Article

Cite This: J. Phys. Chem. C XXXX, XXX, XXX−XXX pubs.acs.org/JPCC

Design and Testing Model Cobalt Catalysts for Reactions Involving

CO2 and H2O
Carlos A. C. Perez,†,‡ Neuman S. de Resende,‡ Vera M. M. Salim,‡ and Martin Schmal*,‡
†
Instituto Federal do Rio de Janeiro, Campus Nilópolis, Nilópolis, 26530-060, RJ, Brazil
‡
Chemical Engineering Program (PEQ), NUCAT, Centro de Tecnologia, Universidade Federal do Rio de Janeiro, Ilha do Fundão,
Rio de Janeiro, 21941-914, RJ, Brazil

ABSTRACT: In this work, we developed experimental model of cobalt

See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

heterogeneous catalyst, inserting the active phase into a nanoporous material

ordered with the purpose of structurally elucidating the processes that occur
during the interaction of the model with molecules. We synthesized the
models by inserting cobalt species into a Faujasite zeolite and performed
Downloaded via Martin Schmal on November 23, 2018 at 13:18:40 (UTC).

detailed studies of the complexes formed in the interaction of these models

with water and carbon dioxide emphasizing electron densities. Regarding the
interaction of Faujasite with water, X-ray diffraction showed the physical
adsorption of a dense monomolecular film in the nanoporous space. There
was no evidence of water entering the sodalite cage. The cell parameter
increased with the degree of hydration for Faujasite containing cobalt (CoY).
In the case of dehydrated material, cobalt occupied the sites within the
hexagonal prism (I) and the sodalite cage (I′). The carbon dioxide adsorbs
bridges between adjacent compensation cations located in II. We tested the
catalyst models for the oxidation of carbon monoxide and the combustion of hydrogen at low temperatures. The CoY and CoYt
models showed different performances in these reactions. The electron density map obtained during the exposure of the CoY
model to the oxidation reaction conditions of the CO showed the cobalt within the smallest cavities of the zeolite. In the
combustion of hydrogen in CoY, there was consumption of reagents in the temperature range 30−100 °C.

■ INTRODUCTION
Natural gas is an important source for different processes and
in particular for production of hydrogen for fuel cells.
Reforming is used to produce synthesis gas, CO + H2 with
an H2/CO ratio about 2. Fuel cells need pure hydrogen and
PROX is the most used process. In applied catalysis, such as
PROX, we used different metals and supports.1 The stream
contains about 60% H2 and CO + O2. One observes CO
oxidation at low temperatures and H2 oxidation forming
respectively H2O and CO2. On the other hand, carbon dioxide
and water are the lowest free energy substances in the cycles of
generation and use of energy, which are also produced by
converting chemicals from molecules, by burning of fuels, and
by the breathing of living beings.
Nowadays, the activation and transformation of CO2 and
H2O molecules can also be a way to control the emission of
carbon dioxide. Although high activation energy is required,
the reorganization of chemical bonds allows transforming them
into higher value-added compounds. Indeed, according to
Izumi,2 solar energy could be a favorable economic option in
the photocatalytic conversion of CO2 into fuel.
However, the CO2 molecule is thermodynamically stable
and is very difficult to activate because it requires modification
of its electronic structure for a chemical reaction. The
interactions between CO2 molecules and the metals are too
weak, as shown by Freund3 in adsorption studies on single
crystals. Although there are exceptions, such as iron, nickel,
and alkali-promoted metals, there are still contradictions about
the spontaneous activation of CO2 by other metals. De la Peña
O’Shea et al.4 showed through density functional theory
(DFT) that CO2 can interact with cobalt surfaces. Their DFT
calculations predicted the structure sensibility and suggested
that cobalt nanoparticles terminated by (110) and (100)
surfaces were active for reactions involving CO2 and facilitate
its molecular dissociation.
In contrast to the activation of CO2, there are only a few
reports related to the activation of water on metals. Gong5
calculated the formation of water from OH and H on cobalt
surfaces, registering that the energy barrier was of the order of
155 kJ/mol on Co (111) surface. Moreover, Tang6 and
Gohake,7 in water formation on Cu (110) and Cu (111)
surfaces, found energy barriers of 60 and 120 kJ/mol,
respectively. They also observed lower energy values when a
monolayer of oxygen atoms covered these surfaces. This study
demonstrated the influence of oxygen at the surface on
activation of water molecules. Van Grootel8 studied the
activation of water on rhodium surfaces, considering the effect
Special Issue: Hans-Joachim Freund and Joachim Sauer Festschrift
Received: July 31, 2018
Revised: October 26, 2018
Published: November 12, 2018

