Characterization of iron catalysts prepared by chemical vapor deposition on nonzeolitic supports

Article abstract of DOI:10.1021/jp040068r

Chemical vapor deposition (CVD) of FeCl₃ has been used to deposit Fe³⁺ ions on the surface of sulfated zirconia (SZ) and silica-alumina (SA). Upon exposure to FeCl₃ vapor most Br??nsted acid sites and silanol groups are replaced by Fe, as evidenced by IR. With SZ the concentration of the acid sites and thus the retention of Fe increase with the sulfate loading up to a??45% of a monolayer, followed by an abrupt decrease at higher loadings. This indicates condensation of sulfate groups to polysulfates, which is in line with a lower number of Br??nsted sites per sulfate. Release of HCl due to the reaction of Br??nsted sites with FeCl₃ peaks at 85?°C for SZ but only at 345?°C for SA. After replacing Cl^- by OH^- and calcining, the materials were tested as De-NOx catalysts and characterized by temperature-programed reduction (TPR) with H₂ or CO. Mononuclear and dinuclear oxo-ions of Fe coexist with Fe oxide particles in calcined Fe/SA, resulting in a low selectivity for NO_x reduction. During reduction of Fe/SA up to 800?°C, a significant fraction of the Fe forms a chemical compound with SA, possibly an aluminate. In Fe/SZ the Fe dramatically increases the reducibility of the sulfate groups, from 57% partial reduction to SO₂ in the absence of Fe, to 90% deep reduction to S^2- ions in its presence. Formation of Fe sulfide is indicated by the enhanced sulfur retention upon reduction. Fe/SZ is active for NO_x reduction with isobutane. Catalysts with low Fe content that are prepared by controlled sublimation are superior to those prepared by impregnation. At 450?°C and GHSV = 30 000 h^-1, 65% of NO_x is reduced to N₂ in excess O₂.

Full text of DOI:10.1021/jp040068r

J. Phys. Chem. B 2005, 109, 2055-2063
2055
Characterization of Iron Catalysts Prepared by Chemical Vapor Deposition on Nonzeolitic
Supports†
Juan D. Henao, Bin Wen,§ and Wolfgang M. H. Sachtler*
Center for Catalysis and Surface Science, Institute for EnVironmental Catalysis, Northwestern UniVersity,
2137 Tech DriVe, EVanston, Illinois 60208
ReceiVed: January 27, 2004
Chemical vapor deposition (CVD) of FeCl3 has been used to deposit Fe3+ ions on the surface of sulfated
zirconia (SZ) and silica-alumina (SA). Upon exposure to FeCl3 vapor most Brønsted acid sites and silanol
groups are replaced by Fe, as evidenced by IR. With SZ the concentration of the acid sites and thus the
retention of Fe increase with the sulfate loading up to ∼45% of a monolayer, followed by an abrupt decrease
at higher loadings. This indicates condensation of sulfate groups to polysulfates, which is in line with a lower
number of Brønsted sites per sulfate. Release of HCl due to the reaction of Brønsted sites with FeCl3 peaks
at 85 °C for SZ but only at 345 °C for SA. After replacing Cl- by OH- and calcining, the materials were
tested as De-NOx catalysts and characterized by temperature-programed reduction (TPR) with H2 or CO.
Mononuclear and dinuclear oxo-ions of Fe coexist with Fe oxide particles in calcined Fe/SA, resulting in a
low selectivity for NOx reduction. During reduction of Fe/SA up to 800 °C, a significant fraction of the Fe
forms a chemical compound with SA, possibly an aluminate. In Fe/SZ the Fe dramatically increases the
reducibility of the sulfate groups, from 57% partial reduction to SO2 in the absence of Fe, to 90% deep
reduction to S2- ions in its presence. Formation of Fe sulfide is indicated by the enhanced sulfur retention
upon reduction. Fe/SZ is active for NOx reduction with isobutane. Catalysts with low Fe content that are
prepared by controlled sublimation are superior to those prepared by impregnation. At 450 °C and GHSV )
30 000 h-1, 65% of NOx is reduced to N2 in excess O2.
Introduction
CVD technique using FeCl3 vapor interacting with the acid sites
of H-MFI leads to γ ) 1.0, a value easily obtained experimen-
tally, as IR data confirm that all Brønsted protons of H-MFI
are swiftly replaced by [FeCl2]+ ions:
Heterogeneous catalysts, exposing transition metal ions as
active sites on high-surface-area supports, are usually prepared
by impregnation, coprecipitation, or ion exchange from aqueous
solution. In recent years two novel techniques, chemical vapor
deposition (CVD) and solid state ion exchange (SSI), have been
added to this toolbox.1,2 They have mainly been applied to
zeolites, for instance, Fe/MFI and Co/MFI (the MFI zeolite is
often called ZSM-5), which are remarkably active and selective
catalysts for the selective reduction of NO and NO2 to N2 with
hydrocarbons as reductants.3-6 As these catalysts operate in an
atmosphere containing O2 and H2O vapor in large excess over
NOx, the demands on activity and selectivity are very high. Since
it would be desirable to achieve a similar performance with
supports of higher thermal stability, the question arises whether
the same preparation techniques are also applicable to high-
surface-area supports such as silica-alumina or sulfated zirco-
nia. The present paper addresses this issue.
For zeolites it is certain that catalysts prepared by CVD or
SSI expose sites that are qualitatively different from those
obtained by ion exchange from solution or by impregnation.3,7,8
As the number of exchange positions in a zeolite is given by
the number of Al-centered tetrahedra, N(Al), conventional ion
exchange of an Fe3+ solution with a Na-zeolite is limited to a
ratio of Fe to Al ions N(Fe)/N(Al) ) γ ) 1/3. In contrast, the
H+zeol + (FeCl3)vapor f [FeCl2]+zeol + HClvapor
(1)
As most zeolites have lattice imperfections and silanol groups
leading to additional acid sites, the experimental value of γ will
exceed unity. This confirms that for the CVD technique the
strong Brønsted acidity characteristic of zeolites is not required,
which is an additional argument for exploring the potential of
applying this preparation technique to less strong solid acids.7
Exchange of the Cl- ions against OH- groups by hydrother-
mal treatment transforms [FeCl2]+ ions to Fe-hydroxo ions
which can reorganize themselves during calcination to oxo-ions,
Fe oxide nanoparticles, and bare ions. Techniques exist to limit
the formation of the usually undesirable oxide particles far below
the amounts obtained by conventional impregnation.3,9 Oxo-
ions and oxide particles are also formed on amorphous supports
such as Al2O3 when using traditional impregnation, but they
readily form solid solutions or aluminates which make it
impossible to distinguish the oxygen ligated to the active metal
from that of the support. For CVD-prepared Fe/MFI catalysts,
however, isotope exchange with gaseous 18O2 distinguishes
clearly between the oxygen atoms in oxo-ions and those of the
zeolite lattice.10 In addition to the higher value of γ, this
propensity that zeolite-supported oxo-ions have their own
oxygen ligands which are not shared with the supporting oxide
could be a major cause of the specific catalytic qualities of CVD-
prepared transition metal/zeolite catalysts. An objective of the
† Part of the special issue “Michel Boudart Festschrift”.
