells.5 In spite of these
but less common for molecular catalysts.16,20 There is a simi-
larity in the trend in reactivity for molecular catalysts and for
metal catalysts.
A similar correlation is obtained when
comparing log of activities as (log i)E versus the M(n)+/M(nꢀ1)+
redox potential of the catalysts. Volcano correlations depict two
binding energy conditions: one leg of the volcano, with a posi-
tive slope corresponding to the strong binding catalysts and the
other volcano leg corresponding to the weak binding region.16
According to the literature, all FeN4s appear on the strong
binding leg of the volcano and catalyze ORR via 4 electrons
involving the splitting of the O–O bond.16 On the other leg of the
volcano are the catalysts that catalyze ORR via 2 electrons like
Co and Ni complexes. According to linear scale relationships
strong O2 binding will correlate linearly with strong peroxide
binding so this could explain the 4 eꢀ selectivity of catalysts
a
´
Centro de Nanotecnologıa Aplicada, Facultad de Ciencias, Universidad Mayor,
Santiago, Chile
b
´
´
Facultad de Quımica y Biologıa, Universidad de Santiago de Chile, Santiago, Chile.
E-mail: jose.zagal@usach.cl; federico.tasca@usach.cl
cCenter for the Development of Nanoscience and Nanotechnology, CEDENNA,
Santiago, Chile
d
´
´
Instituto de Quımica Fısica “Rocasolano”, CSIC, Madrid, Spain
e
´
´
Departamento de Ciencias Fısicas, Universidad Andres Bello, Sazie 2212, 837-0136
Santiago, Chile
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c9ta09125d
24776 | J. Mater. Chem. A, 2019, 7, 24776–24783
This journal is © The Royal Society of Chemistry 2019

View Article Online
Communication
Journal of Materials Chemistry A
sitting on the strong binding leg of a volcano correlation.16,21 In chamber at the so X-ray spectroscopy (SXS) beam-line end-
this case the peroxide intermediate that can form during ORR station at the Brazilian Synchrotron Light Laboratory. As re-
remains bound to the active site, facilitating its further reduc- ported before,24 room temperature O2 treatment was conducted
tion to H2O or OHꢀ. In this work we have investigated a new in the preparation chamber using a pressure of 20 psi of 5% O2
catalyst for ORR, iron-hexadecauorophthalocyanine (16(F) + 95% He over two hours. The spectra were collected using an
FePc) that has powerful electron-withdrawing F residues that InSb (111) double-crystal monochromator at a xed photon
hypothetically should have the highest activity among all the energy of 1840 eV. The hemispherical electron analyser (PHOI-
MN4 complexes and therefore would be sited at the top of the BOS HSA500 150 R6) was set at a pass energy of 30 eV, and the
volcano correlation. The complex was characterized by EPR and energy step was 0.1 eV. The base pressure used inside the
XPS spectroscopy in the presence and absence of O2. Experi- chamber was around 2.0 ꢂ 10ꢀ9 mbar. The monochromator
mental O2–Fe binding energies and calculated ones, as well as photon energy calibration was done at the Si K edge (1839 eV).
electrochemical characterization conrm the highest activity of An additional calibration of the analyser's energy was per-
this complex towards the ORR.
formed using standard Au foil (Au 4f7/2 peak at 84.0 eV). Curve
tting of the core XPS lines was carried out using Casa XPS
soware using a Gaussian–Lorentzian product function and
a nonlinear Shirley background subtraction.
2. Experimental
Iron 1,2,3,4,8,9,10,11,15,16,17,18,22,23,24,25-hexadeca(uoro)
Spin-polarized density functional theory (DFT) calculations
phthalocyanine (16(F)FePc) was obtained from Luminescence were performed with the Quantum-ESPRESSO ab initio
Technology Corp (New Taipei City, Taiwan) and used as package.25 The dispersive (p-stacking) interaction of 16(F)FePc
provided. N,N-dimethylformamide (DMF), isopropyl alcohol, adsorbed on the carbon nanotube (CNT) was included through
NaOH, H2SO4, K2HPO4 were obtained from Sigma (St. Louis, the van der Waals exchange-correlation functional (vdW-DF2).26
USA). Pt catalyst, 20% Pt supported on Vulcan XC-72 carbon Kohn–Sham eigen-functions were expanded on a plane-waves
(Pt20%Vulcan), was from Fuel Cell Store (College Station, basis set where the interaction between valence electrons and
Texas). Aqueous solutions were prepared using Milli Q water ion cores are described by ultra-so pseudo-potentials.27
(Millipore, Inc.). Double-walled CNT were obtained from Converged results have been achieved by using cut-off energies
Nanocyl (Sambreville, Belgium). Double-walled CNT were used of 35 Ry on plane waves and of 280 Ry on the electronic density.
instead of multi- or single-walled CNT because of their higher The 16(F)FePc-CNT complex was simulated in a large unit cell,
purity (>90%).22,23 CNT were used as support because of their containing up to 345 atoms, with periodic boundary conditions.
