fundamental study of noble metal and transition metal …
Post on 18-Apr-2022
7 Views
Preview:
TRANSCRIPT
The Pennsylvania State University
The Graduate School
Department of Chemistry
FUNDAMENTAL STUDY OF NOBLE METAL AND TRANSITION METAL
OXIDE CLUSTERS IN THE PRESENCE OF CO:
INSIGHT INTO MECHANISMS OF HETEROGENEOUS CATALYSIS
A Thesis in
Chemistry
by
Nelly Moore Reilly
© 2007 Nelly Moore Reilly
Submitted in Partial Fulfillment of the Requirements
for the Degree of
Doctor of Philosophy
May 2007
The thesis of Nelly Moore Reilly was reviewed and approved* by the following:
A. Welford Castleman, Jr. Eberly Distinguished Chair in Science Evan Pugh Professor of Chemistry and Physics Thesis Advisor Chair of Committee
James B. Anderson Evan Pugh Professor of Chemisty and Physics
Nicholas Winograd Evan Pugh Professor of Chemistry
Robert J. Santoro George L. Guillet Professor of Mechanical Engineering
Ayusman Sen Professor of Chemistry Head of the Department of Chemistry
*Signatures are on file in the Graduate School
iii
ABSTRACT
The conversion of carbon monoxide (CO) to carbon dioxide (CO2) is one of the
most important catalytic reactions. Fundamental studies aimed at providing information
to aid in the design of more efficient and selective catalysts for the abatement of this
harmful environmental pollutant were conducted. Gas phase metal oxide cluster studies
are a valuable complementary technique to surface methods for elucidating the
mechanistic details and active sites of catalytic systems. To this end, gas phase studies
were utilized to uncover possible species responsible for the increased activity and
selectivity of noble metal and transition metal nanoparticle catalysts. A guided ion beam
mass spectrometer was employed to study the dissociation patterns and interactions with
CO of mass selected clusters. This provides information into specific reaction pathways
based on cluster size, ionic charge and oxidation state, stoichiometry and coordinative
saturation.
There has been recent interest in the activity of gold nanoparticles toward the
oxidation of CO and studies herein have focused on the effects of charge state in
elucidating the mechanism. Most reactions of gold oxide cations underwent oxygen
replacement by CO, however evidence of other reaction pathways including CO
oxidation for both Au4O2+ and Au4O3
+ species were also observed. Compared to anionic
gold oxide clusters of the same stoichiometry, CO oxidation was previously found to be
the dominant reaction channel. Anionic clusters were governed by a structure-reactivity
relationship in which a peripheral oxygen atom was the most efficient reaction center for
effecting CO oxidation. Density functional calculations found molecularly bound oxygen
iv
in the ground state structure of cationic clusters in which oxygen atom replacement by
CO was observed in experiments. High barriers to O-O bond dissociation suggest the
role of ionic charge may be an important factor governing the reaction. A mechanism
based on a triplet-singlet spin state transition was proposed that may provide a pathway
with lower barriers for O2 dissociation in order to explain the experimentally observed
reaction products.
Next, studies involving mixed metal oxide clusters were undertaken to probe to
effects of catalyst-dopant interactions and gain a deeper understanding of the principles
governing the oxidation of CO. Comparison of the reactions of pure metal oxides with
CO to experiments of bimetallic oxides can aid in determining the influence that each
individual metal exhibits and the changes that occur in reactivity when another metal is
present. Studies involving silver-gold dimer oxide clusters were conducted and provide
insight into the role of charge transfer between different metals. AgAuO1,2+ species
showed similar reaction products as pure gold oxide cationic clusters, while AgAuO3,4+
reacted similar to pure silver oxide clusters with the same stoichiometry. Only AgAuO2-
exhibited oxygen atom transfer which was different from reactions of either pure metal
oxide. This provides experimental evidence of density functional calculations that
previously predicted the reaction of AgAu- with O2 and CO promotes CO oxidation.
Studies concerning 3d transition metals were conducted to investigate their
activity as common catalytic support materials and their ability toward direct oxidation of
CO. Structural elucidation was conducted through CID experiments and density
functional calculations were performed for iron oxide ionic clusters. Reactions of iron
v
oxide cations with CO were compared to anionic clusters. Both ionic charge states were
shown promote CO oxidation; however, attachment of CO onto the metal cluster was
only observed for cationic clusters. Calculated energy profiles of the reaction
demonstrated that the oxidation efficiency was governed by the strength of oxygen
binding to the iron atom. The results provide important insight into the nature of the
active site for condensed phase catalysis, shedding light on the role of charge state for CO
oxidation in the presence of iron oxide clusters.
The behavior of several other 3d transition metal oxide clusters, cobalt, nickel and
copper, were undertaken to examine possible periodic trends. The dissociation patterns
and reactivity in the presence of CO for these metal oxides were similar to iron oxides for
each respective charge state. The most active stoichiometry for CO oxidation was MxOy+
where y=x for cationic clusters and MxOy- where y=x+1 for anionic clusters with the
exception of Cu3O3-. The same number of Cu atoms and oxygen atoms was a particularly
active stoichiometry. This may be a consequence of the electron density for copper,
which possesses a full orbital shell. Comparison of cations and anions showed that
cations were more efficient in their particular reaction channels requiring less CO
reactant gas to observe reaction products.
vi
TABLE OF CONTENTS
LIST OF FIGURES ................................................................................................. ix
LIST OF TABLES...................................................................................................xiii
ACKNOWLEDGEMENTS......................................................................................xiv
Chapter 1 Introduction .............................................................................................1
1.1 The Oxidation of Carbon Monoxide ............................................................1 1.2 Clusters as Models.......................................................................................2 1.3 Insight into Alternative Catalytic Material ...................................................4
1.3.1 Gold Oxides ......................................................................................5 1.3.2 Bimetallic Oxides ..............................................................................5 1.3.3 Transition Metal Oxides ....................................................................6
1.4 Synopsis......................................................................................................7 References ........................................................................................................9
Chapter 2 Experimental Setup ..................................................................................14
2.1 Introduction.................................................................................................14 2.2 Vacuum system ...........................................................................................14 2.3 Laser Vaporization Source...........................................................................17 2.4 Guided Ion Beam Setup...............................................................................20 2.5 Detection and Data Acquisition ...................................................................23 References ........................................................................................................25
Chapter 3 Comprehensive Study of the Reactivity of Gold Oxide Cluster Cations (AuxOy
+; x=1-4, y=1-5) in the Presence of Carbon Monoxide............................26
3.1 Introduction.................................................................................................26 3.2 Experimental Methods.................................................................................28 3.3 Results ........................................................................................................28 3.4 Discussion...................................................................................................32
3.4.1 CO Replacement Reactions ...............................................................32 3.4.2 Au Fragmentation/Replacement.........................................................42 3.4.3 Oxidation...........................................................................................44
3.5 Comparison to Anionic Gold Oxides ...........................................................45 3.6 Conclusions.................................................................................................46 References ........................................................................................................47
Chapter 4 Reactivity of Ag2Oy+,- and Bimetallic AgAuOy
+,- (y=1-4) Ionic Clusters with CO ............................................................................................................50
vii
4.1 Introduction.................................................................................................50 4.2 Experimental ...............................................................................................52 4.3 Results ........................................................................................................52
4.3.1 Pure Silver Oxide Ions.......................................................................53 4.3.2 Bimetallic Ag-Au Oxide Ions ............................................................56
4.4 Discussion...................................................................................................62 4.5 Conclusion ..................................................................................................64 References ........................................................................................................65
Chapter 5 Influence of Charge State on the Reaction of FeO3+,- with Carbon
Monoxide..........................................................................................................67
5.1 Introduction.................................................................................................67 5.2 Methods ......................................................................................................68 5.3 Results and Discussion ................................................................................70 5.4 Conclusion ..................................................................................................79 References ........................................................................................................80
Chapter 6 Experimental and Theoretical Study of the Structure and Reactivity of Fe1-2O≤6¯ Clusters with CO................................................................................82
6.1 Introduction.................................................................................................82 6.2 Methods ......................................................................................................84 6.3 Results and Discussion ................................................................................86
6.3.1 CID Fragmentation............................................................................90 6.3.2 Reactions with CO and N2 .................................................................98
6.4 Conclusion ..................................................................................................104 References ........................................................................................................105
Chapter 7 Experimental and Theoretical Study of the Structure and Reactivity of Fe1-2O1-5
+ Clusters with CO...............................................................................107
7.1 Introduction.................................................................................................107 7.2 Methods ......................................................................................................109 7.3 Results and Discussion ................................................................................111
7.3.1 CID Studies .......................................................................................114 7.3.2 Reactivity Studies..............................................................................118
7.4 Conclusion ..................................................................................................126 References ........................................................................................................127
Chapter 8 Gas Phase Study of Cobalt, Nickel, and Copper Oxide Cluster Ions: Dissociation Patterns and Trends in Reactivity with CO ....................................129
8.1 Introduction.................................................................................................129 8.2 Experimental ...............................................................................................132
viii
8.3 Results and Discussion ................................................................................133 8.3.1 Collision Induced Dissociation Studies ..............................................133 8.3.2 Reactivity Studies..............................................................................145
8.4 Comparison to Bulk.....................................................................................156 8.5 Conclusion ..................................................................................................157 References ........................................................................................................159
Chapter 9 General Conclusions ................................................................................163
9.1 Major Findings ............................................................................................164 9.2 Future Studies .............................................................................................167 References ........................................................................................................169
Appendix Kinetic Analysis of the Reaction between (V2O5)1,2+ and Ethylene ..........170
A.1 Abstract ......................................................................................................170 A.2 Introduction................................................................................................171 A.3 Experimental and Computational Methods .................................................174 A.4 Results and Discussion ...............................................................................175 A.5 Conclusion .................................................................................................192 References ........................................................................................................195
ix
LIST OF FIGURES
Figure 2-1: Guided Ion Beam Mass Spectrometer employed in gas phase metal oxide studies. ....................................................................................................15
Figure 2-2: Schematic setup of the dual rod LaVa source which includes the beam splitting mirror and conical nozzle. ..........................................................19
Figure 3-1: A typical mass distribution for cationic gold oxide clusters. The first gold oxide in each series is labeled with the addition of one oxygen atom to subsequent peaks...............................................................................................29
Figure 3-2: Predicted lowest energy structures for AuxOy+ where x=1-2, y=1-4.
Source: Figure courtesy of the group of Professor V. Bonačić-Koutecký at the Humboldt University in Berlin.....................................................................33
Figure 3-3: Branching ratios for a) AuO+, b) Au2O+, and c) Au3O
+ which all show a large O atom replacement by CO reaction channel.................................36
Figure 3-4: Logarithm of the intensity ratio of ln[Ir/Io] versus CO reactant concentration for a) AuO+, b) Au2O
+, and c) Au3O+...........................................37
Figure 3-5: Branching ratio for AuO3+ containing a peripheral and molecular
oxygen unit in the lowest energy structure calculation. ......................................39
Figure 3-6: Branching ratio for AuO2+ which shows a selective pathway for O
atom replacement by CO. ..................................................................................39
Figure 3-7: Branching ratios for a) Au4O2+ and b) Au4O3
+ and their respective products with increasing carbon monoxide reactant gas.....................................43
Figure 4-1: Branching ratios for a) Au2O2+, b) AgAuO2
+, c) Ag2O2+, and d)
AgAuO2- showing the products in the reaction with CO reactant gas. ................55
Figure 4-2: a) Typical anionic silver-gold oxide mass distribution and b) cationic silver-gold oxide mass distribution obtained from dual rod LaVa source. ..........57
Figure 4-3: Branching ratios for a) Au2O+ and b) AgAuO+ with increasing CO
pressure.............................................................................................................59
Figure 4-4: Plots of ln[Ir/Io] versus CO reactant concentration for a) Au2O+ and b)
AgAuO+ ............................................................................................................60
Figure 5-1: Branching ratios for a) FeO3+ and b) FeO3
- with increasing CO pressure.............................................................................................................71
x
Figure 5-2: Ground state geometries for O2, CO, CO2, and FeOy+,- clusters.
Source: Figure courtesy of the group of Professor S. N. Khanna at VCU. ..........73
Figure 5-3: a) Graph of the energy required to remove an O or O2 from the FeOy+
cluster. b) Graph of the energy required to remove an O, O2, O- and O2
- from the FeOy
- cluster. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.................................................................................................75
Figure 5-4: Profile of the reaction energy for FeO3+ with CO including the
corresponding change in energy, ΔE, for each reaction step. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU...................................76
Figure 5-5: Profile of the reaction energy for FeO3- with CO including the
corresponding change in energy, ΔE, for each reaction step. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU...................................78
Figure 6-1: A typical mass distribution produced for iron oxide anionic clusters. The first iron oxide in each series with subsequent peaks in the series having one additional oxygen atom...............................................................................87
Figure 6-2: The ground state geometries of O2, CO, FeO1-3- and FexOy
- clusters. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU. ..........89
Figure 6-3: Graphs of the fragmentation energies for atomic and molecular oxygen loss in a) FeOy
- and b) Fe2Oy- clusters. c) Graph of the fragmentation
energy for neutral metal and metal oxide loss from Fe2Oy- clusters. ...................93
Figure 6-4: The change in energy (ΔE) for the dissociation pathway of parent cluster, Fe2O4
- producing FeO3- and FeO2
-. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU. .........................................................95
Figure 6-5: The change in energy (ΔE) for the dissociation pathway of parent cluster Fe2O3
- producing FeO3- and FeO2
-. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU. .........................................................97
Figure 6-6: The branching ratios for a) FeO2- and b) Fe2O3
- with increasing CO reactant gas pressure..........................................................................................100
Figure 6-7: The change in energy (ΔE) for the reaction pathway of Fe2O3- with
CO. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU....101
Figure 6-8: The branching ratios for a) FeO4- and b) Fe2O6
- with increasing CO pressure, both clusters show molecular oxygen loss...........................................103
xi
Figure 7-1: Typical mass distribution for iron oxide cation clusters produced when employing a) a 27 mm conical nozzle and b) a 51 mm conical nozzle at the exit of the source. ........................................................................................112
Figure 7-2: The ground state geometries for a) FexOy+ clusters and b) Fe(CO)1-2
+ and FexOyCO+ clusters. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU….. .......................................................................................116
Figure 7-3: Graphs of the fragmentation energies for atomic and molecular oxygen loss in a) FeOy
+ and b) Fe2Oy+ clusters. c) Graph of the fragmentation
energy for neutral metal and metal oxide loss from Fe2Oy+ clusters. ..................119
Figure 7-4: Branching ratios for a) FeO+, b) FeO2+, c) Fe2O
+, and d) Fe2O2+
reacted with CO.. ..............................................................................................123
Figure 7-5: Branching ratio for Fe2O3+ reacted with CO. .........................................123
Figure 7-6: Profile of the reaction energy for FeO2+ with CO including the
corresponding change in energy, ΔE, for each reaction step. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU...................................124
Figure 7-7: Branching ratio for FeO4+ reacted with CO showing at higher
pressures two CO molecules become attached to bare iron. ...............................125
Figure 8-1: Typical anionic mass distributions for a) Nickel oxides, b) Cobalt oxides, and c) Copper oxides. ............................................................................134
Figure 8-2: Typical cationic mass distributions for a) Nickel oxides and b) Cobalt oxides................................................................................................................135
Figure 8-3: Branching ratios for a) Co2O3-, b) Ni2O3
-, and c) Cu3O3- reacted with
carbon monoxide. ..............................................................................................147
Figure 8-4: Branching ratios for a) CoO2+ and b) NiO3
+ reacted with carbon monoxide.. ........................................................................................................148
Figure 8-5: Branching ratios for a) NiO2- and b) NiO2
+ reacted with carbon monoxide.. ........................................................................................................151
Figure 1-1: General scheme of the energetic profile and structural transformation of the mechanism for the reaction between (V2O5)n=1,2
+ and C2H4. Source: Figure courtesy of the group of Professor V. Bonačić-Koutecký at the Humboldt Unisversity in Berlin………………………………………………....173
Figure 1-2: Snapshots of the ab initio MD trajectory calculated using DFT methods for the reaction of a) V2O5
+ with C2H4 and b) V4O10+ with C2H4
xii
leading to the formation of acetaldehyde. Source: Figure courtesy of the group of Professor V. Bonačić-Koutecký at the Humboldt Unisversity in Berlin. ...............................................................................................................177
Figure 1-3: Branching ratios for the energetic analysis of V2O5+ with ethylene
conducted under the following pressures: a) 0.1 mTorr, b) 0.2 mTorr, c) 0.3 mTorr, d) 0.4 mTorr, e) 0.5 mTorr and f) 0.6 mTorr of ethylene........................178
Figure 1-4: Branching ratios for the energetic analysis of V4O10+ with ethylene
conducted under the following pressures: a) 0.1 mTorr, b) 0.2 mTorr, c) 0.3 mTorr, d) 0.4 mTorr, e) 0.5 mTorr and f) 0.6 mTorr of ethylene. .......................179
Figure 1-5: a) Ratio of product ion intensity, Ip, to total ion intensity, I0, versus the reaction cell gas pressure for reactions of V2O5
+ with ethylene at ECM= 0 eV and b) at ECM= 2.0 eV. .................................................................................183
Figure 1-6: a) The rate constant versus energy for V2O5+ with ethylene calculated
from cross section data. b) The rate constant versus energy for V4O10+ with
ethylene calculated from cross section data........................................................183
Figure 1-7: a) Logarithm of the intensity ratio [V2O5+/V2O5
++ΣIp] versus the concentration of C2H4 calculated as an ideal gas at ECM=0 eV and b) at ECM= 2 eV. .................................................................................................................185
Figure 1-8: a) The rate constant versus center-of-mass energy for the reaction of V2O5
+ with ethylene calculated from the velocity of the ions and reaction time and b) The rate constant versus center-of-mass energy for the reaction of V4O10
+ with ethylene calculated from the velocity of the ions and reaction time...................................................................................................................185
Figure 1-9: a) Energy dependence of the total reaction cross sections for V2O5+
with ethylene for the theoretical model (solid line) and experimental measurements (diamonds). b) Energy dependence of the total reaction cross sections for V4O10
+ with ethylene. .....................................................................193
Figure 1-10: Dependence of the reaction cross section (σ(E)) on the energy for different values of the barrier for hydrogen transfer: 1.45 eV (black line), 3.75 eV (blue line), 3.85 eV (green line), 3.88 eV (red line). Source: Figure courtesy of the group of Professor V. Bonačić-Koutecký at the Humboldt Unisversity in Berlin. ........................................................................................193
xiii
LIST OF TABLES
Table 3-1: Reaction products for AuxOy+ with CO reactant gas and nitrogen
collision gas. .....................................................................................................31
Table 4-1: Product list for the dimer of pure silver, pure gold, and mixed Ag-Au oxide clusters in the presence of CO..................................................................54
Table 6-1: The CID fragmentation channels, calculated dissociation energies (DE), and reaction products with CO and N2 for Fe1,2O2-6
- clusters. ...................92
Table 7-1: Results of CID studies showing the fragmentation channels for selected Fe1-2Oy
+ clusters...................................................................................117
Table 7-2: Reaction products for the reaction between selected iron oxide cluster cations and carbon monoxide…………………………………………………. ..120
Table 8-1: CID fragmentation channels for selected MxOy- (M=Co, Ni, Cu)
anions................................................................................................................137
Table 8-2: CID fragmentation channels for selected MxOy+ (M=Co, Ni) cations. ....138
Table 8-3: Reaction products for selected MxOy- (M=Ni, Co, Cu) anions with CO
and N2. ..............................................................................................................140
Table 8-4: Reaction products for MxOy+ (M=Co, Ni) cations with CO and N2.........142
xiv
ACKNOWLEDGEMENTS
I would like to thank my thesis advisor Professor A. Welford Castleman, Jr. for
giving me the opportunity to grow as a scientist and a person. I would also like to thank
past and present members of the Castleman group for providing support, knowledge, and
helpful discussions, especially my lab partners Dina Justes and Grant Johnson. Also, I
thank Richard Bell for help in troubleshooting the instrument and Michele Kimble for
providing her knowledge and friendship. I also acknowledge and thank the members of
my thesis committee, Professor James B. Anderson, Professor Nicholas Winograd, and
Professor Robert J. Santoro.
I would like to thank my collaborators, Christian Bürgel, Roland Mitrić, and
Professor Vlasta Bonačić-Koutecký at the Humboldt University in Berlin, Germany and
J. Ulises Reveles and Professor Shiv N. Khanna at Virginia Commonwealth University in
Richmond, Virginia. Their theoretical calculations and insightful discussions have
contributed to a greater understanding of the experimental studies herein.
Most importantly I want to express my gratitude to my family who have always
supported and encouraged me. Finally, I dedicate this thesis to my husband, Brian Reilly.
He has stood by all my decisions and kept me going through the rough times.
Chapter 1
Introduction
1.1 The Oxidation of Carbon Monoxide
Carbon monoxide (CO) is a harmful atmospheric pollutant produced by
incomplete combustion of fossil fuels. The conversion of CO to CO2 is one of the most
important catalytic reactions pertaining to environmental and industrial applications. For
example, the ability to remove byproducts such as CO in fuel cell systems through the
water-gas shift (WGS) reaction at low temperatures and at fast reaction rates would aid in
the development of lower cost fuel alternatives (1).
From a fundamental point of view, the energy gain in going from CO to CO2 is
5.85 eV while the energy required to break the O-O bond in O2 is 5.23 eV. Therefore
based on the pure energetics of the reaction, the conversion of CO to CO2 should occur
spontaneously in the presence of O2. It has been suggested that the rate limiting step in
this process is the activation of O2 because the breaking of the O-O bond involves a large
energetic barrier (2). Catalysts that can lower the reaction barrier and hence promote the
reaction are needed to assist the conversion. The basic idea of the catalyst is then to
weaken/break the O-O bond through charge transfer and/or bonding, thus facilitating the
reaction. The ability to tailor the design of more energy efficient catalysts for the
oxidation of CO however remains elusive and fundamental studies on the active sites and
reaction mechanisms would aid in creating more selective catalysts.
2
1.2 Clusters as Models
Studies on the activity and selectivity of heterogeneous catalysts are highly
inconsistent depending on the chosen preparation method (3-5) and molecular level
insight into the active sites and catalytic mechanisms is difficult employing surface
science techniques alone. Gas phase studies are a complementary method which offer the
opportunity to study isolated potential catalytic active sites in the absence of solvent
effects and surface inhomogenieties that often complicate condensed phase catalytic
research (6-7). To this end, there has been a growing acceptance of the use of gas phase
metal oxide clusters to model the reactions that occur over metal oxide surfaces.
Muetterties and Witko et al. have suggested that a metal oxide surface may be viewed as
a collection of clusters and isolated reaction sites (8-9). In addition, Somorjai has
described a collection of atoms to be “cluster-like” (10), further establishing the value
that cluster studies may have in elucidating molecular level mechanisms.
Numerous correlations between the products of gas phase experiments and
heterogeneous catalysis have been uncovered (9-12). In our own laboratory, significant
correlations between gas phase experiments and analogous condensed phase reactions
have been obtained. In one study, vanadium and niobium oxide cluster cations with
methanol produced the dehydrogenation of methanol yielding neutral formaldehyde (13).
A major product produced over condensed phase vanadia surfaces is formaldehyde (14).
Another study involving a comprehensive experimental and theoretical evaluation of
vanadium oxides observed that only (V2O5)1,2+ clusters transfer an oxygen atom to
ethylene reactant gas, producing acetaldehyde (12, 15). Acetaldehyde has been produced
during ethylene oxidation over vanadia surfaces with V2O5 sites as the proposed active
3
sites (16). In addition, differences between vanadium oxide clusters and niobium and
tantalum oxides have been elucidated in reactions with n-Butane (17-18). Similar to
analogous bulk catalytic systems (19), V2O5+ transfers an oxygen atom to the
hydrocarbon, while Nb2O5+ and Ta2O5
+ exhibit C-C cracking as the dominant reaction
pathway.
Beyond our own work (12-13, 15, 17-18, 20), additional support for the
correlation between gas phase and condensed phase studies is being realized. Zamaraev
et al. studied the oxidation of methanol on MoxOy+ clusters and observed similar
reactions occurring over the corresponding molybdenum-oxygen sites in homogeneous
and heterogeneous catalysts (11). Gas phase studies conducted by Plattner and
coworkers found C-H activation on cationic iridium (III) complexes (21-22). Their
studies were used to design a more efficient and reactive solution phase catalyst
containing Ir (23). Wallace and Whetten and Wöste and coworkers have combined gas
phase experiments and theoretical calculations to elucidate information on the full
catalytic cycle of CO oxidation with proposed reaction steps (2, 24-25). Schwarz et al.
has also demonstrated a full gas phase catalytic cycle for the oxidation of NO in the
presence of platinum oxide cation clusters (26). The ability to probe the active sites and
reaction intermediates of complex catalytic reactions through gas phase ion-molecule
studies has aided many researchers in unraveling the related mechanisms (26-29).
Furthermore, Armentrout and coworkers have revealed ion-molecule reaction kinetics
yielding information on the thermochemistry, structure, and bond energy of transition
metal clusters (30).
4
Investigating the nature of catalytic active sites is of utmost importance for
understanding the reaction mechanisms and designing better catalytic systems. Gas
phase studies are especially amenable to probing ionic charge state and electron density,
two factors that have been proposed to assist in activating reactions (31-34). Studies by
Grzybowska-Świerkosz have focused on the role of ionic centers that arise on certain
sites in vanadium catalysts and the properties of these charged sites (35). It has been
pointed out by Davis that further investigation of ionic sites will expand our
understanding of active sites in catalytic systems (36). Therefore the above mentioned
findings illustrate the significance of gas phase ionic cluster studies in elucidating such
effects as cluster size, stoichiometry, charge and oxidation state, and composition on
reactivity. These studies provide a useful means to gain important insight into the basic
processes involved in heterogeneous catalysis.
1.3 Insight into Alternative Catalytic Material
The most common catalytic materials employed in the oxidation of CO presently
are Pdn and Ptn nanoparticles; however these materials are expensive and require high
operating temperatures (1, 37-38). New efforts are being put forth to find suitable
alternative materials with good active lifetimes that are selective toward the oxidation of
CO.
5
1.3.1 Gold Oxides
Bulk gold has long been known to be catalytically inert (39-40). However,
findings by Haruta and coworkers demonstrated that nanosized gold particles deposited
onto select metal oxides exhibit high activities for the oxidation of carbon monoxide (39,
41-42). These studies also established that the catalytic activity was highly dependant on
the type of metal oxide support and the gold particle size.
Past studies have focused on elucidating the structures of charged gold through
ion mobility measurements and theoretical methods (43-47). The adsorption properties
(48-51) and reactivity of O2 and CO with gold clusters have also been extensively
explored to uncover potential active sites and mechanistic details in the oxidation of CO
(52-54). Several researchers suggest that charge transfer between gold and adsorbed O2
allows for elongation and activation of the O-O bond (2, 55). However, there is no
consensus on the active charge state for the catalytic conversion, thus further
investigation is warranted.
1.3.2 Bimetallic Oxides
The introduction of another metal into a catalytic system can greatly impact the
structure and electronic properties of the catalyst (56-57). In a study by Metiu and
coworkers on AgmAun clusters, replacing active sites with either Au or Ag resulted in a
change in the bond strength (58). This change in strength was directly related to the
LUMO energy; the lower the LUMO energy, the stronger the bond energy became. The
ability to tune the properties by changing the number of atoms of each component in
6
mixed clusters may be particularly valuable for designing new catalytic materials with
enhanced activity. There is a great disparity in the information available for pure metal
oxide systems compared to bimetallic systems and further experiments on bimetallic
reactivity may aid in elucidating the active sites and reaction mechanisms responsible for
increased catalytic activity.
1.3.3 Transition Metal Oxides
An alternate approach to overcome the O2 rate limiting step is to use systems that
contain O atoms without the O-O bonds. Transition metal oxides with oxygen atoms
bound only to the transition metal without O-O bonding offer an attractive possibility.
Indeed, recent experiments on iron oxide nanoparticles show that it may be possible to
convert CO to CO2 at low temperatures in the presence or absence of O2 (59-62).
Theoretical studies on small clusters containing less than 10 atoms reveal that the ease in
structural rearrangement at small sizes almost eliminates the reaction barriers (63-64).