* Corresponding author. Phone: (847) 491-5263. Fax: (847) 467-1018.
E-mail: wmhs@northwestern.edu.
§ Present address: Nano Dynamics Inc., 901 Fuhrmann Blvd., Buffalo,
NY 14203.
10.1021/jp040068r CCC: $30.25 © 2005 American Chemical Society
Published on Web 05/29/2004

2056 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Henao et al.
present study is to find whether similar catalytic qualities can
be attained with nonzeolite systems by preparing them with the
CVD technique.
For comparison, two Fe/SZ samples were prepared by
impregnation. In this case, iron was loaded as iron(II) oxalate
dihydrate (Aldrich, 99%) dissolved in water; 0.3 mL of this
solution was dosed per gram of solid. The materials were dried
overnight and calcined in an O2 flow at 600 °C for 2 h. These
samples will further be labeled as Fe/SZ-1-I and Fe/SZ-5-I;
their Fe content is equal to that of the CVD-prepared samples
Fe/SZ-1 and Fe/SZ-5, respectively.
The notable Brønsted acidity of silica-alumina makes this
material suitable for anchoring iron by CVD. Varying the silica/
alumina ratio permits modification of its acidity and catalytic
activity.11-13 Sulfated zirconia is an even stronger acid; both
Lewis and Brønsted acid sites are exposed. One OH group per
sulfate is bound through its H atom to a surface oxygen,14 while
sulfur retains its oxidation state of +6.15 Sulfated zirconia is
usually prepared by impregnation of Zr(OH)4 with (NH4)2SO4
or H2SO4 followed by drying and calcination at 500-750 °C.16
The structure, texture, and catalytic activity of SZ depend on
the pH used to precipitate Zr(OH)4, the source of the sulfate
and its concentration in the solution, and the calcination
temperature.17-19
In the present work, the CVD technique is used in the
preparation of Fe/SZ and Fe/SA. Temperature-programed
reduction, FTIR, and XRD have been used to characterize the
anchored iron and other structural elements related to the
catalytic activity. Catalytic tests were done on the reduction of
NOx with isobutane in the presence of excess oxygen and water
vapor.
Characterization. FTIR was used to determine the effect of
iron deposition on the acid sites; data were acquired with a
Thermo Nicolet Nexus 670 FTIR spectrometer equipped with
a DTGS detector by recording 150 scans with a resolution of 1
cm-1. Sample wafers of 8-10 mg/cm2 were placed into a flow-
through quartz cell sealed with NaCl windows. They were
calcined in O2 at 450 °C before analysis; spectra of materials
at different stages of preparation were collected in the transmit-
tance mode in a dry He flow at room temperature.
In order to identify the oxidation state of the iron, H2-TPR
and CO-TPR measurements were performed. For H2-TPR the
samples were put in a quartz reactor, calcined in flowing O2 at
500 °C for 2 h, and then cooled to room temperature. Flowing
oxygen was then replaced by Ar for 1 h before admitting the
reducing mixture (5% H2/Ar at 40 mL/min). The sample was
heated under this reduction gas from 25 to 800 °C at a rate of
8 °C/min, while the H2 consumption was monitored with a Gow-
Mac thermal conductivity detector (TCD) located downstream
of a cold trap. For Fe/SZ an n-pentane/N2 trap (-131 °C) was
used to retain H2S and SO2, whereas a dry ice trap (-78.5 °C)
was used to retain the water generated by reduction of Fe/SA.
CO-TPR was only done with Fe/SA; the sample was first
calcined in O2 at 500 °C for 2 h, cooled to room temperature,
purged with He, and then contacted with a CO/Ar (2%) flow at
75 mL/min. A Dycor M200 mass spectrometer (MS) was used
to monitor the consumption of CO (m/e ) 28, 14) and the
evolution of CO2 (m/e ) 44, 28), H2O (m/e ) 18), and H2 (m/e
) 2), while the temperature was increased from 25 to 760 °C
at a ramp of 8 °C/min. The contribution of CO2 to the m/e )
28 signal was subtracted before calculating the CO consump-
tion.22 The TCD and MS detectors were calibrated for H2 and
CO consumption by using known amounts of CuO. Tempera-
ture-programed oxidation (TPO) in an O2/He (5%) mixture was
carried out between 25 and 760 °C to quantify any carbon
deposits.
Experimental Section
Catalyst Preparation. Commercial silica-alumina (SA)
(Aldrich, 76% SiO2, 13% Al2O3) with a BET surface area of
475 m2/g has been used as a support. Sulfated zirconia (SZ)
was prepared by stirring a slurry of 32 g of Zr(OH)4 (Magnesium
Elektron Inc., 70.4% ZrO2; BET area 100 m2/g) and 25 mL of
an aqueous solution of sulfuric acid (Mallinckrodt, 96.1%) at
80 °C until a dry powder was obtained (usually after 1 h). Seven
samples were prepared by varying the H2SO4 concentration
while keeping the amount of Zr(OH)4 constant. The materials
were dried overnight at 110 °C, then calcined in O2 at 600 °C
for 4 h. At this temperature tetragonal ZrO2 is formed, while
decomposition of the SO4 groups is prevented.20,21
Temperature-programed surface reaction (TPSR) runs were
done with SA and SZ to establish the temperatures for maximum
rate of iron deposition from the FeCl3 vapor. An amount of 1
g of either support (SA or SZ) was loaded in a U-shaped quartz
reactor and heated for 1 h at 550 °C in an Ar flow. After cooling
the sample to room temperature, 0.5 g of FeCl3 was loaded in
the opposite side of the U-shaped reactor. Under a flow of Ar
through both beds the reactor was heated from 25 to 500 °C
with a ramp of 5 °C/min. The HCl produced was quantified by
using a Dycor M200 mass spectrometer (MS).