˚
high surface to volume ratio and the possibility to support A metallic single-walled CNT with (8,8) chiral indexes and 11 A
higher amounts of catalyst.17 To modify electrodes an ink of in diameter was considered. The sampling in the irreducible
16(F)FePc-CNT was obtained by dispersing 1 mg of 16(F)FePc part of the Brillouin zone was restricted to the G point. The
with 1 mg of CNT to obtain 16(F)FePc-CNT in a solution of 75% complex was fully relaxed until the force on each atomic
volume isopropyl alcohol and 25% water. The dispersion was component was less than 0.01 eV Aꢀ1. The ORR catalytic activity
˚
sonicated for 1 h and collected by ltering and washing to of 16(F)FePc-CNT complex was estimated by assessing the
eliminate excess of 16(F)FePc which would not adsorb on the minimum energy path for the O2 dissociation aer being
CNT walls. An ink of Pt catalyst was obtained by dispersion and adsorbed on the Fe atom, using the nudged elastic band (NEB)
sonication of 1 mg of catalyst in 1 mL of similar solvent. 20 mL of method.28
catalyst ink were loaded onto the surface of the electrode and
le to dry (nal loading was 0.1 mg cmꢀ2). The counter elec-
trode was a 10 cm long 0.1 cm diameter Pt spiral wire. As
3. Results and discussion
reference electrode an Origalys (Rillieux-la-Pape, France) 3 M In the volcano correlation reported previously (see graphical
KCl Hg|HgCl was used and all the potentials are reported vs. abstract with the exclusion of the apex), there were no catalysts
this electrode. The working electrode was an edge-plane pyro- sitting on the apex of the plot. Hypothetically, a catalyst could
lytic rotating disk graphite (5 mm diameter and 4 mm thick), exist that has the proper M(III)/(II) redox potential or M–O2
mounted in Teon (external diameter of 7.50 mm, internal binding energy to stand at the top of the correlation, so essen-
diameter of 6.50 mm) and was rotated using a MSR rotator from tially this could be the best possible catalyst for the reaction. In
Pine Instruments (Durham, NC, USA). The graphite disk was this work iron-hexadecauorophthalocyanine, 16(F)FePc,
renewed prior to modication with 800 grit emery paper. Elec- adsorbed on carbon nanotubes (i.e. 16(F)FePc-CNT) was tested.
trochemical experiments were carried out on an Autolab The synthesis of 16(F)FePc was rst reported in 1969 by Jones
PGSTAT 302N with a dual mode bipotentiostat module (Utrecht, and Twigg,29 but, to the best of our knowledge, the electro-
The Netherlands). In RRDE experiments, the ring potential was chemistry of this compound and its catalytic properties for the
set to 0.65 V vs. Hg|HgCl (saturated KCl). For EPR and XPS ORR have never been reported. This particular complex has
measurements a dried powder of MN4-CNT was used. EPR more powerful electron-withdrawing substituents on the
spectra were collected at 298 K with a Bruker EMX-1572 spec- phthalocyanine ligand, so the Fe(III)/(II) is shied to more
trometer working at ꢁ9.39 GHz (X-band). For the XPS positive potentials compared to all previously reported cata-
measurements, the sample powder of 16(F)FePc-CNT was lysts.1,13,16,17 Fig. 1 illustrates the conguration of a 16(F)FePc
spread out on carbon tape and introduced into the analysis molecule adsorbed at on the graphene layer of a CNT or
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A, 2019, 7, 24776–24783 | 24777

View Article Online
Journal of Materials Chemistry A
Communication
Fig. 1 Schematic representation of 16(F)FePc adsorbed on CNT and electrochemical characterization by cyclic voltammetry in N2 saturated
media and by linear sweep voltammetry in O2 saturated media of electrodes modified with 16(F)FePc-CNT (blue and red lines) or Pt20%Vulcan
(black line). Conditions (a) 0.1 M NaOH; (b) 0.1 M H2SO4. Cyclic voltammetry recorded at 100 mV sꢀ1; linear sweep voltammetry recorded at 5 mV
s
ꢀ1, 1200 rpm.
a graphite electrode and typical cyclic voltammogram of Fe(III)/(II) redox peak of MN4 usually shows a 60 mV pHꢀ1 unit
graphite electrodes modied with 0.1 mg cmꢀ2 of 16(F)FePc- dependence in the wide range of pH 13–1 as the process
CNT, in N2 saturated 0.1 M NaOH (Fig. 1(a)) or H2SO4 solu- involves the transfer of an OHꢀ ion as [Fe(III)OH]ad + e $
ꢀ
tions (Fig. 1(b)). For comparison, we report the linear sweep [Fe(II)]ad + OH(aq)
.