© XXXX American Chemical Society A DOI: 10.1021/acs.jpcc.8b07416

J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C
of the surface topology and the presence of oxygen
preadsorbed on the activation process. For this system, authors
claim that the activation of water does not depend either on
the coordination atoms of rhodium surface and adsorbed
oxygen species. Liu et al.9 found by ambient pressure angle
resolved X-ray photoelectron spectroscopy that a polycrystal-
line Co surface adsorbed CO2 as carbonates and coadsorbed
CO2 and H2O molecules forming methoxy and formate
species.
Liao et al.10 studied photocatalytic water splitting using
sunlight over cobalt nanoparticles. The process generated
stoichiometric values of H2 and O2 without external electric
polarization or sacrificial reagents, which makes the reaction
simple and less expensive. Their results showing a quantum
efficiency of 5% under visible radiation without cocatalysts
were significant since the conventional catalysts for water-
splitting have quantum efficiencies of 0.1−0.3%. Cobalt oxide
nanoparticles-based materials are promising photocatalysts for
reactions involving the formation of oxygen and hydrogen11
and artificial photosynthesis.12
On the other hand, Bonenfant et al.13 showed that the
structure influences the adsorption of CO2 on zeolite materials.
The presence and type of exchanged cations, besides the
topology and ration Si/Al in the structure induced higher or
lower basicity and differentiated electrical fields in the space
cavities, provoking greater interaction with CO2 molecules,
since they have quadrupole moments. The capacity of zeolite
interaction with molecules depend on their topology, like
geometry and channel sizes, or shape selectivity which allows
to introduce nanoparticles and influence the molecules and
their transition states during the reaction. Faujasites have
approximately half of volume of the cell unit empty. The
presence of species with charge in the zeolite structure
produces electrical fields, which are enough to polarize
confined molecules in the nanopores for activation of the
molecules. In our previous study,14 we investigated the
interaction of water with sodium Faujasite and determined
the sodium and water sorption sites as a function of water
content. We found that water forms a physisorbed shell-like
layer inside the supercages, located at a distance of 2.8 Å from
the pore wall oxygens. For water-saturated Faujasite, this shell
had a surface density comparable to liquid water. The ion
exchange sites and ordered water found inside the super cage
are within this layer. These results, although limited by the use
of a conventional lab source X-ray diffraction, were confirmed
by a more detailed synchrotron study that found water forming
tetrahedral clusters and ice-like hexamers inside the Faujasite
cages15 and also investigated the thermal behavior of these
structures.
These results lead us to propose ordered nanoporous solids
as model supports for heterogeneous catalysts. We recently
showed14 by using in situ X-ray diffraction (XRD) that the
crystal symmetry of a zeolite allows us to investigate the
structure of adsorbate-nanoporous solid complex in atomic
detail. Inspecting the experimental electron density within the
pores allows investigating the structural features of physisorbed
matter inside the Faujasite nanochannels and nanocages. Our
proposal is not intended to be an alternative to the well-
established surface science methods to model heterogeneous
catalysts, but we think of it as a complementary approach. For
instance, by hosting active sites in the nanopores of a model-
support, we lack the chance to “see” them by the current
analytical surface science techniques. Therefore, we can
B DOI: 10.1021/acs.jpcc.8b07416
Article
include porosity and surface curvature effects in the model.
For a nanoporous solid, its representative “surface” is also part
of this bulk structure.
Heterogeneous model catalysts are traditionally made on
single crystal surfaces in ultrahigh vacuum (UHV), an ideal
condition to investigate them through current surface science
techniques. Although this approach provided an atomic-scale
understanding of many catalyst surfaces,16 it can be performed
in an environment quite different from that of industrial
processes, where pressures are at least 12 orders of magnitude
higher. In the language of model catalysis, this is called the
pressure gap.17 The role of porosity is also overlooked in these
model systems, since they are made on flat surfaces. The
topological difference between model and industrial catalysts is
called the material gap.17 For the ordered nanoporous model
system, there is no need to concern about the pressure gap,
since structural studies by X-rays and/or neutrons can be
performed under the typical pressure, temperature and flow
conditions of industrial processes.
We choose a Faujasite (Y) zeolite as a model support,
because its structure has a three-dimensional system of
interconnected nanosized channels and cavities that represent
50% (by volume) of the material. The sizes of these channels
and cavities are big enough for the transit of small molecules.
Faujasite is an aluminosilicate framework made by (Si,Al)O4
tetrahedra linked by their corners. This structure is built by
sodalite cages (6846), each one tetrahedrally linked to another
four by the hexagonal prisms (46). Each silicon ion substitution
by aluminum causes a unit negative charge in the framework.
In order to turn the structure electrically neutral, positive ions
sit at specific sites, these are called extra framework cation sites.
Faujasite topology and extra framework cation site nomencla-
ture18 are shown in Figure 1. Site I is at the center of the
Figure 1. Faujasite topology and extra framework sites.16
hexagonal prism at x = 0, y = 0, z = 0; site I′ is inside the
sodalite cage, near the aperture to the prism; site II at the
center of the hexagonal window between sodalite and