SA and seven SZ samples of different sulfate content were
submitted to isothermal CVD of FeCl3 at the previously
determined optimum temperature. Typically, 10 g of sample, 3
g of FeCl3, and 120 mL/min of Ar were used. While [FeCl2]+
ions were formed at the surface, the released HCl was absorbed
in an NaOH solution. The reaction stopped when all reactive
protons were replaced by [FeCl2]+ ions; the excess FeCl3 vapor
passed through the solid bed without being retained.
The Fe-loaded samples were washed with 2-fold deionized
water to replace Cl- ions by OH- groups, then dried at 110 °C
and calcined in flowing O2 at 500 °C (600 °C for SZ) for 2 h.
Due to the different sulfate loads in each of the SZ samples,
their capacity to retain iron during sublimation was found to
be different. We will refer to them as Fe/SZ-j, with j ) 1, 2,
..., 7. Inductively coupled plasma atomic emission spectroscopy
(ICP-AES) shows that the Fe/S ratio in the Fe/SZ samples varies
between 0.10 and 0.75, whereas the Fe/Al ratio in Fe/SA was
∼0.25.
XRD patterns were recorded in a Rigaku DMAX diffracto-
meter, using Ni-filtered Cu KR radiation (λ ) 1.540 598 Å);
scanning was conducted from 2θ ) 5° to 2θ ) 70° at a scan
step of 0.1°. Surface areas were determined with a Micromeritics
ASAP 2010 liquid nitrogen analyzer, after degassing the samples
overnight at 0.1 Pa and 350 °C.
Catalytic Tests. Catalytic reduction of NOx with isobutane
was studied in a microflow reactor using a 200 mL/min feed
gas containing 0.1% of a NO + NO2 mixture (from a cylinder
that originally contained pure NO2), 0.1% i-C4H10, 2% O2, and
1% H2O. The actual NO/NO2 ratio at the catalyst depends, of
course, on the temperature and will be higher at high temper-
ature. Catalytic data will, therefore, be expressed in terms of
NOx conversion and N2 yield, with NOx ) NO + NO2. The
catalyst was either 550 mg of Fe/SZ or 200 mg of Fe/SA,
resulting in a GHSV of 30 000 h-1. In a typical test, the sample
was loaded in a U-shaped quartz reactor, calcined in flowing
O2 at 500 °C for 2 h, then cooled to 150 °C before feeding the
reaction mixture. The effluent gas was monitored while the
temperature was raised stepwise from 250 to 550 °C. Sampling
was performed 30 min after reaching each preset temperature.

Iron by CVD
J. Phys. Chem. B, Vol. 109, No. 6, 2005 2057
Figure 1. HCl generated by CVD of FeCl3 onto (a) SZ-1 and (b) SA as determined by TPSR experiments. MS signal normalized per unit surface
area.
Reaction products were analyzed on-line by GC-TCD using
Alltech 13X molecular sieve and Porapak Q columns. To
prevent water condensation, all lines were kept at 80-100 °C
by means of heating tapes. An ice-water trap was positioned
before the GC to condense most of the water vapor and protect
the packed columns. The N2 yield was calculated from the N2
formation, while the hydrocarbon combustion was quantified
from the amounts of CO and CO2.
Results
To determine the most favorable temperatures for deposition
of FeCl3 onto SZ or SA, preliminary TPSR runs were carried
out by mass spectrometrically monitoring the HCl that is
produced when FeCl3 vapor reacts with surface protons. The
results are shown in Figure 1; the features are characteristic of
Figure 2. Effect of sulfur content on (a) sulfated zirconia capacity to
the reactivity of the acid sites toward FeCl3. Sulfated zirconia
retain iron and (b) sulfate surface coverage.
displays two types of H+ sites: one type is very active at a
relatively low temperature; sites of the other type are less active
but much more numerous, giving the dominant TPR peak at
285 °C. The width of the feature between 60 and 200 °C
suggests a rather wide distribution in reactivity of the former
sites. The TPSR profile of SA is dominated by one large,
asymmetric feature, peaking at 345 °C; the shoulder at 300 °C
is representative of the Brønsted sites.
TABLE 1: Chemical Composition of Catalysts Prepared by
CVD and Impregnation
molar ratio
sample
wt (%) Fe
Fe/S
Zr/S
Fe/Al
Fe/SZ-1
Fe/SZ-2
Fe/SZ-3
Fe/SZ-4
Fe/SZ-5
Fe/SZ-6
Fe/SZ-7
Fe/SZ-1-I
Fe/SZ-5-I
Fe/SA
0.19
0.26
0.46
0.50
0.61
0.64
0.74
0.16
0.59
2.41
0.10
0.14
0.25
0.75
0.54
0.44
0.45
0.05
0.19
19.7
24.1
24.1
65.4
39.2
30.4
23.6
12.4
13.2
On the basis of these results, sulfated zirconia and silica-
alumina were exposed to FeCl3 vapor at 285 °C and 345 °C,
respectively. Saturation was attained after 3 h for SA and after
4-5 h for SZ. The capacity of SZ samples to retain iron from
the vapor depends on their sulfate load. Once saturation is
achieved, no further uptake of iron or change in color is
registered, even if exposure to the FeCl3 vapor is continued for
4 additional hours. Values of the ICP-AES chemical analysis,
including the impregnated samples, are compiled in Table 1.
Figure 2 shows how the sulfate content in SZ is crucial for
its capacity to retain iron. The iron uptake increases with sulfate
content; it reaches a maximum at 30 × 10-5 mol SO42-/g
catalyst (0.96% S) but precipitously decreases at higher sulfate
concentrations. At this stage the surface coverage of zirconia
with sulfate is 45% of the monolayer coverage calculated for a
size per sulfate group of 0.25 nm2. Higher coverage of the
surface with sulfate results in much lower retention of Fe from
FeCl3.
0.24
band in different spectra reflect, in part, different thicknesses
of these wafers and should not be used for quantitative
evaluations. The bands corresponding to the OH groups vanish
after CVD, indicating complete substitution of H+ by [FeCl2]+,
both in SZ and SA. Trace a in Figure 3A shows the spectrum
for SZ-1 after calcination at 450 °C in flowing O2. It exhibits
the characteristics typical of sulfated zirconia and metal-
promoted sulfated zirconia;15,23,24 the band at 3645 cm-1 is
assigned to its strong Brønsted acid sites.25,26 In Figure 3B trace
a shows the spectrum of SA, where the band at 3743 cm-1
represents free hydroxyl groups, while the broad band between
3690 and 3530 cm-1 is due to the Brønsted acid sites.11,27,28
Exposure to FeCl3 affects both the Brønsted sites and the
IR results are shown in Figure 3. As each spectrum was
measured with a different wafer, different intensities of a given

2058 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Henao et al.