31 MN4 complexes were known to be unstable
voltammetry of electrodes modied with Pt20%Vulcan (black in acid during the ORR process. It is probable that demetalla-
line), the industrial standard catalyst for the ORR. The Fe(III)/(II) tion of the metal site occurs because of the production of H2O2
redox peaks appear at potentials close to 1 and +0.75 V vs. RHE as a result of the ORR in acid media. In a recent publication, the
at pH 13 and 1 respectively. Table 1 reports these values as well activity of penta-coordinated FePc catalyst at pH 1 was re-
as the values for the Fe(II)/(I) redox peak. Because of the strong ported.32 The h pyridine axial coordination pulls electron
electron-withdrawing F residues, which are distributed in all density from the Fe centre, lowering the separation between the
peripheral and non-peripheral positions in the phthalocyanine energies of the frontier orbitals of the donor (Fe) and the
ligand 16(F)FePc exhibits the most positive redox potential for acceptor (O2), resulting in increased activity towards the ORR.
the Fe(III)/(II) couple among the previously studied FeN4 (e.g. Following the idea that the activity of MN4 can be increased also
0.9 V for iron hexadecachloro phthalocyanine 16(Cl)FePc17 and in acid media with the increasing electron-withdrawing power
0.95 V for Fe tetrapyridinoporphyrazine, at pH 13 (ref. 30)). The of the residues in the phthalocyanine macrocycle, the 16(F)
Table 1 Electrochemical parameters of 16(F)FePc-CNT. Redox potential of Fe(II)/(I) or Fe(III)/(II); number of electrons calculated from the
Koutecky–Levich equation; Tafel slope; surface coverage (Ƭ); onset from the O2 reduction polarization curves
E(M(I)/(II))
(V vs. RHE)
E(M(II)/(III))
(V vs. RHE)
ONSET
(V vs. RHE)
Catalyst
Media
nꢃ eꢀ
Tafel (V)
Ƭ (mol ꢂ cmꢀ2
)
16(F)FePc-CNT
16(F)FePc-CNT
Pt20%Vulcan
Pt20%Vulcan
0.1 M NaOH
0.1 M H2SO4
0.1 M NaOH
0.1 M H2SO4
0.750
0.610
—
0.99
0.75
—
4.05
3.45
4.01
4.01
ꢀ0.046
ꢀ0.073
ꢀ0.047
ꢀ0.048
1.18 ꢂ 10ꢀ9
2.71 ꢂ 10ꢀ9
4.50 ꢂ 10ꢀ10
2.83 ꢂ 10ꢀ10
1.055
0.810
0.995
0.955
—
—
24778 | J. Mater. Chem. A, 2019, 7, 24776–24783
This journal is © The Royal Society of Chemistry 2019

View Article Online
Communication
Journal of Materials Chemistry A
Fig. 2 Polarization curves and relative H2O2 oxidation curves recorded at electrodes modified with 16(F)FePc-CNT (disks) and Pt ring electrode.
Conditions: (a and c) O2 saturated 0.1 M NaOH solution. (b and d) O2 saturated 0.1 M H2SO4 solution. Rotating ring disk electrode velocity: 200,
400, 800, 1200, and 1600 rpm, 5 mV sꢀ1; Pt ring polarized at 1.3 V vs. RHE.