supercage and sites III and III′ are in the supercage, close to
the sodalite square window.
The objective of this work was to design and test
heterogeneous model catalysts made by dispersing cobalt
into a nanoporous zeolite Faujasite. We focus on in situ
methods that can reveal the structure of the adsorbate-
nanoporous complex during water and carbon dioxide
interaction with isolated cobalt ions and cobalt oxide
supported nanoparticles.
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C
Table 1. Chemical Composition and Formulas
%
sample Na2O CoO Al2O3
NaY 13.6(2) − 22.4(2)
CoY 4.23(3) 10.8(2) 22.6(2)
CoYt 9.93(2) 9.89(2) 21.3(2)
■ MATERIALS AND METHODS
The material used was a commercial sodium saturated
Faujasite in powder form (code CBV100) from Zeolyst
International.
The Faujasite, in powder form, was initially saturated with
sodium, is the model support called NaY. Cobalt incorporation
was made by ion exchange in an aqueous solution of cobalt
nitrate hexahydrate at a concentration of 0.1 molar, producing
a model named CoY. Another cobalt oxide−nanoparticle
model was produced by treatment of the CoY with 0.1 molar
sodium hydroxide solution, following the procedure suggested
by Tang et al.19 The latter was named CoYt.
Chemical Analysis. Chemical analyses were performed by
X-ray fluorescence (XRF), using a Rigaku RIX3100 sequential
X-ray spectrometer with Rh tube (4 kW). The zeolite powder
was pressed into a pellet and calcined at 500 °C in a muffle
furnace during 4 h before XRF analysis. The pellets were
maintained at ∼200 °C and in vacuum (13 mPa) before and
during XRF spectroscopy, in order to minimize adsorption of
species from the atmosphere.
In Situ X-ray Diffraction (Conventional and Synchro-
tron Sources). We use two experimental setups, a conven-
tional Cu Kα (λ = 1.54056 Å) source X-ray diffractometer and
another installed in the X-ray powder diffraction (XPD)
beamline at Brazilian Synchrotron Light Laboratory (LNLS).
Both have reaction chambers installed on the goniometers and
allow in situ data collection in the temperature range from
ambient to 900 °C under gas flow. The conventional setup has
been described in detail elsewhere.14 For the synchrotron data
collection, we tuned the beam energy to 7.5 keV, (λ = 1.6501
Å) which is slightly less than the CoK absorption edge (7.712
keV). This precaution avoided fluorescence by cobalt atoms.
We use a scanning linear Mythen 1024 channel detector and
vacuum path between both monochromator and chamber and
chamber and detector. Exiting gases were analyzed by a
quadrupole mass spectrometer Pfeiffer model Prisma 100.
Simulations and Rietveld refinements of the powder
diffraction data were performed by using the Fullprof Suite,20
Prima program21 for maximum entropy calculations, and
Vesta22,23 for crystal structure drawing.
Ultraviolet−Visible Absorption Spectroscopy (UV−
Vis). UV−visible diffuse reflectance spectroscopy (DRS)
analyses were performed using Varian Cary 5000 UV−vis
spectrometer with a reduction chamber coupled to a Harrick
diffuse reflectance accessory. Prior to analysis, the samples
were dried at 300 °C under He flow (30 mL/min) for 30 min
and then reduced at 300 °C, 400 or 500 °C using 50 mL/min
flow rate of a reducing mixture containing 10% H2/He (v/v)
for 1 h. The reflectance measurements were in the UV−visible
region (200−800 nm) scanning the whole range. The
contribution of the support was removed by dividing the
catalyst reflectance by the support reflectance measured in the
same reduction temperature, before calculating the “Kubelka−
Munk” function, F(R).
C DOI: 10.1021/acs.jpcc.8b07416
Article
SiO2 formula e/cell
63.8(3) Na56Al56Si136O384 6320
62.4(4) Co19Na18Al56Si136O384 6415
58.9(4) (Co3O4)4Co6Na43Al56Si130O384 6579
■ RESULTS AND DISCUSSION
Chemical analysis. The unit cell composition in Table 1
was calculated from the X-ray fluorescence data. The total
number of electrons in the unit cell (e/cell) was obtained by
summing the atomic numbers from the formula components.
The zeolite NaY is fully saturated by Na+ ions, as the atomic
ratio Na/Al determined of 1:1, resulting in electrostatic
equilibrium structure. The CoY structure shows partial
substitution of sodium by cobalt. The structural charge
neutrality is achieved if we suppose that the cobalt ions are
divalent. There are 19 divalent (Co2+) and 18 monovalent ions
(Na+) inside the Faujasite unit cell, which means 56 positive
charges to compensate for the 56 Al negative framework
charges. However, after treatment with NaOH, it was observed
the incorporation of an amount of Na and no significant loss of
cobalt ions (CoYt). In fact, the NaOH treatment caused the
precipitation of cobalt hydroxide in the zeolite cages,
considering the cobalt moved from the position of charge
compensation to form another structure. About six atoms of
cobalt per unit cell remained in the compensation sites, while
12 Co atoms migrate to form cobalt oxide inside the zeolite
channels. After calcination, we identified the presence of a
nanocrystalline Co3O4 phase by X-ray diffraction as we will
show forward in this text.
UV−Visible Spectroscopy. Figure 2 shows UV−visible
spectrum of CoY sample at different temperatures (30, 100,
Figure 2. UV−vis absorption spectra of CoY after heating in He.
200, 300, and 400 °C). As the temperature increases, besides
the band at 520 nm attributed to cobalt in octahedral
coordination (Co2+ (H2O)6), the spectrum exhibited bands at
534, 584, and 653 nm, assigned to Co2+ species in tetrahedral
coordination.23 Moreover, the band at 300 nm corresponds to
the Co2+ species in trigonal coordination.
Figure 3 shows the spectra of CoYt sample also at 30, 100,
200, 300, and 400 °C. The absorption band at 300 nm
observed at temperatures lower than 300 °C is assigned to the
trigonal coordination of cobalt.23 Above this temperature, the
absorption is significantly higher due to the transformation of
hydroxide (Co(OH)2) to the oxide (Co3O4).
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