Figure 5. XRD patterns of (a) SZ-1, (b) SZ-7, (c) Fe/SZ-1, and (d)
Fe/SZ-7 after calcination in O2 flow at 600 °C.
TABLE 2: H2 Consumption in TPR Experiments over SZ-1,
SZ-7, Fe/SZ-1, and Fe/SZ-7 Samples
Fe
S
6+ to S4+
(%)
S
6+ to S2- peak temp
(%) (°C)
sample (wt %) H2/(S+Fe)
SZ-1
SZ-7
Fe/SZ-1 0.19
Fe/SZ-7 0.74
0
0
2.1
2.1
3.2
2.8
57
57
10
11
43
43
90
89
738
732
499
486
reduction of the sulfate groups, as previous experiments showed
that no reduction of unsulfated zirconia takes place in this
temperature region.31 In the presence of Fe this peak is shifted
to lower temperature by more than 200 °C, indicating that Fe
strongly catalyzes the reduction of sulfate groups. The peak
positions for Fe/SZ-1 and Fe/SZ-7 (see Table 2) suggest that
the reduction temperature varies with the Fe load and is lower
for higher iron loads. This has been confirmed for other Fe/SZ
samples not shown here.
Figure 3. Infrared spectra of SZ-1 (Panel A) and SA (Panel B): (a)
fresh, (b) after CVD, and (c) after CVD and hydrolysis. All samples
were calcined at 450 °C in O2 flow.
Integration of profiles a and b, and those not shown of SZ-7
and Fe/SZ-7, provides the data in Table 2. While SZ-1 and
SZ-7 consume roughly four H atoms per S atom (H2/S ) 2.1),
the hydrogen consumption in Fe/SZ-1 and Fe/SZ-7 is much
higher, suggesting complete reduction of the Fe3+ ions and most
of the sulfate to sulfide. As previous work showed that SO2
and S2- ions are preferred reduction products of sulfate groups
in SZ, the results are rationalized by assuming that no other
reduction products are formed. This leads to the result that
sulfate groups in SZ-1 and SZ-7 are reduced by ∼57% to SO2
and by 43% to sulfide ions. Upon applying the same assumption
to the samples loaded with iron, it follows that ∼90% of the
surface sulfate groups are reduced to S2- ions.
XRD patterns for iron-free and iron-doped SZ-1 and SZ-7
in Figure 5 indicate that both supports were present in the
tetragonal modification (2θ ) 30.3°); a small amount of
monoclinic zirconia (2θ ) 28.7°) was also detected in Fe/SZ-
1. Formation of FeSO4 (2θ ) 18.1°, 19.8°), Zr(SO4)2 (2θ )
13.6°, 20.5°), hematite (2θ ) 33.3°, 35.7°), wu¨stite (ferrous
oxide; 2θ ) 36.1°, 41.9°), or magnetite (2θ ) 30.1°, 35.4°)
was not apparent from the XRD data. The CVD method is
known to prevent the formation of large iron oxide clusters.
The XRD patterns for the samples prepared by impregnation,
that is, Fe/SZ-1-I and Fe/SZ-5-I, have the same characteristics
as those of Figure 5.
Figure 4. H2-TPR profiles of (a) SZ-1, (b) Fe/SZ-1 after calcination
in O2 flow at 500 °C.
terminal silanol groups, lowering their band intensity to almost
zero, as is clear from traces b in Figure 3. Subsequent washing
and calcination regenerates the OH sites. In the case of sulfated
zirconia, a new band arises at 3770 cm-1 (see Figure 3A, trace
c) which is similar to that of the terminal OH groups (3780
cm-1) on sulfate-free zirconia.29,30 This suggests that upon
washing some OH groups are formed that are similar to those
on sulfate-free zirconia, which is a weaker acid than SZ.
Figure 4 shows the H2-TPR profiles of SZ-1 and Fe/SZ-1;
they are representative of the set of sulfated zirconias used in
this study. The profile of SZ-1 displays one large feature starting
at 500 °C and peaking at 738 °C; this peak is attributed to
H2-TPR profiles of Fe/SA are shown in Figure 6. Profile A
displays six overlapping peaks, some of which may be due to
stepwise reduction of a given species in the original sample.
Blank experiments confirmed the absence of any TPR peaks

Iron by CVD
J. Phys. Chem. B, Vol. 109, No. 6, 2005 2059
Figure 7. CO consumed (Panel A) and CO2 generated (Panel B) by
CO-TPR of fresh Fe/SA. Fresh Fe/SA calcined in O2 flow at 500 °C
before reduction.
Figure 6. H2-TPR profiles of fresh Fe/SA (Panel A) and CO-reduced
Fe/SA (Panel B). Fresh Fe/SA calcined in O2 flow at 500 °C before
reduction.
result of such a second H2-TPR run is shown in Figure 6B.
The difference between profiles A and B in that figure illustrates
the abundance of oxo-ions in calcined Fe/SA. A tentative
assignment of the peaks will be presented in the Discussion.
The materials were probed as catalysts in the reduction of
NO2 with i-C4H10 in a “wet” feed. Figure 8A shows the
temperature dependence of the N2 yield for the samples prepared
by CVD while Figure 8B compares the activities of Fe/SA and
the iron-impregnated samples. Fe/SA is found to have a very
low catalytic activity for this reaction, but some of the Fe/SZ
samples display higher N2 yields. While yields are lower than
those observed with good Fe/MFI catalysts, it is of interest that
the Fe/SZ with the lowest Fe content converts 65% of NOx to
N2 at 450 °C. The N2 yield at 300 °C is comparable to that
achieved with Fe/MFI. This could mean that in Fe/SZ-1 some
Fe is present in a structure similar to that identified for Fe/
MFI, prepared by sublimation, but more research will be needed
to check this. The selective nature of the NOx reduction is also
manifest from the CO2/CO ratio in the product as evidenced
by parts A and B of Figure 9. The selective catalyst Fe/SZ-1
produces a significant amount of CO; even at 550 °C the CO2/
CO ratio remains near unity. This is typical for selective catalysts
with very low concentration of Fe oxide particles which are
selective for deep oxidation to CO2. By contrast, over the SZ-
supported catalyst with the highest metal loading, Fe/SZ-7,
combustion to CO2 is significant. Over this catalyst the COx
contains less than 30% CO at 400 °C and even less at higher
temperature, signaling a catalytic signature reminiscent of Fe
oxide. Comparison of Fe/SZ-1 and Fe/SZ-1-I is quite relevant.