FePc-CNT was tested in acid. Fig. 2 illustrates the ORR polari- pathway. From Tafel plots using currents in the low polarization
zation curves at various rotating speeds in O2 saturated 0.1 M linear region values zꢀ0.046 V were obtained while at pH 1, the
NaOH or H2SO4 solutions. Because of the high redox potential Tafel slopes were zꢀ0.073 V. Slopes close to ꢀ0.04 V indicated
of the Fe(III)/(II) redox couple the onset for the ORR at pH 13 that the rst step involved a fast one-electron transfer followed
appears already at potentials above 1.05 V. The overpotential for by a slow one-electron transfer step while values close to
the ORR is reduced by more than 0.16 V if compared to FePc- ꢀ0.060 V suggested that the rst step involves a fast electron
CNT and to almost 0.1 V if compared to 16(Cl)FePc-CNT transfer followed by a slow rate determining chemical step. The
measured in the same media.17 At pH 1 the onset for the ORR change in ET mechanism going from alkaline solution to acid
appears already at 0.81 V, that is to say a diminution of the solution has been observed before for similar FeN4 complexes.33
overpotential for the ORR of around 0.1 V if compared to FePc-
Turn over frequencies (TOFs) were estimated for the 16(F)
CNT and of 0.05 V if compared to 16(Cl)FePc-CNT in similar FePc-CNT modied electrode from the coverage of the elec-
media.32 In Fig. 2(c) and (d) H2O2 oxidation current recollected trode G and the kinetic current densities at the half wave
at the Pt ring electrode during linear sweep voltammetry at pH potential of the ORR polarization curve, with values ranging
13 and at pH 1 are presented. At pH 1, ve times more H2O2 was from 5.4 sꢀ1 to 3.4 sꢀ1 for pH 13 and pH 1 respectively (values
produced than at pH 13, demonstrating the higher activity of listed in Table 2). At pH 13, TOF values of 5.2 sꢀ1 were calculated
the 16(F)FePc at pH 13 and therefore the lower stability at pH 1. for electrodes modied with 16(Cl)FePc-CNT or the unsub-
If we compare those results with the ones obtained with Pt20% stituted iron phthalocyanine (FePc-CNT).17 At pH 1, TOF values
Vulcan, we can notice that the onset for the ORR at pH 13 for of 0.2 and 0.02 sꢀ1 were calculated for 16(Cl)FePc-CNT and
16(F)FePc occurs at 0.06 V lower overpotentials. Nevertheless it FePc-CNT respectively. Therefore, similar TOF values were ob-
should be considered that the concentration of active Pt on the tained in alkaline media, but at pH 1, one order magnitude
graphite electrode modied with Pt20%Vulcan is almost one higher TOF values were obtained with the 16(F)FePc catalyst
order of magnitude lower than the amount of active Fe(III)/(II) in because of its lower electron density in the metal center and
16(F)FePc-CNT (see Table 1 for the reꢀsumed values).
higher activity towards the ORR. If we compare the TOF values
The extrapolation of the total nꢃ e determined by the Kou- obtained for 16(F)FePc-CNT with the ones obtained for Pt%
tecky–Levich equation for pH 13 and pH 1 are summarized in 20Vulcan, we can notice that the 2 catalyst show similar TOF at
Table 1. The alkaline pH favored 4 eꢀ reduction (4.0 electrons low overpotential (i.e. at 0.9 V vs. RHE) in alkaline media, only
per O2 molecule). Nevertheless, at pH 1, 3.45 electrons were because of the lower onset for the ORR of 16(F)FePc-CNT. If we
transferred per O2 molecule indicating a mixed reaction compare the 2 catalyst close to the limiting current region (i.e. at
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A, 2019, 7, 24776–24783 | 24779

View Article Online
Journal of Materials Chemistry A
Communication
Table 2 TOF calculated at 0.9, 0.8, 0.7 V vs. RHE or at half wave potential. Conditions: O2 saturated, 0.1 M NaOH or H2SO4 solutions
Catalyst
Media
TOF @ 0.9 V (sꢀ1
)
TOF @ 0.8 V (sꢀ1
)
TOF @ 0.7 V (sꢀ1
)
TOF @ hw (sꢀ1
)
16(F)FePc-CNT
16(F)FePc-CNT
Pt20%Vulcan
0.1 M NaOH
0.1 M H2SO4
0.1 M NaOH
0.99
—
0.8
6
11
2.8
18.4
5.4
3.4
11.3
12.1
0.7 V vs. RHE), TOF numbers for Pt%20Vulcan are in the order reported binding energies and resembled the spectra of some
of 18 sꢀ1 while for the 16(F)FePc-CNT catalyst values of 11 and phthalocyanines reported containing only Fe(III).32 It is impor-
2.8 sꢀ1 were obtained at pH 13 and pH 1 respectively, as tant to note, that satellite peaks represent the intermolecular
resumed in Table 2.
charge transfer process, where the intensity of these peaks can
EPR and XPS spectroscopy conrmed that in 16(F)FePc-CNT be related to the amount of charge transferred between the
the electrons are displaced from the Fe centre because of the fragments of the macrocycle and the iron center.38 This
pulling effect of F atoms.34,35 In Fig. 3(a) EPR spectra of 16(F) phenomenon can be correlated with the Franck–Condon effect,
FePc-CNT and 16(F)FePc are reported. The strong electron- where it occurs aer an electron is removed from the sample
withdrawing effect of the F substituents on the electron envi- (photoemission) and the sample suffers an inner-sphere reor-
ronment on the odd spin of the metallic centre of the porphyrin ganization.39,40 Aer the exposition to O2 (Fig. 3(b), red line),
shows a higher value of g (2.06909) if compared to the unsub- a signicant energy shi of 1.5 eV was observed (708.7 eV for Fe
stituted porphyrin (i.e. FePc-CNT) 2.06494 or to the per- 2p3/2). According to previous work, this second contribution
chlorinated FeN4 (i.e. 16(Cl)FePc-CNT) 2.06574.