This is provided that changes in the spin multiplicity are minimal and offer a spin
allowed process. Employing small size clusters therefore may offer the possibility of
spontaneous conversion of CO to CO2.
Since charge transfer and electron density are predicted to play a role in the active
sites of catalysts, it is valuable to gain information about different transition metal
systems each with their own structural and electronic properties. Studies involving
several 3d transition metal clusters, iron, nickel, cobalt, and copper, will provide insight
into the role of additional d electrons on the reactivity toward CO oxidation.
7
Comprehensive studies on transition metals have revealed periodic trends in reaction
energy and the role of d orbitals in bonding (65). The influence of different electron
affinities for each metal composition may provide insight into how regions of electron
deficiency and enhancement affect catalytic activity. Individual studies have probed the
structures, bond dissociation and adsorption energies, and reactivity of iron (66), nickel
(67), cobalt (68), and copper (69). However, there remain many unanswered questions
involving the catalytic efficiency of these metals.
1.4 Synopsis
The following thesis contains an investigation of ionic gold oxides, gold-silver
bimetallic oxides, and several transition metal oxide species toward elucidating the
reactivity for alternative catalytic materials in the conversion of CO to CO2. The
technique employed in the experimental studies is described in Chapter 2. Reaction
pathways for cationic gold oxides are discussed in Chapter 3 and compared to reactions
of anionic clusters with similar stoichiometry. An emphasis on the oxygen atom
replacement pathway and a proposed mechanism is presented where the role of charge
state is an important factor in this observed reaction product. The reactivity of bimetallic
gold-silver dimer oxides are examined in Chapter 4. The ability to tune the properties of
clusters by changing the composition is revealed based on similarities and differences in
the reaction products observed. The structure of ionic iron oxide clusters and charge state
influences in the reaction pathways are explored in Chapters 5-7. Reactivity and collision
induced dissociation studies involving cobalt, nickel, and copper oxide clusters are
8
investigated in Chapter 8. The periodic trends in the first row 3d transition metals are
discussed. Chapter 9 presents the major finding and draws conclusions from the
experiments presented in the previous chapters. Finally, in addition to CO reactivity
studies discussed throughout this thesis, a study describing the reaction between
(V2O5)n=1-2+ and ethylene is presented in the appendix.
9
References
1. Huang, C.; Jiang, R.; Elbaccouch, M.; Muradov, N.; Fenton, J. M. J. Power Sour. 2006, 162, 563.
2. Wallace, W. T.; Whetten, R. L. J. Am. Chem. Soc. 2002, 124, 7499.
3. Bollinger, M. A.; Vannice, M. A. Appl. Catal. B 1996, 8, 417.
4. Fu, Q.; Weber, A.; Flytzani-Stephanopoulos, M. Catal. Lett. 2001, 77, 87.
5. Kung, H. H.; Kung, M. C.; Costello, C. K. J. Catal. 2003, 216, 425.
6. Somorjai, G. A.; Levine, R. D. J. Phys. Chem. B (Comment) 2005, 109, 9853.
7. Böhme, D. K.; Schwarz, H.; Angew. Chem. Int. Ed. 2005, 44, 2336.
8. Muetterties, E. L. Science 1997, 196, 839.
9. Witko, M.; Hermann, K.; Tokarz, R. J. Electron. Spec. Rel. Phenom. 1994, 69, 89.
10. Somorjai, G. A. Introduction to Surface Chemistry and Catalysis John Wiley:
New York, 1994, 402-409.
11. Fialko, E. F.; Kikhtenko, A. V.; Goncharov, V. B.; Zamaraev, K. I. J. Phys. Chem. B 1997, 101, 5772.
12. Zemski, K. A.; Justes, D. R.; Castleman, A. W., Jr. J. Phys. Chem B 2002, 106,
6136.
13. Justes, D. R.; Moore, N. A.; Castleman, A. W., Jr. J. Phys. Chem. B 2004, 108, 3855.
14. Wachs, I. E.; Deo, G.; Juskelis, M. V.; Weckhuysen, B. M. Dynamics of Surface
and Reaction Kinetics in Heterogeneous Catalysis 1997, 305.
15. Justes, D. R.; Mitrić, R.; Moore, N. A.; Bonačić-Koutecký, V.; and Castleman, A. W., Jr. J. Am. Chem. Soc. 2003, 125, 6289.
16. Oyama, S. T.; Middlebrook, A. M.; Somorjai, G. A. J. Phys. Chem. B 1990, 94,
5029.
10
17. Zemski, K. A.; Justes, D. R.; Bell, R. C.; Castleman, A. W., Jr. J. Phys. Chem. A
2001, 105, 4410.
18. Bell, R. C.; Zemski, K. A.; Kerns, K. P.; Deng, H. T.; Castleman, A. W., Jr. J. Phys. Chem. A 1998, 102, 1733.
19. Whittborn, A. M. C.; Costas, M.; Bloomberg, M. R. A.; Siegbahn, P. E. M. J.
Chem. Phys. 1997, 107, 4318.
20. Bell, R. C.; Zemski, K. A.; Castleman, A. W., Jr. J. Phys. Chem. 1999, 103, 1585.
21. Plattner, D. A. Int. J. Mass Spectrom. 2001, 207, 125.
22. Hinderling, C.; Plattner, D. A.; Chen, P. Angew. Chem. Int. Ed. 1997, 36, 243.; Hinderling, C.; Feichtinger, D.; Plattner, D. A.; Chen, P. J. Am. Chem. Soc. 1997, 119, 10793.
23. Luecke, H. R.; Bergman, R. G. J. Am. Chem. Soc. 1997, 119, 11538.
24. Socaciu, L. D.; Hagen, J.; Bernhardt, T. M.; Wöste, L.; Heiz, U.; Häkkinen, H.; Landman, U. J. Am. Chem. Soc. 2003, 125, 10437.
25. Bernhardt, T. M.; Socaciu-Siebert, L. D.; Hagen, J.; Wöste, L. Appl. Catal. A
2005, 291, 190. 26. Eller, K.; Schwarz, H. Chem. Rev. 1991, 91, 1121.
27. Waters, T.; O’Hair, R. A. J.; Wedd, A. G. J. Am. Chem. Soc. 2003, 125, 3384.
28. Hagan, J.; Socaciu, L. D.; Elijazyfer, M.; Heiz, U.; Bernhardt, T. M.; Wöste, L. Phys. Chem. Chem. Phys. 2002, 4, 1707.
29. Kinne, M.; Heidenreich, A.; Rademann, K. Angew. Chem. Int. Ed. 1998, 37,
2509. 30. Armentrout, P. B. Int. J. Mass Specrom. 2000, 200, 219.; Xu, J.; Rogers, M. T.;
Griffin, J. B.; Armentrout, P. B. J. Chem. Phys. 1998, 108, 9339.
31. Bielañski, A.; Haber, J. Oxygen in Catalysis Marcel Dekker, Inc.: New York 1991.
32. Costello, C. K.; Kung, M. C.; Oh, O.-S.; Kung, H. H. Appl. Catal. A 2002, 232,
159.
11
33. Oh, O.-S.; Costello, C. K.; Cheung, C.; Kung, H. H.; Kung, M. C. Studies of Surface Science and Catalysis-Catalyst Deactivation 2001 (J. J. Spivey, G. W. Roberts, and B. H. Davis, Eds.) Elsevier: Amsterdam 2001, 375-381.
34. Hodge, N. A.; Kiely, C. J.; Whyman, R.; Siddiqui, R. H.; Hutchings, G. J.;
Pankhurst, Q. A.; Wagner, F. E.; Rajaram, R. R.; Golunski, S. E. Catal. Today 2002, 72, 133.
35. Grzybowska-Świerkosz, B. Appl. Catal. A 1997, 157, 409.
36. Davis, R. J. Science 2003, 301, 926.
37. Stolcic, D.; Fischer, M.; Ganteför, B.; Kim, Y. D.; Sun, Q.; Jena, P. J. Am. Chem. Soc. 2003, 125, 2848.
38. Gong, X.-Q.; Hu, P.; Raval, R. J. Chem. Phys. 2003, 119, 6324.
39. M. Haruta, Catalysis Today 1997, 36, 153.
40. Bond, G. C.; Thompson, D. T. Catal. Rev.-Sci. Eng. 1999, 41, 319.
41. Sanchez, A.; Abbet, S.; Heiz, U.; Schneider, W.-D.; Häkkinen, H.; Barnett, R. N.; Landman, U. J. Phys. Chem. A 1999, 103, 9573.
42. Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M. J.; Delmon, B.
J. Catal. 1993, 144, 175.
43. Walker, A. V. J. Chem. Phys. 2005, 122, 094310.
44. Furche, R.; Ahlrichs, R.; Weis, P.; Jacob, C.; Gilb, S.; Bierweiler, T.; Kappes, M. M. J. Chem. Phys. 2002, 117, 6982.
45. Okumura, M.; Kitagawa, Y.; Haruta, M.; Yamaguchi, K. Appl. Catal. A 2005, 291, 37.
46. Wu, X.; Senapati, L.; Nayak, S. K.; Selloni, A.; Hajaligol, M. J. Chem. Phys.
2002, 117, 4010.
47. Gilb, S.; Weis, P.; Furche, F.; Ahlrichs, R.; Kappas, M. M. J. Chem. Phys. 2002, 116, 4094.
48. Ding, X.; Li, Z.; Yang, J.; Hou, J. G.; Zhu, Q. J Chem. Phys. 2004, 121, 2558.
49. Fernández, E. M.; Balbás, L. C. J. Phys. Chem. B 2006, 110, 10449.
12
50. Vindigni, F.; Manzoli, M.; Chiorino, A.; Tabakova, T.; Boccuzzi, F. J. Phys. Chem. B 2006, 110, 23329.
51. Boccuzzi, F.; Chiorino, A. J. Phys. Chem. B 2000, 104, 5414.
52. Lopez, N.; Nørskov, J. K. J. Am. Chem. Soc. 2002, 124, 11262.
53. Bond, G. C.; Thompson, D. T. Gold Bull. 2000, 33, 41.
54. Lee, T. H.; Ervin, K. M. J. Phys. Chem. 1994, 98, 10023.
55. Prestianni, A.; Martorana, A.; Labat, R.; Ciofini, I.; Adamo, C. J. Phys. Chem. B 2006, 110, 12240.
56. Comotti, M.; Li, W. C.; Spliethoff, B.; Schüth, F. J. Am. Chem. Soc. 2006, 128,
917.
57. Neukermans, S.; Janssens, E.; Tanaka, H.; Silverans, R. E.; Lievens, P. Phys. Rev. Lett. 2003, 90, 033401.
58. Chretien, S.; Gordon, M. S.; Metiu, H. J. Chem. Phys. 2004, 121, 9931.
59. Yumura, T.; Amenomori, T.; Kagawa, Y.; Yoshizawa, K. J. Phys. Chem. A 2002, 106, 621.
60. Lin, H.-Y.; Chen, Y.-W.; Wang, W.-J. J. Nanopart. Res. 2005, 7, 249.
61. Li, P.; Miser, D. E.; Rabiei, S;. Yadav, R. T.; Hajaligol, M. R. Appl. Catal. B
2003, 43, 151.
62. PalDey, S.; Gedevanishvili, S.; Zhang, W.; Rasouli, F. Appl. Catal. B 2005, 56, 241.
63. Reddy, B. V.; Khanna, S. N. Phys. Rev. Lett. 2004, 93, 068301.
64. Reddy, B. V.; Rasouli, R.; Hajaligol, M. R.; Khanna, S. N. Fuel 2004, 83, 1537.
65. Wittborn, A. M. C.; Costas, M. Blomberg, M. R. A.; Siegbahn, P. E. M. J Chem. Phys. 1997, 107, 4318.; Cox, D. M.; Reichmann, K. C.; Trevor, D. J.; Kaldor, A. J. Chem. Phys. 1988, 88, 111.; Morse, M. D.; Geusic, M. E.; Heath, J. R.; Smalley, R. E. J. Chem. Phys. 1985, 83, 2293.; Föhlisch, A.; Nyberg, M.; Bennich, P.; Triguero, L.; Hasselström, J.; Karis, O.; Pettersson, L. G. M.; Nilsson, A. J. Chem. Phys. 2000, 112, 1946.
13
66. Yumura, T.; Amenomori, T.; Kagawa, Y.; Yoshizawa, K. J. Phys. Chem. A 2002 106, 621.; Li, X.-Q.; Zhang, W.-X. Langmuir 2006, 22, 4638.; Wu, K-C.; Tung, Y-L.; Chen, Y-L.; Chen, Y-W. Appl. Catal. B 2004, 53, 111.; Lin, H-Y.; Chen, Y-W.; Wang, W-J. J. Nanopart. Res. 2005, 7, 249.; Guczi, L.; Frey, K.; Beck, A.; Petõ, G.; Daróczi, C. S.; Kruse, N.; Chenakin, S. Appl. Catal. A 2005, 291, 116.; PalDey, S.; Gedevanishvili, S.; Zhang, W.; Rasouli, F. Appl. Catal. B 2005, 56, 241.
67. Hintz, P. A.; Ervin, K. M. J. Chem. Phys. 1994, 100, 5715.; Hintz, P. A.; Ervin,
K. M. J. Chem. Phys. 1995, 103, 7897.; Lian, L.; Su, C.-X.; Armentrout, P. B. J. Chem. Phys. 1992, 96, 7542.; Fielicke, A.; von Helden, G.; Meijer, G.; Pedersen, D. B.; Simard, B.; Rayner, D. M. J. Chem. Phys. 2006, 124, 194305.; Stevens, A. E.; Feigerle, C. S.; Lineberger, W. C. J. Am. Chem. Soc. 1982, 104, 5026.; Parks, E. K.; Kerns, K. P.; Riley, S. J. J. Chem. Phys. 2000, 112, 3384.; Kerns, K. P.; Parks, E. K.; Riley, S. J. J. Chem. Phys. 2000, 112, 3394.; Vajda, Š.; Wolf, S.; Leisner, T.; Busolt, U.; Wöste, L. H.; Wales, D. J. J. Chem. Phys. 1997, 107, 3492.
68. Russon, L. M.; Heidecke, S. A.; Birke, M. K.; Conceicao, J.; Morse, M. D.;
Armentrout, P. B. J. Chem. Phys. 1994, 100, 4747.; Kapiloff, E.; Ervin, K. M. J. Phys. Chem. A 1997, 101, 8460.; Guo, B. C.; Kerns, K. P.; Castleman, A. W., Jr. J. Chem. Phys. 1992, 96, 8177.; Guo, B. C.; Kerns, K. P.; Castleman, A. W., Jr. J. Phys. Chem. 1992, 96, 6931.
69. Nagase, K.; Zheng, Y.; Kodama, Y.; Kakuta, J. J. Catal. 1999, 187, 123.; Huang,
T.-J.; Tsai, D.-H. Catal. Lett. 2003, 87, 173.; Sueyoshi, T.; Sasaki, T.; Iwasawa, Y. Chem. Phys. Lett. 1995, 241, 189.; Leuchtner, R. E.; Harms, A. C.; Castleman, A. W., Jr. J. Chem. Phys. 1990, 92, 6527.; Jernigan, G. G.; Somorjai, G. A. J. Catal. 1994, 147, 567.
Chapter 2
Experimental Setup
2.1 Introduction
The experiments discussed herein were conducted on a guided ion beam (GIB)
mass spectrometer. The GIB mass spectrometer is capable of studying the reactivity of
ion-molecule reactions and collision induced dissociation (CID) of mass selected cluster
ions. The design of this apparatus has been described in detail previously (1-2).
Figure 2-1 shows a schematic of the GIB with the different chambers and individual
components labeled. The instrument consists of a laser vaporization source, quadrupole
mass selector and analyzer, octopole reaction cell, a vacuum system, and a detection and
data collection system. Below is a discussion of the individual components and the
conditions with which the transition metal and noble metal oxide experiments were
conducted.
2.2 Vacuum system
The GIB apparatus has three regions, the source, reaction, and detection, each
with its own vacuum pumps in order to maintain differential pumping. The source region
is pumped by a 6” diffusion pump (Varian VHS-6) with a pumping speed of 3000 L/s and
backed by a Sargent-Welch mechanical pump (model 1397) which pumps 500 L/min.
15
Figure 2-1: Guided Ion Beam Mass Spectrometer employed in gas phase metal oxide studies.
Gas Inlet
OctopoleReaction Cell
LaserVaporization
Source
Deflectors CEM
Diffusion Pump Diffusion Pump Diffusion Pump
Quadrupole 2Quadrupole 1
DiffusionPump
Conversion dynode
MKS Baratron
Source
Reaction
Detection
Skimmer
Gas Inlet
OctopoleReaction Cell
LaserVaporization
Source
Deflectors CEM
Diffusion Pump Diffusion Pump Diffusion Pump
Quadrupole 2Quadrupole 1
DiffusionPump
Conversion dynode
MKS Baratron
Source
Reaction
Detection
Skimmer
16
This mechanical pump has a switching valve that allows it to rough-out the mixing tank
or chamber as needed. The source region operates at ~5x10-5 Torr and is monitored by a
thermocouple gauge (Fil-tech, model G-100-P). The reaction chamber region is
evacuated by a 6” diffusion pump (Varian VHS-6) and backed with a Sargent-Welch
mechanical pump (model 1397), similar to the source region. The operating pressure
inside the reaction chamber is ~6x10-7 Torr monitored using an ionization gauge and ion
gauge controller (Granville Phillips, model 270). The skimmer serves to separate the
source and reaction chamber effectively creating two differentially pumped chambers.
The detection chamber uses both a 10” diffusion pump (Varian HS-10) with an
effective pumping speed of 4200 L/s and a 6” diffusion pump (Varian NRC-6) having a
pumping speed of 3000 L/s. The 10” pump has a freon cooled baffle and is backed by a
Sargent-Welch mechanical pump, while the 6” is backed by a direct drive mechanical
pump (Varian SD-700) having a pumping speed of 765 L/min. The operating pressure
inside the detection chamber is ~6x10-7 Torr and is monitored by an ionization gauge and
ion gauge controller (Granville Phillips, model 270).
All the diffusion pumps have a manual gate valve to separate the pump from the
chamber. An in-house built safety system monitors the pressure and temperature of the
diffusion pumps. If the vacuum pressure increases to above 10-4 the safety will switch off
the main power. Also, if the pressure of the cooling water decreases below the set
threshold (20 psi) a McDonnell flow switch (FS-1), installed in-line with the cooling
lines, will switch off the main power.
17
All the mechanical pumps and the freon cooling baffle are mounted on spring
isolators (Kinetics Noise Control, FDS 1-35 & FDS-70). This is to dampen the vibrations
of these pumps and allow for stability of the components in the chamber.
2.3 Laser Vaporization Source
The guided ion beam apparatus is coupled to a laser vaporization source (LaVa)
employed in the formation of the ionic clusters and based on the design of Smalley et al.
(3-4). Utilizing a LaVa source, metal oxide clusters of varying stoichiometries can be
produced by ablating a rotating and translating metal rod with the second harmonic (532
nm) of a Nd:YAG laser (Spectra-Physics, INDI-50, 20 Hz). The laser is focused by a 20
cm focal length lens onto a 1/8” metal rod which is continuously rotated at a rate of 1-1.5
revolutions per minute to ensure a fresh metal ablation surface each time the laser is
pulsed. The laser fluence is adjusted between 5 to 20 mJ/pulse depending on the metal
and size of clusters being studied. Generally lower laser power is used for gold because
it is a soft metal, whereas silver is run at higher laser power. Also, it is generally
recognized that lower energy pulses produce larger metal oxide ions and higher energy
pulses generate smaller metal oxides due to the higher energy atomizing the metal.
A pulse valve (General Valve, Series 9, 28 V) is employed to deliver a carrier gas
composed of a mixture of high purity oxygen (99.999%, MG Industries) and ultra high
purity helium (99.9999%, MG Industries). The ratio of oxygen to helium is dependent on
the metal cluster ion being studied and ranges from 1% for iron anionic clusters to 80%
for anionic gold clusters. The pulse valve is operated by a home built pulse valve driver
18
that delivers a pulse of mixed gas into the source region with a fixed backing pressure,
~5.5 to 6.5 atm. The timing of the pulse valve and laser pulse is controlled by a home-
built 20 Hz pattern generator. The pulse valve delay ranges from 20 to 150 μs while the
laser pulse delays range between 300 to 450 μs. It has generally been found that larger
mass clusters, like gold, use a longer gas delay time and shorter laser delay to maximize
the overlap between the pulse nozzle being open (240 μs) and when the laser pulse hits
the rod.
As the mixture of gas is pulsed it is expanded in a waiting room before it passes
over the ablated rod and interacts with the vaporized metal plasma composed of metal
atoms and ions. The helium in the carrier gas serves to induce clustering and thermalize
the particles through collisions. A conical expansion nozzle (shown in Figure 2-2 for the
dual rod source) placed at the exit of the source region aids in the clustering process by
allowing more third-body collisions. Two different nozzles were employed depending on
the metal being studied. For anionic gold oxides, a 51 millimeter long conical nozzle
with a 2 mm inner diameter and a 12 mm conical expansion cone of 30 degree internal
angle was used, while cationic iron oxides had a wider cluster distribution using a 27
millimeter long nozzle composed of a 15 mm long channel with a 4 mm diameter and 12
mm conical expansion cone with a 30 degree total internal angle. When the clusters exit
the nozzle they rapidly expand into the high vacuum region (~10-7 Torr) and undergo
supersonic expansion, which cools the clusters internally to near thermal energy. The
clusters then travel 25 cm in a field free region and are collimated through a 3mm
skimmer having a 50 degree total internal angle.
19
Figure 2-2: Schematic setup of the dual rod LaVa source which includes the beam splitting mirror and conical nozzle.
Pulse nozzle Conical
Expansion
Nozzle
2 rods
Pulse nozzle Conical
Expansion
Nozzle
2 rods
20
A LaVa source able to hold two metal rods and ablate them using one laser was
built based on the design of Wagner et al. (5) in order to study the effects of bimetallic
clusters. An expanded view of the dual rod LaVa source is shown in Figure 2-2. A half
circular mirror was used to split the laser beam from a single laser into two beams able to
ablate two separate rotating and translating 1/4” metal rods. All other aspects of this dual
rod source are similar to the single metal LaVa source describe herein. This source was
only employed for the mixed Ag-Au oxide experiments described in Chapter 4.
2.4 Guided Ion Beam Setup
The mass spectrometer used in the present studies consists of three
radiofrequency components, a mass selecting quadrupole, an octopole reaction cell, and a
mass analyzing quadrupole, shown schematically in Figure 2-1. Each of the components
is aligned with the beam of a HeNe laser which creates the path the ion beam travels
through the instrument.
The molecular ion beam created by the LaVa source is further collimated by a set
of electrostatic lenses and two pairs of ion deflectors before being directed into the first
quadrupole. All the lenses and deflectors here and further down the setup are powered by
205B-01 Bertan power supplies (0-1 kV, 0-30 mA). Each lens is individually set to
optimize the ion intensity and kept at the lowest possible voltages in order to avoid
transferring energy to the ions.
Both quadrupoles used in the experiments are the same dimensions having four
cylindrical stainless steel rods, 9.5 mm in diameter and 20 cm in length. Each rod is
21
symmetrically aligned around a 0.8 cm diameter circle giving an optimal quadrupolar
field for the ions to pass through. The quadrupoles are each mounted in a vented
stainless steel housing that is also used to attach the electrostatic lenses. An external
oscillator is used to produce the rf voltage in order to avoid phase shifts between the two
quadrupoles. A radio frequency of 880 kHz is generated and this signal is amplified by
separate 300 W power supplies (Extrel, model QC-150). These power supplies are
connected through RG-8 coaxial cables connected to feedthroughs on the outside of the
chamber. In addition to the rf component, a dc potential is applied to opposite pairs of
rods and controlled by a mass programmer (Extranuclear Laboratories, model 091-3).
The detectable mass range for the quadrupoles employed in these experiments is 0-4000
amu, which is determined by the rf signal and radius of the rods.
The first quadrupole is used to mass select the cluster ion of interest. By applying
a dc potential to the opposite pairs of rods the quadrupole can act as a mass filter to only
transmit ions of specific a mass/charge (m/e) ratio (6-8). This dc potential in combination
with the rf voltage and angular frequency of the oscillating potential determine which
ions have stable trajectories through the quadrupole. Once the ion is selected in the first
quadrupole it is directed into the octopole reaction cell by a second set of electrostatic
lenses.
The guided ion beam apparatus functions similar to the triple quadrupole, initially
introduced by Teloy and Gerlich (9). However, it employs an octopole reaction cell
instead of a quadrupole. The octopole ion guide, based on the design of Armentrout et al.
(10) using the approach of Denison (11) for calculating the rf potential of non-ideal
multipoles, has eight circular stainless steel rods, each 40.6 cm long and 3.57 mm in
22
diameter. The rods are set in a circle with a radius of 5.18 mm. The rf and dc potentials
are supplied by a home-built rf generator based on the design of Anderson (12). The
generator is powered by an adjustable 0 to 1000 V, 100 mA power supply (Bertan, model
915-1 P). The dc float voltage is set to zero in reactivity studies where the ions
experience near thermal energies; however, in CID experiments the voltage on the
octopole rods may be increased. This voltage is supplied by a bipolar 0-1 kV, 0-15 mA
power supply (Bertan, model 230-01F).
There are several advantages to employing an octopole in these experiments. The
octopole has a higher trapping efficiency, allowing 4π collection of ionic products, and
does not impart significant off axis energy that would scatter the ions and decrease signal.
The ion collection efficiency depends on the cell geometry and direction of scattered
products and is determined using eq 2-1 for the effective radial potential energy (Ueff) (9,
12),
s
n
oo
oeff U
r
r
rm
VqnrU +⎥
⎦
⎤⎢⎣
⎡=
−22
22
222
4)(
ω. (2-1)
The number of poles is n, q is the charge, m is the mass of the ion, Vo is the rf potential on
the rods, r is the radial distance from the multipole axis, and ro is the inner circle radius of
the rods.
The ion-molecule reactions occur in a stainless steel reaction cell, 11.4 cm long
with a 5 cm diameter, within the octopole rods. The calculated effective path length of
the cell was determined to be 12.9 cm using the trapezoidal pressure fall-off
approximation (2, 13). The neutral reactant gas pressure, carbon monoxide in the present
studies, is monitored by a Bertan capacitance manometer (MKS) controlled by a signal
23
conditioner (model 270). Pressures range between 0-10 mTorr for cationic and 0-20
mTorr for anionic cluster studies.
The reaction products generated in the octopole are then focused by a third set of
electrostatic lenses into the second quadrupole. The second quadrupole, employed as a
mass analyzer, is operated in rf mode only in order to separate ions with differing m/e
ratios. The products are then directed into the channel electron multiplier.
In addition to reactivity studies, CID studies were conducted to aid in the cluster’s
structural determination and bond energy strength. In these experiments the dc float
voltage on the octopole rods was increased to raise the kinetic energy of the ions in the
reaction cell. Teflon screws were used to hold the octopole and second quadrupole in
their housing to effectively isolate them from ground so that voltages could be applied.
An inert gas, xenon, was added to the reaction cell under single collision conditions,
~0.09 mTorr. The mass analyzing quadrupole was then floated by a dc potential. In
order to attract the product ions, the potential of the quadrupole needed to be greater than
the birth potential of the ions, ie. the potential of the octopole. The third set of lenses,
extraction lenses, was increased to attract the ions also. By increasing the pole bias and
voltage on the lenses, signal during these studies greatly diminished, therefore only metal
oxides with significant signal initially were studied.
2.5 Detection and Data Acquisition
The mass analyzed product ions are then directed into a channel electron
multiplier, CEM, (Detector Technology, Inc., model 402A-H) by a single electrostatic
24
lens. The CEM has a conversion dynode which is used to attract the ions with a high
voltage, ±5000 V. The channeltron operates at the same polarity for positive and
negative ions, in which secondary electrons and secondary positive ions are detected,
respectively. In the case of cations, the impact of the cation with the dynode creates a
cascade of secondary electrons that are then accelerated by the gradient electrostatic field
(14). The CEM operates with a voltage drop from approximately -2300 V on the front to
ground on the rear of the channeltron. This decrease directs the secondary electrons
through the detector generating a large gain of approximately 107 to 108. Similarly for
anions, a +5000 V dynode attracts the negative ions, creates a cascade of secondary
positive ions which are then accelerated to the -2300 V to ground drop across the
channeltron. The CEM is operated in pulse counting mode to distinguish spurious noise
from input signal.