Table 1 shows that both have a very similar iron content, but
for SA. Formation of iron oxide nanoparticles that are not
detectable by XRD is indicated by the TPR data. To get more
information on the nature of the iron species, we decided to
also use CO-TPR, as this method discriminates between Fe oxo-
ions (and oxide particles) and bare Fe ions. Reduction of bare
Fe ions by H2 implies removal from Fe of positive charge which
ends up in protons. Reduction of neutral oxide particles implies
removal of oxygen only; this can be done either by hydrogen
or by carbon monoxide. Reduction of Fe oxo-ions requires
removal of both oxygen and charge. These two phenomena can
be separated by first using CO-TPR to remove oxygen, followed
by H2-TPR to remove the charge.
Figure 7 shows the CO consumption and the CO2 formation
during CO-TPR. The results indicate the presence of large
amounts of oxo-ions, indicating dinuclear (or multinuclear)
oxygen-bridged species in which two Fe3+ are easily reduced
to Fe2+ by removal of the bridging oxygen. An inherent
complication of the CO-TPR technique is that water gas shift
and the Boudouard reaction will also consume CO and produce
CO2. In the present work the temperature was kept below the
value where the Boudouard reaction is known to take place.
Formation of H2 was monitored, and it was found that the peak
at 527 °C is largely caused by water gas shift; this peak will
further be disregarded in the analysis of the oxidation states of
iron. The first two peaks correspond to true reduction of iron
oxide particles and iron oxo-ions.
As reduction of Fe oxide particles with CO will give Fe0,
but that of Fe oxo-ions will create bare Fe ions, one can
discriminate between both types by H2-TPR after CO-TPR. The

2060 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Henao et al.
Figure 8. N2 yield for Fe/SZ samples prepared by CVD (Panel A)
and Fe/SA and Fe/SZ-I-1 (Panel B). Reaction mixture: 0.1% NO2,
Figure 9. CO (Panel A) and CO2 (Panel B) yield for samples prepared
by CVD and impregnation. Reaction mixture: 0.1% NO2, 0.1% i-C4H10,
0.1% i-C4H10, 2% O2, 1% H2O; GHSV ) 30 000 h-1
.
2% O2, 1% H2O; GHSV ) 30 000 h-1
.
Figure 8 reveals that below 350 °C the catalyst prepared by
sublimation gives a N2 yield near 50%, while the N2 yield of
the impregnated sample remains below 20% at this temperature.
a small number of sites release HCl at much lower temperature.
The pattern reflects the microscopic heterogeneity inherent to
the impregnation process of zirconium hydroxide with H2SO4.
A new finding is that a very high surface coverage of sulfate
groups on zirconia leads to an abrupt drop of its propensity to
retain Fe from FeCl3. This suggests that at surface coverages
above 45% neighboring sulfate groups undergo condensation
reactions of the type
Discussion
Whereas a very high acid strength seems not to be a
prerequisite for the deposition of Fe from FeCl3, the present
data show that acidity does matter. Deposition of Fe and release
of HCl take place at a lower temperature on the support with
the higher acidity, sulfated zirconia, than on silica alumina. The
former solid acid exposes a small number of strong acid sites
that are apparently responsible for Fe retention and HCl release
below 100 °C. Silica-alumina is able to retain more Fe at higher
temperature, in agreement with its high surface concentration
of active OH groups. The IR spectra in Figure 3B confirm that
the terminal hydroxyl groups in SA contribute most to the
production of HCl during CVD. These groups appear at a higher
wavenumber (Figure 3B) and release HCl from FeCl3 at higher
temperature (Figure 1). Integration of trace a and of the
deconvoluted trace b in Figure 1 shows that SZ-1 has roughly
twice as many Brønsted sites per unit area as SA; the surface
density of silanol groups on SA is roughly equal to that of the
Brønsted sites on SZ-1. The presence of two well-defined peaks
for sulfated zirconia in Figure 1 indicates that two kinds of
Brønsted sites are present. It has been proposed14,32 that the
remarkable acidity of SZ is due to the existence of Brønsted
and Lewis site pairs; their density reaches a maximum at a
Brønsted/Lewis ratio near unity.33 In addition to these groups,
-S-OH + HO- S- f -S-O-S- + H2O
(2)
This type of chemistry is known for phosphate groups, which
change from ortho via meta to pyrophosphate, and for chromates
giving dichromate. We are not aware of previous reports of such
condensation of adjacent sulfate groups decorating the surface
of zirconia. That such condensation takes place is also in line
with the IR data (not shown) of SZ-1 and SZ-7, that is, two
materials with a high and a relatively low sulfate loading (see
Figure 2). With the use of the band at ∼1025 cm-1 of the OS-O
stretching mode34 as an internal standard, the relative concentra-
tion of Brønsted sites (∼3645 cm-1) per sulfur can be calculated
for both materials. This estimate shows that the number of
Brønsted sites per sulfur atom is lower by a factor of 2 for SZ-1
(with high sulfate loading) than for SZ-7 (see Table 5),
indicating that appreciable condensation of adjacent sulfate
groups takes place at the sulfate coverage where the Fe retention
from FeCl3 decreases precipitously.
The effect of iron on the reduction of sulfated zirconia is
rather dramatic and very similar to that of platinum in Pt/SZ.31

Iron by CVD
J. Phys. Chem. B, Vol. 109, No. 6, 2005 2061
TABLE 3: Normalized H2 and CO Consumption from TPR
Experiments over Fe/SAa
the overall reduction will increase with decreasing distance
between Fe (or Fe-sulfide) clusters and sulfate groups and thus
with increasing Fe loading, as is indeed observed.