agrees with the presence of Fe(II).24,32 Moreover, aer the expo-
XPS in the presence and the absence of O2 coordinated to the sure to O2 an increased Fe(II)/Fe(III) ratio was observed (from
Fe metal center was performed to verify the electronic structure 1.27 ratio when reduced to 1.38 ratio when oxidized, Fig. S2†).
of the 16(F)FePc-CNT. Distinct peaks in F 1s, Fe 2p, O 1s, N 1s These results suggest the presence of a more stable Fe–O2
and C 1s regions, conrmed the attachment of 16(F)FePc on the adduct.24,32,41,42 It can be concluded that aer the exposition to
external surface of the CNT (Fig. S1†). The atomic composition O2, an increased character of Fe(II) exists due to the presence of
of 16(F)FePc-CNT, between the Fe 2p peak and F 1s peak, was electron-withdrawing groups, that can tune the electron density
obtained within experimental error 1 : 16.8. The Fe 2p region towards the molecular fragments. When O2 binds to the metal
was analyzed for providing further insights into the catalytic in MN4, its 2p electrons interact with the partially lled
behavior of 16(F)FePc-CNT (Fig. 3(b) and S2†). Two broad spin– d orbitals of the same. These processes are accompanied by
orbit doublet Fe 2p peaks indicated the presence of Fe in intermolecular electron transfer, in which O2 accepts charge
different valence states, with the main component being density from the partially lled d orbitals of the metal via back-
a mixture of Fe(II) and Fe(III).32,36,37 The binding energy of the Fe bonding to the p* antibonding orbital and donates charge from
2p3/2 photoelectron line of 16(F)FePc-CNT was located at a lled p molecular orbital to a half-lled dz orbital of the
710.2 eV, and a satellite peak (shake-up) at 7.4 eV higher was metal.32
also observed, which is easily detectable and does not overlap
Results from theoretical calculations show that 16(F)FePc-
with Fe 2p peaks. These peaks were in good agreement with CNT complex can exist in three spin states, which is due to
Fig. 3 (a) EPR spectrum of 16(F)FePc-CNT. (b) XPS spectrum of 16(F)FePc-CNT before (black line) and after (red line) treatment with O2.
24780 | J. Mater. Chem. A, 2019, 7, 24776–24783
This journal is © The Royal Society of Chemistry 2019

View Article Online
Communication
Journal of Materials Chemistry A
Fig. 4 (a) Equilibrium geometry for O2 adsorbed on the metal center of the 16(F)FePc-CNT, complex in the end-on configuration. (b) Minimum
energy path for the O2 dissociation after its adsorption on the metal center of the 16(F)FePc-CNT complex. Calculations performed with the NEB
method.
the Fe metal centre. The most stable s state has two unpaired breakage one O atom would be stabilized by a hydrogen bond to
electrons (magnetic moment m ¼ 2 mB). The states with moment one of the C atoms before the release of a hydroxyl anion. The
m ¼ 0 and m ¼ 4 mB are also rather stable but with a much most stable state of the 16(F)FePc-CNT adduct has a magnetic
higher energy of 0.232 and 0.197 eV, respectively (results in moment of m ¼ 0, which means that the spin of the metal centre
Table 2). In Fig. 4(a) is shown the equilibrium geometry of the couples antiferromagnetically with the spin of the O2 molecule.
O2 molecule adsorbed on the metal centre of the 16(F)FePc-CNT The adduct is also stable with the m ¼ 2 mB and m ¼ 4 mB spin
complex. The 16(F)FePc molecule rmly sticks to the CNT states with 0.042 and 0.196 eV higher in energy, respectively.
external walls through p-stacking or van der Waals interactions.