during calcination, apart from those that belong to the Fd3̅ m

Figure 3. UV−vis absorption spectra of CoYt after heating in He.
Electron Density Map by in Situ XRD. The model
catalyst must have free nanopores for the transit of molecules,
so the inner surfaces were initially cleaned by calcination. In
sequence, we carried out the adsorption of probe molecules
(CO2 and H2O) and after, we performed the in situ reaction of
CO oxidation and water formation. Experimental electron
densities were obtained by in situ XRD for all these steps.
Calcination of the Models. The diffraction patterns of NaY,
CoY and CoYt after calcination are displayed in Figure 4. The
diffraction pattern of the model support NaY (Figure 4a) is
characteristic of a Faujasite structure with space group Fd3̅ m
and matches the corresponding reference pattern.24 As
calcination temperature increases, the cell parameter increases.
No additional peaks were observed in the diffraction patterns
space group assigned to the NaY structure, evidencing no
symmetry changes upon heating.
Besides the Fd3̅ m symmetry, the diffractogram of the CoY
model evidenced significant alterations of relative peak
intensities when compared to that of NaY. These are probably
due to the different electron densities between cobalt and
sodium and/or different site locations. The exchange of cobalt
ions preserves Faujasite lattice symmetry, and no other phase
appeared. Further, the diffraction peak widths of the NaY and
CoY are similar, meaning that the cobalt ionic exchange does
not alter the crystal domain sizes. In short, NaY and CoY
diffraction patterns exhibited a single crystalline phase with
Faujasite topology. Besides that, phase, CoYt showed
segregated nanoparticles (Co3O4) with mean domain size
around 30 Å. During calcination, cobalt hydroxide precipitated
in the larger cavities and as the temperature increased, Co3O4
nanoparticles were formed.
Noteworthy is that in the calcination step the intensities of
the peaks of all patterns increased significantly in the initial
Bragg angular range (2θ<16°), while for Bragg angles greater
the 40° remained unchanged. Therefore, the question then
arises is what kind of information can be obtained from the
intensities for these lower Bragg angles. It probably can be
related to the sorption of molecules in the structure of the
porous material.
The Rietveld refinement method25 was used to refine the
structure and to calculate the cell parameters, occupancy, and
positions or exchangeable cations. In the case of the CoYt
model, with the introduction of the Co3O4 phase, it was also
possible to determine the weight fraction of this phase and
estimate the mean crystallite domain size of the generated
oxides in the zeolite pores around 30 Å. Table 2 presents the
cell parameters, extra framework cation distribution of the
calcined samples.

Table 2. Rietveld Refined Parameters after Calcination at

400° C (λ = 1.65306 Å)a
NaY CoY CoYt
cell parameter a (Å) 24.7769(1) 24.5597(1) 24.7328(1)
scale factor S 7.69 × 10−5 7.21 × 10−5 7.13 × 10−5
cation I 7.5 Na 12.4 Co 4.0 Co
distribution
I′ 15.6 Na 3.8 Co 14.8 Na
II 29.4 Na 17.0 Na 28.4 Na
refinement RBragg 5.78 4.15 3.26
agreement
indexesb
Rp 10.2 9.80 10.3
Rwp 12.8 10.8 11.0
χ2 2.06 2.96 2.86
a
Atomic coordinates (origin is at inversion center −1): (I) 0, 0, 0;
(I′) x, x, x (x ∼ 0.06); (II) x, x, x (x ∼ 0.24). bRBragg = Σ|Iobs(hkl) −
I c a l c (hkl)|/Σ|I o bs (hkl)|; R w p = {Σw i [Y i (obs) − Y i (cal)] 2 /
Σwi[Yi(obs)]2}1/2; χ2 = (Rwp/Rexp) (see McCusker et al.25).