H2/Fe or CO/Fe (mol ratio)
This could have important consequences for the durability
of such catalysts. It is well-known that sulfate groups protect
the metastable tetragonal phase of zirconia against transforma-
tion to the more stable, but for catalysis less desirable,
monoclinic modification.16,17,20 If catalysts are reduced and
subsequently reoxidized, the sulfur which is retained can
continue to exert its protective function. Likewise, the acid sites
associated to sulfate groups will be regenerated. If, however,
sulfur escapes as SO2 during a reducing cycle, the protective
function of the sulfate and its effect on catalyst acidity will be
permanently lost.
experiment
sample
fresh
CO reduced
peak I peaks II-IV peaks V-VI
H2-TPR
H2-TPR
0.05
0.51
0.28
0.41
0.44
peak I
0.06
peak II
0.23
peak III
CO-TPR
fresh
a Refer to Figures 6 and 7 for peak labeling.
TABLE 4: Effect of H2 Reduction on Sulfur Content of
SZ-1 and Fe/SZ-1
before H2-TPR
Fe/S S/Zr
after H2-TPR
Fe/S S/Zr
sample
As to Fe/SA, six deconvoluted H2-TPR peaks in Figure 6
indicate six discernible states of Fe, some of which may be the
result of stepwise reduction of one original species. The CO-
TPR data show that after calcination most of the iron is present
as oxide particles and/or oxo-ions. As to their nature, it is known
that dinuclear ions [HO-Fe-O-Fe-OH]2+ can be easily
reduced to [HO-Fe-0-Fe-OH]2+, but mononuclear [Fed
O]+ ions, if isolated and immobile, would give highly unstable
Fe+. We therefore assume that such ions are only reducible if
they can migrate and dimerize or, more probably, dock on other
Fe oxide or Fe metal particles. As ion migration over the surface
of an oxide is known to be strongly facilitated by H2O
ligands,37-39 such reduction of docked [FedO]+ ions is expected
to be more facile under conditions of H2-TPR when H2O is
formed than under CO-TPR conditions. This is confirmed by
the present data. The TPR data in Figures 6 and 7 can be
tentatively interpreted on this basis. We assign the peaks at 376
and 434 °C in Figure 7 to the reduction of (1) oxide nanopar-
ticles to Fe0 and (2) dinuclear oxo-ions to bare Fe2+ ions. The
larger number of peaks in Figure 6A then shows that besides
these two species there are bare Fe3+ ions that can only be
reduced by H2. There also appear to be mononuclear Fe oxo-
ions present that require water or OH ligands to migrate and
dock before they can be reduced to Fe0.
SZ-1
Fe/SZ-1
0
0.10
0.08
0.05
0
0.13
0.02
0.04
TABLE 5: Brønsted Sites Per Sulfur Atom on SZ-1 and
SZ-7a
peak area (absorbance × cm-1
)
ratio
sample
1025 cm-1
3645 cm-1
Brønsted/total sulfate
SZ-1
SZ-7
181.4
135.8
67.1
106.8
0.4
0.8
a Normalized by using the area of the peak at ∼1025 cm-1
,
corresponding to the OS-O stretching mode(s) of all surface sulfates.34
In both systems reduction of the transition metal facilitates
dissociative chemisorption of H2. Whereas reduction by H2 of
Fe-free SZ-1 requires a high temperature and transforms ∼57%
of the sulfate groups to SO2 but only ∼43% to S2-, the presence
of Fe induces a spectacular effect on the temperature and the
extent of reduction, as ∼90% of the surface sulfate groups in
Fe/SZ are reduced to S2- ions. The fact that H2 is reversibly
adsorbed on zirconia without any bond cleavage35 illustrates
the high resistance of that material against chemical reduction.
This situation is completely changed when Fe is present; both
the position of the TPR peaks and the level of sulfate reduction
are strongly affected. As with Pt/SZ,36 dissociation of H2 takes
place, followed by surface migration of atomic hydrogen leading
to reduction of sulfur S6+ to S2- (see Table 2). The relative
position of the peaks in Figure 4 indicates that once sites are
formed where H2 molecules can dissociate, swift reduction of
nearby sulfate groups will follow.
Figure 6B can be deconvoluted to three peaks at 381, 454,
and 597 °C, which we tentatively attribute to the reduction of
(1) bare Fe3+ ions, (2) Fe2+ ions that were created by the
preceding CO reduction of dinuclear Fe3+ oxo-ions, and (3)
mononuclear Fe oxo-ions that cannot be reduced by CO because
they are immobile in the absence of hydroxyl or water ligands.
Admittedly, this interpretation is speculative and requires
independent verification.
Reducing sulfur in sulfate ions to sulfide ions or H2S results
in a high hydrogen consumption ratio of Hcons/S ) 8:
SO42- + 4H2 f 4H2O + S2-
(3)
Adding up the H2/Fe ratios in the first row of Table 3 gives
a value of 0.96, whereas a value of 1.5 would be expected for
the total reduction of all Fe from Fe3+ to Fe0. This finding shows
that a significant fraction of the Fe in SA remains unreduced.
Apparently this Fe is stabilized in the support; a solid state
reaction takes place either with the silica or the alumina or both.
From the standard Gibbs free energy it appears that formation
of an aluminate is more favorable than formation of a
silicate:40
which is much larger than the H2 consumption per Fe (Hcons/Fe
) 3) in the reduction of Fe2O3
Fe2O3 + 3 H2 f 2Fe + 3 H2O
(4)
It is, therefore, not meaningful to derive a value for the final
oxidation state of Fe from the TPR data. However, the fact that
Fe strongly promotes the reduction of the sulfate groups in a
way similar to platinum shows that a significant fraction of the
iron must be present as Fe0 or as an iron sulfide that is capable
of dissociating H2 molecules. Chemical analysis of SZ-1 and
Fe/SZ-1 before and after H2-TPR (see Table 4) shows that with
SZ-1 much sulfur has been removed (mainly as SO2), but in
Fe/SZ-1, 70-80% of the sulfur is retained. This indicates that
with Fe/SZ-1, iron sulfide is formed. As dissociative chemi-
sorption of H2 will have a lower activation energy than migration
of the H atoms over the surface of sulfated zirconia, the rate of
Fe2SiO4 + 2Al2O3 H 2FeAl2O4 + SiO2,
4G0 ) -11 kcal/mol (5)
It is well-known that Al2O3 provides octahedral holes in the
oxygen sublattice which are attractive for transition metal ions
such as Cr3+ in gems. We therefore assume that this escape
route protects the Fe against reduction to Fe0. Phase diagrams
show that solid solutions exist of FeAl2O4 and Fe3O4; the
formation of this aluminate and similar Co and Cu aluminates

2062 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Henao et al.
has been reported by Bolt.41 Formation of Cu aluminate in Cu/
ZSM-5 was shown by Yan et al.42,43
the group of catalysts prepared by sublimation the materials
with low Fe content perform best. Results confirm that some
iron sites selectively catalyze NOx reduction to N2. Other sites,
such as Fe oxide particles, promote the combustion of i-C4H10.