The results for O2 binding to the metallic centre of the (16F)
The binding energy of 16(F)FePc on the CNT is quite strong FePc-CNT are shown in Table 3. In the most stable spin state, m
(2.078 eV, Table 3) and comparable to covalent bonds. The most ¼ 0, the binding energy of O2 is 0.655 eV and its bond length of
˚
stable structure of the complex has a magnetic moment of m ¼ 2 1.322 A, which is 6.4% longer than that of in the O2 gas-phase.
mB. Other allowed spin states are m ¼ 0 and m ¼ 4 mB with For the other allowed spin states, the O2 binding energy
binding energies of 2.116 and 2.518 eV, respectively. Because of decreases, conrming that the m ¼ 0 state is the most stable
the strong p-stacking interaction between 16(F)FePc and the spin coupling of the adduct. An energy barrier of 1.2 eV for the
CNT, the 16(F)FePc molecule follows the CNT curvature with dissociation energy of O2 when adsorbed on the metal centre
˚
a distance from the CNT surface of 3.36 A. O2 is adsorbed on the was calculated through the NEB method, as was the minimum
Fe metal centre with an end-on conguration, with O2 tilted to energy path (Fig. 4(b)). In the saddle point, the O2 molecule
the plane of the 16(F)FePc molecule, as shown in Fig. 4. The binds with both Fe and C atoms, showing that the dissociation
possibility that O2 binds to the metal centre by a side-on occurs on the surface of the 16(F)FePc-CNT catalyst. The energy
conguration (with O–O parallel to the plane of the 16(F)FePc barrier for O2 dissociation on a Pt(111) surface has been
molecule) was also considered. However, this geometry was calculated to be of 0.7 eV.43 The ORR energy barrier obtained by
shown to be unstable and relaxed to the end-on one. The end-on 16(F)FePc-CNT, of 1.2 eV, suggests a good performance
orientation promotes the splitting of the O–O bond involving comparable to Pt. The binding energy for the FeN4 and the
a lower energy transition state and therefore lower activation CoN4 complexes in the presence of CNT was calculated and
energy for the process. From simulation aer O–O bond reported in Table 4. Activities for the same complexes for the
Table 3 Theoretical results for the binding energy of the 16(F)FePc molecule adsorbed on a (8,8) CNT, for the allowed spin states (m). Total
energy (Etotal) of 16(F)FePc-F-CNT complex, also with an adsorbed O2 molecule (O2-16(F)FePc-CNT) for the allowed spin states. E0 is the lowest
energy state. Binding energy (Eb) of O2 adsorbed on the metallic center of the 16(F)FePc-CNT complex for the allowed spin states. dFePc-F-CNT
,
dFe–N4, dO–O, and dFe–O are bond distance between respective atoms
16(F)FePc-CNT
O2–16(F)FePc-CNT
Etotal (eV)
˚
˚
˚
˚
m (mB)
Eb 16(F)FePc (eV)
dFePc–F-CNT (A)
dFe–N4 (A)
Etotal (eV)
Eb (O2) (eV)
dO–O (A)
dFe–O (A)
0
2
4
ꢀ2.116
ꢀ2.078
ꢀ2.518
3.358
3.359
3.359
1.947
1.954
1.955
E0 + 0.232
E0
E0 + 0.197
E0
ꢀ0.655
ꢀ0.612
ꢀ0.458
1.322
1.322
1.315
1.832
1.890
2.064
E0 + 0.042
E0 + 0.196
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A, 2019, 7, 24776–24783 | 24781

View Article Online
Journal of Materials Chemistry A
Communication
Table 4 Theoretical results for the magnetic moment (m) and O2
binding energy (Ebind), for FePc-CNT, 16(Cl(FePc))-CNT, CoPc-CNT,
16(F)CoPc-CNT and 8b(2-Et-C6H11O)CoPc-CNT in their lowest
energy state
acknowledge the laboratory of free radicals for use of the EPR
(USACH) and CONICYT-FONDEQUIPEQM140060. G. A. thanks
´
FONDECYT-Iniciacion No. 11170879. The authors thank Dr
Ingrid Ponce for fruitful advices on the experimental proce-
dures for the XPS experiments and the LNLS-CNPEM (Brazil) for
the project 20170843 (SXS beam-line).
Compound
m (mB)
O2 Ebind (eV)
16(F)fePc-CNT
FePc-CNT
16(Cl)FePc-CNT
16(F)CoPc
0
0
0
1
1
1
ꢀ0.655
ꢀ0.730
ꢀ0.660
ꢀ0.325
ꢀ0.380
ꢀ0.345
References
CoPc
8b(2-Et-C6H11O)CoPc
1 J. H. Zagal, S. Griveau, J. F. Silva, T. Nyokong and F. Bedioui,
Coord. Chem. Rev., 2010, 254, 2755–2791.
2 A. B. Sorokin, Chem. Rev., 2013, 113, 8152–8191.
3 H. Lu and N. Kobayashi, Chem. Rev., 2016, 116, 6184–6261.
4 R. Jasinski, Nature, 1964, 201, 1212–1213.
5 T. Sun, B. Tian, J. Lu and C. Su, J. Mater. Chem. A, 2017, 5,
18933–18950.