The raw signal is filtered and sent through a 50 MHz preamplifier-discriminator
pad (MIT, model F-100T). The ion pulses are counted by a linear ratemeter (Mech-
Tronics, model 777) and also sent to a multi-channel scalar (MCS) card
(Tennelec/Nucleus, Inc., MCS-II) installed in a PC. Before entering the MCS card, the
signal is sent through a home-built time-resolved mass gate that filters random noise and
fast ions. The data is converted into mass spectra and analyzed by Grams/32 software
package.
25
References
1. Bell, R. C.; Zemski, K. A.; Justes, D. R.; Castleman, A. W., Jr. J. Chem. Phys. 2001, 114, 798.
2. Bell, R. C. Ph.D. Thesis Dissertation, The Pennsylvania State University 2000.
3. Smalley, R. E. Laser Chem. 1983, 2, 167.
4. Dietz, T. G.; Duncan, M. A.; Powers, D. E.; Smalley, R. E. J. Chem. Phys. 1981, 74, 6511.
5. Wagner, R. L.; Vann, W. D.; Castleman, A. W., Jr. Rev. Sci. Instrum. 1997, 68,
3010.
6. Lawson, G.; Todd, J. F. J. Chem. Brit. 1972, 8, 373.
7. Miller, P. E.; Denton, M. B. J. Chem. Ed. 1986, 63, 617.
8. Campana, J. E. Int. J. Mass Spec. Ion Physics 1980, 33, 101.
9. Teloy, E.; Gerlich, D. Chem. Phys. 1974, 4, 417.
10. Ervin, K. M.; Armentrout, P. B. J. Chem. Phys. 1985, 83, 166.
11. Denison, D. R. J. Vac. Sci. Technol. 1971, 8, 266.
12. Landau, L. D.; Lifshitz, E. M. Mechanics, 3rd Ed. Oxford: New York 1976, 3, 53.
13. Rosion, S.; Rabi, I. I. Phys. Rev. 1935, 48, 373.; Amdur, I.; Jordon, J. E. Adv. Chem. Phys. 1966, 10, 29.
14. Strobel, H. A.; Heineman, W. R. Chemical Instrumentation: A Systematic Approach John Wiley and Sons: New York, 1989, 805.
Chapter 3
Comprehensive Study of the Reactivity of Gold Oxide Cluster Cations (AuxOy+;
x=1-4, y=1-5) in the Presence of Carbon Monoxide
3.1 Introduction
Nano-scale gold particles are shown to be catalytically active species useful for
the oxidation of carbon monoxide to carbon dioxide (1) at low temperatures (2-3).
However, the reaction mechanism and intermediates of active gold species are still under
much debate. Controversy exists over the role of charge state in gold nanoparticles and
most past studies have focused on the anionic gold species (4-5). For anionic gold, the
existence of surface oxygen vacancies in support material enables the transfer of
electrons from the support to the gold (6). Therefore, it has been suggested that anionic
gold is the active site for weakening/breaking molecular oxygen and allowing for CO
oxidation to proceed (4, 7-8).
However, size-selected cationic Au deposition studies showed clusters as small as
Au3 promote the oxidation of CO (9). Numerous studies have provided evidence of both
metallic and cationic gold species present in working catalysts (10-16). Corma and
coworkers observed both cationic and metallic gold played a role in CO oxidation and
their findings established that the cationic species were the catalytic active sites (17).
They proposed Au3+ was an active site for the oxidation of CO, leaving both Au+ and Au0
reduced species. While some studies advocate the influence of Au+ sites toward
27
oxidation activity (18-19), others report that CO on Au+ sites are detrimental to the
oxidation of CO (20).
Fierro-Gonzalez and Gates prepared a gold catalyst and employed XANES
spectroscopy to show the reduction of Au3+ to Au+ in the oxidation of CO, with no
metallic Au being produced (13). Kinetic studies of this catalyst containing only cationic
gold exhibited a decrease in the reaction rate for CO oxidation versus catalysts containing
both cationic and metallic species. In other studies, it was proposed that an electron is
transferred from the metal to the support; this was represented by an FTIR absorbance
spectra band corresponding to CO absorbed to Auδ+ (21). These positive gold sites were
found to form strong Au-CO bonds which consequently decreased the oxidation activity.
Therefore, the influence of charge density and charge transfer between the gold and metal
oxide support plays an important role in catalytic activity (22-26); however, there is little
agreement on the role positive gold sites play in the activity of CO oxidation.
Previously, anionic gold monomer through trimer oxide cluster experiments
conducted in our laboratory found structure-reactivity relationships for CO oxidation and
CO association reactions (27-30). A comparison of their reaction pathways is included in
the discussion herein. This chapter addresses mass selected gold ions reacted with CO,
adding to prior investigations utilizing a flow tube reactor (31). The results offer
important insight into reaction mechanisms responsible for driving condensed phase
catalysis, with particular emphasis on the role of charge state.
28
3.2 Experimental Methods
Gas-phase cluster studies were performed using a guided ion beam mass
spectrometer coupled to a laser vaporization source, detailed in Chapter 2. Briefly, the
second harmonic of a Nd:YAG laser was used to ablate a rotating and translating gold
rod (SurePure Chemetals, 99.99% purity). At a predetermined time, oxygen seeded in
helium (~20%) was pulsed over the rod, forming hot dense plasma. A 51 millimeter
conical expansion nozzle with a 2 mm channel diameter was placed at the exit of the
source to allow for more collisions and clustering. The clusters underwent supersonic
expansion in a field free region before being passed through a 3 mm skimmer which
created a molecular beam. The cooled clusters were then focused by a set of electrostatic
lenses and deflectors into the first quadrupole. Each cluster species was individually
selected in the first quadrupole and passed through a second set of lenses into the
octopole reaction cell where they were decelerated to thermal energy. Reactant gas,
ranging from 0-10 mTorr, was added to the reaction cell and the pressure was monitored
by a MKS baratron. The products were directed through a third set of lenses to a second
quadrupole where they were mass analyzed and finally detected using a channel electron
multiplier.
3.3 Results
The laser vaporization source employed in these studies allows for the
atomization of oxygen and creates gold oxide clusters with atomic as well as molecular
oxygen.
29
Figure 3-1: A typical mass distribution for cationic gold oxide clusters. The first gold oxide in each series is labeled with the addition of one oxygen atom to subsequent peaks.
150 300 450 600 750 900
Au+ A
u 2+
AuO
2+
Au 3
+
Au 2
O3+
Au 4
O2+
Au 3
O4+
Au 4
O4+
Mass (amu)
Ion
Inte
nsity
(ar
b. u
nits
)
150 300 450 600 750 900
Au+ A
u 2+
AuO
2+
Au 3
+
Au 2
O3+
Au 4
O2+
Au 3
O4+
Au 4
O4+
Mass (amu)
Ion
Inte
nsity
(ar
b. u
nits
)
30
Figure 3-1 shows the typical cluster mass distribution at near thermal energy for gold
monomer through tetramer oxide species. The present reactivity studies include an
investigation of AuO1-5+, Au2O1-5
+, Au3O1-5+, and Au4O2-5
+ species. Au4O+ was the only
atomic O species not produced (see Figure 3-1).
Table 3-1 lists the products of each cluster species in descending order of
intensity reacted with either carbon monoxide or nitrogen. For reaction verification each
cluster was exposed to nitrogen gas under the same experimental conditions of pressure
and energy as CO reactant gas. Any products obtained with nitrogen as the collision gas,
are assumed to represent fragmentation products. These studies aid in identifying
different structural elements and verifying whether atomic or molecular oxygen was
adsorbed to the gold cluster. Oxygen rich clusters, AuO4+, AuO5
+, Au2O2+, Au2O3
+,
Au2O4+, Au2O5
+, Au3O2+, Au3O3
+, Au3O4+, Au3O5
+, Au4O2+, Au4O4
+, and Au4O5+, with
molecular O2 units dissociated an oxygen molecule at higher N2 pressures. Clusters with
more than five oxygen atoms were assumed to contain weakly bound O2 units that
fragmented to form more stable, less oxygen rich clusters. An oxygen atom was lost in
reactions with N2 for the following cluster species: AuO2+, AuO4
+, AuO5+, Au2O5
+, and
Au3O+.
Notice from Table 3-1 that most of the products of reactions with carbon
monoxide contain CO attached to the metal oxide cation. Adsorption of CO onto anionic
gold oxide clusters was not observed for monomer or dimer species and only Au3OCO-
species was observed during previous anionic studies with CO (27-29).
31
Table 3-1: Reaction products for AuxOy+ with CO reactant gas and nitrogen collision gas.
AuxOy+
(x, y) Products with CO
Products with N2
AuxOy+
(x,y) Products with CO
Products with N2
(1,1) AuCO+ none (3,1) Au3CO+ none
(1,2) AuOCO+ none (3,2) Au3+
Au3CO+ Au3
+
(1,3) AuOCO+ Au(CO)2
+ AuCO+
AuO+a
none (3,3) Au3O+
Au3OCO+
Au2OCO+ Au3CO+
Au3O2CO+a
Au3O+
(1,4) AuO2CO+
AuCO+ AuOCO+
AuO2+ (3,4) Au3CO+
Au3+
Au3OCO+
Au3O2+
Au3+
(1,5) AuO3CO+ AuO4
+ AuO2CO+
AuO3+a
AuO3+
AuO4+
(3,5) Au3O3+
Au3CO+ Au3O2CO+
Au3O+a
Au3O3+
Au3O+
(2,1) Au2CO+
Au2+a
none (4,2) Au3CO+ Au4
+ Au4OCO+
Au4O+
Au3+a
Au4+
(2,2) Au2CO+
Au2+
Au2+ (4,3)
Au4O
+
Au3OCO+
Au4O2CO+ Au4O2
+ Au3
+a
Au4O+
(2,3) Au2CO+
AuOCO+ Au2O
+ Au(CO)2
+
Au2O2CO+a
Au2O+ (4,4) Au4CO+
Au3CO+
Au4O+
Au4O3CO+ Au4OCO+a
Au4O2+
Au4+
(2,4) Au2CO+ Au2O2+ (4,5) Au4O3
+ Au4O2CO+
Au4O2+
Au3+a
Au3CO+a
Au4O3+
(2,5) Au2OCO+ Au2O3
+ Au2O4
+
Au2O3+
aDenotes a minor product channel whose relative intensity is less than 1% and not shown in the branching ratios for clarity.
32
3.4 Discussion
For anionic gold, structure-reactivity relationships were developed for the three
different reaction centers identified, including a peripheral oxygen, bridging oxygen, and
molecular oxygen, in order to show the reactions with CO were largely structure driven.
Structural calculations for small cationic gold monomer and dimer oxides conducted by
the Bonačić-Koutecký research group found either molecular oxygen or bridging oxygen
bound to gold in the calculated lowest energy structure (see Figure 3-2). As shown in
Figure 3-2, AuO+ was the only cluster with a peripheral oxygen atom, while AuO3+
contained both peripheral and molecular oxygen. Au2O+ and Au2O3
+ both have a
bridging oxygen atom and even numbered oxygen species, AunO2,4+, have structures with
molecular oxygen attached. The following sections are presented according to the
reactions that the clusters undergo, including oxygen replacement by CO, Au
fragmentation, and CO oxidation, with structural and energetic implications to the
reaction pathways being addressed. According to the observed reactions, the role of
charge density may be an important driving force for cationic reactions.
3.4.1 CO Replacement Reactions
The most common reaction product observed for cationic gold oxide clusters was
the replacement of I) an oxygen atom, or II) molecular O2 group by CO.
I) Oxygen Atom Replacement
It is energetically demanding to break the strong O-O bond in gold oxides that
contain an O2 subunit based on the structures of Figure 3-2.
33
Figure 3-2: Predicted lowest energy structures for AuxOy
+ where x=1-2, y=1-4. The superscripts represent the spin multiplicity. Source: Figure courtesy of the group of Professor V. Bonačić-Koutecký at the Humboldt University in Berlin.
34
Therefore, according to the three structural features identified in anionic studies (27), it
was expected that O atom replacement only occur in clusters that had a peripheral oxygen
atom.
Peripheral Oxygen: The only cluster shown to have oxygen bound to a peripheral
site on cationic gold was AuO+. Figure 3-3(a) shows the branching ratio for AuO+ which
displays a selective reaction channel for O atom replacement by CO. Atomic oxygen
replacement on AuO+ easily occurs as the lone pair of CO is drawn to the positive charge
of the gold atom. This reaction channel was therefore energetically possible and very
selective. It should be noted that CO oxidation was also an energetically feasible channel
since the positive charge on gold would bind oxygen weaker than anionic gold centers.
However, oxidation was not observed for this cluster experimentally.
The CO replacement reaction rate, assuming pseudo-first order kinetics, was
calculated based on the decrease in parent intensity versus the reactant concentration,
shown in eq 3-1.
[ ]tRkI
I
o
r =⎥⎦
⎤⎢⎣
⎡ln (3-1)
Here, Ir and Io is the parent ion intensity with and without the addition of CO reactant
gas, respectively. The pressure of the CO reactant molecule, [R], is calculated as a
concentration, and t is the time the ion spends interacting in the octopole reaction cell.
Figure 3-4 shows the plot of ln[Ir/Io] versus the concentration of the reactant ion, in which
the slope is equal to –kt. In order to calculate the time the ion spends in the reaction cell,
the velocity of the ions is calculated. Since no kinetic energy is added into the reaction
35
cell, the velocity depends only on the supersonic expansion velocity. The supersonic
expansion velocity is given by eq 3-2,
2/1
2
11 ⎥
⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
⎟⎠⎞
⎜⎝⎛ −+
=T
Ts
Mm
RTMv
γγ
,
(3-2)
where MT is the mach number, m is the mass of the ion, and γ is the specific heat capacity
ratio (32). In this case, the heat capacity was calculated for atoms because the carrier gas
was mostly comprised of helium being pulsed through the nozzle. The velocity is equal
to the time divided by the length of the reaction cell, 0.129 cm. Finally, the slope divided
by the time reveals a rate constant of 7.6x10-12 ± 0.5 cm3s-1 for AuO+ as shown in Figure
3-4(a). This reaction rate was calculated from the average of several experiments
collected on different days.
For AuO3+ species, oxygen is bound both peripherally and molecularly. The
reaction products, Au(CO)2+ and AuCO+, lose both of these oxygen subunits as shown in
the branching ratio of Figure 3-5. The binding energy of O2 is approximately 0.6 eV and
the binding of the O atom is much stronger at 1.4 eV (32). The overall energy required to
remove both O2 and O is therefore 2 eV. This pathway is slightly energetically possible
because the binding of CO onto the gold cluster is approximately 2.2 eV. Thus the
energetic pathway of this reaction is feasible because the gain in binding CO is greater
than the energy needed to break both bonds. It can be seen in Figure 3-5 that the product
ion, AuOCO+, is more intense than the other two since only one O2 bond is broken to
form this product.
36
Figure 3-3: Branching ratios for a) AuO+, b) Au2O
+, and c) Au3O+ which all show a
large O atom replacement by CO reaction channel.
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
AuO+
AuCO+
AuO+
AuCO+
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au3O+
Au3CO+
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au3O+
Au3CO+
Au3O+
Au3CO+
Au3O+
Au3CO+
a)
b)
c)
Rel
ativ
e In
tens
ity
Rel
ativ
e In
tens
ity
Rel
ativ
e In
tens
ity
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au2O+
Au2CO+
Au2O+
Au2CO+
b)
37
Figure 3-4: Logarithm of the intensity ratio [Ir/Io] versus CO reactant concentrationcalculated as an ideal gas at a) AuO+, b) Au2O
+, and c) Au3O+.
y = -2.26x10-15xR2 = 0.826
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
b)
y = -4.48x10-15x
R2 = 0.993
-1.6
-1.4-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
c)
y = -4.23x10-15xR2= 0.994
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
a)
y = -2.26x10-15xR2 = 0.826
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
b)
y = -2.26x10-15xR2 = 0.826
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
y = -2.26x10-15xR2 = 0.826
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
b)
y = -4.48x10-15x
R2 = 0.993
-1.6
-1.4-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
c)
y = -4.48x10-15x
R2 = 0.993
-1.6
-1.4-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
c)
y = -4.23x10-15xR2= 0.994
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
a)
y = -4.23x10-15xR2= 0.994
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
a)
38
Bridging Oxygen: For Au2O+ and Au3O
+, shown in Figure 3-2, the oxygen is
bound at a bridging site between two gold atoms. The barriers for removing bridging
oxygen are high and it was expected that no reaction would occur based on the study of
anionic gold clusters containing bridging oxygen (27). A selective reaction channel for O
atom replacement was observed as shown in the branching ratios of Figure 3-3(b-c). This
reaction channel was very selective and a transfer of charges may play a large role in the
observed products. It is possible that oxygen is only bound electrostatically or that
charge density may lower the barriers for O atom removal through creating a strong Au-C
bond. A comparison of the rate of O atom replacement by CO with peripheral oxygen
versus bridging oxygen reveals a slight decrease in the reaction rate. The plots in Figure
3-4 for Au2O+ and Au3O
+ show a kinetic rate of 3.6x10-12 ± 1.6 cm3s-1 and 2.8x10-12 ± 3.1
cm3s-1, respectively, for replacement of the bridging O atom. Therefore the calculated
rate constants are decreased by a factor of 2 compared to AuO+ which contains a
peripheral oxygen atom. However, the large standard deviation in the rate constant
indicates that the kinetic rate is relatively unaffected and the reaction proceeds.
Molecular Oxygen: The cluster species that undergo atomic oxygen replacement
by CO and also contain adsorbed molecular oxygen are AuO2+, AuO4
+, and Au3O4+.
These species are unique because they seemingly have the ability to break the strong O-O
bond in O2 subunits attached to gold. The branching ratio for AuO2+ is shown in Figure
3-6, and is presented below as a test case. Consider the reaction energies of the following
equations for neutral species:
CO + O2 CO2 + O, (3-3)
Au-O2 + CO Au-CO2 + O. (3-4)
39
Figure 3-5: Branching ratio for AuO3
+ containing a peripheral and molecular oxygen unit in the lowest energy structure calculation. Notice that CO is attached to all the reaction products observed.
Figure 3-6: Branching ratio for AuO2
+ which shows a selective pathway for O atom replacement by CO.
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
AuO3+
AuOCO+
Au(CO)2+
AuCO+
Rel
ativ
e In
tens
ity
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
AuO3+
AuOCO+
Au(CO)2+
AuCO+
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
AuO3+
AuOCO+
Au(CO)2+
AuCO+
Rel
ativ
e In
tens
ity
0
0.25
0.5
0.75
1
0 4 8 12
CO (mTorr)
AuO2+
AuCO2+
0
0.25
0.5
0.75
1
0 4 8 12
CO (mTorr)
AuO2+
AuCO2+
Rel
ativ
e In
tens
ity
40
Equations 3-3 and 3-4 are exothermic by 0.4 eV and 0.6 eV, respectively. Since Au-O2
and CO form AuCO2 in an exothermic reaction, the reaction may proceed if any energetic
barriers are overcome. For eq 3-3 there is a 2.3 eV barrier without gold present and a 4
eV barrier if gold is attached (eq 3-4). The positive charge density may act in lowering
the barriers of this reaction by triplet-singlet spin state flipping (33). This process may
account for the breaking of the strong O-O bond by removing the large barrier to CO2
formation and replacing it with two smaller barriers that are surmountable (less than 0.2
eV). In order for spin state flipping to occur, both singlet and triplet spin states must
cross or lie close in energy on the potential energy surface. A transition to the singlet spin
state would increase the reactivity of the species and account for our experimental
findings. It is interesting to note recent theoretical studies by Vach et al. found mild
molecular collisions were able to induce triplet to singlet oxygen transitions explained by
a nonadiabatic “ladder climbing” mechanism (34). It is also important to point out that
no dimer clusters exhibited the O-O bond breaking. This may be due to the fact that
dimer species possess doublet ground states which are harder to induce transitions.
Other researchers have observed intermediate species that correspond to products
observed herein. Studies by Bernhardt and coworkers describe a proposed catalytic cycle
for anionic gold dimers in the presence of O2 and CO and reveal that Au2CO2- is a short
lived weakly bound species (35). Although this species is not observed in the anionic
case, for cations the strong Au+-C bonds formed may allow for a more stable
intermediate species to be observed in experiments. Additionally, a photoelectron spectra
of anionic, neutral, and cationic silver dimers found a temperature dependent
intermediate of the cationic silver dimer can dissociate O2 and bind atomic oxygen (36).
41
Finally, Kung and coworkers observed an IR peak which they assigned to an OC-Auδ+-
Oδ- species upon CO adsorption and O2 exposure (3). The appearance of products with O
atom loss further substantiates the observed mechanisms in the present studies as
intermediate species. Further calculations are being conducted by our collaborators in the
Bonaćič-Koutecký research group to explore the possible energetic pathways.
II) Molecular Oxygen Replacement
The dominant reaction channel for oxygen rich gold clusters is O2 replacement by
CO. This reaction pathway is observed for the following cluster species: AuO3+, AuO4
+,
AuO5+, Au2O2
+, Au2O3+, Au2O4
+, Au2O5+, Au3O2
+, Au3O3+, Au3O4
+, Au3O5+, Au4O3
+,
Au4O4+, Au4O5
+. By comparing collisional products observed in studies with N2 with CO
products (Table 3-1), this reaction channel was identified. CO association was also a
possible reaction pathway in the case of oxygen rich clusters, AuO4+ and AuO5
+, since O2
was lost under nitrogen. However, the percentage of O2 loss observed with nitrogen was
lower than products produced with CO reactant gas. This is evidence that the chemical
reaction, O2 replacement by CO, occurs. As described previously the bond energy for O2
on Au is ~0.6 eV which is overcome from the energy released in binding CO to the
cluster.
A maximum of two CO molecules bound to gold, as demonstrated in Figure 3-5
for AuO3+, was observed. The only products identified that associate two CO molecules
were Au(CO)2+ from AuO2
+, AuO3+, Au2O
+, and Au2O4+ cluster species and Au3O(CO)2
+
produced from Au4O5+. Both AuCO+ and Au(CO)2
+ were found to be more stable
species than Au and CO separately as shown in density functional theory (DFT)
calculations (37-38). Products with a maximum of two CO molecules bound to the gold
42
clusters were also observed in flow tube reactor experiments (31). Previous flow tube
studies verified the initial appearance of AuCO+, Au(CO)2+, AuO2CO+, Au2CO+, and
Au2OCO+ species and with additional CO reactant gas each of these species decreased or
disappeared (31). While there was no observance of the addition of Au(CO)3+ or species
with increasing numbers of CO attached, these species with CO attached may be key
reaction intermediates. The presence of both oxygen and carbon monoxide on a gold
cluster is important in CO oxidation for changing the electronic structure of the cluster
and allowing for the activation of bonds (39).
The clusters most selective for replacement of an oxygen atom or molecule by CO
however, may not be good candidates in the design of more active catalysts. When CO
binds to a Au cluster, it may displace the oxygen necessary to participate in the CO
oxidation. Studies have shown that too much CO can poison active gold sites (6, 9). If
CO binds to the cluster first, it may block O2 activation. It is interesting to note that on
free gold anions, O2 and CO has been found to cooperatively co-adsorb to the cluster with
each preferring different sites (40). From the observed CO attachment products and the
strength of Au+-CO bonds, these studies may show that O2 and CO prefer the same site to
adsorb, thus creating a competitive adsorption and decreasing the oxidation channel.
3.4.2 Au Fragmentation/Replacement
The fragmentation of a Au atom and replacement by CO was also observed in
select larger cluster species including Au4O2+ and Au4O3
+ shown in Figure 3-7.
43
Figure 3-7: Branching ratios for a) Au4O2+ and b) Au4O3
+ and their respective products with increasing carbon monoxide reactant gas.
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au4O3+
Au4O+
Au3OCO+
Au4O2CO+
Au4O2+R
elat
ive
Inte
nsit
y
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au4O3+
Au4O+
Au3OCO+
Au4O2CO+
Au4O2+R
elat
ive
Inte
nsit
yR
elat
ive
Inte
nsit
y
0
0.25
0.5
0.75
1
0 3 6 9 12CO (mTorr)
Au4O2+
Au4+
Au3CO+
Au4OCO+
Au4O+R
elat
ive
Inte
nsit
y
0
0.25
0.5
0.75
1
0 3 6 9 12CO (mTorr)
Au4O2+
Au4+
Au3CO+
Au4OCO+
Au4O+
)
)
a)
b)
44
There was dissociation of Au, AuO, AuO2, AuO3, AuO4, and AuO5 from the selected
cluster, which was not directly observed in our experiments. Assignments were based on
the selected cluster minus the detected ionic product. For example, Figure 3-7 shows
Au3OCO+ as a product from Au4O3+ suggesting fragmentation of neutral AuO2. No
fragmentation products were observed with nitrogen experiments under similar
experimental conditions of pressure and energy; therefore, CO was responsible for
activating the fragmentation of a Au-O or Au-Au bond in the cluster. It is reasonable for
Au to fragment from the selected metal cluster since Au-C bonds are stronger than Au-
Au bonds (33). Therefore these reaction products show that the weaker bond (Au-Au)
was broken and replaced by the stronger bond (Au-C), further evidence that Au-N was
not a strong enough bond to fragment the cluster.
3.4.3 Oxidation
Au4O2+ and Au4O3
+ were identified to be slightly active toward the oxidation of
CO; however, this was not the dominant reaction pathway as shown in the branching
ratios of Figure 3-7. Calculations on larger cationic gold clusters were not conducted;
therefore, the structures for these two species are not known. However, according to the
monomer and dimer structures in Figure 3-2, Au4O2+ and Au4O3
+ are expected to contain
bridging and molecular oxygen groups.
Heiz, Landman and coworkers showed anionic gold clusters were very size
dependent and that Au8 deposited on MgO was the smallest size cluster to catalyze the
oxidation of CO (41). Here, at least four gold atoms are necessary in order to observe the
45
oxidation reaction with a positive charge. These studies show the importance of size on
reactivity and the ability of charge to redistribute on larger clusters of gold to possibly aid
in the oxidation of CO.
3.5 Comparison to Anionic Gold Oxides
As discussed throughout this chapter, there were remarkable differences between
small anionic and cationic gold oxide clusters and the reaction pathways observed in the
presence of CO. For anionic monomer and dimer gold oxide species, AuO-, AuO3-,
Au2O-, Au2O3
-, and Au2O4-, CO oxidation was found to occur (27, 29); whereas, Au-CO
bonds were prevalent for cationic clusters. Both AuCO+ and Au(CO)2+ were calculated
by DFT to be more stable species than Au and CO separately (37-38) due to shorter CO
bond lengths (42). Gold carbonyl cations are generally easier to form compared to anions
because of the ability to form strong Au+-CO bonds (5, 43). The lone pair electrons on
CO bind to positive gold which has deficient electron density allowing for σ donation
from the sσ* orbital of CO into the metal σ orbital (44). Carbon monoxide saturation
studies of bare cationic gold clusters showed that uptake of CO was dependent on the
cluster size, geometry and available binding sites while anionic cluster uptake of CO was
governed by different mechanisms (38). Weaker CO bonds are formed with anionic gold
due repulsion between the filled metal s orbital and CO sσ* orbital (44). This difference
in the behavior of anionic and cationic gold is an example of how charge density affects
the reactivity of small gold clusters.
46
Cations also were more efficient in their respective reaction pathways compared
to gold anionic clusters. As the kinetic analysis of oxygen replacement by CO shows a
rate constant on the order of 10-12 cm3s-1, the anionic gold oxides that underwent CO
oxidation were on the order of 10-13 cm3s-1 (27).
3.6 Conclusions
This chapter provides support for the condensed phase catalyst study of Gates et
al., which showed that cationic gold alone was able to oxidize CO to CO2 (13). Gas phase
studies of cationic gold clusters found both Au4O2+ and Au4O3
+ species to be the only
active stoichiometries in the oxidation of CO to CO2. Although this reaction channel was
identified, it was not a very selective pathway. Competition between other pathways
including replacement by CO reduced the efficiency of the oxidation channel. Further
calculations need to be conducted and a study involving larger clusters would help
elucidate trends and establish a relationship for the oxygen atom transfer pathway. Small
cationic gold oxide clusters were found to be less active and selective in the oxidation
mechanism than their anionic counterparts.