Sulfate appears to enhance the selectivity of catalysts with low
Fe load.
Under redox conditions the SZ materials that are decorated
with Fe are expected to display a better durability than those
without a transition metal, because in the reduction mode sulfur
is better retained and thus will exert its protective function
against transformation to monoclinic zirconia in the oxidation
mode.
On silica-alumina, the CVD technique deposits iron in the
form of dinuclear and mononuclear iron oxo-ions, in addition
to oxide particles. Upon reduction at elevated temperature some
36% of the Fe forms a solid compound, presumably iron
aluminate, in which Fe2+ is stabilized against further reduction.
The low N2 yield with Fe/SA contrasts strongly with the high
yield, exceeding the 80% obtained with Fe/MFI tested in the
same lab with the same equipment. Clearly, applying the same
CVD technique is no warranty for obtaining catalysts with
similar performance. A probable chemical cause for the strik-
ingly different performance of Fe catalysts using crystalline or
amorphous supports is suggested by the TPR data. In the case
of Fe/MFI they show that one type of site prevails; the physical
characterization by a number of research groups suggests that
the selective centers are dinuclear oxo-complexes of iron. In
the case of Fe/SA, however, the TPR data in the present paper
show a wide variety of Fe sites. The inferior catalytic selectivity
for NOx reduction is presumably due to the high reactivity of
the most selective sites which form an aluminate with the
support or agglomerate to oxide particles. It thus seems that
the higher stability of the crystalline support and the fairly
localized nature of the acid sites in zeolites are favorable factors
for the highly selectivity of catalysts of this type.
As to the catalytic activity for NOx reduction, the data show
the behavior expected for catalysts containing a variety of Fe
species. It is well-known that Fe oxide particles catalyze the
combustion of hydrocarbons and, therefore, display a low
selectivity for NOx reduction in the presence of oxygen in
excess. With Fe/MFI our previous work showed that N2 yields
exceeding 80% have been obtained near 350 °C; dinuclear
oxygen-bridged Fe ions are believed to be responsible for this
selective catalysis. The present data show that with Fe/SA the
maximum N2 yield is 10% at 350 °C. To understand this striking
difference two points should be considered: First, comparison
of the H2-TPR and CO-TPR spectra in ref 44 and the present
paper reveals significant differences in the population of Fe sites.
With Fe/MFI a sharp H2-TPR peak at 400 °C identifies dinuclear
oxygen-bridged Fe sites, and a second peak above 600 °C has
been ascribed to Fe oxide clusters. The CO-TPR pattern of
calcined Fe/MFI shows one predominant sharp peak at 400 °C,
ascribed to the same dinuclear oxo-ion as the H2-TPR peak at
400 °C. The H2-TPR pattern of Fe/SA in Figure 6 of the present
work is markedly different; its deconvolution reveals six Fe
species of potentially catalytic relevance. The second point is
the escape of some Fe complexes from the surface of SA to its
interior. The present data suggest that the most selective sites
preferentially disappear, when Fe aluminate is formed. As the
catalytic tests were done at a rather high partial pressure of water
and the catalyst was held at each temperature for considerable
time, it makes sense that under these conditions elusive oxo-
complexes react with the support and form a stable aluminate,
whereas in the case of the crystalline support that solid state
reaction will have a much lower rate.
The catalytic data of the SZ-based materials display a
different picture. Samples with high Fe loading appear, again,
to display a signature typical of Fe oxide particles. In contrast,
the Fe/SZ catalysts with low Fe loading show a remarkable
selectivity for NOx reduction and deserve further study. Some
of the samples have a yield of 65%. The fairly high selectivity
of Fe/SZ-1 in comparison to Fe/SZ-7 is encouraging as it could
indicate that the presence of sulfate prevents excessive combus-
tion. A similar effect of sulfate content on combustion activity
has been observed with Pd/SZ in NO reduction by CH4.45 In
Fe/SZ-1, the sulfur and iron contents provide sufficient sites
for NO2 activation but prevent excessive isobutane combustion.
Interestingly, the mass balance shows that below 350 °C the
carbon oxides are mainly due to oxidation by NO2, not the O2
which is present in large excess.
Acknowledgment. This work was supported by the EMSI
program of the National Science Foundation and the U.S.
Department of Energy Office of Science (CHE-9810378) at the
Northwestern University Institute for Environmental Catalysis.
It made use of central facilities supported by the MRSEC
program of the National Science Foundation (DMR-0076097)
at the Materials Research Center of Northwestern University.
References and Notes
(1) Iwasawa, Y. In Handbook of Heterogeneous Catalysis; Ertl, G.,
Kno¨zinger, H., Weitkamp, J., Eds.; VCH: Weinheim, 1997; Vol. 2, pp
853-873.
(2) Menon, P. G.; Delmon, B. In Handbook of Heterogeneous Catalysis;
Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; VCH: Weinheim, 1997; Vol.
1, pp 100-118.
Conclusions
The CVD method permits complete replacement by Fe of
the Brønsted sites and silanol groups in sulfated-zirconia and
silica-alumina. The rate of this substitution correlates with the
acid strength. The iron uptake by sulfated zirconia during CVD
initially increases with sulfate content but drops precipitously
above a critical coverage near 45% of monolayer. This indicates
condensation of sulfate ions to “pyro” or polysulfate groups.
Iron has a dramatic effect on the reducibility of the sulfate
groups in sulfated zirconia. It strongly lowers the reduction
temperature and significantly enhances the depth of sulfate
reduction inducing formation of iron sulfide. The performance
of sulfated zirconia-supported iron in NOx reduction with
isobutane in a wet feed reveals a strong effect of the preparation
procedure and the Fe loading. Catalysts prepared by sublimation
are more active than those prepared by impregnation. Within
(3) Gao, Z.-X.; Kim, H.-S.; Sun, Q.; Stair, P. C.; Sachtler, W. M. H.
J. Phys. Chem. B 2001, 105, 6186-6190.
(4) Wang, X.; Chen, H.-Y.; Sachtler, W. M. H. J. Catal. 2001, 197,
281-291.
(5) Schwartz, V.; Prins, R.; Wang, X.; Sachtler, W. M. H. J. Phys.