6 J. Yang, F. Toshimitsu, Z. Yang, T. Fujigaya and
N. Nakashima, J. Mater. Chem. A, 2017, 5, 1184–1191.
7 J. Yang, J. Tao, T. Isomura, H. Yanagi, I. Moriguchi and
N. Nakashima, Carbon, 2019, 145, 565–571.
8 K. Kumar, P. Gairola, M. Lions, N. Ranjbar-Sahraie,
M. Mermoux, L. Dubau, A. Zitolo, F. Jaouen and
F. Maillard, ACS Catal., 2018, 8, 11264–11276.
9 N. Ramaswamy, U. Tylus, Q. Jia and S. Mukerjee, J. Am. Chem.
Soc., 2013, 135, 15443–15449.
ORR were extracted from previous publications.17,44 The values
are reported in the graphical abstract where the activity of the
complexes at ꢀ0.255 V vs. the binding energy are correlated.
From these values a volcano correlation is reported in the
graphical abstract. The calculated binding energy of O2 to Fe
and the measured electrocatalytic activity of the complex agree
with positioning this complex on the top of a volcano correla-
tion where binding energy and activities are reported for several
MN4 complexes.
4. Conclusions
Fe hexadecauorophthalocyanine (16(F)FePc) was studied as an
electrocatalyst for the ORR. Because of uorine substituents on
the phthalocyanine ligand, electrons are dislocated from the Fe
center. EPR and XPS spectroscopy conrmed the pulling effect
of Fꢀ atoms. Aer exposition of the catalyst to O2 a signicant
energy shi of 1.5 eV was observed during XPS experiments,
indicating the presence of a stable Fe–O2 adduct. 16(F)FePc-
CNT exhibited the most positive redox potential for the Fe(III)/
(II) couple and the highest activity towards the ORR among
previously studied MN4. DFT calculations for the O2–M binding
energy and the minimum energy path for the O2 dissociation
were performed for various MN4. If the electrocatalytic activities
of the MN4 at constant driving force of the electrode versus the
M–O2 binding energy of each particular catalyst were compared
a volcano-shaped curve was obtained. 16(F)FePc-CNT sits at the
top of this volcano and is the most active MN4 catalyst for the
ORR.
10 J. H. Zagal, I. Ponce, D. Baez, R. Venegas, J. Pavez, M. Paez
and M. Gulppi, Electrochem. Solid-State Lett., 2012, 15, B90–
B92.
˜
11 J. H. Zagal, F. Javier Recio, C. A. Gutierrez, C. Zuniga,
´
M. A. Paez and C. A. Caro, Electrochem. Commun., 2014, 41,
24–26.
12 A. A. Gewirth, J. A. Varnell and A. M. DiAscro, Chem. Rev.,
2018, 118, 2313–2339.
13 X. Huang and J. T. Groves, Chem. Rev., 2018, 118, 2491–2553.
14 N. Zion, A. Friedman, N. Levy and L. Elbaz, Adv. Mater., 2018,
30, 1800406.
15 M. J. Workman, A. Serov, L.-k. Tsui, P. Atanassov and
K. Artyushkova, ACS Energy Lett., 2017, 2, 1489–1493.
16 J. H. Zagal and M. T. M. Koper, Angew. Chem., Int. Ed., 2016,
55, 14510–14521.
17 R. Venegas, F. J. Recio, C. Zuniga, M. Viera, M.-P. Oyarzun,
N. Silva, K. Neira, J. F. Marco, J. H. Zagal and F. Tasca,
Phys. Chem. Chem. Phys., 2017, 19, 20441–20450.
18 C. Linares-Flores, J. Espinoza-Vergara, J. H. Zagal and
R. Arratia-Perez, Chem. Phys. Lett., 2014, 614, 176–180.
19 J. Masa, K. Ozoemena, W. Schuhmann and J. H. Zagal, J.
Porphyrins Phthalocyanines, 2012, 16, 761–784.
20 M. Busch, N. B. Halck, U. I. Kramm, S. Siahrostami, P. Krtil
and J. Rossmeisl, Nano Energy, 2016, 29, 126–135.
Conflicts of interest
The authors state that there are no conicts to declare.