Although the Faujasite framework is relatively rigid, the unit

cell contracts when cobalt ions replace sodium, as the ionic
Figure 4. Rietveld refinement of the models (dark, experiments; red, radius of Co2+ is smaller than that of Na+. Therefore, the cell
calculated; and blue, difference of observed and calculated parameter of CoY is small than of NaY. The CoYt model
intensities): (a) NaY calcined at 400 °C, (b) CoY calcined at 400 showed cell parameter close to that of NaY. The incorporation
°C, and (c) CoYt calcined at 400 °C, highlighting the presence of of sodium during the treatment with hydroxide allowed the
Co3O4 nanoparticles (∼30 Å). replacement of the sodium in the cationic sites because a
D DOI: 10.1021/acs.jpcc.8b07416
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C
significant part of cobalt ions moved from these sites and
formed Co3O4 particles. These results agree with the chemical
composition consideration presented above.
The electron density of the NaY structure calcined at 400 °C
is displayed in Figure 5, showing the framework and the ion
Figure 5. Projection of electron density of NaY calcined at 400 °C.
exchange sites. Na occupies sites I, I′ and II in the super cage,
coordinated with three oxygens on the hexagonal sodalite in
connection with the Nano channels of the system. Some
residual water molecules were observed in the center of the
sodalite cage, as confirmed by diffuse density in this site.
Figure 6 displays the map density of CoY, showing more
concentration of electrons on sites I and I′ and less on site II,
Figure 6. Projection of electron density of CoY calcined at 400 °C.
confirming the preference of small and more charged ions to sit
at positions of higher coordination. Figure 7 shows the map
density of CoYt, which is very similar to the NaY, with less
density on sites I and I′ and higher electron concentration on
sites II.
Indeed, the calcination temperature was enough to clean the
channels, except for the sodalite cage, because of residual water
molecules inside, as mention above. As observed, the
nanochannels did not show electron density, except on sites II.
Figure 7. Projection of electron density of CoYt calcined at 400 °C.
E DOI: 10.1021/acs.jpcc.8b07416
Article
On the other hand, the cobalt is localized in a well-defined
position of the CoY model, confined in the center of the
hexagonal prism (I), coordinated with six framework oxygens.
This location is inaccessible to the small molecules. However,
the CoYt model contains Co3O4 nanoparticles on reachable
places for molecules with kinetic diameters less than 7.4 Å,
which is size of the smallest aperture of these channels. In fact,
the original NaY sample contains sodium in the sodalite cage
(sites I and I′) as well as on the super cage, coordinated with
the hexagonal ring (site II).
Interaction of Water Molecules with the Model Catalysts.
First, the sample was calcined and then cooled to room
temperature in the presence of He. Therefore, water molecules
were added through a saturator at a He flow of 50 mL/min and
the saturation temperature of 10 °C, during sequential intervals
of 10 min. The amount of water adsorbed was calculated from
the water vapor pressure at the saturation temperature.
We reported in our previous publication the results for the
interaction of water molecules on the NaY material.14 Here we
present the interaction of water with the CoY and CoYt
models. The in situ diffractogram for the CoY catalyst are
displayed in Figure 8.
Figure 8. Sequence of diffractograms during hydration of CoY.
As the cobalt model (CoY) interacts with water, the cell
parameters increased, causing the peaks shifted to smaller
Bragg angles. This behavior is quite different from the unit cell
contraction observed in NaY hydration.14
The dehydrated CoY patterns are similar to that from the
calcined material, indicating no residual molecules sorbed into
nanopores after cooling process. The refinement and the
electron density map exhibited a preferential occupation of site
(I) where cobalt is coordinated at a distance 2.224(8) Å from
the framework O1. The location of the Co2+ ion in the
hexagonal prism causes a T−O1 bond length elongation to
1.716(9) Å and a reduction in length of the T−O2 bond to
1.598(9) Å, creating a stress in the structure. In the dehydrated
material, site I′ is partially occupied, as well as site II, in the
window between the super cage and the sodalite cage. The
existence of these two vacancies probably facilitates the
transport of ions between the sodalite and the super cage
through the six-membered ring that communicates these
cavities.
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C
Figure 9 shows the sequence of density observed in the
regions of the hexagonal prism and the sodalite cage during the
Figure 9. Densities in the region between the hexagonal prism and
sodalite cage during progressive hydration of Faujasite exchanged with
cobalt: (a) CoY·0H2O; (b) CoY·50H2O; (c) CoY·100H2O; (d) CoY·
150H2O; (e) CoY·250H2O; (f) CoY·350H2O.
initial stages of hydration of CoY. In the first instance, the
interaction with water (Figure 9a), the cobalt migrated from
the center of the hexagonal prism (I) toward the sodalite cage
(I′) where it coordinates with three framework oxygens O1 and
one water molecule OW1 located in the center of the cage,
forming an approximate tetrahedral arrangement. The exiting
of the cobalt from the hexagonal prism relieves the T−O1 and
T−O2 stress, causing an increase of the cell parameter. With
the continued hydration of the CoY model, more water enters
in the sodalite. Compensation ions begin to coordinate with
the three framework oxygens and three OW1 water molecules.
The latter, totalizing four per cage, are at a mean distance of
2.77(2) Å between them and of 2.46(2) Å of the internal
cations to this cage (Figure 9.d).
Once the amount of water is enough to complete the film in
the nanopores, a fraction of the cobalt ions leaves the sodalite
cages, crosses the six-membered window and migrates to the
super cage, where it initially coordinates with three O3
framework oxygens and three water molecules OW2 in
approximately octahedral geometry. The electron density
map also shows the presence of cobalt in the second water
layer. In this context, it becomes structurally similar to the
Co(H2O)62+ ion in solution, but it is heterogeneous.
Indeed, the water causes the cobalt ions to leave the
positions in the smallest caves because of its coordination with
F DOI: 10.1021/acs.jpcc.8b07416
Article
the oxygen and to occupy positions in the super cage. The
water interacts with the solid, generating structures at the
interface that promotes the movement of the ions, going from
the smaller cages through the space of the super cages.
The pattern (not shown) of the CoYt model after saturation
presented two phases: Faujasite and Co3O4. The maximum
amount of physisorbed water in the supercage of the Faujasite
is lower than in NaY, as evidenced in Table 3, as cobalt is
present in the channels as Co3O4 nanoparticles, decreasing the
sorption capacity. The amount of water inside (OW5) is also
lower than those observed on NaY and CoY, which confirms
that Co3O4 occupies fully the spaces in the supercage.
Interaction of CO2 Molecules with the Model
Catalysts. The experimental procedure for CO2 adsorption
study was similar to that done on water absorption. The
samples were previously calcined at 400 °C for 4 h, and CO2
sorption was done at room temperature and atmospheric
pressure in the in situ X-ray diffractometer. The patterns were
obtained successively until saturation.
The diffractogram of NaY during the CO2 sorption (Figure
10) evidenced the preservation of the crystalline structure of
the solid and the presence of a scattered background, which is
related to the gas phase scattering. The refinement calculations
(Table 4) showed less variation of the cell parameters of this
model with the CO2 sorption, compared with the water
sorption parameters.
The electron density map of NaY·CO2 (Figure 11)
evidenced preferential sorption in the largest cage, between
both adjacent sites of Na II, forming apparently a bridged bond
like Na II···OCO···Na II. This configuration was also
observed for molecular sorption of CO2 on Faujasites,
according to Maurin26 and observed experimentally through
neutron diffraction by Wong-Ng et al.27 Since the linear
dimension of the CO2 molecule is 2.32 Å and the ionic
distance of Na−O around 2.4 Å, the two oxygen terminals of
CO2 must be simultaneously coordinated with the adjacent
compensation cations, and the distance between them should
be around 7.1 Å. In the NaY Faujasite, the distances between
nearest Na II are 9.6 Å, which do not allow the bridged
sorption of CO2 between these two cations.28
The electron density shape in the region between the cations
is apparently bent, which indicates that the CO2 molecule were
no longer linear. It is important to note that the electron
densities obtained by diffraction are a space and time average
in the various unit cells, so there is a probability that one of the
oxygens of the CO2 molecule coordinate with a Na II cation
leaving the other end of the molecule free to perform rotations,
and is preferably aligned in the direction of the electric field
produced by the adjacent cation. Statistically, the molecule can
coordinate with any of the two cations, so the curvature
appearance of the CO2 electron density format is probably
caused by the average of these configurations. As seen, the
sorption of CO2 in the CoY model cannot move cobalt out of
the smaller cavities. This model has greater amount of
vacancies in II, which turns the amount of sorption lower
than on NaY. On models NaY and CoYt, the site II is
completely occupied.
In Situ Reactions. The cobalt model catalysts were tested
with the CO oxidation in the diffractometer chamber, coupled
to a mass spectrometer, analyzing simultaneously the
diffraction patterns. The CO oxidation reaction was conducted
in the temperature range of 20 to 300 °C at 20 °C intervals.
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