These experiments focused on the oxygen replacement by CO reaction pathway.
Findings show O2 molecules were fragmented to form products with a stoichiometry of
AuxOy-1CO+. Although the energetic pathways of this O atom replacement channel are
not clear, a possible mechanism to account for the experimental findings was proposed.
Therefore the charge state and the strength of CO binding have a great effect on the
observed reaction pathways.
47
References
1. Haruta, M.; Yamada, N.; Kobayashi, T.; Iijima, S. J. Catal. 1989, 115, 301.
2. Min, B. K.; Alemozafar, A. R.; Pinnaduwage, D.; Deng, X.; Friend, C. M. J. Phys. Chem. B 2006, 110, 19833.
3. Henao, J. D.; Caputo, T.; Yang, J. H.; Kung, M. C.; Kung, H. H. J. Phys. Chem. B 2006, 110, 8689.
4. Böhme, D. K.; Schwarz, H. Angew. Chem. Int. Ed. 2005, 44, 2336.
5. Okumura, M.; Kitagawa, Y.; Haruta, M.; Yamaguchi, K. Appl. Catal. A 2005, 291, 37.
6. Yoon, B.; Häkkinen, H.; Landman, U.; Wörz, A. S.; Antonietti, J. M.; Abbet, S.; Judai, K.; Heiz, U. Science 2005, 307, 403.
7. Yoon, B.; Häkkinen, H.; Landman, U. J. Phys. Chem. A 2003, 107, 4066.
8. Stolcic, D.; Fischer, M.; Ganteför, G.; Kim, Y. D.; Sun, Q.; Jena, P. J. Am. Chem.
Soc. 2003, 125, 2848.
9. Lee, S.; Fan, C.; Wu, T.; Anderson, S. L. J. Am. Chem. Soc. 2004, 126, 5682.
10. Guzman, J.; Gates, B. C. J. Am. Chem. Soc. 2004, 126, 2672.
11. Bond, G. C.; Thompson, D. T. Gold Bulletin 2000, 33, 41.
12. Guzman, J.; Gates, B. C. J. Phys. Chem. B 2002, 106, 7659.
13. Fierro-Gonzalez, J. C.; Gates, B. C. J. Phys. Chem. B 2004, 108, 12529.
14. Fu, Q.; Deng, W.; Saltsburg, H.; Flytzani-Stephanopoulos, M. Appl. Catal. B 2005, 56, 57.
15. Costello, C. K.; Yang, J. H.; Law, H. Y.; Wang, Y.; Lin, J.-N.; Marks, L. D.;
Kung, M. C.; Kung, H. H. Appl. Catal. A 2003, 243, 15.
16. Costello, C. K.; Kung, M. C.; Oh, H.-S.; Wang, Y.; Kung, H. H. Appl. Catal. A 2002, 232, 159.
48
17. Guzman, J.; Carrettin, S.; Corma, A. J. Am. Chem. Soc. 2005, 127, 3286.
18. Minicó, S.; Scirè, S.; Crisafulli, C.; Visco, A. M.; Galvagno, S. Catal. Lett. 1997, 47, 273.
19. Margitfalvi, J. L.; Fási, A.; Hegedűs, M.; Lónyi, F.; Gőbölös, S.; Bogdanchikova,
N. Catal. Today 2002, 72, 157.
20. Dekkers, M. A. P.; Lippits, M. J.; Nieuwenhuys, B. E. Catal. Lett. 1998, 56, 195.
21. Boccuzzi, F.; Cerrato, G.; Pinna, F.; Strukel, G. J. Phys. Chem. B 1998, 102, 5733.
22. Kung, H. H.; Kung, M. C.; Costello, C. K. J. Catal. 2003, 216, 425.
23. Haruta, M.; Daté, M. Appl. Catal. A 2001, 222, 427.
24. Oh, H. S.; Costello, C. K.; Cheung, C.; Kung, H. H.; Kung, M. C. Stud. Surf. Sci. Catal. 2001, 139, 375.
25. Fierro-Gonzalez, J. C.; Gates, B. C. Langmuir 2005, 21, 5693.
26. Arrii, S.; Morfin, F.; Renouprez, A. J.; Rousset, J. L. J. Am. Chem. Soc. 2004, 126, 1199.
27. Kimble, M. L.; Moore, N. A.; Johnson, G. E.; Castleman, A. W., Jr.; Bürgel, C.;
Mitrić, R.; Bonaćič-Koutecký, V. J. Chem. Phys. 2006, 125, 204311.
28. Kimble, M. L.; Bürgel, C.; Moore, N. A.; Mitrić, R.; Bonaćič-Koutecký, V.; Castleman, A. W., Jr. “Is O2 Dissociation Sufficient to Effect the Oxidation of CO on Anionic Gold Oxide Clusters?” in prep.
29. Kimble, M. L.; Castleman, A. W., Jr.; Mitrić, R.; Bürgel, C.; Bonaćič-Koutecký,
V. J. Am. Chem. Soc. 2004, 126, 2526.
30. Kimble, M. L. Ph. D. Thesis The Pennsylvania State University, 2005.
31. Kimble, M. L.; Castleman, A. W., Jr. Int. J. Mass Spec. 2004, 233, 99.
32. Anderson, J. B.; Fenn, J. B. Phys. Fluids 1965, 8, 780.
33. Correspondence with collaborators of the Bonaćič-Koutecký research group.
34. Vach, H.; Nguyen, R.-N. V.; Timerghazin, Q. K.; Peslherbe, G. H. Phys. Rev. Lett. 2006, 97, 143402.
49
35. Bernhardt, T. M.; Socaciu-Siebert, L. D.; Hagen, J.; Wöste, L. Appl. Catal. A
2005, 291, 170.
36. Socaciu-Siebert, L. D.; Hagen, J.; Le Roux, J.; Popolan, D.; Vaida, M.; Vajda, S.; Bernhardt, T. M.; Wöste, L. Phys. Chem. Chem. Phys. 2005, 7, 2706.
37. Liang, B.; Andrews, L. J. Phys. Chem. A 2000, 104, 9156.
38. Fielicke, A.; von Helden, G.; Meijer, G.; Pederson, D. B.; Simard, B.; Rayner, D.
J. Am. Chem. Soc. 2005, 127, 8416.
39. Sterrer, M.; Yulikov, M.; Risse, T.; Freund, H. J.; Carrasco, J.; Illas, F.; Di Valentin, C.; Giordano, L.; Pacchioni, G. Angew. Chem. Int. Ed. 2006, 45, 1.
40. Wallace, W. T.; Whetten, R. L. J. Am. Chem. Soc. 2002, 124, 7499.
41. Sanchez, A.; Abbet, S.; Heiz, U.; Schneider, W.-D.; Häkkinen, H.; Barnett, R. N.;
Landman, U. J. Phys. Chem. A 1999, 103, 9573.
42. Wu, X.; Senapati, L.; Nayak, S. K.; Selloni, A.; Hajaligol, M. J. Chem. Phys. 2002, 117, 4010.
43. Prestianni, A.; Martorana, A.; Labat, F.; Ciofini, I.; Adamo, C. J. Phys. Chem. B
2006, 110, 12240.
44. Bernhardt, T. M. Int. J. Mass Spec. 2005, 243, 1.
Chapter 4
Reactivity of Ag2Oy+,- and Bimetallic AgAuOy
+,- (y=1-4) Ionic Clusters with CO
4.1 Introduction
In contrast to the widely investigated structural, electronic, and reactivity
properties of pure gold (1-3) and silver clusters (4-7), there is considerably less
information available for bimetallic silver-gold clusters (8-12). Calculations by Bonačić-
Koutecký and coworkers illustrated the role of charge transfer on catalytic activity and
the differences in s and d electrons for bonding in bimetallic systems compared with pure
gold or silver (13). Due to the larger relativistic effects of gold, where the 6s and 5d
orbitals lie close in energy, there is significant contribution from the d electrons. For
silver clusters, the s and d shells are energetically well separated and findings for
bimetallic Ag-Au clusters exhibit structures similar to pure silver (14-15).
It has been theoretically predicted that the interaction of AgAu- with O2 and CO
would promote the oxidation of CO (13-14). Charge transfer from Ag to Au would
create negatively charged gold subunits (13). Negatively charged gold was expected to
be the active site in the oxidation of CO. Indeed, previous experiments of pure gold
anions found the oxidation of CO to CO2 on Au2O-, Au2O3
- and Au2O4- (16-17).
However, positively charged Ag species may provide an active site that enhances CO
oxidation through O-O bond activation which has been observed in previous pure silver
studies (5, 18). Wang and coworkers found a synergistic effect between silver and gold
51
that resulted in higher catalytic activity with silver proposed to be a key factor in oxygen
activation (19). Although reaction schemes for the oxidation of CO are proposed, no
agreement has been reached. In fact, experiments by Bernhardt et al. have reported no
reactive behavior towards CO with Au2-, Ag2
- or AgAu- (4).
Kappes et al. have studied cationic AgmAun+ clusters with CO and observed
cluster carbonyls of the form AgmAunCO+ (11). They also revealed that CO binds to the
gold atoms, with binding energies ranging from 0.77 to 1.09 eV. Calculations of surface
Ag(100) find CO adsorbs with lower energies ranging from 0.12 to 0.19 eV (7).
Structural calculations of larger clusters, AgmAun0,+,- where n+m ≥ 3, revealed gold atoms
to be embedded on Ag in exposed positions and also found hetero-bonding more
prevelant than homo-nuclear bonding in order to enable the withdraw of charge from Ag
atoms (13).
Through experiments and theory some information on the properties of mixed
clusters has been achieved; however, the active sites and mechanisms for the reactivity of
silver-gold clusters are still not fully understood. Comparison of the reactions of pure
metal oxides with CO to experiments of bimetallic oxides can aid in determining the
influence that each individual metal exhibits and the changes that occur in reactivity
when another metal is present. This chapter focuses on elucidating the role of charge
state and composition of Ag-Au oxide clusters in the presence of CO.
52
4.2 Experimental
Experiments were conducted utilizing a guided ion beam mass spectrometer
explained in detail in Chapter 2. Briefly, the clusters were formed using the modified
laser vaporization source which employed the second harmonic of a Nd:YAG laser to
ablate two rotating and translating metal rods (gold, silver, SurePure Chemetals). A half
circular mirror was used to split the laser beam from a single laser into two beams able to
ablate two separate metal rods. Oxygen, approximately 20% for cations and 50% for
anion clusters, seeded in helium was pulsed over the two rods. After exiting the source
through a 51 millimeter conical expansion nozzle, the clusters were cooled by supersonic
expansion in a field free region. The clusters were then passed through a 3 mm skimmer
and directed by a set of electrostatic lenses and deflectors into the first quadrupole for
mass selection of the reactant species. The cluster of interest was focused through a
second set of lenses into the octopole reaction cell. CO reactant gas, ranging in pressure
from 0-10 mTorr for cationic and 0-20 mTorr for anionic clusters, was added to the
reaction cell. The pressure was monitored using a MKS baratron. The remaining
reactant and product clusters were then focused by a third set of lenses into a second
quadrupole for mass analysis and detected by a channel electron multiplier.
4.3 Results
Experiments of mixed Ag-Au dimer oxide clusters reveal similarities and
differences in reaction products compared to the dimer of either pure silver or gold oxide
clusters. Pure silver ionic clusters are presented first followed by bimetallic Ag-Au oxide
53
clusters. For completeness, gold oxide reactions were compared here to pure silver and
bimetallic clusters in the discussion that follows. Pure gold oxide anions were reported
previously (16-17, 20) and detailed information on cationic clusters was discussed in
Chapter 3.
4.3.1 Pure Silver Oxide Ions
The reaction products of pure silver dimers are shown in Table 4-1. The dominant
reaction pathway for cationic silver dimer oxides was oxygen atom replacement by CO.
The product stoichiometry of AgxOy-1CO+ was observed for Ag2O2+, Ag2O3
+ and Ag2O4+.
The following equation, eq 4-1, shows that at multiple collision conditions, the oxidation
of CO may occur on silver cations through cooperative effects.
Ag2O2+ + CO Ag2CO2
+ + CO2 (4-1)
Thereby, two CO molecules may be able to dissociate the strong O-O bond as a result of
multiple collisions and excess CO molecules. This accounts for the first reaction product
observed in Figure 4-1. The following equation, eq 4-2, describes a further reaction
channel that results from the product Ag2CO2+ shown in eq 4-1, as CO2 desorbs from the
silver cation.
Ag2CO2+ Ag2
+ + CO2 (4-2)
This desorption reaction may be possible as the Ag2CO2+ molecule has time to relax and
rearrange. This also gives rise to the other product, Ag2+, detected in the branching ratio.
The fact that two products were observed in the spectra signifies that this reaction may
occur sequentially.
54
Table 4-1: Product list for the dimer of pure silver, pure gold, and mixed Ag-Au oxide clusters in the presence of CO.
Cations Anions
Ag2Oy+ Au2Oy
+ AgAuOy+ Ag2Oy
- Au2Oy- AgAuOy
-
y=1 AgCO+ Au2CO+ AgAuCO+
Ag2+
y=2 Ag2 + Au2
+ AgAu + y=2 No rxn No rxn AgAuO-
Ag2CO2+ Au2CO+ AgAu-
y=3 Ag2O2CO+ Au2CO+ AgAuO2CO+ y=3 AgO- Au2O2- AgAuO-
Ag2O+ Au2O
+ AgAuCO+ AgAu-
AuOCO+ AgAuO+
Au(CO)2+ AgO2
+
y=4 Ag2O3CO+ Au2CO+ AgAuO3CO+ y=4 Au2O3- AgAuO2
-
Ag2CO+ AgAuO2+
Ag2O(CO)2+
Ag2O2(CO)2+
55
Figure 4-1: Branching ratios for a) Au2O2
+, b) AgAuO2+, c) Ag2O2
+, and d) AgAuO2-
showing the products in the reaction with CO reactant gas.
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au2O2+
Au2CO+
Au2+
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
AgAuO2+
AgAu+
0.96
0.98
1
0 7 14 21 28
CO (mTorr)
0
0.01
0.02
AgAuO2-
AgAu-
AgAuO-
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Ag2O2+
Ag2CO2+
Ag2+
Rel
ativ
e In
tens
ity
a)
b)
c)
Rel
ativ
e In
tens
ity
Rel
ativ
e In
tens
ityR
elat
ive
Inte
nsit
y (P
aren
t ion
) Relative Intensity (Product ions)
d)
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au2O2+
Au2CO+
Au2+
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
AgAuO2+
AgAu+
0.96
0.98
1
0 7 14 21 28
CO (mTorr)
0
0.01
0.02
AgAuO2-
AgAu-
AgAuO-
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Ag2O2+
Ag2CO2+
Ag2+
Rel
ativ
e In
tens
ity
a)
b)
c)
Rel
ativ
e In
tens
ity
Rel
ativ
e In
tens
ityR
elat
ive
Inte
nsit
y (P
aren
t ion
) Relative Intensity (Product ions)
d)
56
This sequential mechanism is in contrast to the proposed singlet to triplet transition
discussed in Chapter 3 for AuO2+. First, the doublet ground spin state hinders the
proposed spin transition and no intermediate was observed as in the present case for
silver dimers.
Oxygen atom transfer was observed for Ag2O+ species reacted with CO, while no
oxygen atom was lost with N2. The positive charge on silver may weaken the Ag-O bond
to allow for easy formation of CO2. However, this was a very minor channel. The
dominant channel for Ag2O+ was AgO replacement by CO in which the weaker Ag-Ag
bond was broken forming a stronger Ag-C bond.
Finally, no reaction was detected for Ag2O2- and Ag2O3
- cluster species showed
O2 loss.
4.3.2 Bimetallic Ag-Au Oxide Ions
Typical mass distributions for anionic and cationic clusters presented in Figure 4-
2 show pure silver, pure gold, and mixed silver-gold oxides in the spectrum. The spectra
demonstrate the ability to form mixed dimer cluster species, specifically for clusters of
AgAuOx-,+ where x=1-5, employing the dual rod source described in Chapter 2. The laser
vaporization source allowed for dissociated and molecular oxygen to adsorb onto the
clusters. This enabled the study of Ag-Au oxide clusters with oxygen atoms bound.
Previous calculations have been conducted on the interaction of Ag-Au clusters and
molecular oxygen, but have not focused on dissociated oxygen (14).
57
Figure 4-2: a) Typical anionic silver-gold oxide mass distribution and b) cationic silver-gold oxide mass distribution obtained from dual rod LaVa source.
50 150 250 350 450
150 200 250 300 350 400
AgO
4-
AuO
-AgO
6-
AuO
2-A
uO3-
Ag 2O
3-A
uO4-
Ag 2O
5-
Ag 2O
6-
AuA
gO-
Ag 2O
7-
AuA
gO2-
AuA
gO3-
Ag 2O
8-
AuA
gO4-
AuA
gO5-
AgO
3+
AgO
+
Ag+
AgO
2+
AgO
4+A
u+
AuO
+
AuA
g+
Ag 2
O+
AuA
gO+
AuA
gO3+
AuO
2+
AuA
gO4+
Au 2+
Ag 2O
2+
Ag 2O
3+A
uO3+
Ag 2O
4+
AuA
gO5+
Mass (amu)
Arb
itrar
y In
tens
ity
Arb
itrar
y In
tens
ity
a)
b)
50 150 250 350 450
150 200 250 300 350 400
AgO
4-
AuO
-AgO
6-
AuO
2-A
uO3-
Ag 2O
3-A
uO4-
Ag 2O
5-
Ag 2O
6-
AuA
gO-
Ag 2O
7-
AuA
gO2-
AuA
gO3-
Ag 2O
8-
AuA
gO4-
AuA
gO5-
AgO
3+
AgO
+
Ag+
AgO
2+
AgO
4+A
u+
AuO
+
AuA
g+
Ag 2
O+
AuA
gO+
AuA
gO3+
AuO
2+
AuA
gO4+
Au 2+
Ag 2O
2+
Ag 2O
3+A
uO3+
Ag 2O
4+
AuA
gO5+
Mass (amu)
Arb
itrar
y In
tens
ity
AgO
+
Ag+
AgO
2+
AgO
4+A
u+
AuO
+
AuA
g+
Ag 2
O+
AuA
gO+
AuA
gO3+
AuO
2+
AuA
gO4+
Au 2+
Ag 2O
2+
Ag 2O
3+A
uO3+
Ag 2O
4+
AuA
gO5+
Mass (amu)
Arb
itrar
y In
tens
ity
Arb
itrar
y In
tens
ity
a)
b)
58
The present experiments therefore provide a report of bimetallic Ag-Au adsorbed with O
and O2 and the interactions in the presence of CO. Table 4-1 also lists the products for
mixed silver-gold cationic and anionic dimer clusters reacted with CO. Below, each
ionic dimer cluster is presented according to the amount of oxygen attached to the metal
ion.
AgAuO+,-: For a predissociated oxygen atom attached to cationic Ag-Au, the
bimetallic species behaved similar to the gold dimer oxide. Both Au2O+ and AgAuO+
underwent O atom replacement by a CO molecule. It can be seen in Figure 4-3 the
branching ratio of Au2O+ was more reactive toward O atom replacement compared to the
mixed AgAuO+. A kinetic analysis of the replacement rate constant value reveals the
presence of silver decreased the efficiency of the reaction channel. Figure 4-4 shows
plots of ln[Ir/Io] versus the CO reactant concentration in which the slope is equal to –kt.
Solving for the kinetic rate constant showed Au2O+ has a rate of 3.6x10-12 ± 1.6 cm3s-1
while AgAuO+ is only 1.5x10-12 ± 0.2 cm3s-1. This decrease in the rate may be a
consequence of charge transfer from silver to gold, creating more electron density around
gold and a weaker CO bond.
None of the anionic clusters with one oxygen atom were formed in our laser
ablation source and therefore are not reported.
AgAuO2+,-: The most studied interaction in the literature is of AgAu with O2. The
present studies found that the AgAu+ cation behaved similar to both pure gold and pure
silver with an O2 loss reaction channel shown in Figure 4-1. It is interesting to note that
both pure gold and pure silver also underwent other reaction pathways which were more
dominant than O2 loss. Au2O2+, besides losing O2, also formed Au2CO+.
59
Figure 4-3: Branching ratios for a) Au2O
+ and b) AgAuO+ with increasing CO pressure.
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au2O+
Au2CO+
0
0.25
0.5
0.75
1
0 2 4 6 8CO (mTorr)
AgAuO+
AgAuCO+
Rel
ativ
e In
tens
ity
Rel
ativ
e In
tens
ity
a)
b)
0
0.25
0.5
0.75
1
0 2 4 6 8 10
CO (mTorr)
Au2O+
Au2CO+
0
0.25
0.5
0.75
1
0 2 4 6 8CO (mTorr)
AgAuO+
AgAuCO+
Rel
ativ
e In
tens
ity
Rel
ativ
e In
tens
ity
a)
b)
60
Figure 4-4: Plots of ln[Ir/Io] versus CO reactant concentration for a) Au2O
+ and b) AgAuO+
y = -9.73x10-16xR
2= 0.874
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0
Conc [CO] 1014 mc/cc
ln[I
r/Io]
y = -2.26x10-15xR2 = 0.826
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
a)
b)
y = -9.73x10-16xR
2= 0.874
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0
Conc [CO] 1014 mc/cc
ln[I
r/Io]
y = -2.26x10-15xR2 = 0.826
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
a)
y = -2.26x10-15xR2 = 0.826
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Conc [CO] 1014 mc/cc
ln[I
r/Io]
a)
b)
61
In the case of pure silver, an oxygen atom replacement product was also detected
(Ag2CO2+), as described above in the previous section. Therefore the bimetallic cluster
may aid in enhancing the selectivity one of the reaction channels, O2 loss.
For anionic clusters, Figure 4-1 shows the branching ratio of AgAuO2- which
revealed an oxygen atom transfer pathway and also O2 loss channel. It is difficult
determine whether this is a sequential reaction mechanism, in which AgAuO- transfers its
first oxygen to CO, followed by its second oxygen atom, shown below in eqs 4-3 and 4-4.
AgAuO2- + CO AgAuO- + CO2 (4-3)
AgAuO- + CO AgAu- + CO2 (4-4)
The small intensity of the products and the fact that this reaction occurs very slowly
complicates the ability to definitively assign the mechanism. The behavior of the mixed
metal is different from pure gold in which no reaction is observed and pure silver which
showed no reaction was detected at the highest pressures.
AgAuO3+,-: As the Ag-Au cluster becomes more oxygen rich, there is a shift in the
behavior toward pure silver oxide reactivity. For cationic Ag2O3+ and AgAuO3
+ clusters,
the major product was Ag2O2CO+ and AgAuO2CO+, respectively. This corresponds to an
oxygen atom replacement by a CO molecule and could be an intermediate species for CO
oxidation to CO2. These two clusters also show Ag2O+ and AuAgO+ for silver and
bimetallic species which suggests dissociation of a CO2 molecule for the metal center.
The bimetallic Ag-Au cluster also had a reaction product corresponding to neutral AuO
loss. The AgO2+ product species may arise because the Ag-Au bond is weaker in energy
then the Au-C bond. The dissociation of a metal atom has been observed with larger
62
mixed clusters, Ag1-3Au1-3+, by Neumaier et al. (11). They showed that the CO binding
energies were greater than the activation energy for losing a metal atom.
In the case of anions, mixed AgAuO3- behaved similar to Ag2O3
- with both
species showing loss of molecular oxygen. On the other hand, pure gold oxide produced
oxygen atom transfer to CO.
AgAuO4+,-: Both cationic Ag2O4
+ and AgAuO4+ clusters underwent O atom
replacement by a CO molecule. For the silver-gold cluster, AgAuO2+ was also observed.
In the case of anionic AgAuO4- an O2 loss is the only reaction product detected.
This does not correspond with either pure metal, as silver signal was too low to obtain
and the gold anion dimer underwent oxygen atom transfer.
4.4 Discussion
For both AgAuO+ and AgAuO2+, the bimetallic species exhibit products that
correspond to the reactivity of pure gold dimer oxide. There is evidence, however, that
the reactivity of the mixed species was affected by the presence of silver. For the
reaction of AgAuO2+, there was only one reaction product similar to that detected in pure
gold or silver oxide reactions with CO. Silver may have influenced the gold toward a
selective pathway for O2 loss. This supports the existence of structural and electronic
changes for catalytic systems that contain two different metals and the ability to tune the
properties of a catalyst based on composition.
In the case of AgAuO3+ and AgAuO4
+ when more oxygen is surrounding the
cluster, the bimetallic species produced similar products to pure silver dimer oxides. This
63
may be the result of charge redistribution for highly coordinated silver. It has been
shown that as more oxygen binds to silver, silver induces an O2 bond elongation which
results in activation (4, 21). This would account for the increased dominance by silver as
O-O bonds become weaker and more active.
The major reaction channel observed for AgAuO3-4+ was O atom replacement by
CO. This was an important reaction pathway exhibited by Ag2O2-4+ and select cationic
gold monomer oxide clusters. This pathway for gold oxides was proposed to be the result
of charge density effects which allowed for a spin state transition. However, as outlined
above for silver oxides, the mechanism for silver and bimetallic oxide clusters may differ
from gold oxides. The appearance of other products in the branching ratio suggests that O
atom replacement products may be intermediate species in the oxidation of CO. Other
studies have shown the existence of AgxOy-1CO+ and AuOCO+ intermediate species (5,
18, 22). It is also possible that multiple CO molecules induce cooperative effects to
promote the observed reactions. It has been shown in gold clusters by Whetten and
coworkers that when multiple CO molecules adsorb onto the cluster, changes in the
electronic structure occur (23).
Since the Ag2O4- products with CO are unknown, it is hypothesized that the
pattern in mixed cluster behavior would be similar to the silver behavior. In the case of
AgAuO3+ and AgAuO3
- both reaction products were analogous to silver reactions. It is
logical that since AgAuO4+ corresponds to silver products, that AgAuO4
- would also
likely follow this trend.
64
4.5 Conclusion
The bimetallic Ag-Au oxide studies show that oxygen deficient cationic clusters
behave similarly to pure gold dimers while clusters that are more oxygen rich have
comparable reaction products to pure silver dimers. For anionic clusters, there was less
connection between the pure and mixed clusters. Trends in the behavior have been
hypothesized; however, most often the Ag-Au species behaved differently then either
pure dimer oxides.
Collaboration with the Bonaćič-Koutecký research group to determine possible
structure-reactivity relationships that may be present for several cluster species is
ongoing. The study of larger mixed clusters with multiple gold and silver atoms would
shed light on the effects of size and stoichiometry for the observed reaction pathways.
65
References
1. Böhme, D. K.; Schwarz, H. Angew. Chem. Int. Ed. 2005, 44, 2336.
2. Okumura, M.; Kitagawa, Y.; Haruta, M.; Yamaguchi, K. Appl. Catal. A 2005, 291, 37.
3. Gilb, S.; Weis, P.; Furche, F.; Ahlrichs, R.; Kappes, M. J. Chem. Phys. 2002, 116,
4094.
4. Bernhardt, T. M.; Socaciu-Siebert, L. D.; Hagen, J.; Wöste, L. Appl. Catal. A 2005, 291, 170.
5. Socaciu, L. D.; Hagen, J.; Heiz, U.; Bernhardt, T. M.; Leisner, T.; Wöste, L.
Chem. Phys. Lett. 2001, 340, 282.
6. Weis, P.; Bierweiler, T.; Gilb, S.; Kappes, M. Chem. Phys. Lett. 2002, 355, 355.
7. Qin, C.; Sremaniak, L. S.; Whitten, J. L. J. Phys. Chem. B 2006, 110, 11272.
8. Negishi, Y.; Nakamura, Y.; Nakajima, A.; Kaya, K. J. Chem. Phys. 2001, 115, 3657.
9. Mitrić, R.; Hartmann, M.; Stanca, B.; Bonaćič-Koutecký, V.; Fantucci, P. J. Phys.
Chem. A 2001, 105, 8892.
10. Bauschlicher, C. W., Jr.; Langhoff, S. R.; Partridge, H. J. Chem. Phys. 1989, 91, 2412.
11. Neumaier, M.; Weigend, F.; Hampe, O.; Kappes, M. M. J. Chem. Phys. 2006,
125, 104308.