Acknowledgements
F. T. thanks for nancial support the Fondecyt Project 1181840, 21 M. T. M. Koper, J. Electroanal. Chem., 2011, 660, 254–260.
and Proyecto Basale Dicyt. J. H. Z. thanks the nancial support 22 A. Morozan, S. Campidelli, A. Filoramo, B. Jousselme and
of Millenium Project RC120001, Project Anillo ACT 1412 and
Dicyt-USACH, Fondecyt 1140199. W. O. thanks nancial 23 F. Tasca, W. Harreither, R. Ludwig, J. J. Gooding and
support from Fondecyt Project 1170480 and the Power- L. Gorton, Anal. Chem., 2011, 83, 3042–3049.
ed@NLHPC supercomputing infrastructure of the NLHPC 24 A. Pizarro, G. Abarca, C. Gutierrez-Ceron, D. Cortes-
S. Palacin, Carbon, 2011, 49, 4839–4847.
(ECM-02). C. A. thanks CONICYT grant FB0807. The authors
Arriagada, F. Bernardi, C. Berrios, J. F. Silva,
24782 | J. Mater. Chem. A, 2019, 7, 24776–24783
This journal is © The Royal Society of Chemistry 2019

View Article Online
Communication
Journal of Materials Chemistry A
M. C. Rezende, J. H. Zagal, R. Onate and I. Ponce, ACS Catal., 33 J. Zagal, P. Bindra and E. Yeager, J. Electrochem. Soc., 1980,
2018, 8, 8406–8419. 127, 1506–1517.
25 S. B. P. Giannozzi, N. Bonini, M. Calandra, R. Car, 34 H. Fujii, T. Yoshimura and H. Kamada, Inorg. Chem., 1996,
C. Cavazzoni, D. Ceresoli, G. L. Chiarotti, M. Cococcioni, 35, 2373–2377.
I. Dabo, A. Dal Corso, S. de Gironcoli, S. Fabris, G. Fratesi, 35 A. MacCragh, C. B. Storm and W. S. Koski, J. Am. Chem. Soc.,
R. Gebauer, U. Gerstmann, C. Gougoussis, A. Kokalj, 1965, 87, 1470–1476.
M. Lazzeri, L. Martin-Samos, N. Marzari, F. Mauri, 36 V. S. Vasan and R. J. Cross, J. Chem. Phys., 1982, 77, 4507–
R. Mazzarello, S. Paolini, A. Pasquarello, L. Paulatto, 4514.
C. Sbraccia, S. Scandolo, G. Sclauzero, A. P. Seitsonen, 37 M. C. Biesinger, B. P. Payne, A. P. Grosvenor, L. W. M. Lau,
A. Smogunov, P. Umari and R. M. Wentzcovitch, J. Phys.:
Condens. Matter, 2009, 21, 395502.
A. R. Gerson and R. S. C. Smart, Appl. Surf. Sci., 2011, 257,
2717–2730.
´
26 K. Lee, E. D. Murray, L. Kong, B. I. Lundqvist and 38 R. W. Bigelow, K. Y. Law, D. H. K. Pan and H. J. Freund, J.
D. C. Langreth, Phys. Rev. B: Condens. Matter Mater. Phys.,
2010, 82, 081101.
Electron Spectrosc. Relat. Phenom., 1988, 46, 1–17.
39 J. Conradie and E. Erasmus, Polyhedron, 2016, 119, 142–150.
27 K. F. Garrity, J. W. Bennett, K. M. Rabe and D. Vanderbilt, 40 E. Erasmus, J. Electron Spectrosc. Relat. Phenom., 2018, 223,
Comput. Mater. Sci., 2014, 81, 446–452. 84–88.
28 B. L. U. G. Henkelman and H. A. Jonsson, J. Chem. Phys., 41 R. Cao, R. Thapa, H. Kim, X. Xu, M. Gyu Kim, Q. Li, N. Park,
2000, 113, 9901–9904. M. Liu and J. Cho, Nat. Commun., 2013, 4, 2076–2082.
29 J. G. Jones and M. V. Twigg, Inorg. Chem., 1969, 8, 2018–2019. 42 A. B. P. Lever, Inorg. Chem., 1990, 29, 1271–1285.
30 A. A. Tanaka, C. Fierro, D. A. Scherson and E. Yeager, Mater. 43 W. Orellana, J. Phys. Chem. C, 2013, 117, 9812–9818.
´˜
44 J. Riquelme, K. Neira, J. F. Marco, P. Hermosilla-Ibanez,
Chem. Phys., 1989, 22, 431–456.
´
31 J. Zagal, M. Paez, A. A. Tanaka, J. R. dos Santos and
W. Orellana, J. H. Zagal and F. Tasca, Electrochim. Acta,
2018, 265, 547–555.
C. A. Linkous, J. Electroanal. Chem., 1992, 339, 13–30.
32 R. Venegas, F. J. Recio, J. Riquelme, K. Neira, J. F. Marco,
I. Ponce, J. H. Zagal and F. Tasca, J. Mater. Chem. A, 2017,
5, 12054–12059.
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A, 2019, 7, 24776–24783 | 24783