Table 3. Rietveld Refinement of the Saturated Models with H2Oa

NaY CoY CoYt
cell parameter a (Å) 24.6641(1) 24.6370(4) 24.6190(3)
scale factor (×10−6) S 1.433(7) 0.79441(7) 0.75951(7)
population of sites (multiplicity) I(16) 7.8(2) − 2.4(2) Na
I′(32) 16.0(4) 17.2(3) Na 22.5(2) Na
II(32) 13.8(2) − 10.4(2) Na
II*(32) 6.8(4) 5.2(2) Co 9.1(3) Na
III(48) 11.1(2) 3.3(2) Co 8.6(2) Na
9.7(3) Co
ΣNa/ΣCo 55.5(6) 17.2(3) Na 53 Na
18.2(5) Co
population of sites (multiplicity) OW1(32) 8.5(2) 25.9(2) 15.7 (2)
OW2(96) 11.1(9) 12.0(2) 16.0(2)
OW3(96) 30.9(5) 29.5(3) 30.2(4)
OW4(96) 43.6(2) 38.0(2) 35.6(2)
OW5(32) 28.0(5) 16.7(5) 11.6(5)
Zs (e) 1601(23) 1424(15) 1060(11)
Σwater 282(5) 264 215
refinement agreement indexes RBragg (%) 3.60 4.76 4.26
RF (%) 3.32 4.58 4.48
Rp (%) 5.87 4.27 4.37
Rwp (%) 7.5 5.47 6.20
χ2 1.42 1.47 1.90
a
Atomic coordinates (origin is at inversion center −1): (I) 0, 0, 0; (I′) x, x, x (x = 0.06); (II) x, x, x (x = 0.24); (II*) x, x, x (x = 0.28); (III) x, x, z
(x = 0.375, z = 0.195); (OW1:) x, x, x (x = 0.166); (OW2) x, x, z (x = 0.265, z = 0.375); (OW3) x, x, z (x = 0.06, z = 0.442); (OW4) x, −x, 1/2 (x =
0.07); (OW5) x, x, x (x = 0.414).

Table 4. Rietveld Refinement Parameters for All Models

Saturated with CO2a
NaY CoY CoYt
cell parameter a (Å) 24.7441(1) 24.4689(1) 24.7286(1)
scale factor S 0.5366(6) 0.3827(5) 0.4055(7)
(×10−6)
population of I(16) 7.1(2) Na 9.4(3) Co 5.4(2) Na
sites
(multiplicity)
I′(32) 15.5(3) Na 8.1(3) Co 18.2(2) Na
II(32) 31.5(2) Na 18.1(3) Na 31.5(2) Na
ΣNaΣCo 54.1(4) Na 18.1(3) Na 55.1(3) Na
17.5(4) Co
OW1(32) 11.0(2) 17.8(2) 16.2 (2)
OW3(96) 17.6(4) 29.5(3) 10.9(4)
C(48) 55.7(3) 32.4(3) 47.4(5)
Figure 10. Rietveld refinement of NaY saturated with CO2. OC(96) 73.5(6) 50.3(6) 68.0(8)
ΣCO2 36.7(3) 25.0(3) 34.1(4)
refinement RBragg (%) 4.60 4.76 4.63
The composition of the blend was 2% CO, 1% O2 and the agreement
balance of He. indexes
Figure 12 shows the conversion for the CoY and CoYt RF (%) 3.94 4.13 3.89
catalysts for different temperatures. Note that at 200 °C the Rp (%) 12.5 11.1 10.1
conversion of the CoYt catalyst is almost 95%, while the CoY Rwp (%) 15.5 15.1 13.4
was practically inactive in this temperature range. When we χ2 1.62 1.47 1.53
compare these results with our experimental results on real
a
Atomic coordinates (origin is at inversion center −1): (I) 0, 0, 0;
catalyst,1 for the cobalt ferrite nanoparticles in the preferential (I′) x, x, x (x = 0.06); (II) x, x, x (x = 0.24); (III) x, x, z (x = 0.375, z
oxidation of CO, we observe that for the same temperature = 0.195); (OW1) x, x, x (x = 0.166); (OW3) x, x, z (x = 0.06, z =
0.442); (OW5) x, x, x (x = 0.414); (C) x, x, z (x = 0.375, z = 0.195).
(200 °C) the conversion was about 80%.
We calculated the electronic density map of the model
catalyst CoY, from the experimental reaction conditions that 300 °C. Indeed, these regions have limited access to the
evidenced the presence of cobalt on the center of the reactants, and therefore, it justifies the low activity of this
hexagonal prism and on the sodalite cavities throughout the model catalyst. On the other hand, the CoYt model catalyst
analyzed temperature did not change. Figure 13 displays the presents Co3O4 nanoparticles in the nanochannels, allowing
electron density map of CoY obtained for the CO oxidation at the reagents to reach the active phase.
G DOI: 10.1021/acs.jpcc.8b07416
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

Figure 11. Electron density map of NaY saturated with CO2.