12. Cottancin, E.; Lermé, J.; Gaudry, M. Pellarin, M; Vialle, J. L.; Broyer, M. Phys. Rev. B 2000, 62, 5179.
13. Bonačić-Koutecký, V.; Burda, J.; Mitrić, R.; Ge, M.; Zampella, G.; Fantucci, P. J.
Chem. Phys. 2002, 117, 3120.
14. Mitrić, R.; Bürgel, C.; Burda, J.; Bonaćič-Koutecký, V.; Fantucci, P. Eur. Phys. J. D 2003, 24, 41.
66
15. Lee, H. M.; Ge, M.; Sahu, B. R.; Tarakeshwar, P.; Kim, K. S. J. Phys. Chem. B 2003, 107, 9994.
16. Kimble, M. L.; Moore, N. A.; Johnson, G. E.; Castleman, A. W., Jr.; Bürgel, C.;
Mitrić, R.; Bonaćič-Koutecký, V. J. Chem. Phys. 2006, 125, 204311.
17. Kimble, M. L.; Castleman, A. W., Jr.; Mitrić, R.; Bürgel, C.; Bonaćič-Koutecký, V. J. Am. Chem. Soc. 2004, 126, 2526.
18. Socaciu-Siebert, L. D.; Hagen, J.; Le Roux, J.; Popolan, D.; Vaida, M.; Vajda, Š.;
Bernhardt, T. M.; Wöste, L. Phys. Chem. Chem. Phys. 2005, 7, 2706.
19. Wang, A. Q.; Chang, C. M.; Mou, C. Y. J. Phys. Chem. B 2005, 109, 18860.; Liu, J. H.; Wang, A. Q.; Lin, H. P.; Mou, C. Y. J. Phys. Chem. B 2005, 109, 40.; Wang, A. Q.; Liu, J. H.; Lin, S. D.; Lin, T. S.; Mou, C. Y. J. Catal. 2005, 233, 186.
20. Kimble, M. L.; Bürgel, C.; Moore, N. A.; Mitrić, R.; Bonaćič-Koutecký, V.;
Castleman, A. W., Jr. “Is O2 Dissociation Sufficient to Effect the Oxidation of CO on Anionic Gold Oxide Clusters?” in prep.
21. Hagen, J.; Socaciu, L. D.; Le Roux, J.; Popolan, D.; Bernhardt, T. M.; Wöste, L.;
Mitrić, R.; Noack, H.; Bonaćič-Koutecký, V. J. Am. Chem. Soc. 2004, 126, 3442.
22. Henao, J. D.; Caputo, T.; Yang, J. H.; Kung, M. C; Kung, H. H. J. Phys. Chem. B 2006, 110, 8689.
23. Wallace, W. T.; Whetten, R. L. J. Am. Chem. Soc. 2002, 124, 7499.
Chapter 5
Influence of Charge State on the Reaction of FeO3+,- with Carbon Monoxide
5.1 Introduction
Stimulated by recent theoretical findings regarding the activity of iron oxide
clusters for CO oxidation (1-2), a systematic experimental and theoretical investigation of
FexOy+,- clusters with CO was undertaken. Recent experiments on iron oxide
nanoparticles show that it may be possible to convert CO to CO2 at low temperatures in
the presence or absence of O2 (3-6). Transition metal oxides with oxygen atoms bound
only to the transition metal without O-O bonding offer an attractive possibility to
overcome the need for a catalyst to weaken/break the O-O bond through charge transfer
and/or bonding. In the past, Pdn and Ptn nanoparticles, and more recently charge Aun
clusters, have attracted attention (7). In comparison to precious metal based catalysts
used in effecting the oxidation of CO, iron oxide is a much more affordable and readily
available material.
The basic issue addressed herein is whether it is possible to use the stored oxygen
in highly oxidized and charged metal clusters to convert CO to CO2. Note that such a
possibility avoids the use of external O2 and hence may allow CO conversion in oxygen
free environments. The ratio of oxygen atoms to metal atoms in bulk metal oxides is
generally less than two. It is, however, possible to generate metal-oxygen clusters with
oxygen to metal ratios much higher than in the bulk (8). Further, unlike bulk, small
(Reproduced in part with permission from Chem. Phys. Lett. 2007, 435, 295-300. Copyright 2007 Elseveir.)
68
clusters offer the possibility of different ionic charge states. Numerous studies involving
charged iron and iron oxide clusters have been able to provide fundamental information
corresponding to bulk phase iron processes (8-12).
In this chapter, FeO3 is presented as the test case. Studies of both the cationic and
anionic clusters show that charge has a dramatic effect on the oxidation behavior. As
opposed to the case of Aun clusters where the findings have been that anions are more
effective in converting CO (7), here FeO3+ cations are far more efficient than the anions.
Also, the reaction pathways for cations are different from those for anions. In most cases,
the reaction proceeds without barriers and hence may allow CO oxidation at all
temperatures.
5.2 Methods
Experiments were conducted using a guided ion beam mass spectrometer
explained in detail in Chapter 2, and described here briefly. The clusters were formed by
laser vaporization employing the second harmonic of a Nd:YAG laser to ablate a rotating
and translating iron rod (PVD Materials Corp., 99.95% purity). Oxygen (1%) seeded in
helium was pulsed over the iron rod into the hot metal plasma where the clusters form.
After exiting the source through a 27 millimeter conical expansion nozzle, the clusters
were cooled by supersonic expansion in a field free region. The clusters were then
passed through a 3 mm skimmer and directed by a set of electrostatic lenses and
deflectors into the first quadrupole for mass selection of the reactant species. The cluster
of interest was focused through a second set of lenses into the octopole reaction cell. CO
69
reactant gas, ranging in pressure from 0-10 mTorr for cationic and 0-20 mTorr for
anionic clusters, was added to the reaction cell. The pressure was monitored by a MKS
baratron. The remaining reactant and product clusters were then focused by a third set of
lenses into a second quadrupole for mass analysis and detected by a channel electron
multiplier.
Theoretical studies were carried out by the Khanna research group using a first
principles electronic structure scheme employing a density functional formalism (13).
The exchange correlation effects were incorporated through a generalized gradient
approximation (GGA) functional as proposed by Perdew, Burke and Ernzerhof (14). The
electronic structure was determined using a linear combination of atomic orbitals
molecular orbital approach. The wave function for the cluster was constructed by a linear
combination of Gaussian type orbitals centered at the atomic positions in the cluster. The
actual calculations were performed using the deMon2k software (15). In this
implementation, an auxiliary function set is used for the variational fitting of the
Coulomb potential (16-17). The numerical integration of the exchange-correlation
energy and potential were then performed on an adaptive grid (18). In the present
studies, we employed the double zeta valence polarized (DZVP) basis sets (19) for C and
O and the Wachters-F basis set (20-22) for Fe. The GEN-A2 auxiliary function set for C
and O and the GEN-A2* auxiliary function set for Fe were used. To determine the
ground state, the configuration space was sampled by starting from several initial
configurations and optimizing the geometry by moving atoms in the direction of forces
until they dropped below a threshold value. Since transition metal atoms are marked by
non-zero spin multiplicities, the ground state determination included investigation over
70
several spin multiplicities. The geometries were optimized without any symmetry
constraint using delocalized internal coordinates with the rational function optimization
(RFO) and the Broyden, Fletcher, Goldfarb and Shanno (BFGS) update in the deMon2K
program (23).
5.3 Results and Discussion
Investigation into the reaction products for both charge states includes a look at
the observed experimental products, cluster structures, and kinetics involved in specific
reaction pathways. The branching ratios for FeO3+,- in the presence of increasing CO
pressure are shown in Figure 5-1. This figure shows the intensity of the parent ion as
well as all the product ions. Notice that the maximum CO pressure for the anion species
is twice that for the cation species. The experimental results reveal three important
findings. (1) The FeO3+ is far more reactive than the FeO3
-. Even though the maximum
pressure in the cation reaction is half the amount used in the anion reaction, the intensity
of the products is far higher for the cations. (2) For FeO3+, the reaction proceeds to the
final stages forming FeO2+, FeO+, and Fe+ as the products. For anions, only FeO2
- is
observed. (3) Only one reaction product with association of CO is observed, namely
FeOCO+. The intensity of this product is initially higher than that of FeO2+ indicating that
it is produced in the initial stages of reaction.
71
0.5
0.75
1
0 2.5 5 7.5 10CO (mTorr)
0
0.05
0.1
0.5
0.75
1
0 5 10 15 20
CO (mTorr)
0
0.05
0.1
FeO3+
FeOCO+
FeO+
Fe+
FeO2+
Rel
ativ
e In
tens
ity
(par
ent
ion)
Relative Intensity (product ions)
Rel
ativ
e In
tens
ity
(par
ent
ion)
Relative Intensity (product ion)
FeO3-
FeO2-
0.5
0.75
1
0 2.5 5 7.5 10CO (mTorr)
0
0.05
0.1
0.5
0.75
1
0 5 10 15 20
CO (mTorr)
0
0.05
0.1
FeO3+
FeOCO+
FeO+
Fe+
FeO2+
Rel
ativ
e In
tens
ity
(par
ent
ion)
Relative Intensity (product ions)
Rel
ativ
e In
tens
ity
(par
ent
ion)
Relative Intensity (product ion)
FeO3-
FeO2-
)
)
Figure 5-1: Branching ratios for a) FeO3
+ and b) FeO3- with increasing CO pressure. The
relative selected ion intensity is plotted on the left-hand axis and the relative product intensities on the right-hand axis. Note that the pressure used in the reactions of cationic FeO3
+ is half the amount used in reactions of anionic FeO3- to acquire reaction products
of the same order of magnitude.
a)
b)
72
Further evidence for this comes from the observation that its intensity is higher than that
of FeO2+ suggesting that it is not the product of association of CO to previously formed
FeO+ but that it is produced directly from the reaction of FeO3+. Theoretical studies on
the ground state geometries of FeOy+,- and the reaction pathways were carried out to shed
light on some of these findings.
Figure 5-2 shows the ground state geometries of FeOy+,- and FeOyCO+,- clusters
containing up to three oxygen atoms. For the FeOy+, clusters, note that in the lowest
energy state, there are no O-O bonds and that all the oxygen atoms are bound to the metal
atom. Our studies on larger sizes of metal clusters indicate that O-O bonds do form at
higher coverage. These will be presented in Chapters 6 and 7. A Mulliken population
analysis of the overall charge density was carried out to identify the location of missing
charge in cation clusters, extra charge in anion species, and any other charge transfers.
The calculated Mulliken charges are marked below the individual atoms. Irrespective of
the charge state, Figure 5-2 shows that there is a charge transfer from iron sites to the
oxygen atoms indicating that the FeO bond is stabilized by such transfers. The change in
the charge on oxygen, therefore, leads to a change in the strength of the Fe-O bond. To
show this more quantitatively, the energy required to remove an O atom and an O2
molecule from the various clusters was calculated. (It should be noted that for some
anionic clusters, it takes less energy to remove O-. In this work, however, the primary
focus is on neutral O and O2.) The relevant energy values are shown in Figure 5-3. Note
that the energy required to remove a single O atom from cationic clusters is significantly
lower than the energy required for removing an O atom from the corresponding anion of
the same size.
73
Figure 5-2: Ground state geometries for O2, CO, CO2, and FeOy
+,- clusters. The bond lengths are given in Angstroms and superscripts indicate the spin multiplicity. The Mulliken charges are marked below each atom. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
74
Even more puzzling is the trend in the energy to remove an O2 molecule. For cations, at a
given size, the energy to remove O2 is far less than to remove O atoms. On the other
hand, it takes approximately the same energy to remove an O2 or an O atom in FeO3-.
These trends are rooted in the fact that the oxides are stabilized by a charge transfer from
Fe to oxygen (see Figure 5-2). The addition of an extra electron facilitates the O-
formation and hence leads to more strongly bound species.
In order to understand the observed products more clearly, the changes in energy
and the various reaction products are depicted in the energy diagram shown in Figure 5-4.
Starting from FeO3+, a CO molecule was placed around the cluster in various
configurations and the system was allowed to relax to determine the ground state
geometry. What was interesting is that the CO molecule approached an O atom leading to
the formation of CO2. As Figure 5-2 shows, the ground state of FeO3CO+ corresponds to
a CO2 molecule bound to FeO2+. To further examine any reaction barrier for the
formation of CO2, the reaction path for the entire process was calculated. Starting from
well separated FeO3+ and CO, the total energy as a function of Fe-C separation was
calculated. For each separation, the position of Fe and that of C were fixed and the
remaining atoms were allowed to relax without any symmetry constraint.
The CO molecule approached an O atom of the FeO3+ cluster, without any barrier,
resulting in the formation of the complex. A profile of the reaction intermediates and
their energies are given in Figure 5-4. The heat of formation is sufficient to drive the
reaction product FeO3CO+ to two possible channels. The first one involves the loss of
CO2 while the other corresponds to the loss of an O2 leaving behind the FeOCO+
complex.
75
Figure 5-3: a) Graph of the energy required to remove an O or O2 from the FeOy+ cluster.
b) Graph of the energy required to remove an O, O2, O- and O2
- from the FeOy- cluster.
Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
b)
a)
76
ΔE (eV) 2FeO3
+ + 1CO 2FeO3CO+ -4.29 2FeO3CO+ 2FeO2
+ + 1CO2 1.64 2FeO2
+ + 1CO 6FeO2CO+ -3.98 6FeO2CO+ 6FeO+ + 1CO2 1.31 6FeO+ + 1CO 4FeOCO+ -2.88 4FeOCO+ 4Fe+ + 1CO2 1.10 2FeO3CO+ 4FeOCO+ + 3O2 2.66
Figure 5-4: Profile of the reaction energy for FeO3+ with CO including the
corresponding change in energy, ΔE, for each reaction step. The superscripts indicate spin multiplicity. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
77
Indeed, experimentally we observe both these products as they lie close in energy and are
lower in energy than the initial reactants.
Starting from FeO2+, an additional CO was allowed to approach the cluster from
various directions. As shown in Figure 5-2, the most favorable path corresponds to a CO
approaching one of the O atoms, forming a CO2 molecule, and resulting in the FeOCO2+
complex. A detailed analysis of the reaction path again showed no reaction barriers. The
heat of formation for this process is 3.98 eV. One can again consider two possibilities,
namely (1) the loss of CO2 leading to FeO+ or (2) the loss of O leading to FeOCO+.
However, our calculations show that the energy required to remove an O atom is 4.73 eV.
Hence, the logical choice is the loss of the CO2 molecule which requires only 1.25 eV.
This is further evidence that the FeOCO+ species observed in Figure 5-1 arises from the
earlier pathway and not this latter channel. It is important to note that the ground state of
FeO+ has a spin multiplicity of 6 while FeO2+ has a spin multiplicity of 2. Hence the
formation of the ground state of FeO+ requires a change in total spin multiplicity.
Starting from the ground state of FeO+, a CO was again allowed to approach the
molecule. The ground state indicates the formation of the CO2 molecule. Again, the
detailed reaction path indicates that there is no barrier, leading finally, to the appearance
of Fe+ as found in the experiments.
Figure 5-5 shows the energetic pathways for the corresponding processes for the
anions. As mentioned before, the removal of an O atom from the anions requires much
more energy than from cations. Hence the energy gain in the formation of CO2 is not as
large as for cations. This is clearly seen in Figure 5-5 that shows that the heats of
formations are significantly reduced.
78
ΔE (eV) 4FeO3
- + 1CO 2FeO3CO- -1.96 2FeO3CO- 4FeO2
- + 1CO2 1.44 4FeO2
- + 1CO 4FeO2CO- -2.47 4FeO2CO+ 4FeO- + 1CO2 2.19 4FeO- + 1CO 4FeOCO- -2.35 4FeOCO- 4Fe- + 1CO2 2.22 4FeOCO- + 1CO 2FeOCOCO- -2.28 2FeOCOCO- 4FeCO- + 1CO2 2.58
Figure 5-5: Profile of the reaction energy for FeO3- with CO including the corresponding
change in energy, ΔE, for each reaction step. The superscripts indicate spin multiplicity. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
79
Starting from FeO3-, our studies indicate that the approaching CO molecule does not
encounter a barrier. In this case, however, the heat of reaction is only 1.96 eV as opposed
to 4.29 eV for FeO3+. Consequently, the reaction channel involving the loss of an O2
molecule that was feasible for the cation, is energetically prohibited. The only pathway is
the formation of FeO2-. This is in agreement with the experimental findings that do not
detect any FeOCO- product. We also found that while the oxidation of FeO2- and FeO-
does not involve any appreciable barrier, the heats of formation are consistently low.
Combined with the low concentration of FeO2-, this may account for the fact that we do
not see any other products in significant amounts in our experiments and that one
probably needs to go to much higher amounts of CO to see subsequent products.
5.4 Conclusion
It is demonstrated that the charge state of a FeO3 cluster has a strong effect on its
ability to oxidize CO to CO2 using stored oxygen. An analysis of the reaction pathways
provides evidence that the observed dependence on the charge state is driven by the
binding energy of oxygen to the metal atom. As the clusters are stabilized by the charge
transfer from metal to oxygen, more energy is required to remove oxygen from anionic
clusters. For cations, the positive charge on the iron atom weakens the Fe-O bonds that
are formed and enables higher reactivity. We therefore provide an economical alternative
for the direct oxidation of CO using highly oxidized iron clusters thus eliminating the
need for costly precious metals to effect the oxidation of CO.
80
References
1. Reddy, B. V.; Khanna, S. N. Phys. Rev. Lett. 2004, 93, 068301.
2. Reddy, B. V.; Rasouli, R.; Hajaligol, M. R.; Khanna, S. N. Fuel 2004, 83, 1537.
3. Yumura, T.; Amenomori, T.; Kagawa, Y.; Yoshizawa, K. J. Phys. Chem. A 2002, 106, 621.
4. Lin, H.-Y.; Chen, Y.-W.; Wang, W.-J. J. Nanopart. Res. 2005, 7, 249.
5. Li, P.; Miser, D. E.; Rabiei, S.; Yadav, R. T.; Hajaligol, M. R. Appl. Catal. B 2003, 43, 151.
6. PalDey, S.; Gedevanishvili, S.; Zhang, W.; Rasouli, F. Appl. Catal. B 2005, 56, 241.
7. Böhme, D. K.; Schwarz, H. Angew. Chem. Int. Ed. 2005, 44, 2336.
8. Schröder, D.; Jackson, P.; Schwarz, H. Eur. J. Inorg. Chem. 2000, 2000, 1171.
9. Wang, L.-S.; Wu, H.; Desai, S. R. Phys. Rev. Lett. 1996, 76, 4853.
10. Schröder, D.; Schwarz, H.; Clemmer, D.; Chen, Y.; Armentrout, P. B.; Baranov, V. I.; Böhme, D. K. Inter. J. Mass Spec: Ion Proc. 1997, 161, 175.
11. Schultz, R. H.; Crellin, K. C.; Armentrout, P. B. J. Am. Chem. Soc. 1991, 113, 8590.
12. Armentrout, P. B.; Koizumi, H.; MacKenna, M. J. Phys. Chem. A 2005, 109,
11365.
13. Kohn, W.; Sham, L. J. Phys. Rev. 1965, 140, A1133.
14. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865.
15. Köster, A. M.; Calaminici, P.; Casida, M.; Flores-Moreno, R.; Geudtner, G.; Goursot, A.; Heine, T.; Ipatov, A.; Janetzko, F.; Patchkovskii, S.; Reveles, J. U.; Vela, A.; Salahub, D. R. deMon2k V. 1.08, The international deMon developers community, 2005, http://www.deMon-software.com.
81
16. Dunlap, B. I.; Connolly, J. W. D.; Sabin, J. R. J. Chem. Phys. 1979, 71, 4993.
17. Mintmire, J. W.; Dunlap, B. I. Phys. Rev. A 1982, 25, 88.
18. Köster, A. M.; Flores-Moreno, R.; Reveles, J. U. J. Chem. Phys. 2004, 121, 681.
19. Godbout, N.; Salahub, D. R.; Andzelm, J.; Wimmer, E. Can. J. Chem. 1992, 70, 560.
20. Wachters, A. J. H. J. Chem. Phys. 1970, 52, 1033.
21. Wachters, A. J. H. IBM Tech. Rept. 1969, RJ584; F exponents: Bauschlicher, C. W., Jr.
22. Langhoff, S. R.; Barnes, L. A. J. Chem. Phys. 1989, 91, 2399.
23. Reveles, J. U.; Köster, A. M. J. Comp. Chem. 2004, 25, 1109.
Chapter 6
Experimental and Theoretical Study of the Structure and Reactivity of Fe1-2O≤6¯
Clusters with CO
6.1 Introduction
Transition metal catalysts, specifically those composed of iron nanoparticles, have
been employed in industrial and biological processes as well as pollution abatement
applications (1). Lin et al. observed nano-sized iron oxides to be highly active for CO
oxidation at low temperatures (2). In addition to other studies probing the oxidation of
carbon monoxide with iron (3), iron oxides showed activity for the oxidation of methane
(4) and various hydrocarbons (5). Iron oxides are also promising because they exhibit
good catalytic lifetimes and resistance to high concentrations of moisture and CO2, which
often poison catalysts (6). Therefore, iron oxides may become economical alternatives to
costly precious metal catalysts.
Studies involving iron and iron oxide have been conducted previously employing
theoretical calculations (7-9) and photoelectron spectroscopy experiments (10-12). FeO,
Fe2O3 and Fe3O4 are stable stoichiometries in the condensed phase, thus they are the most
studied species (10, 13). It was suggested by Li and coworkers that Fe2O3 was the most
active species for CO oxidation, showing that Fe2O3 behaved as a catalyst in the presence
of O2 as well as a reagent in the absence of oxygen, oxidizing CO to CO2 (13).
Sequential reduction of Fe2O3 in the absence of O2 produced Fe3O4, FeO, and Fe species.
(Reproduced in part with permission from J. Phys. Chem. A 2007, Accepted. Copyright 2007 American Chemical Society.)
83
Previous calculations of the reaction of neutral Fe2O3 with CO have proposed a
mechanism with a viable energetic pathway to produce CO2 (14-16). Through
cooperative effects, the adsorption of CO onto the cluster was predicted to weaken a Fe-
O bond making this oxygen available to a second CO for easy formation of CO2.
Iron oxide layers are commonly employed in catalytic supports and have shown
active participation in the reactions (6, 17). Iron is unique and more active than other
supports because it is easily reduced and provides anion vacancy formation due to a
highly disordered structure (18, 19). Anion vacancies are important in the catalytic
process because they stabilize the supported clusters and transfer charge density.
Schubert et al. studied several different transition metal supports and found that iron
oxides were able to adsorb large amounts of oxygen (17). In addition to the adsorbed
oxygen participating in the reaction, they proposed a mechanism that also allowed for the
oxide support to supply the oxygen for the reaction. The ability of iron oxide to function
both ways greatly enhanced the oxidation activity.
This chapter presents a joint experimental and theoretical study that aids in
uncovering species with increased activity and selectivity for the oxidation of CO to CO2.
The results provide important insight into the nature of the active sites responsible for
condensed phase catalysis, shedding light on the role of cluster size, charge and oxidation
state, composition, and stoichiometry toward CO oxidation in the presence of iron oxide.
This chapter focuses on the anionic iron oxide behavior, and Chapter 7 addresses charge
state and electron density effects on the structure and bonding of small cationic iron oxide
clusters.
84
6.2 Methods
Gas-phase cluster studies were performed using a guided ion beam mass
spectrometer coupled to a laser vaporization source, explained previously (20) and in
Chapter 2. Briefly, the second harmonic of a Nd:YAG laser was used to ablate a rotating
and translating iron rod (PVD Materials Corp., 99.95% purity). At a predetermined time,
oxygen seeded in helium (~1%) was pulsed over the rod, forming dense, hot plasma. A
27 millimeter (mm) conical expansion nozzle was placed at the exit of the source
allowing for more third body collisions and aiding in larger cluster formation. The
nozzle was composed of a 15 mm long channel with a 4 mm diameter and 12 mm conical
expansion cone with a 30° total internal angle. The clusters then underwent supersonic
expansion in a field free region before passing through a 3 mm skimmer which created a
molecular beam. The cooled clusters were then focused by a set of electrostatic lenses
and deflectors into the first quadrupole. Each cluster species was individually selected in
the first quadrupole and directed through a second set of lenses into the octopole reaction
cell. Carbon monoxide reactant gas, ranging from 0-20 mTorr, was added to the reaction
cell and the pressure was monitored by a MKS baratron. The products were focused by a
third set of lenses to a second quadrupole where they were mass analyzed and finally
detected using a channel electron multiplier. Studies were also conducted with nitrogen
in the reaction cell under the same conditions of pressure and energy for verification of a
chemical reaction with carbon monoxide. Since both CO and N2 are of the same nominal
mass, experiments with N2 aided in identifying collisional fragmentation. Collision
induced dissociation (CID) experiments were conducted to study the fragmentation
85
patterns of the iron oxide clusters. In these experiments, inert xenon gas was introduced
into the reaction cell under single collision conditions (0.09 mTorr) while the kinetic
energy of the ions in the octopole reaction cell was slowly raised from 0 to 40 eV
laboratory frame energy. Experiments with slightly higher collision pressures, 0.2 mTorr
of Xe, were conducted to find sequential fragmentation patterns under multiple collision
conditions.
The theoretical studies were carried out using two different numerical schemes
that were developed within a density functional formalism by the Khanna research group
(21). The exchange and correlation effects were incorporated through the generalized
gradient approximation (GGA) via a functional proposed by of Perdew, Burke and
Ernzerhof (22). The electronic structure was determined using a linear combination of
atomic orbitals molecular orbital approach. The wave function for the cluster was
constructed by a linear combination of Gaussian type orbitals centered at the atomic
positions in the cluster. The actual calculations employed two different numerical
programs. Most calculations were carried using the Naval Research Laboratory
Molecular Orbital Library (NRLMOL) set of codes developed by Pederson and
coworkers (23-25). For these calculations, a 5s, 4p and 3d basis set for the C and O
atoms and 7s, 5p and 4d basis for the Fe atom was employed (25). In each case, the basis
set was supplemented by a diffuse Gaussian. Supplementary calculations were carried
out using the deMon2K software (26) in order to eliminate any uncertainties associated
with the choice of basis set or the numerical procedure. In these studies, a gradient
corrected density functional (22) and the double zeta valence polarized (DZVP) basis sets
(27) for C and O and the Wachters-F basis set (28) for Fe was employed. The GEN-A2
86
auxiliary function set for C and O and the GEN-A2* auxiliary function set for Fe were
used. For each cluster structure, the configuration space was sampled by starting from
several initial configurations. Then the geometry was optimized by moving atoms in the
direction of forces until they dropped below a threshold value. Since transition metal
atoms are marked by non-zero spin multiplicities, the calculations included optimizing
the spin multiplicities of each cluster.
6.3 Results and Discussion
The anionic iron oxide cluster distribution with both dissociated and molecular
oxygen adsorbed at near thermal energy is shown in Figure 6-1. Expansion gas mixtures
ranging from 1%-20% oxygen were investigated, with the lowest percentage of oxygen in
helium providing the broadest cluster distribution. Various conditions at the exit of the
source were also examined and experiments conducted with no conical expansion nozzle
produced only monomer iron oxide species. A conical nozzle placed at the end of the
source region before the field free expansion region allowed for more collisions and thus
better iron oxide clustering. A 22 mm nozzle was used because the longer nozzle (55
mm) produced species that were more oxygen rich possessing large numbers of
molecularly bound O2 units. We did not generate any stoichiometric FexOy- clusters with
x=y having the same number of iron and oxygen atoms. It is interesting to note that
Bernstein and coworkers produced neutral iron oxide clusters with oxygen deficient
stoichiometries by varying the oxygen concentration in the system (29).
87
Figure 6-1: A typical mass distribution produced for iron oxide anionic clusters. The first iron oxide in each series with subsequent peaks in the series having one additional oxygen atom.