Chem. B 2002, 106, 7210-7217.
(6) Wang, X.; Chen, H.-Y.; Sachtler, W. M. H. Appl. Catal. B 2001,
29, 47-60.
(7) Chen, H.-Y.; Sachtler, W. M. H. Catal. Today 1998, 42, 73-83.
(8) Wang, X.; Chen, H.-Y.; Sachtler, W. M. H. Appl. Catal. B 2000,
26, L227-239.
(9) Chen, H.-Y.; Wang, X; Sachtler, W. M. H. Phys. Chem. Chem.
Phys. 2000, 2, 3083-3090.
(10) Voskoboinikov, T. V.; Chen, H.-Y.; Sachtler, W. M. H. J. Mol.
Catal. A 2000, 155, 155-168.
(11) Mapes, J. E.; Eischens, R. P. J. Phys. Chem. 1954, 58, 1059-
1062.
(12) Parry, E. P. J. Catal. 1963, 2, 371-379.
(13) Wu, C. Y.; Hall, W. K. J. Catal. 1970, 19, 277-291.

Iron by CVD
J. Phys. Chem. B, Vol. 109, No. 6, 2005 2063
(14) Adeeva, V.; de Haan, J. W.; Ja¨nchen, J.; Lei, G. D.; Schu¨nemann,
V.; van de Ven, L. J. M.; Sachtler, W. M. H.; van Santen, R. A. J. Catal.
1995, 151, 364-372.
(30) Morterra, C.; Cerrato, G.; Ferroni, L. J. Chem. Soc., Faraday Trans.
1995, 91 125-132.
(31) Xu, B.-Q.; Sachtler, W. M. H. J. Catal. 1997, 167, 224-233.
(32) Lunsford, J. H.; Sang, H.; Campbell, S. M.; Liang, C.-H.; Anthony,
R. G. Catal. Lett. 1994, 27, 305-314.
(15) Jin, T.; Yamaguchi, T.; Tanabe, K. J. Phys. Chem. 1986, 90, 4794-
4796.
(16) Arata, K. In AdVances in Catalysis; Eley, D. D., Pines, H., Weisz,
P. B. Eds.; Academic Press: San Diego, 1990; Vol. 37, pp 165-211.
(17) Corma, A.; Fornes, V.; Juan-Rajadell, M. I.; Lopez-Nieto, J. M.
Appl. Catal. A 1994, 116, 151-163.
(33) Nascimento, P.; Akratopoulou, C.; Oszagyan, M.; Coudurier, G.;
Travers, C.; Joly, J. F.; Vedrine, J. C. In Studies in Surface Science and
CatalysissNew Frontiers in Catalysis; Guczi, L., Solymosi, F., Te´te´nyi,
P., Eds.; Elsevier: Amsterdam, 1993; Vol.75, pp 1185-1194.
(34) Morterra, C.; Cerrato, G.; Pinna, F.; Signoretto, M.; Strukul, G. J.
Catal. 1994, 149, 181-188.
(35) Hoang, D. L.; Lieske, H. Catal. Lett. 1994, 27, 33-42.
(36) Xu, B.-Q.; Sachtler, W. M. H. In SpilloVer and Migration of Surface
Species on Catalysts; Li, C., Xin, Q., Eds.; Elsevier: Amsterdam, 1997;
pp 229-240.
(37) Homeyer, S. T.; Sachtler, W. M. H. J. Catal. 1989, 117, 91-101.
(38) Augustine, S. M.; Sachtler, W. M. H. J. Catal. 1989, 116, 184-
194.
(18) Waqif, M.; Bachelier, J.; Saur, O.; Lavalley, J.-C. J. Mol. Catal.
1992, 72, 127-138.
(19) Hino, M.; Arata, K. J. Chem. Soc., Chem. Commun. 1980, 851-
852.
(20) Tanabe, K.; Hattori, H. In Handbook of Heterogeneous Catalysis;
Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; VCH: Weinheim, 1997; Vol.
1, pp 404-412.
(21) Chen, F. R.; Coudurier, G.; Joly, J.-F.; Vedrine, J. C. J. Catal.
1993, 143, 616-626.
(22) Cornu, A.; Massot, R. Compilation of Mass Spectral Data;
Heyden: London, 1966; pp 1C-2C.
(39) Carvill, B. T.; Lerner, B. A.; Zhang, Z.; Sachtler, W. M. H. J. Catal.
1993, 143, 314-317.
(23) Yamaguchi, T. Appl. Catal. 1990, 61, 1-25.
(24) Morterra, C.; Cerrato, G.; Pinna, F.; Signoretto, M. J. Catal. 1995,
157, 109-123.
(25) Reimer, T.; Spielbauer, D.; Hunger, M.; Mekhemer, G. A. H.;
Kno¨zinger, H. J. Chem. Soc., Chem. Commun. 1994, 10, 1181-1182.
(26) Kustov, L. M.; Kazansky, V. B.; Figueras, F.; Tichit, D. J. Catal.
1994, 150, 143-149.
(27) Morimoto, T.; Imai, J.; Nagao, M. J. Phys. Chem. 1974, 78, 704-
708.
(28) Rouxhet, P. G.; Sempels, R. E. J. Chem. Soc., Faraday Trans. 1974,
70, 2021-2032.
(29) Yamaguchi, T.; Nakano, Y.; Tanabe, K. B. Chem. Soc. Jpn. 1978,
51, 2482-2487.
(40) Barin, I.; Knacke, O. Thermochemical Properties of Inorganic
Substances; Springer-Verlag: New York, 1973; pp 32, 304-306, 690.
(41) Bolt, P. H. Transition Metal Aluminate Formation in Alumina-
supported Model Catalyst: High-energy Ion Beam Analysis of Interfacial
Processes, Ph.D. Thesis, Universiteit Utrecht, Utrecht, 1994; pp 63-94.
(42) Yan, J. Y.; Lei, G.-D.; Sachtler, W. M. H.; Kung, H. H. J. Catal.
1996, 161, 43-54.
(43) Yan, J. Y.; Sachtler, W. M. H.; Kung, H. H. Catal. Today 1997,
33, 279-290.
(44) Voskoboinikov, T.; Chen, H.-Y.; Sachtler, W. M. H. Appl. Catal.
B 1998, 19, 279-287.
(45) Chin, Y.-H.; Pisanu, A.; Serventi, L.; Alvarez, W. E.; Resasco, D.
E. Catal. Today 1999, 54, 419-429.