Figure 12. Conversion of CO for model catalysts CoY and CoYt.

Figure 13. Electron density map of CoY obtained for the CO

oxidation at 300 °C. The cobalt atoms are shown inside the sodalite
cages.

Similarly, we measured the structure of CoYt using in situ X-

ray diffraction, for similar reaction conditions. The quantitative
analyses of the diffractograms were done using Rietveld
method and the domain sizes calculated at different temper-
atures, as presented in Table 5. It shows that particles remain
as spinel structure of Co3O4, without sintering.

Table 5. Mass Fraction and Domain Sizes of Co3O4 for

CoYt model Figure 14. Mass spectra and electron density maps during the
temperature (°C) mass fraction of Co3O4(%) domain size of Co3O4 (Å)
combustion of hydrogen for CoY and CoYt model catalysts.
30 7.1(4) 31(1)
100 7.3(5) 30(1) displayed in Figure 14a, without significant formation of H2O.
200 7.2(4) 31(1) However, the electron density map showed that during the
300 7.1(3) 33(1) reaction water condensates in the Faujasite nano channels, and
released above 200 °C, as shown in Figure 14a.
On the other hand, for the CoYt model the reaction
The combustion of hydrogen (2H2 + O2 → 2H2O) was occurred with the formation of water, released slowly at low
performed in a similar way for both models. The reaction temperature, or completely above 200 °C, without water
conditions in the in situ reactor chamber were: 3% H2+1.5%O2 accumulation in the channels (Figure 14 b). The reaction was
balanced with He, at flow rate 100 mL/min and temperatures completed at 300 °C without modification of the structure and
between 30 and 300 °C. the electron density, in good agreement with the experimental
results.