0 100 200 300 400 500 600
Mass (amu)
Ion
Inte
nsit
y (A
rb. U
nits
)
FeO
3-
Fe2O
3-
Fe3O
5-
Fe4O
7-
Fe5O
9-
Fe6O
10-
Fe7O
12-
0 100 200 300 400 500 600
Mass (amu)
Ion
Inte
nsit
y (A
rb. U
nits
)
FeO
3-
Fe2O
3-
Fe3O
5-
Fe4O
7-
Fe5O
9-
Fe6O
10-
Fe7O
12-
88
They made oxygen deficient iron clusters employing a 0.1% oxygen mixture. However,
in the present source this concentration produced a narrower mass distribution of iron
oxide clusters than the 1% oxygen mixture. The following studies include FeO2-4- and
Fe2O3-6- anionic species, mostly with an oxygen-rich metal to oxygen ratio of x/y <1.
Structures. Figure 6-2 shows the calculated optimized lowest energy structures for
the FeO1-4- and Fe2O2-6
- clusters. For iron oxide clusters containing a single Fe atom,
oxygen atoms bind directly to the metal with no molecular oxygen units. The maximum
coordination number of Fe was four in the tetrahedral form of FeO4-. For FeO5
- (not
shown), the fifth oxygen attaches to an oxygen atom forming an O2 unit. In the case of
FeO6- (not shown), an FeO4
- unit with a O2 molecule is bound to an oxygen atom at a
distance of 3.1 Å. A higher energy isomer for FeO4- possessing a molecular oxygen
subunit was 0.78 eV above ground state energy. The FeO bond lengths for FeO1-4- are
relatively similar. The spin multiplicity changes from quartet in FeO1-3- species to
doublet for FeO4-.
In the iron oxide clusters containing two Fe atoms, a basic ring structure
composed of Fe2O2- with oxygen bridging each iron atom is formed. Fe2O3
- and Fe2O4-
have their third and fourth oxygen atoms attached to iron outside the ring. The addition
of oxygen atoms to the cluster shortens the FeO bond length from 1.85 Å to 1.79Å within
the ring and 1.65 Å to 1.62 Å outside. Maximum coordination was obtained for the
Fe2O6- cluster with two extra oxygen atoms bound directly to each Fe atom. A higher
energy isomer of Fe2O6- with an O2 subunit was found 1.23 eV above the ground state.
An anti-ferromagnetic spin coupling was found for Fe2O2-, Fe2O3
-, Fe2O4- and Fe2O5
-
cluster species.
89
Figure 6-2: The ground state geometries of O2, CO, FeO1-3
- and FexOy- clusters. The
bond lengths are given in Angstroms and the superscripts indicate the spin multiplicity. The arrows indicate the spin polarization at the Fe atoms for the Fe2Oy
- clusters. The Mulliken charges are marked below each atom. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
90
However, the Fe sites in Fe2O6- were found to be coupled ferromagnetically. Similar
progressions of the exchange coupling with oxidation have been reported for the case of
Cr2On clusters (30).
6.3.1 CID Fragmentation
Collision induced dissociation studies were undertaken with inert xenon gas under
single (0.09 mTorr) and multiple (0.2 mTorr) collision conditions and the results are
listed in Table 6-1. The fragmentation of selected clusters and their corresponding neutral
loss aids in elucidating the structures of charged clusters and obtaining the general order
of bond strength. Trends have been established between the calculated dissociation
energies (DE) and the order of experimental fragmentation products, which are discussed
below.
For FeOy- clusters containing a single Fe atom, the observed fragmentation
channels with Xe correspond to the loss of an O atom or O2 molecule. For FeO2- and
FeO3-, the only dissociation observed was loss of an atomic O. For FeO4
-, the loss of O2
is the lowest energy pathway. At higher kinetic energies, FeO4- lost an O atom. More
interesting are the fragmentation channels for Fe2Oy- clusters containing two Fe atoms. In
the case of Fe2O3-, the product channel that is observed at the lowest collision energy is
loss of an O atom. In contrast, the product channel that is observed at the lowest collision
energy for Fe2O4- is loss of FeO. For Fe2O5
-, the lowest energy channel involves the loss
of a FeO2 and in Fe2O6-, the loss of an O2 molecule is the lowest channel. In each case,
91
other dissociation channels are observed as the kinetic energy is increased. The basic
question is then whether these pathways mirror the energetics of fragmentation.
To answer this question, the energies required to remove an O, O2, O-, and O2
-
from FeOy- and Fe2Oy
- clusters and the energies for fragmentation pathways leading to
the production of Fe, FeO, FeO2, and FeO3 in the case of Fe2Oy- clusters were calculated.
The results of these investigations are plotted in Figure 6-3. The calculated dissociation
energies for various reaction channels corresponding to the observed experimental
products are recorded in Table 6-1. The following is a discussion of the trends in the
theoretical results reported in Figure 6-3.
The current experiments detect anionic species, therefore the loss of neutral O and
O2 from the mass selected cluster can be experimentally observed. For FeOy- clusters,
one notices that the energy to remove an O atom or an O2 molecule decreases with
increasing oxygen coverage. As an example, while it takes 5.99 eV to remove an O atom
from FeO2- only 3.86 eV is needed to remove an O atom from FeO4
-. This decrease
makes the removal of O2 a favorable channel at higher coverage particularly since the
removed O atom can gain energy by combining to form O2 molecules. Indeed, for FeO4-,
the energy to remove an O2 molecule is noticeably less than to remove an O atom. While
for FeO5-, it will take only 0.95 eV to remove an O2 molecule. This is generally
consistent with the observed fragmentation patterns in Table 6-1 which shows the loss of
O as the preferred pathway for FeO2- and FeO3
- while the lowest energy pathway for
FeO4- fragmentation is the loss of an O2 molecule.
For Fe2Oy- clusters, the fragmentation pathways include the production of Fe,
FeO, FeO2 and possibly FeO3 fragments.
92
Table 6-1: The collision induced dissociation fragmentation channels, calculated dissociation energies (DE), and reaction products with CO and N2 for Fe1,2O2-6
- clusters.
FexOy-
(x,y) Products with Xea
Neutral(s) Lostb
DE (eV)
Products with CO
Products with N2
1,2 1,1 0,1 6.03 1,1 No products
1,3 1,2 0,1 5.8 1,2 No products
1,4 1,2 1,3
0,2 0,1
3.47 3.87
1,2
1,3c 1,2c
2,3 2,2 1,2 1,3
0,1 1,1 1,0
6.10 3.70 3.41
2,2 No products
2,4 1,3 1,2 2,3
1,1 1,2 0,1
3.94 4.68 6.03
2,3
1,3 No
products
2,5 1,3 2,4 2,3
1,2 0,1 0,2c
3.44 4.56 4.39
2,4
1,3 No
products
2,6 2,4 2,5 1,3
0,2d
0,1 1,3c
2.90 4.55 3.22
2,4
2,5c 2,4 ≈ 2,5c
aFragmentation channels are shown in order of observation with increasing collision energy. bNeutral loss is assigned based on the difference between the selected cluster and fragment ion formed. cDescribes channels taking place under multiple collision conditions. dDissociation occurs at near thermal energies.
93
Figure 6-3: Graphs of the fragmentation energies for atomic and molecular oxygen loss in a) FeOy
- and b) Fe2Oy- clusters. c) Graph of the fragmentation energy for neutral metal
and metal oxide loss from Fe2Oy- clusters.
0
1
2
3
4
5
6
7
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
2
3
4
5
6
7
8
1 2 3 4 5 6# of Oxygen atoms
DE
(eV
)
2
3
4
5
6
7
8
9
1 2 3 4 5 6# of Oxygen atoms
DE
(eV
)
Fe2Oy- Fe2Oy-1
- + OFe2Oy
- Fe2Oy-2- + O2
Fe2Oy- Fe2Oy-1 + O-
Fe2Oy- Fe2Oy-2 + O2
-
Fe2Oy- Fe2Oy
- + FeFe2Oy
- Fe2Oy-1- + FeO
Fe2Oy- Fe2Oy-2
- + FeO2Fe2Oy
- Fe2Oy-3- + FeO3
FeOy- FeOy-1
- + OFeOy- FeOy-2
- + O2FeOy- FeOy-1 + O-
FeOy- FeOy-2 + O2-
a)
b)
c)
0
1
2
3
4
5
6
7
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
2
3
4
5
6
7
8
1 2 3 4 5 6# of Oxygen atoms
DE
(eV
)
2
3
4
5
6
7
8
9
1 2 3 4 5 6# of Oxygen atoms
DE
(eV
)
Fe2Oy- Fe2Oy-1
- + OFe2Oy
- Fe2Oy-2- + O2
Fe2Oy- Fe2Oy-1 + O-
Fe2Oy- Fe2Oy-2 + O2
-
Fe2Oy- Fe2Oy
- + FeFe2Oy
- Fe2Oy-1- + FeO
Fe2Oy- Fe2Oy-2
- + FeO2Fe2Oy
- Fe2Oy-3- + FeO3
FeOy- FeOy-1
- + OFeOy- FeOy-2
- + O2FeOy- FeOy-1 + O-
FeOy- FeOy-2 + O2-
FeOy- FeOy-1
- + OFeOy- FeOy-2
- + O2FeOy- FeOy-1 + O-
FeOy- FeOy-2 + O2-
a)
b)
c)
94
These dissociation energies are shown graphically in Figure 6-3. The dissociation energy
values associated with experimental products are recorded in Table 6-1. It is interesting
to note that the loss of O2 reaches an energetic minimum for Fe2O6- as the energy needed
to break two Fe-O bonds is likely to be overcome by the exothermic formation of O2
releasing 6.21 eV of energy. Oxygen recombination was also observed in CID studies
previously conducted in our laboratory with V2O5+ (31). The structure of V2O5
+
contained only atomic oxygen bonds to the metal (32); however, dissociation of O2 was
observed.
Fragmentation of Fe-Fe bonds occurred in the dimer clusters, mostly at higher
energies or under multiple collision conditions. The core structure of the dimer clusters
after fragmentation was either FeO3- or FeO2
-, which represent the stable building blocks
of the larger iron oxide clusters. Fragmentation of neutral FeO from the Fe2O4- cluster
requires 3.94 eV of overall energy as shown in Figure 6-4. The dissociation of Fe2O4-
begins with the breaking of the bond between an Fe atom and a bridging O atom, which
changes its coordination to become a terminal O atom, requiring 0.81 eV. Breaking the
Fe-Fe bond and internal rotation of an FeO2 subunit then leads to a chain structure and
requires an additional 0.29 eV. From this open chain structure it takes an additional 2.84
eV to lose neutral FeO and 3.59 eV to lose FeO2. The energy for neutral FeO2
fragmentation from Fe2O5- reaches a low of 3.44 eV. Figure 6-3 shows that more energy
is needed to fragment neutral FeO2 from Fe2O4- than neutral FeO; however, this order is
reversed in the case of Fe2O5-. Our experiments verify these theoretical results as
dissociation of FeO from Fe2O4- is observed at lower energy while fragmentation of FeO2
required additional energy.
95
Figure 6-4: The change in energy (ΔE) for the dissociation pathway of parent cluster, Fe2O4
- producing FeO3- and FeO2
-. The superscripts indicate the spin multiplicity. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
96
These two clusters are unique because they form the stable fragment FeO3- prior to
oxygen loss. This indicates that the Fe-O bonds for Fe2O4- and Fe2O5
- are very strong,
and indeed they require 6.03 eV and 4.56 eV to break the Fe-O bond, respectively.
In the case of Fe2O3-, loss of an oxygen atom is the first fragment observed. This
fragment does not follow the pattern of dissociation energies found in Table 6-1. Figure
6-5 shows the change in energy for the dissociation pathways for neutral Fe and FeO loss.
In both pathways multiple bond breaks are required for neutral Fe and FeO loss. In the
first step a bridging Fe-O bond is broken and the oxygen either bonds to the Fe atom that
is less coordinated requiring 0.59 eV or stays on the iron atom with an oxygen atom
already attached requiring 0.90 eV of energy. At this stage in the reaction path, spin is
conserved in the doublet state for the later neutral FeO dissociation pathway. From the
[2Fe2O3-] transition state, the bonds are relaxed, and followed by Fe-Fe dissociation to
form a linear chain of alternating Fe-O bonds. Finally, neutral FeO is released with an
additional 3.08 eV of energy required. The other pathway for neutral Fe dissociation
also requires a spin change and bond rearrangements. Subsequent neutral Fe loss from
the cluster requires an additional 2.14 eV of energy. The bond breaking and
rearrangement of the atoms along the pathway causes these neutral fragments to occur
after O atom loss.
While a comparison of the two pathways may indicate that the loss of Fe is
energetically more favorable, the pathway for the loss of FeO requires more stable
intermediate steps. We believe that this accounts for the FeO fragment observed before
Fe loss.
97
Figure 6-5: The change in energy (ΔE) for the dissociation pathway of parent cluster Fe2O3
- producing FeO3- and FeO2
-. The superscripts indicate the spin multiplicity. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
98
Further, the numerous bond breaking steps and rearrangement of the atoms along the
pathways may make these FeO and Fe loss mechanisms kinetically unfavorable
compared to the energetically unfavorable O atom loss channel. In both cases (loss of
FeO and loss of Fe), the need to undergo a spin transition from a doublet state to an octet
intermediate state is expected to slow the reaction considerably (33).
It is, therefore, possible that spin changes slow down the rate of dissociation and
FeO fragmentation pathways occur after oxygen atom dissociation from the cluster.
Additionally the calculated dissociation pathways for Fe2O5- and Fe2O6
- can be explained
with the same reasoning. For Fe2O6-, the fragmentation order observed with CID differs
from the calculated dissociation energies. For the same reasons discussed above for
Fe2O3-, FeO3 loss occurs after O atom loss. Multiple bonds fragmenting in FeO3 requires
more time than one Fe-O bond break and therefore is observed after O atom loss.
6.3.2 Reactions with CO and N2
The reaction products for each cluster species are recorded in descending order of
relative product intensity in Table 6-1. Theoretical studies indicate that the atomization
energy of CO and CO2 are 11.63 and 17.97 eV respectively. The formation of CO2 is
therefore energetically feasible in cases where it takes less than 6.34 eV to remove an O
atom from the cluster. However, the presence of reaction barriers can prevent the
formation of CO2. This is the reason that CO2 formation does not occur in the presence
of O2 even though the binding energy of O2 is only 6.20 eV. The energy calculated to
remove an O from the cluster species, FeO2-, FeO3
-, FeO4-, Fe2O3
-, Fe2O4-, and Fe2O5
- is
99
less than 6.34 eV and they are all active toward oxygen atom transfer to CO. This
process occurs as a dominant reaction channel for most anionic species. Since neutral
species are not detected, the presence of FexOy-1- products suggests that oxygen is
transferred from iron oxide clusters to CO to produce neutral CO2.
Figure 6-6 shows the branching ratios for FeO2- and Fe2O3
- in the presence of
increasing CO pressure. These cluster stoichiometries are the most efficient iron oxide
anions to affect the CO oxidation reaction channel. Notice that both clusters are
composed of one more oxygen atom than the number of iron atoms. The energy
calculated to remove an oxygen atom from these two clusters is approximately 6 eV.
Therefore the total energy needed is more than can be supplied through thermal collisions
alone. This is supported by the fact that no oxygen atom loss products are observed in
nitrogen studies conducted under the same experimental conditions (see Table 6-1).
Confirmation of the chemical reactivity of iron oxide anionic clusters with CO, therefore
can be seen by comparing the products with nitrogen.
The reaction paths for FeO2-, FeO3
-, FeO4-, Fe2O3
- and Fe2O4- with CO were
calculated in order to understand the different reactivity they displayed. For the most
active clusters, FeO2- and Fe2O3
-, the reaction was found to proceed without barriers and
follow a spin allowed path. The reaction path for Fe2O3- is shown in Figure 6-7. The first
CO can attach to the Fe site or approach the O atom to form CO2. Our studies indicate
that the more stable configuration corresponds to CO attached to the Fe site. This is
consistent with the electron donating behavior of CO and the partial positive charge
present on the Fe site as revealed by the Mulliken population calculation.
100
Figure 6-6: The branching ratios for a) FeO2
- and b) Fe2O3- with increasing CO reactant
gas pressure. Notice that both clusters are selective for the oxygen atom transfer reaction pathway.
0.8
0.9
1
0 5 10 15 20CO (mTorr)
0
0.1
0.2
0.8
0.9
1
0 5 10 15 20
CO (mTorr)
0
0.1
0.2
FeO2-
FeO-
Rel
ativ
e In
tens
ity
(par
ent)
Relative Intensity (product)
a)
Fe2O3-
Fe2O2-
b)
Rel
ativ
e In
tens
ity
(par
ent)
Relative Intensity (product)0.8
0.9
1
0 5 10 15 20CO (mTorr)
0
0.1
0.2
0.8
0.9
1
0 5 10 15 20
CO (mTorr)
0
0.1
0.2
FeO2-
FeO-
Rel
ativ
e In
tens
ity
(par
ent)
Relative Intensity (product)
a)
Fe2O3-
Fe2O2-
b)
Rel
ativ
e In
tens
ity
(par
ent)
Relative Intensity (product)
0.8
0.9
1
0 5 10 15 20
CO (mTorr)
0
0.1
0.2
FeO2-
FeO-
Rel
ativ
e In
tens
ity
(par
ent)
Relative Intensity (product)
a)
Fe2O3-
Fe2O2-
b)
Rel
ativ
e In
tens
ity
(par
ent)
Relative Intensity (product)
101
Figure 6-7: The change in energy (ΔE) for the reaction pathway of Fe2O3
- with CO. The superscripts indicate the spin multiplicity. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
102
The subsequent CO then attaches to the O site forming CO2 as shown in Figure 6-7.
Previous studies on neutral Fe2O3 also obtained a similar reaction path (8).
On the other hand, for FeO3- and FeO4
-, the reactions with CO involve a change of
spin which may slow the reactions. For the Fe2O4- the reaction proceeds without barriers
and follows a spin allowed path. However, in the initial step of the reaction there is a
gain in energy of only 1.54 eV for the absorption of the CO molecule compared with a
gain of 4.20 eV for the absorption of two CO molecules in the Fe2O3- cluster. This is
because one of the Fe sites in Fe2O3- is coordinated to only two O atoms and the first CO
binds to this metal site. The difference in absorption energy explains the difference in
reactivity observed in the two clusters.
It is important to note that the reaction pathways involve intermediate species and
consequently involve structural rearrangements. For small clusters, the reduced size is
amenable to these rearrangements and consequently, the reaction barriers are far less than
for the case of bulk surfaces where the extended geometry is less flexible towards
structural changes.
Another reaction channel observed was the loss of molecular oxygen from FeO4-
and Fe2O6-. Branching ratios of these clusters are shown in Figure 6-8 with O2 loss as the
dominant reaction channel followed by minor O atom loss. In the case of FeO4-, it is
assumed that at the high CO pressures in which FeO2- is detected as a product, the loss of
molecular oxygen can be attributed to collision energy that promotes the cluster to the
structure containing an intact molecular oxygen unit. The same will apply for the
formation of the higher energy isomer of Fe2O6- with a molecular oxygen unit. Both
Fe2O6- and FeO4
- showed O2 loss in the presence of nitrogen as seen in Table 6-1.
103
Figure 6-8: The branching ratios for a) FeO4
- and b) Fe2O6- with increasing CO pressure,
both clusters show molecular oxygen loss.
0.8
0.9
1
0 5 10 15 20CO (mTorr)
0
0.1
0.2
0.8
0.9
1
0 7 14 21CO (mTorr)
0
0.1
0.2
Fe2O6-
Fe2O4-
Fe2O5-
FeO4-
FeO2-
FeO3-
a)
b)
Rel
ativ
e In
tens
ity
(par
ent) R
elative Intensity (products)
Rel
ativ
e In
tens
ity
(par
ent)
Relative Intensity (products)
0.8
0.9
1
0 5 10 15 20CO (mTorr)
0
0.1
0.2
0.8
0.9
1
0 7 14 21CO (mTorr)
0
0.1
0.2
Fe2O6-
Fe2O4-
Fe2O5-
FeO4-
FeO2-
FeO3-
a)
b)
Rel
ativ
e In
tens
ity
(par
ent) R
elative Intensity (products)
Rel
ativ
e In
tens
ity
(par
ent)
Relative Intensity (products)
104
These products were observed at higher nitrogen pressure under conditions where
multiple collisions are expected.
6.4 Conclusion
It is shown that the nature of fragments produced in collision induced dissociation
is governed not only by the overall energetics but also by the nature and multiplicity of
the intermediate motifs marking the fragmentation. For FeOy- clusters, the present studies
show that while the loss of atomic O is the favored channel at low energies, the loss of O2
is the dominant pathway at higher coverage. In the case of Fe2Oy- clusters, the lowest
fragmentation pathways involve the loss of Fe or FeO units except for Fe2O6-, where the
lowest energy pathway is the loss of an O2 molecule. Small anionic iron oxide clusters
are shown to enable the oxidation of CO at near thermal energies. The most active and
selective iron oxides were composed of one more oxygen atom then iron atom. The
increased reactivity is attributed to the following: (1) the energy required to remove an O
atom in these clusters is less than the gain in energy to form CO2 making the oxidation
thermodynamically feasible and (2) the reduced size (compared to bulk species) allows
structural rearrangements that eliminate the high reaction barriers and make the oxidation
kinetically possible.
105
References
1. Li, X.-Q.; Zhang, W.-X. Langmuir 2006, 22, 4638.
2. Lin, H-Y.; Chen, Y-W.; Wang, W-J. J. Nanopart. Res. 2005, 7, 249.
3. Yumura, T.; Amenomori, T.; Kagawa, Y.; Yoshizawa, K. J. Phys. Chem. A 2002, 106, 621.
4. Tanaka, S.; Nakagawa, K.; Kanezaki, E.; Katoh, M.; Murai, K.-I.; Moriga, T.;
Nakabayashi, I.; Sugiyama, S.; Kidoguchi, Y.; Miwa, K. J. Japan Petrol. Inst. 2005, 48, 223.
5. Schröder, D.; Schwarz, H.; Clemmer, D.; Chen, Y.; Armentrout, P. B.; Baranov,
V. I.; Böhme, D. K. Inter. J. Mass Spec: Ion Proc. 1997, 161,175. 6. Wu, K-C.; Tung, Y-L.; Chen, Y-L.; Chen, Y-W. Appl. Catal. B 2004, 53, 111.
7. Shiroishi, H.; Oda, T.; Hamada, I.; Fujima, N. Euro. Phys. J. D 2003, 24, 85.
8. Jones, N. O.; Reddy, B. V.; Rasouli, F.; Khanna, S. N. Phys. Rev. B 2005, 72, 165411.
9. Shiroishi, H.; Oda, T.; Hamada, I.; Fujima, N. Polyhed. 2005, 24, 2472.
10. Wu., H.; Desai, S. R.; Wang, L.-S. J. Am. Chem. Soc. 1996, 118, 5296.
11. Wang, L.-S. “Photoionization and Photodetechmant Part II.” Adv. Series in Phys. Chem. Cheuk-Yiu, N. (ed.) Word Scientific: River Edge, NJ 2000, 10B, pp 854-957.
12. Wang, L.-S.; Wu, H.; Desai, S. R. Phys. Rev. Lett. 1996, 76, 4853.
13. Li, P.; Miser, D. E.; Rabiei, S.; Yadav, R. T.; Hajaligol, M. R. Appl. Catal. B 2003, 43, 151.
14. Reddy, B. V.; Rasouli, F.; Hajaligol, M. R.; Khanna, S. N. Fuel 2004, 83, 1537.
15. Reddy, B. V.; Rasouli, F.; Hajaligol, M. R.; Khanna S. N. Chem. Phys. Lett. 2004, 384, 242.
16. Reddy, B. V.; Khanna, S. N. Phys. Rev. Lett. 2004, 93, 068301.
106
17. Schubert, M. M.; Hackenberg, S.; van Veen, A. C.; Muhler, M.; Plzak, V.; Behm, R. J. J. Catal. 2001, 197, 113.
18. Kozlov, A. I.; Kozlova, A. P.; Liu, H.; Iwasawa, Y. Appl. Catal. A 1999, 182, 9.
19. Lin, H.-Y.; Chen, Y.-W. Ind. Eng. Chem. Res. 2005, 44, 4569.
20. Bell, R. C.; Zemski, K. A.; Justes, D. R.; Castleman, A. W., Jr. J. Chem. Phys. 2001, 114, 798.
21. Kohn W.; Sham, L. J. Phys. Rev. 1965, 140, A1133.
22. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865.
23. Pederson M. R.; Jackson, K. A. Phys. Rev. B 1990, 41, 7453.
24. Jackson K.; Pederson, M. R. Phys. Rev. B 1990, 42, 3276.
25. Porezag D.; Pederson, M. R. Phys. Rev. A 1999, 60, 2840.
26. Köster, A. M.; Calaminici, P.; Casida, M. E.; Flores-Moreno, R.; Geudtner. G.; Goursot, A.; Heine, T.; Ipatov, A.; Janetzko, F.; del Campo, M. J.; Patchkovskii, S.; Reveles, J. U.; Vela, A.; Salahub, D. R. deMon2k V. 2.2.6, The international deMon developers community 2006, http://www.deMon-software.com; Köster, A. M.; Flores-Moreno, R.; Reveles, J. U. J. Chem. Phys. 2004, 121, 681; Reveles, J. U.; Köster, A. M. J. Comp. Chem. 2005, 25, 1109.
27. Godbout, N.; Salahub, D. R.; Andzelm, J.; Wimmer, E. Can. J. Chem. 1992, 70,
560.
28. Wachters, A. J. H. J. Chem. Phys. 1970, 52, 1033.; Wachters, A. J. H. IBM Tech. Rept. 1969, RJ584; F exponents: Bauschlicher, C. W., Jr.; Langhoff, S. R.; Barnes, L. A. J. Chem. Phys. 1989, 91, 2399.
29. Shin, D. N.; Matsuda, Y.; Bernstein, E. R. J. Chem. Phys. 2004, 120, 4150.
30. Reddy, B. V.; Khanna, S. N. Phys. Rev. Lett. 1999, 83, 3170.
31. Bell, R. C.; Zemski, K. A.; Kerns, K. P.; Deng, H. T.; Castleman, A. W., Jr. J.
Phys. Chem. A 1998, 102, 1733.
32. Justes, D. R.; Mitrić, R.; Moore, N. A., Bonačić-Koutecký, V.; Castleman, A. W., Jr. J. Am. Chem. Soc. 2003, 125, 6289.
33. Schwarz, H. Int. J. Mass Spec. 2004, 237, 74.
107
Chapter 7
Experimental and Theoretical Study of the Structure and Reactivity of Fe1-2O1-5+
Clusters with CO
7.1 Introduction
It is particularly important to understand the influence of ionic charge state on the
oxidation of CO with iron oxides. There is much debate over the role that charge transfer
plays on the active sites in supported condensed phase catalysts and a better
understanding of the mechanism governing the reaction may eventually lead to the design
of more economical and efficient catalysts. Previous studies of iron oxide layers
supported on gold have uncovered both Fe2O3 and FeO species present, with Fe3+ and
Fe2+ oxidation states, respectively, in the oxidation of carbon monoxide (1). PalDey and
coworkers employed XRD to show that mixed catalysts containing Fe2O3 exhibited
defect sites in the form of cationic vacancies (2). These vacancies, which constitute
regions of charge deficiency, may be important as active sites for CO oxidation.
Schröder et al. have employed three gas-phase techniques, ion cyclotron
resonance (ICR), guided ion beam (GIB), and selected ion flow tube (SIFT) mass
spectroscopy, to probe the reaction of FeO+ toward the oxidation of hydrocarbons (3).
Their results provide evidence for the importance of gas-phase studies which show
similar reactivity between all three methods and provide further understanding of the
molecular level mechanisms involved in condensed phase oxidation reactions. Studies by
108
Armentrout and coworkers have investigated the thermochemistry and bond energies of
gas-phase iron cations with CO and CO2 (4, 5), while Schwarz et al. have classified
dissociation patterns of small iron oxide cations (6). Ionic gas-phase studies are becoming
an acceptable means for elucidating the fundamental mechanisms and specific reaction
steps of catalytic reactions.
Previous theoretical calculations have predicted the thermodynamics and reaction
steps that occur over iron oxide catalysts. Findings by Yumura et al. for the oxidation of
formaldehyde with FeO+ showed that complete oxidation to CO2 was exothermic and
involved a carbon monoxide intermediate complex, H2O-Fe+-CO (7). Other theoretical
studies of the structure and reactivity of small iron oxide clusters have predicted weakly
bound oxygen units to be active in effecting the oxidation of CO to CO2 (8, 9). A
mechanism for the simultaneous oxidation of CO and reduction of NO utilizing iron
oxide species with regeneration of active sites was proposed (9, 10). This shows the
potential importance of iron oxide species in the abatement of harmful pollutants.
This chapter presents a joint experimental and theoretical study of the role of
charge state on the structure and reactivity of monomer and dimer iron oxide cluster
cations. Most previous studies of iron oxides have focused specifically on neutral FeO
and Fe2O3 as these are the most common bulk phase stoichiometries. Herein, studies of
cationic clusters of various size and stoichiometry were explored in order to uncover the
reaction pathways and associated thermodynamics.
109
7.2 Methods
Gas-phase iron oxide cluster studies were carried out employing a guided ion
beam mass spectrometer coupled to a laser vaporization source (LAVA), presented in
detail in Chapter 2. Briefly, the second harmonic of a Nd:YAG laser was used to ablate a
rotating and translating iron rod (PVD Materials Corp., 99.95% purity) ensuring a fresh
surface was ablated with each laser pulse. At a predetermined time, oxygen seeded in
helium (~1%) was pulsed over the rod, forming dense, hot plasma. A conical expansion
nozzle was placed at the exit of the source to allow for more collisions and aid in cluster
formation. Two different nozzle were used, one having a total length of 27 millimeters
with a 4 mm diameter, 15 mm long channel with a 30° total internal angle, 12 mm long
expansion cone. The other nozzle was 51 millimeters in total length with a 2 mm inner
diameter and similar cone dimensions as the other nozzle. The clusters were then allowed
to pass into a field free region where they underwent supersonic expansion before being
collimated through a 3 mm skimmer which created a molecular beam of iron oxide
clusters. The cooled clusters were focused through a set of electrostatic lenses and
deflectors into the first quadrupole. Each cluster species was individually selected in the
first quadrupole and directed through a second set of electrostatic lenses into the octopole
reaction cell. Carbon monoxide reactant gas was added to the reaction cell and the
pressure, ranging from 0-12 mTorr, was monitored by a MKS baratron. The products
were directed through a third set of lenses to a second quadrupole where they were mass
analyzed and finally detected using a channel electron multiplier. Studies were also
conducted employing inert nitrogen gas in the octopole reaction cell under the same
110
conditions of pressure and energy as carbon monoxide. These studies verify the chemical
reaction products with carbon monoxide, since both CO and N2 are of the same nominal
mass. Products with nitrogen also serve to elucidate whether oxygen is bound atomically
or molecularly to the iron atoms, which can then be compared to the calculated structures
and oxygen bond strengths. Also, collision induced dissociation (CID) experiments were
conducted to study the fragmentation patterns of the iron oxide clusters. In these
experiments, xenon gas was introduced into the octopole reaction cell at single collision
conditions of 0.09 mTorr while the kinetic energy in the collision cell was slowly
increased from 0 to 40 eV lab-frame energy. Slightly higher collision pressures of 0.2
mTorr Xe were also conducted to find sequential fragmentation patterns at multiple
collision conditions.
In order to help interpret the findings, theoretical studies were carried out by the
Khanna research group using two different numerical schemes that were developed
within a density functional formalism (11). The exchange and correlation effects were
incorporated through the generalized gradient approximation (GGA) via a functional
proposed by Perdew, Burke and Ernzerhof (12). The electronic structure was determined
using a linear combination of atomic orbitals molecular orbital approach. The wave
function for the cluster was constructed by a linear combination of Gaussian type orbitals
centered at the atomic positions in the cluster. The actual calculations employed two
different numerical programs. Most calculations were carried using the Naval Research
Laboratory Molecular Orbital Library (NRLMOL) set of codes developed by Pederson
and co-workers (13-15). For these calculations, a 5s, 4p and 3d basis set for the C and O
atoms and 7s, 5p and 4d basis for the Fe atom was employed (15). In each case, the basis
111
set was supplemented by a diffuse Gaussian. For more details, the reader is referred to
the original papers (13-15). Supplementary calculations were also carried out using the
deMon2K software (16) in order to eliminate any uncertainties associated with the choice
of basis set or the numerical procedure. In these studies, a gradient corrected density
functional (12) and the double zeta valence polarized (DZVP) basis sets (17) for C and O
and the Wachters-F basis set (18) was employed for Fe. The GEN-A2 auxiliary function
set for C and O and the GEN-A2* auxiliary function set for Fe were used. For each
cluster structure, the configuration space was sampled by starting from several initial
configurations. Then the geometry was optimized by moving atoms in the direction of
forces until they dropped below a threshold value. Since transition metal atoms are
marked by non-zero spin multiplicities, the calculations included optimizing the spin
multiplicities of each cluster.
7.3 Results and Discussion
Various cluster formation conditions were examined at the exit of the source
region and experiments conducted with no expansion nozzle produced monomer iron
species dominated by Fe+. Figure 7-1(a) shows a typical iron oxide cation cluster
distribution obtained with the 27 mm conical expansion nozzle in which monomer iron
oxide clusters were predominant. Figure 7-1(b) shows the distribution obtained with a
longer, 51 mm conical nozzle. The longer nozzle allowed for more third-body collisions
and better clustering which produced a higher intensity of dimer iron oxide clusters. It is
interesting how the source conditions greatly affected the mass distribution.
112
Figure 7-1: Typical mass distribution for iron oxide cation clusters produced when employing a) a 27 mm conical nozzle and b) a 51 mm conical nozzle at the exit of the source.
0 50 100 150 200
0 50 100 150 200
Fe+
FeO
+
FeO
2+
FeO
3+
FeO
4+
FeO
3+
Fe 2
O+
FeO
5+
FeO
6+
FeO
8+
Fe 2
O2+
Fe 2
O3+
Fe 2
O4+
FeO
8+
Mass (amu)
Inte
nsit
y (a
rb. u
nits
)
Mass (amu)
Inte
nsit
y (a
rb. u
nits
)
(a)
(b)Fe
O4+
0 50 100 150 200
0 50 100 150 200
Fe+
FeO
+
FeO
2+
FeO
3+
FeO
4+
FeO
3+
Fe 2
O+
FeO
5+
FeO
6+
FeO
8+
Fe 2
O2+
Fe 2
O3+
Fe 2
O4+
FeO
8+
Mass (amu)
Inte
nsit
y (a
rb. u
nits
)
Mass (amu)
Inte
nsit
y (a
rb. u
nits
)
(a)
(b)Fe
O4+
113
Therefore both nozzles were employed for these experiments depending on the cation
cluster selected for study. Cationic FexOy+ clusters with a stoichiometry of x=y along
with species that were oxygen rich with a ratio of x/y <1 were produced. The following
studies were conducted on FeO1-10+ and Fe2O1-4,8,10
+ cationic species.
To understand the specific reaction pathways producing each of the products
observed, CID experiments and DFT calculations were undertaken to aid in structure
identification of cationic iron oxide clusters. A discussion of the observed reaction
products with CO follows.
Structures. Figure 7-2 shows the ground state structures of FexOy+
clusters and
geometries of Fe(CO)1-2+ and FexOyCO+. Similar to the anions in Chapter 6, cationic iron
oxide clusters containing a single Fe atom have a maximum coordination number of four
oxygen atoms bound directly to the metal in the tetrahedral form of FeO4+. FeO+ has a
bond length of 1.64 Å. For FeO2+ and FeO3
+ the bond length is reduced to 1.56 Å and
FeO4+ presents three bond lengths of 1.58 Å and one bond length of 1.64 Å. For FeO5
+,
the fifth oxygen attaches to an oxygen atom forming an O2 unit with an elongated Fe-O2
bond at 2.16 Å. The spin multiplicity changes from sextet in FeO+ to doublet in FeO2-5+.
The iron dimer has a bond length of 2.18 Å and a spin multiplicity of octet. In
Fe2Oy+ clusters, a basic ring structure composed of Fe2O2
+, bridging each Fe atom, is
formed. For Fe2Oy+ with y>3, the oxygen atoms attach to iron outside the ring. Maximum
coordination was obtained for the Fe2O6+ cluster with two extra oxygen atoms bound
directly to each Fe atom. The Fe-Fe bond length increases to around 2.50 Å with the
exception of the Fe2O2+ cluster that contains a Fe-Fe bond length of 2.31 Å. The Fe-O
bond length ranges from 1.76 to 1.81 Å within the ring and 1.50 to 1.57 Å outside. The
114
spin multiplicity is octet of Fe2O+ and changes to doublet and quartet states. An anti-
ferromagnetic spin coupling was found for Fe2O2+, Fe2O3
+ and Fe2O4+ cluster species and
the Fe sites in Fe2O5+ and Fe2O6
+ were found to be coupled ferromagnetically.
The structures for carbon monoxide bound to the iron oxide clusters were also
calculated. CO binds more strongly to an iron atom with a bond length of 1.91 Å in
Fe(CO)2+ whereas if CO binds to an oxygen atom the bond length is stretched to 2.04 Å
in FeO2CO+. These bond strengths are shown to be important in the observed reactions
which reflect the stability of the intermediates.
7.3.1 CID Studies
A systematic energy resolved CID study employing inert xenon gas was
undertaken. Table 7-1 lists the fragmentation products in order of increasing kinetic
energy for CID studies undertaken with xenon gas under single collision conditions. The
fragmentation patterns of these clusters help to identify common structural features. The
major product identified was loss of atomic oxygen. Most clusters lost an oxygen atom at
higher collision energies, however Fe2O3+ lost an O atom at near thermal energy. This is
evidence that Fe2O3+ contains a weakly bound O atom in the ground state structure.
Further implications of this weakly bound oxygen atom will be discussed in terms of the
reactivity below.
Loss of molecular oxygen from oxygen rich clusters having a stoichiometry of
FexOy+ (y≥x+2) was another major fragmentation product. If enough kinetic energy is
input into the cluster through raising the DC voltage of the octopole, then vibrational
115
bond rearrangement and exothermic fragmentation occurs, given the large gain in energy
for forming O2. This is likely the case in iron clusters not saturated with oxygen atoms.
Oxygen recombination was observed for vanadium oxide clusters containing only metal-
oxygen atom bonds in previous CID experiments (19). Extremely oxygen-rich clusters,
for example FeO10+ and FeO9
+, lost successive O2 molecules at near thermal energy
uncovering a FeO+ core structure. During mass selection of these clusters it was observed
that an oxygen molecule would dissociate without the addition of a collision gas, which
suggests that these clusters were meta-stable with loosely associated O2 units. Indeed,
Figure 7-2 shows a maximum of 4 oxygen atoms can bind directly to the iron metal
center.
For the dimer cluster series, specifically Fe2O2+ and Fe2O4
+, fragmentation
involved a Fe-Fe bond dissociation to form FeO2+ and FeO4
+, respectively. For Fe2O2-4+
subsequent dissociation with increasing kinetic energy revealed a basic core of FeO2+.
For oxygen-rich clusters, Fe2O8+ and Fe2O10
+, a stable core structure of Fe2O2+ was
revealed after collisional loss of oxygen molecules. There was no Fe-Fe fragmentation in
these clusters and, therefore, it is assumed that the larger oxygen rich clusters dissociate
O2 units and, subsequently have similar reactivity as the underlying core cluster. The
building blocks of small iron oxide cations were identified as FeO+, FeO2+, and Fe2O2
+.
General trends can be established between the calculated DE and the order of
experimental CID products. The graphs in Figure 7-3 show that as the number of oxygen
atoms surrounding iron increases, the dissociation energies for neutral loss of O and O2
decreases. It is interesting to note Schwarz and coworkers showed that neutral O, FeO,
and FeO2 loss was dependent on the iron to oxygen ratio of the parent cluster (6).
116
Figure 7-2: The ground state geometries for a) FexOy+ clusters and b) for Fe(CO)1-2
+ and FexOyCO+ clusters. The bond lengths are given in Angstroms and the superscripts indicate the spin multiplicity. The arrows indicate spin polarization at the Fe atoms for Fe2Oy
+ and Fe2OyCO+. The Mulliken charges are marked below each atom.
(a)
(b)
117
Table 7-1: Results of CID studies showing the fragmentation channels for selected Fe1-2Oy+
clusters
FexOy+
(x,y) Products with Xea
Neutral(s) Lostb
FexOy+
(x,y) Products with Xea
Neutral(s) Lostb
1,1 1,0 0,1 2,1 2,0 0,1 1,2 1,0
1,1 0,2 0,1
2,2 2,1 2,0 1,2
0,1
0,2 1,0c
1,3 1,2 1,1 1,0
0,1 0,2 0,3
2,3 2,2 2,1 2,0 1,2
0,1d
0,2
0,3c 1,1c
1,4 1,2 1,3 1,1 1,0
0,2d 0,1 0,3 0,4
2,4 2,2 2,3 1,4 1,2
0,2 0,1
1,0 1,2
1,5 1,3 1,4 1,2 1,1
0,2d
0,1 0,3 0,4
2,8 2,6 2,4 2,2 2,3
0,2d 0,4 0,6 0,5
1,6 1,4 1,2 1,5 1,1
0,2d 0,4 0,1 0,5
2,10 2,8 2,6 2,4 2,2
0,2d 0,4d 0,6 0,8
1,7 1,5 1,3 1,2 1,1
0,2d 0,4 0,5 0,6
1,8 1,6 1,4 1,2 1,7
0,2d
0,4d 0,6 0,1
1,9 1,7 1,5 1,4 1,3 1,1
0,2d 0,4d 0,5d 0,6 0,8
1,10 1,8 1,6 1,4 1,2 1,9 1,1
0,2d
0,4d
0,6d 0,8 0,1 0,9
aFragmentation channels are shown in order of observation with increasing collision energy. bThe neutral loss is assigned based on the difference between the selected cluster and fragment ion formed. cDescribes channels taking place under multiple collision conditions. dRepresents that dissociation occurs at near thermal energies.
118
In oxygen-rich clusters with a metal to oxygen ratio of 1>x/y>2/3, neutral loss of FeO2
and O2 became more dominate. For Fe2O2+ and Fe2O3
+, it can be seen in Figure 7-3, that
the energy needed to fragment an oxygen atom is less than 1 eV greater than the energy
needed to fragment O2. Consequently, under the low kinetic energies in which these
products were experimentally observed, the overall energy needed to fragment two O
atom bonds is not overcome. These trends are used to describe the CO oxidation and
oxygen replacement by CO reaction pathways for FexOy+ clusters.
7.3.2 Reactivity Studies
The products of reactions of FeO1-5+ and Fe2O1-5
+ cluster species in the presence
of either carbon monoxide or nitrogen are shown in Table 7-2. The dissociation energies
associated with the CO product channels are also listed. For iron clusters with up to three
oxygen atoms, oxygen atom transfer was an observed product channel. Iron species
FeO+ and Fe2O+, with one O atom, were selective for CO oxidation. This was confirmed
in the nitrogen studies which showed no loss of an oxygen atom under the same
experimental pressures as CO reactant gas. Figure 7-4 displays the branching ratios for
FeO1,2+ and Fe2O1,2
+ clusters which all show oxygen atom transfer as the dominant
reaction channel in the presence of CO. Therefore, only clusters with a FexO+ and FexOy
+
(y=x+1) cluster stoichiometry showed atomic oxygen atom transfer to CO producing
neutral CO2 as the major reaction pathway. Considering the energy required to remove
an oxygen atom from iron, and the gain in energy from forming CO2, shows the reaction
to be overall exothermic for these clusters.
119
Figure 7-3: Graphs of the fragmentation energies for atomic and molecular oxygen loss in a) FeOy
+ and b) Fe2Oy+ clusters. c) Graph of the fragmentation energy for neutral metal
and metal oxide loss from Fe2Oy+ clusters.
23456789
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
0
1
2
3
4
5
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
0
1
2
3
4
5
6
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
FeOy+ FeOy-1
+ + OFeOy
+ FeOy-2+ + O2
Fe2Oy+ Fe2Oy-1
+ + OFe2Oy
+ Fe2Oy-2+ + O
Fe2Oy+ Fe2Oy
+ + FeFe2Oy
+ Fe2Oy-1+ + FeO
Fe2Oy+ Fe2Oy-2
+ + FeO2Fe2Oy
+ Fe2Oy-3+ + FeO3
a)
b)
c)
23456789
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
0
1
2
3
4
5
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
0
1
2
3
4
5
6
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
FeOy+ FeOy-1
+ + OFeOy
+ FeOy-2+ + O2
Fe2Oy+ Fe2Oy-1
+ + OFe2Oy
+ Fe2Oy-2+ + O
Fe2Oy+ Fe2Oy
+ + FeFe2Oy
+ Fe2Oy-1+ + FeO
Fe2Oy+ Fe2Oy-2
+ + FeO2Fe2Oy
+ Fe2Oy-3+ + FeO3
a)
b)
c)
0
1
2
3
4
5
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
0
1
2
3
4
5
6
1 2 3 4 5 6
# of Oxygen atoms
DE
(eV
)
FeOy+ FeOy-1
+ + OFeOy
+ FeOy-2+ + O2
Fe2Oy+ Fe2Oy-1
+ + OFe2Oy
+ Fe2Oy-2+ + O
Fe2Oy+ Fe2Oy
+ + FeFe2Oy
+ Fe2Oy-1+ + FeO
Fe2Oy+ Fe2Oy-2
+ + FeO2Fe2Oy
+ Fe2Oy-3+ + FeO3
a)
b)
c)
120
Table 7-2: Reaction products for the reaction between selected iron oxide cluster cationsand carbon monoxide.
FexOy+
(x,y) Products with CO
DE (eV)
Products with N2
1,1 Fe+ 1.78 No products 1,2 FeO+
FeCO+ 2.72 0.31
No products
1,3 FeO+
FeOCO+
Fe+
FeO2+
5.36 1.66 0.65 2.64
FeO+
1,4 FeO2CO+
Fe(CO)2+
FeO+
3.75 1.99 2.49
FeO2+
Fe+
1,5 FeO3+
FeO3CO+ 1.16 3.06
FeO3+
2,1 Fe2+ 0.69 No products
2,2 Fe2O+
Fe2+
Fe2CO+
0.87 1.56 5.54
Fe2O+
2,3 Fe2O2+ 2.12 Fe2O2
+
2,4 Fe2O2CO+ Fe2O2
+ Fe2O3
+ FeO4CO+
6.17 4.25 2.12
Fe2O2+
Fe2+
2,5 Fe2O3+
Fe2O2+
Fe2O+
Fe2O3CO+
4.7 6.83 7.7
7.78
Fe2O3+
121
Other clusters undergoing oxygen atom transfer were FeO3+, Fe2O3
+, and Fe2O4+;
however, these species had competing reaction channels. In the studies involving
nitrogen, oxygen atom loss products were detected under conditions typically considered
to involve multiple collisions.
The Fe2O3+ cluster species was the most active and selective cluster toward the
transfer of an oxygen atom as shown in Figure 7-5. Studies with inert gas used in CID,
have also produced an atomic oxygen loss product at near thermal energies. This
suggests that there is a very low barrier for removing an oxygen atom from Fe2O3+ and
that the energy gained from forming CO2 allows the reaction to proceed. The increased
activity of Fe2O3+ is a consequence of the structure shown in Figure 7-2 and also the
bonding strength of the oxygen atom to iron. The cationic metal center binds the oxygen
atom more weakly than the anionic iron as shown in Chapter 6 (20). Fe2O3 is the most
studied species because it is stable in the condensed phase and thought to be the active
site for many reactions (21). It is not surprising our experiments show Fe2O3+ promotes
CO oxidation as studies of neutral Fe2O3 nanoparticles have shown iron to work as both a
catalyst in the presence of oxygen and a direct oxidant by losing lattice oxygen in the
absence of oxygen (22).
A secondary reaction observed for iron oxide cluster cations was CO bound to the
positive metal center. In the case of FeO2+, a product species corresponding to FeCO+
could arise from several different pathways including O2 replacement by CO or
association of CO after collisional O2 loss. From the branching ratio for FeO2+ in Figure
7-4(b), the product species FeCO+ occurs at low CO pressures supporting the O2
replacement by CO reaction channel. It is energetically feasible to remove oxygen and
122
bind CO since losing an O2 requires only 2.0 eV, providing for the energy gain in
forming an O2 molecule. The subsequent energy gain in binding CO is greater than 4 eV.
Although the product FeO+ could be an intermediate species in a successive reaction, the
loss of an oxygen atom from FeO+ requires more energy than the gain for binding CO.
This is evidence that these two species occur during separate reactions. Indeed, Figure 7-
6 shows the calculated profile for FeO2+ in which FeCO+ and FeO+ follow separate
reaction paths after the initial binding of CO.
Oxygen molecule replacement by CO was a prominent reaction channel observed
for many oxygen-rich clusters, including FeO3+, FeO4
+, FeO5+, Fe2O4
+. Most O2
dissociation energies for monomer iron oxide clusters were less than 1.4 eV. Figure 7-7
shows this reaction pathway for FeO4+, where the product species, FeO2CO+ resulted.
FeO4+ is also unique in that it also showed two CO molecules attached to the metal
center. Armentrout and co-workers have studied bond energies of gas-phase iron
carbonyls and found the strongest bond dissociation energy (BDE) to be for the second
CO ligand (4,5) This supports the high stability of Fe-CO bonds formed in this reaction.
In the case of FeO2+, FeO4
+, and Fe2O2+, carbon monoxide was observed to be bound to
bare iron, Fe1-2(CO)x+. It is not surprising that bare iron clusters were detected as
products from FeO+, FeO3+, Fe2O
+, and Fe2O2+ species, since iron is easily reduced. No
iron carbides were observed, which suggests that all the carbon present was oxidized
(20).
123
Figure 7-4: Branching ratios for a) FeO+ b) FeO2
+ c) Fe2O+ and d) Fe2O2
+ reacted with CO. Note the left-hand axis corresponds only to the selected ion, parent species as it decreases with CO gas. The right-hand axis tracks all product cluster species that increase with the addition of CO gas.
Figure 7-5: Branching ratio for Fe2O3
+ reacted with CO. Note the right-hand scale for product ions in Fe2O3
+ is double the amount in the branching ratios of Figure 7-4.
0.6
0.8
1
0 2.5 5 7.5 10CO (mTorr)
0
0.1
0.2
0.6
0.8
1
0 2.5 5 7.5 10
CO (mTorr)
0
0.1
0.2
0.8
0.9
1
0 2.5 5 7.5 10CO (mTorr)
0
0.1
0.2
FeO+
Fe+
Rel
.Int
. (p
aren
t io
n)
Rel. Int. (product ions)
a)R
el.I
nt. (
pare
nt io
n)
Rel. Int. (product ions)
b)
FeO2+
FeO+
FeCO+
0.8
0.9
1
0 2.5 5 7.5 10
CO (mTorr)
0
0.1
0.2
Fe2O+
Fe2+
Rel
.Int
. (pa
rent
ion)
Rel. Int. (product ions)
c)
Rel
.Int
. (pa
rent
ion)
Rel. Int. (product ion
s)
d)
Fe2O2+
Fe2O+
Fe2+
Fe2CO+
0.6
0.8
1
0 2.5 5 7.5 10CO (mTorr)
0
0.1
0.2
0.6
0.8
1
0 2.5 5 7.5 10
CO (mTorr)
0
0.1
0.2
0.8
0.9
1
0 2.5 5 7.5 10CO (mTorr)
0
0.1
0.2
FeO+
Fe+
Rel
.Int
. (p
aren
t io
n)
Rel. Int. (product ions)
a)R
el.I
nt. (
pare
nt io
n)
Rel. Int. (product ions)
b)
FeO2+
FeO+
FeCO+
0.8
0.9
1
0 2.5 5 7.5 10
CO (mTorr)
0
0.1
0.2
Fe2O+
Fe2+
Rel
.Int
. (pa
rent
ion)
Rel. Int. (product ions)
c)
Rel
.Int
. (pa
rent
ion)
Rel. Int. (product ion
s)
d)
Fe2O2+
Fe2O+
Fe2+
Fe2CO+
0.6
0.8
1
0 4 8 12CO (mTorr)
0
0.2
0.4
Rel
.Int
. (pa
rent
ion)
Rel. Int. (product ions)
Fe2O3+
Fe2O2+
0.6
0.8
1
0 4 8 12CO (mTorr)
0
0.2
0.4
Rel
.Int
. (pa
rent
ion)
Rel. Int. (product ions)
Fe2O3+
Fe2O2+
124
ΔE (eV)
2FeO2+ + 1CO 2OCFeO2
+ -2.33 2OCFeO2
+ 6OCFeO2+ 0.82
6OCFeO2+ 4FeCO+ + 3O2 1.20
2OCFeO2
+ [2FeO2CO+] 1.06
[2FeO2CO+] 2FeO2CO+ -1.80 2FeO2CO+ 6FeO2CO+ -0.90 6FeO2CO+ 6FeO+ + 1CO2 1.26
Figure 7-6: Profile of the reaction energy for FeO2+ with CO including the corresponding
change in energy, ΔE, for each reaction step. The superscripts indicate spin multiplicity. Source: Figure courtesy of the group of Professor S. N. Khanna at VCU.
125
Figure 7-7: Branching ratio for FeO4+ reacted with CO showing at higher pressures two
CO molecules become attached to bare iron.
Rel
.Int
. (pa
rent
ion)
0
0.5
1
0 4 8 12
CO (mTorr)
FeO4+
FeO2CO+
Fe(CO)2+
FeO+
Rel
.Int
. (pa
rent
ion)
0
0.5
1
0 4 8 12
CO (mTorr)
FeO4+
FeO2CO+
Fe(CO)2+
FeO+
126
7.4 Conclusion
Structural calculations and dissociation energies of iron oxide cluster cations
found trends in the bond energy that correlate well with experimental CID fragmentation
products. The observed reaction pathways in the presence of CO were fully accounted
for by the thermodynamic bond energies and the gain in energy from binding CO to the
metal cluster. It is shown for certain stoichiometries that CO oxidation is favorable and
energetically feasible. However, for oxygen rich iron oxide clusters, O2 loss and O2
replacement were observed to be the major reaction channels. Therefore, the reactions of
iron oxide clusters were elucidated on the basis of cluster stoichiometry, oxidation state,
and charge state. These studies provide a foundation for further research in transition
metal oxide clusters as catalytic systems.
127
References
1. Guczi, L.; Frey, K.; Beck, A.; Petõ, G.; Daróczi, C. S.; Kruse, N.; Chenakin, S. Appl. Catal. A 2005, 291, 116.
2. PalDey, S.; Gedevanishvili, S.; Zhang, W.; Rasouli, F. Appl. Catal. B 2005, 56,
241.
3. Schröder, D.; Schwarz, H.; Clemmer, D. E.; Chen, Y.; Armentrout, P. B.; Baranov, V. I.; Böhme, D. K. Int. J. Mass Spec. Ion Proc. 1997, 161, 175.
4. Schultz, R. H.; Crellin, K. C.; Armentrout, P. B. J. Am. Chem. Soc. 1991, 113,
8590.
5. Armentrout, P. B.; Koizumi, H.; MacKenna, M. J. Phys. Chem. A 2005, 109, 11365.
6. Schröder, D.; Jackson, P.; Schwarz, H. Eur. J. Inorg. Chem. 2000, 2000, 1171.
7. Yumura, T.; Amenomori, T.; Kagawa, Y.; Yoshizawa, K. J. Phys. Chem. A 2002,
106, 621.
8. Jones, N. O.; Reddy, B. V.; Rasouli, F.; Khanna, S. N. Phys. Rev. B 2005, 72, 165411.
9. Reddy, B. V.; Rasouli, R.; Hajaligol, M. R.; Khanna, S. N. Fuel 2004, 83, 1537.
10. Reddy, B. V.; Khanna, S. N. Phys. Rev. Lett. 2004, 93, 068301.
11. Kohn W.; Sham, L. J. Phys. Rev. 1965, 140, A1133.
12. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865.
13. Pederson M. R.; Jackson, K. A. Phys. Rev. B 1990, 41, 7453.
14. Jackson K.; Pederson, M. R. Phys. Rev. B 1990, 42, 3276.
15. Porezag D.; Pederson, M. R. Phys. Rev. A 1999, 60, 2840.
128
16. Köster, A. M.; Calaminici, P.; Casida, M. E.; Flores-Moreno, R.; Geudtner. G.; Goursot, A.; Heine, T.; Ipatov, A.; Janetzko, F.; del Campo, M. J.; Patchkovskii, S.; Reveles, J. U.; Vela, A.; Salahub, D. R. deMon2k V. 2.2.6., The international deMon developers community 2006, http://www.deMon-software.com; Köster, A. M.; Flores-Moreno, R.; Reveles, J. U. J. Chem. Phys. 2004, 121, 681; Reveles, J. U.; Köster, A. M. J. Comp. Chem. 2005, 25, 1109.