■
Mass spectra and electron density maps during the
combustion of hydrogen for CoY and CoYt model catalysts
are displayed in Figure 14. CONCLUSIONS
Results from mass spectrometer evidenced the consumption This study showed that it is possible to model heterogeneous
of reactants between 30 and 100 °C for the CoY model, as catalysts with active sites located in ordered nanoporous
H DOI: 10.1021/acs.jpcc.8b07416
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C
structures using special in situ X-ray diffraction with electronic
densities calculations. It also enabled the understanding of the
heterogeneous reactive−catalytic species as a whole. It also
allowed observing the mobility of the active phase that changes
depending on the conditions that the system is exposed for the
adsorption of CO2, the combustion of H2 and the CO
oxidation. The CoY and CoYt models showed different
performances in oxidation of carbon monoxide and the
combustion of hydrogen at low temperatures. The electronic
density obtained for the CoY model under the CO oxidation
reaction conditions showed the cobalt without interaction with
the reactants inside the smallest cages of the zeolite. At the
beginning of the combustion of hydrogen in CoY, the
electronic density evidenced accumulation of water inside the
nanochannels.
The Faujasite that contains cobalt (CoY) indicates the cell
parameter increases with the degree of hydration. Cobalt is
found in the sites within the hexagonal prism (I) and the
sodalite cage (I′) in the dehydrated material. In the first
moments of contact with the saturated mixture H2O/He there
is entry of water in sodalite. At the same time, cobalt migrates
from site I to site I′. The monolayer in the supercage channel
system only forms after sodalite saturation. The presence of
water also promoted the migration of Co2+ from II to positions
within the supercage. The model with nanoparticles (CoYt)
interacts with water in a way similar to the NaY support. The
oxide particles present did not change in quantity or size
during hydration.
■
AUTHOR INFORMATION
Corresponding Author
*(M.S.) Telephone: +55 21 39388352. E-mail: schmal@peq
coppe.ufrj.br.
ORCID
Carlos A. C. Perez: 0000-0003-3564-8109
Neuman S. de Resende: 0000-0002-1061-0772
Martin Schmal: 0000-0001-6816-523X
Notes
The authors declare no competing financial interest.
■ ACKNOWLEDGMENTS
This research used resources of the Brazilian Synchrotron
Light Laboratory (LNLS), an open national facility operated
by the Brazilian Centre for Research in Energy and Materials
(CNPEM) for the Brazilian Ministry for Science, Technology,
Innovations, and Communications (MCTIC). We acknowl-
edge the XPD beamline staff for the assistance during the
experiments.
■
REFERENCES
(1) Chagas, C. A.; de Souza, E. F.; de Carvalho, M.C. N. A.; Martins,
R. L.; Schmal, M. Cobalt ferrite nanoparticles for the preferential
oxidation of CO. Appl. Catal., A 2016, 519, 139−145.
(2) Izumi, Y. Recent advances in the photocatalytic conversion of
carbon dioxide to fuels with water and/or hydrogen using solar energy
and beyond. Coord. Chem. Rev. 2013, 257, 171−186.
(3) Freund, H.-J.; Roberts, M. W. Surface chemistry of carbon
dioxide. Surf. Sci. Rep. 1996, 25, 225−273.
(4) de la Peña O’Shea, V. A.; González, S.; Illas, F.; Fierro, J. L. G.
Evidence for spontaneous CO2 activation on cobalt surfaces. Chem.
Phys. Lett. 2008, 454, 262−268.
I DOI: 10.1021/acs.jpcc.8b07416
Article
(5) Gong, X.-Q.; Raval, R.; Hu, P. CO dissociation and O removal
on Co(0001): a density functional theory study. Surf. Sci. 2004, 562,
247−256.
(6) Tang, Q.-L.; Chen, Z.-X. Density functional slab model studies
of water adsorption on flat and stepped Cu surfaces. Surf. Sci. 2007,
601, 954−964.
(7) Gokhale, A. A.; Dumesic, J. A.; Mavrikakis, M. On the
mechanism of low-temperature water gas shift reaction on copper. J.
Am. Chem. Soc. 2008, 130, 1402−1414.
(8) van Grootel, P. W.; Hensen, E. J. M.; van Santen, R. A. DFT
study on H2O activation by stepped and planar Rh surfaces. Surf. Sci.
2009, 603, 3275−3281.
(9) Liu, Q.; Han, Y.; Cai, J.; Crumlin, E. J.; Li, Y.; Liu, Z. CO 2
Activation on Cobalt Surface in the Presence of H 2 O: An ambient-
pressure X-ray photoelectron spectroscopy study. Catal. Lett. 2018,
148, 1686−1691.
(10) Liao, L.; Zhang, Q.; Su, Z.; Zhao, Z.; Wang, Y.; Li, Y.; Lu, X.;
Wei, D.; Feng, G.; Yu, Q.; et al. Efficient solar water-splitting using a
nanocrystalline CoO photocatalyst. Nat. Nanotechnol. 2014, 9, 69−74.
(11) Nocera, D. G. The artificial leaf. Acc. Chem. Res. 2012, 45, 767−
776.
(12) Jiao, F.; Frei, H. Nanostructured cobalt oxide clusters in
mesoporous silica as efficient oxygen-evolving catalysts. Angew. Chem.,
Int. Ed. 2009, 48, 1841−1844.
(13) Bonenfant, D.; Kharoune, M.; Niquette, P.; Mimeault, M.;
Hausler, R. Advances in principal factors influencing carbon dioxide
adsorption on zeolites. Sci. Technol. Adv. Mater. 2008, 9 (1), 013007.
(14) Perez, C. A. C.; de Resende, N. S.; Salim, V. M. M.; Schmal, M.
Water interaction in Faujasite probed by in situ X-ray powder
diffraction. J. Phys. Chem. C 2017, 121, 2755−2761.
(15) Hu, M.; Hanson, J. C.; Wang, X. Structure and thermal stability
of (H2O)4 tetrahedron and (H2O)6 hexagon adsorbed on NaY zeolite
studied by synchrotron-based time-resolved X-ray diffraction. Ind.
Eng. Chem. Res. 2018, 57, 4988−4995.
(16) Ertl, G. Reactions at Surfaces: From atoms to complexity.
Angew. Chem., Int. Ed. 2008, 47, 3524−3535.
(17) Freund, H.-J.; Kuhlenbeck, H.; Libuda, J.; Rupprechter, G.;
Bäumer, M.; Hamann, H. Bridging the pressure and materials gaps
between catalysis and surface science: clean and modified oxide
surfaces. Top. Catal. 2001, 15, 201−209.
(18) Frising, T.; Leflaive, P. Extraframework cation distributions in X
and Y Faujasite zeolites: A review. Microporous Mesoporous Mater.
2008, 114, 27−63.
(19) Tang, Q.; Zhang, Q.; Wang, P.; Wang, Y.; Wan, H.
Characterizations of cobalt oxide nanoparticles within Faujasite
zeolites and the formation of metallic cobalt. Chem. Mater. 2004,
16, 1967−1976.
(20) Rodríguez-Carvajal, J. Recent advances in magnetic structure
determination by neutron powder diffraction. Phys. B 1993, 192, 55−
69.
(21) Izumi, F. Structure refinement based on the maximum-entropy
method from powder diffraction data; Transworld Research Network:
2002; Vol. 3, pp 699−726.
(22) Momma, K.; Izumi, F. VESTA 3 for three-dimensional
visualization of crystal, volumetric and morphology data. J. Appl.
Crystallogr. 2011, 44, 1272−1276.
(23) Verberckmoes, A. A.; Weckhuysen, B. M.; Pelgrims, J.;
Schoonheydt, R. A. Diffuse reflectance spectroscopy of dehydrated
cobalt-exchanged Faujasite-type zeolites: a new method for Co2+
siting. J. Phys. Chem. 1995, 99, 15222−15228.
(24) Treacy, M. M. J.; Higgins, J. B. Collection of simulated XRD
powder patterns for zeolites, 5th revised ed.; Elsevier: 2007.
(25) McCusker, L. B.; Von Dreele, R. B.; Cox, D. E.; Louër, D.;
Scardi, P. Rietveld refinement guidelines. J. Appl. Crystallogr. 1999, 32,
36−50.
(26) Maurin, G.; Llewellyn, P. L.; Bell, R. G. Adsorption mechanism
of carbon dioxide in Faujasites: grand canonical Monte Carlo
simulations and microcalorimetry measurements. J. Phys. Chem. B
2005, 109, 16084−16091.
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

(27) Wong-Ng, W.; Kaduk, J. A.; Huang, Q.; Espinal, L.; Li, L.;
Burress, J. W. Investigation of NaY zeolite with adsorbed CO2 by
neutron powder diffraction. Microporous Mesoporous Mater. 2013,
172, 95−104.
(28) Seo, S. M.; Lim, W. T.; Seff, K. Single-crystal structures of fully
and partially dehydrated zeolite Y (FAU, Si/Al = 1.56) largely Co2+
exchanged at pH 5.1. Microporous Mesoporous Mater. 2013, 170, 67−
74.

J DOI: 10.1021/acs.jpcc.8b07416
J. Phys. Chem. C XXXX, XXX, XXX−XXX