17. Godbout, N.; Salahub, D. R.; Andzelm, J.; Wimmer, E. Can. J. Chem. 1992, 70,
560.
18. Wachters, A. J. H. J. Chem. Phys. 1970, 52, 1033.; Wachters, A. J. H. IBM Tech. Rept. 1969, RJ584; F exponents: Bauschlicher, C. W., Jr.; Langhoff, S. R.; Barnes, L. A. J. Chem. Phys. 1989, 91, 2399.
19. Bell, R. C.; Zemski, K. A.; Kerns, K. P.; Deng, H. T.; Castleman, A. W., Jr. J.
Phys. Chem. A 1998, 102, 1733.
20. Reilly, N. M.; Johnson, G. E.; Castleman, A. W., Jr.; Reveles, J. U.; Khanna, S. N. Chem. Phys. Lett. 2007, 435, 295.
21. Abdel Halim, K. S.; Khedr, M. H.; Nasr, M. I.; El-Mansy, A. M. Factors affecting
CO oxidation over nanosized Fe2O3, Materials Research Bulletin 2006, doi:10.1016/j.materresbull.2006.07.009
22. Li, P.; Miser, D. E.; Rabiei, S.; Yadav R. T.; Hajaligol M. R. Appl. Catal. B 2003,
43, 151.
Chapter 8
Gas Phase Study of Cobalt, Nickel, and Copper Oxide Cluster Ions: Dissociation
Patterns and Trends in Reactivity with CO
8.1 Introduction
Transition metals offer a wide range of industrial applications, especially in
heterogeneous catalysis, therefore there is considerable interest in understanding their
reactive behavior. Investigations of transition metals in the bulk phase have been carried
out to describe their reactive and structural properties. However, on an atom by atom
basis, the properties and characteristics of transition metals change dramatically at the
nanoscale (1, 2). In this size regime, each atom may alter the structure, geometry and
electronic properties of a species. Gas phase metal cluster experiments are especially
amenable to studying the behavior of nanoscale materials through mass selected size-
specific experiments that have the ability to isolate potential catalytic active sites. Cluster
studies are able to probe the effects of size, oxidation and charge state, composition and
stoichiometry on cluster reactivity.
Numerous past studies have focused on bare transition metals and their
interactions with oxygen (3, 4), nitrogen (4-7) and carbon monoxide (4, 6, 8-16). Efforts
to elucidate small cluster geometries and to characterize surface binding sites have been
conducted through uptake and saturation experiments using either CO or N2, through
collision induced dissociation (CID) studies with xenon, and by various theoretical
130
structural calculations (17-19). Experiments involving CO have used the saturation limits
of metal sites to elucidate complicated transition metal structures through electron
counting of closed shell orbitals (4, 20, 21). Wang and coworkers (22, 23) and Ganteför
et al. (24) studied the electronic structure of several 3d transition metal oxides with
photoelectron spectroscopy (PES). Collision induced dissociation studies also revealed
information on bond strengths and trends in metal cluster structures (25-29). However, it
is difficult to fully determine the nature of bonding in transition metals because there is
limited information available on the ground state structures of metal oxides (23) and the
electronic properties change when oxygen is bound to the metal (25). The reaction of
transition metals with both oxygen and CO (30-33) therefore provides new insight into
catalytic applications, with particular emphasis on environmental pollution abatement.
One study involving CO reactivity with ozone and oxygen found the activity of CO
oxidation was higher in the presence of ozone with ozone decomposition also occurring
(30).
Many interesting structural and reactive properties arise in transition metals due to
the large number of d electrons and their localized bonding character (19,34-36).
Fielicke et al. investigated metal-CO bond strength and metal carbonyl properties based
on cluster size (37). They employed IR-multiple photon dissociation (IR-MPD)
spectroscopy to study the binding of CO on free nickel and cobalt clusters. Their focus
on electrostatic charge and size effects in small clusters helped identify different
structural binding sites of CO and trends in CO binding to various metal clusters. Several
studies have also tracked the effect of changes in the oxidation state of transition metals
toward the observed reactivity for CO oxidation (38-41). Jernigan et al. found for
131
metallic copper and copper (I), CO oxidation proceeds by the Langmuir-Hinshelwood
mechanism, whereas copper (II) was proposed to operate through a redox cycle
mechanism (39). Furthermore, the activity for CO oxidation was found to decrease with
increasing oxidation state of Cu. There is no consensus in the literature on the nature of
the active sites and mechanisms governing the oxidation of CO in the presence of
transition metal oxides due to complications arising from environmental conditions such
as temperature (40) and humidity (42). Therefore, further investigation of isolated
transition metal oxide systems in the presence of CO would be useful.
Experiments investigating several first row transition metals were undertaken to
determine the physical and chemical characteristics of cobalt, nickel, and copper oxides.
Studies on iron oxides covered in Chapters 5-7 are compared here to elucidate the
periodic trends in binding energy and reactivity toward CO oxidation (43). A
comprehensive study of their reactivity and structures through collision induced
dissociation (CID) experiments reveals how additional d electrons in the system and
changes in electron affinity affect the reactivity of the metal oxide clusters.
This chapter focuses on transition metal elements known to be common dopants
or supports for industrial catalysts. Previously, it was believed that the support materials
used in heterogeneous catalysis were inert substances which did not influence the
reactivity of the catalyst. Recently, however, studies conducted on catalysts with various
metal oxide supports have begun to shed light on the role that the support plays in the
catalytic process (44-47). Metal oxide supports were shown to have considerable impact
over the particle size and activity of catalysts in low temperature oxidation of CO (48). It
has also been proposed that CO adsorbs to the catalyst, while transition metal oxide
132
supports, such as CoOx, provide the necessary oxygen to oxidize CO (49). An
understanding of the influence of the metal oxide support may aid in designing enhanced
materials to deposit catalysts onto, which will increase the activity and selectivity toward
a chosen reaction pathway. These pure metal oxide studies lay the groundwork for
further exploration into the effects of catalyst-support interactions and dopant effects.
8.2 Experimental
All metal oxide cluster studies were conducted on a guided ion beam mass
spectrometry coupled to a laser vaporization source described in detail previously (50)
and in Chapter 2. Briefly, the second harmonic of a Nd:YAG laser was used to ablate a
continuously rotating and translating metal rod. Oxygen seeded in helium (~10%) was
pulsed over the ablated metal rod at a predetermine time to form the metal oxide cluster
ions. A 27 mm conical expansion nozzle was employed to aid in the clustering of these
species. After exiting the nozzle, the ions traversed a field free region, were collimated
by a 3 mm skimmer and focused by a set of electrostatic lenses and deflectors into the
mass selecting quadrupole. Individual cluster ions were mass selected and focused by a
second set of lenses and directed into the octopole reaction cell. Carbon monoxide
reactant gas ranging from 0-10 mTorr for cations and 0-20 mTorr for anions was added
into the reaction cell and the pressure of the gas was monitored by a MKS Baratron. The
remaining reactant and product ions were mass analyzed in the second quadrupole and
detected by a channel electron multiplier. For collision induced dissociation (CID)
experiments, xenon was added to the reaction cell at three different pressures, 0.08
133
mTorr, 0.12 mTorr, and 0.22 mTorr, as the DC voltage on the octopole rods was slowly
increased from thermal to 40 eV laboratory-frame energy. The first two pressures were
calculated to be single collision conditions while the last pressure was at multiple
collision conditions. Sequential fragmentation was determined from the results of
multiple collision products at elevated energies and pressures.
8.3 Results and Discussion
Figure 8-1 shows the typical mass distributions for anionic metal oxide clusters
of nickel, cobalt and copper. These are stable metal oxide species formed in the laser
vaporization source and mass selected for the present studies. The corresponding cationic
mass distributions for nickel and cobalt oxides are shown in Figure 8-2. An analysis of
the fragmentation patterns and insight into the structures of the metal oxide ions in this
study will be presented first through CID experiments. Reactivity studies with carbon
monoxide reactant gas and a discussion based on stoichiometry, composition, and charge
state follows.
8.3.1 Collision Induced Dissociation Studies
Table 8-1 and Table 8-2 show the fragmentation products from CID studies for
cobalt, nickel, and copper oxide anions and cations, respectively.
134
Figure 8-1: Typical anionic mass distributions for a) Nickel oxides, b) Cobalt oxides, and c) Copper oxides.
0 100 200 300 400
0 100 200 300 400 500
CoO
2-
CoO
3-
Co 2O
3-
Co 2O
5-
Co 3O
4-
Co 4O
6-
Co 5O
6-
Ion
Inte
nsit
y (a
rb. u
nits
)
Mass (a.m.u.)
b)
0 100 200 300 400
Mass (a.m.u.)
Ion
Inte
nsit
y (a
rb. u
nits
)
NiO
2-
NiO
3-
Ni 2O
3-
Ni 3O
3- Ni 4O
4-
Ni 5O
5-
a)
Mass (a.m.u.)
Ion
Inte
nsit
y (a
rb. u
nits
)
CuO
2-
Cu 2O
3-
Cu 3O
3-
Cu 4O
4-
Cu 5O
5-
c)
0 100 200 300 4000 100 200 300 400
0 100 200 300 400 5000 100 200 300 400 500
CoO
2-
CoO
3-
Co 2O
3-
Co 2O
5-
Co 3O
4-
Co 4O
6-
Co 5O
6-
Ion
Inte
nsit
y (a
rb. u
nits
)
Mass (a.m.u.)
b)
0 100 200 300 400
Mass (a.m.u.)
Ion
Inte
nsit
y (a
rb. u
nits
)
NiO
2-
NiO
3-
Ni 2O
3-
Ni 3O
3- Ni 4O
4-
Ni 5O
5-
a)
Mass (a.m.u.)
Ion
Inte
nsit
y (a
rb. u
nits
)
CuO
2-
Cu 2O
3-
Cu 3O
3-
Cu 4O
4-
Cu 5O
5-
c)
135
Figure 8-2: Typical cationic mass distributions for a) Nickel oxides and b) Cobalt oxides.
0 50 100 150 200Mass (a.m.u.)
Ion
Inte
nsit
y (a
rb. u
nits
)
NiO
+
Ni+
NiO
2+ NiO
3+
NiO
6+
NiO
5+
NiO
4+
Ni 2O
2+
NiO
7+
Ni 2O
4+
NiO
8+
Ni 2O
+
Ni 2O
3+
a)
0 50 100 150 200 250 300
CoO
+
CoO
2+
O2+
2O2+
CoO
3+ CoO
4+
CoO
5+ Co 2O
2+
CoO
6+
Co 3
+
Co 2
O4+
CoO
8+
CoO
9+
CoO
10+
Co 2
O5+
Co 2O
6+
Co 3O
3+C
o 2O7+
CoO
11+
Co 3
O4+
Co 2O
8+
Co 3O
5+
Co 2
O9+
Mass (a.m.u.)
Inte
nsit
y (a
rbit
rary
uni
ts)
Co+
Co 2O
10+
b)
0 50 100 150 200Mass (a.m.u.)
Ion
Inte
nsit
y (a
rb. u
nits
)
NiO
+
Ni+
NiO
2+ NiO
3+
NiO
6+
NiO
5+
NiO
4+
Ni 2O
2+
NiO
7+
Ni 2O
4+
NiO
8+
Ni 2O
+
Ni 2O
3+
0 50 100 150 200Mass (a.m.u.)
Ion
Inte
nsit
y (a
rb. u
nits
)
NiO
+
Ni+
NiO
2+ NiO
3+
NiO
6+
NiO
5+
NiO
4+
Ni 2O
2+
NiO
7+
Ni 2O
4+
NiO
8+
Ni 2O
+
Ni 2O
3+
a)
0 50 100 150 200 250 300
CoO
+
CoO
2+
O2+
2O2+
CoO
3+ CoO
4+
CoO
5+ Co 2O
2+
CoO
6+
Co 3
+
Co 2
O4+
CoO
8+
CoO
9+
CoO
10+
Co 2
O5+
Co 2O
6+
Co 3O
3+C
o 2O7+
CoO
11+
Co 3
O4+
Co 2O
8+
Co 3O
5+
Co 2
O9+
Mass (a.m.u.)
Inte
nsit
y (a
rbit
rary
uni
ts)
Co+
Co 2O
10+
b)
136
Anionic oxygen rich metal oxide clusters, MO>3-, M2O>4
-, M3O>5-, exhibited molecular
O2 loss at near thermal energies and under single collision conditions. Previous
calculations performed on iron oxide clusters found a maximum saturation of four
oxygen atoms on a metal center (43, 51). For oxide clusters with oxygen atoms bound
directly to the metal, CID experiments have indicated that oxygen recombination takes
place (52). However, the facile loss of O2 indicates that oxygen is loosely bound as either
a peroxide or superoxide unit for cluster species with coordinative saturation. Metal
atoms, MO and MO2 were also common neutral fragments of larger anionic clusters
demonstrating the greater electron affinity of the more oxygen rich fragments produced
by dissociation of these neutral fragments. Sequential fragmentation occurred with select
copper oxides at elevated energies and under multiple collision conditions. Cationic
metal oxide clusters also lost O2 at near thermal energy for MO>4+ and M2O>2
+ clusters
which is consistent with the smaller activation of the O-O bond near charge withdrawing
metal centers.
Cobalt. The anionic cobalt oxides studied were CoO2-3-, Co2O3-5
-, and Co3O3-6-.
Oxygen atom loss was found to be the major fragment for monomer clusters, CoO2- and
CoO3-, occurring at slightly elevated energies. This is in agreement with the theoretical
structural calculations of Uzunova et al. who predict atomically bound oxygen units (53).
Neutral metal atom loss was observed for dimer oxides resulting in a stable CoO2-
subunit, while for trimer oxides, Co2O3- was the most stable fragment species. This in
agreement with reactions of cobalt anions and oxygen performed by Kapiloff and Ervin,
who found CoO2- and Co2O3
- to be the major products at high O2 concentration,
indicating that these structures were particularly stable (4).
137
Table 8-1: CID fragmentation channels for selected MxOy- (M=Co, Ni, Cu) anions.
Cobalt Nickel Copper Cluster MxOy
-
(x,y) Products with Xea
Neutral (s) Lostb
Products with Xea
Neutral (s) Lostb
Products with Xea
Neutral (s) Lostb
1,2 1,1 0,1 1,1 0,1 1,1 0,1 1,3 1,2 0,1 1,1
1,2 0,2 0,1
1,1 1,2
0,2c 0,1
2,3 1,2 2,2
1,1 0,1
1,2 2,2
1,1 0,1
1,2 2,2 2,1
1,1 0,1 0,2
2,4 1,3 1,2 2,3
1,1 1,2 0,1
1,2 2,3 2,2 1,3
1,2 0,1 0,2 1,1
2,2 2,3 1,3 1,2
0,2c 0,1 1,1d 1,2d
2,5 2,3 1,3 2,4 1,2
0,2 1,2 0,1 1,3
2,3 0,2
3,3 2,3 1,0 2,3 1,0 3,4 3,3
2,3 0,1 1,1
3,2 3,3 2,3
0,2 0,1 1,1
3,5 3,3 3,4 2,4 2,3
0,2 0,1
3,3 3,4 2,3
0,2c 0,1 1,2
3,6 3,4 3,5 2,4 3,3 2,3
0,2c 0,1 1,2 0,3 1,3
4,4 4,2 3,3
0,2 1,1
3,3 1,1d
4,5 4,3 4,4 2,3
0,2 0,1 2,2
4,3 4,4 3,3
0,2c 0,1 1,2
aFragmentation channels are shown in order of observation with increasing collision energy. bThe neutral loss is assigned based on the difference between the selected cluster and fragment ion formed. cRepresents that dissociation occurs at near thermal energies. dDescribes channels taking place under multiple collision conditions.
138
Table 8-2: CID fragmentation channels for selected MxOy+ (M=Co, Ni) cations.
Cobalt Nickel Cluster MxOy
+
(x,y) Products with Xea
Neutral (s) Lostb
Products with Xea
Neutral (s) Lostb
1,1 1,0 0,1 1,0 0,1 1,2 1,0
1,1 0,2 0,1
1,0 1,1
0,2 0,1
1,3 1,1 1,2
0,2 0,1
1,1 1,2
0,2 0,1
1,4 1,2
1,3 0,2c
0,1c 1,2 1,3
0,2 0,1
2,2 2,1
2,0 0,1 0,2
2,0 1,3 1,2
0,2c
1,1 1,0
2,3 2,1 0,2c
2,4 2,2
2,0 0,2c
0,4
2,6 2,4
2,2
2,1
1,4 1,2
0,2c
0,4c
0,5c
1,2 1,4
aFragmentation channels are shown in order of observation with increasing collision energy. bThe neutral loss is assigned based on the difference between the selected cluster and fragment ion formed.cRepresents that dissociation occurs at near thermal energies.
139
For larger cobalt oxide anions of Co2O5-, Co3O5
- and Co3O6-, O2 loss was a major product
fragment. O2 loss from larger CoxOy- clusters is consistent with the PES results of
Pramann et al. who predict peroxo bound O2 units for these species (54). In Co2O5- the
ordering of CID products and N2 products shown in Table 8-3 was reversed. It should be
noted that nitrogen products were observed at relatively high pressures, approximately 15
mTorr N2 and the intensity was roughly in the same relative porportion. When energy
was added into the system during collisions with Xe, loss of an oxygen molecule was
observed first, in contrast with N2 collisional studies where an oxygen atom was the first
product observed. This may be due to the sticky nature of the collisions between Xe and
Co2O5- which may better facilitate vibrational energy transfer resulting in a larger
fragment species (55).
The cationic cobalt oxides examined were CoO1-4+ and Co2O2,4,6
+. Both CoO+
and Co2O2+ lost an oxygen atom first at slightly elevated energies. The remaining
monomer clusters lost O2 molecules first then O atoms, with CoO4+ losing both O2 and O
at near thermal energy. The cationic dimer cobalt oxides were unique because they
attached only an even number of oxygen atoms. From these studies it can be assumed
that only molecular oxygen was associated to Co2O4+ and Co2O6
+ because O2 loss
occurred readily under thermal energy. For Co2O4+ and Co2O6
+ the major fragments
were O2 loss followed by the loss of two O2 molecules. The stable fragments tended to
have molecular oxygen units and included CoO2+, CoO4
+ and Co2O2+.
Nickel. The following nickel oxide anions were studied: NiO2-3-, Ni2O3-4
-, Ni3O3-5-
and Ni4O4-5-. Previous experimental work in our laboratory utilizing a metal ion fast flow
reactor has shown similar nickel oxide anion and cation mass distribution (56).
140
Table 8-3: Reaction products for selected MxOy- (M=Ni, Co, Cu) anions with CO and N2.
Cobalt Nickel Copper Cluster MxOy
-
(x,y) Products with CO
Products with N2
Products with CO
Products with N2
Products with CO
Products with N2
1,2 1,1 No products
1,1 No products
1,1 1,0
1,3 1,2 1,2 1,2 No products
1,2
1,1
1,0a
1,1 1,2
2,3 2,2
1,2a 2,1 2,2
1,2a No
products 2,2
2,1a
1,2a
No products
2,4 2,3a No products
2,3
2,2 2,3 2,3
2,2 2,2 2,3
2,5 2,4
2,3
1,3
2,4 2,3 1,3
2,4
2,3 2,1a 1,2a
2,4 2,3
3,3 No Reaction
3,2
3,1a 3,1 2,3
3,4 3,3a
2,4a No
products 3,3 No
products 3,3 3,2
3,3 3,5 3,4
3,3
2,3a
2,4a
No products
3,3
3,4 3,3 3,4
3,6 3,5
3,4
3,3
2,4
2,3a
3,4 3,5
4,4 No Reaction
4,3a
3,3a 4,2 3,3
4,5 4,4
4,3 4,3 4,4
4,4 4,4 4,3 3,3
4,6 4,4a 4,7 4,5
4,6a
aDenotes a minor product channel whose relative intensity is less than 1% of the total reaction products and are not displayed in their respective branching ratios.
141
It is interesting that monomer and dimer anionic species in the present studies are oxygen
rich which is consistent with previous studies showing slightly more oxidized anionic
species. The absence of nickel monomer and dimer clusters in the previous study was
attributed to a slower electron attachment rate. Indeed, the electron affinities of NixOy+,-
increase with increasing size. The fragments occurring at lower energies were mostly O
or O2 units; however, several clusters lost neutral nickel and nickel oxygen. For Ni2O3-
and Ni2O4-, NiO2
- was the first fragment detected. Studies with N2 gas shown in Table 8-
3 verify that this product fragment was initiated through additional kinetic energy added
to the system and did not result from thermal collisions. Since these products were not
observed with N2 at high collision pressures, it suggests that the addition of energy in
CID experiments was needed for any metal bond dissociation. In the case of Ni3O5- and
Ni4O5- the order of CID products follows the same order of N2 products observed.
The nickel oxide cations studied included NiO1-4+ and Ni2O2-3
+. For all of the
clusters except NiO+, O2 was the first fragment observed. The dimer clusters showed this
O2 subunit to desorb at near thermal energy, demonstrating that this fragment was only
weakly bound to the nickel cation. Similar results were observed with N2 studies in
Table 8-4 with the exception of NiO+ and NiO2+. Only Ni2O2
+ exhibited loss of a metal
atom, revealing a stable NiO2+ species at the highest energies. This is consistent with the
results of Vardhan and coworkers who demonstrated Ni atom loss from nickel dioxides
formed at low energy in a molecular beam (57).
Copper. Anionic copper oxide clusters which were studied include CuO2-3-,
Cu2O3-5-, Cu3O3-4
-, and Cu4O4-5-. Oxygen rich clusters that lost O2 readily were CuO3
-,
Cu2O4-, and Cu4O5
-, with Cu2O5- losing molecular oxygen at near thermal energy.
142
Table 8-4: Reaction products for MxOy+ (M=Co, Ni) cations with CO and N2.
Cobalt Nickel Cluster MxOy
+
(x,y) Products with CO
Products with N2
Products with CO
Products with N2
1,1 (1,0) + No products
(1,0) + (1,0) +
1,2 (1,0)CO+
(1,0) +
(1,1)(CO)2+a
No products
(1,0)CO+
(1,1) + (1,1) + (1,0) +
1,3 (1,1)CO+
(1,1) + (1,0) +
(1,1)(CO)2+a
(1,1)CO+
(1,0) +
(1,2) +
(1,0)CO+a
(1,1) +a
(1,1) +
1,4 (1,2)CO+
(1,0)(CO)2+
(1,2) +
(1,0)CO+
(1,0)CO+
(1,2) +
(1,2)CO+
(1,1) +
(1,0) +
2,1 (1,1)CO+ 2,2 (1,0)(CO)2
+
(1,2)CO+ (1,0)(CO)2
+
(1,2)CO+
(1,0)CO+
(1,1)CO+a
(2,0) +
2,3 (2,0)CO+
(2,1)CO+ (1,1)(CO)2
+ (2,2) +
(1,0)(CO)2+
(1,1)CO+ (1,0)CO+
(2,1) +
2,4 (2,0)CO+
(2,2) + (1,0)(CO)2
+
(1,2)CO+
2,6 (2,4) +
(2,2) + (1,0)(CO)2
+
(1,0)CO+a
aDenotes a minor product channel whose relative intensity is less than 1% of the total reaction products and are not displayed in their respective branching ratios.
143
PES studies conducted by Wang et al. have revealed that once the transition metal
reaches its saturated oxidation state of +3 for copper, then molecular O-O units are
formed (58). This supports the facile loss of O2 from oxygen-rich clusters since
dissociative adsorption of oxygen does not occur on saturated transition metals. Loss of
CuO and Cu were the first fragments observed for Cu2O3- and Cu3O3
-, respectively, with
neutral Cu loss for Cu4O4- occurring under multiple collision conditions. Other metal
atom losses at multiple collisions occurred in Cu2O4- which produced a stable CuO2
-
species. For this stable copper dioxide molecule, it has been proposed by Wang et al. that
two isomers exist, OCuO, and Cu(O2), even though they do not observe CuO- at the
wavelengths employed in the PES experiments (23). For Cu3O3- and Cu4O4
- the first
fragmented species was Cu2O3- and Cu3O3
-, respectively. This agrees with CID studies
conducted by Gord et al. on oxygen-deficient and metal and oxygen equivalent copper
oxide anions (25). The reason we may not observe further dissociation products reported
by Gord et al., however, may be due to the short reaction time of our experiments
compared to the trapping ability of the FT-ICR-MS technique. Several of the copper
oxide CID products fragmented differently with Xe than the N2 collisional fragments
shown in Table 8-3. For example, Cu4O5- produced O2 loss at near thermal energy with
Xe, while it was the second product detected in N2 studies. Again, this may be due to
better kinetic to vibrational energy transfer with Xe as characterized by Armentrout et al.
(55).
The mass degeneracy observed in the spectra of copper oxide cations made
identification of individual cluster species difficult. No copper oxide cationic species are
144
reported herein due to the production of oxygen rich metal clusters having mass overlap
with larger metal clusters.
Trends. Similar dissociation patterns were observed for each cluster size;
however, there were some differences in the energy ordering of products within the metal
clusters. For M2O3-, all species lost neutral MO first, followed by O loss. This sequence
is unique because most often oxygen atoms or molecules are observed to fragment first.
It is likely, therefore, that the M-O bond in these clusters is stronger than M-M bonding.
Similar products are detected for M2O4- clusters, but with a different order. Nickel oxides
lose MO2, cobalt oxides lose MO, and copper oxides lose neutral O2 upon collisional
studies at thermal energy. This suggests that the structures and bond strengths are
different between the various metal oxides of this stoichiometry. For cobalt specifically,
loss of an oxygen atom is the third product fragment, compared to the second detected for
both nickel and copper. Therefore we expect this increased M-O bond energy to reflect
in the reactivity studies, discussed below. Other minor differences occur in M3O4- and
M4O4- between nickel and copper oxides, with nickel losing an O atom and NiO more
easily and copper losing O2 and CuO.
The differences in the N2 and CID dissociation patterns occur mostly in the
oxygen rich metal clusters, Co2O5-, Cu2O5
- and Cu4O5-. This difference suggests that
increasing the kinetic energy of the collision will increase the internal energy of the
reactant species resulting in vibrational excitation and geometric changes. If the energy
within the system allows for rearrangement of the structure and two oxygen atoms come
in close contact, then the energy to form an oxygen molecule and allow for O2
dissociation may exceed the energy required to break the metal-oxygen bonds. Other
145
studies have reported on the slight activity observed between transition metals and N2,
however the reaction products were often molecular adsorption of N2 onto the metal
cluster (6). Armentrout and coworkers have observed FexN+ products in reactions
between Fex+ clusters and N2, but only at elevated energies (59). We do not observe
molecular or dissociative adsorption onto any metal clusters and only collisional
fragmentation with N2 occurs in the present experiments. It is not thermodynamically
feasible to break a N-N bond to form NO and we do not believe this chemical reaction is
occurring in these experiments. However, at the high pressures indicative of multiple
collisions, it may be possible to form N2O.
The dissociation patterns of cationic metal oxide exhibited a main product
fragment of O2. The only exception was in the M2O2+ cluster stoichiometry. Here, O2
loss was followed by MO and M loss in nickel clusters, whereas O loss was followed by
O2 loss in cobalt clusters. Although most studies on the structure of transition metal
oxides revealed metal oxygen bonds and not molecular oxygen bonds until coordinative
saturation (60), oxygen recombination was observed. Previously, in our laboratory, we
observed O2 loss in V2O5+ in CID studies (52). Also, N2 and CID dissociation patterns
were observed to differ for only NiO2+, with O atom loss dominant in N2 studies and O2
loss in CID experiments.
8.3.2 Reactivity Studies
Table 8-3 and Table 8-4 present the products of reactions with CO and collisions
with N2 for each of the nickel, cobalt, and copper oxide clusters studied. Figure 8-3
146
shows the branching ratios for the clusters having the most reactive stoichiometries for
oxygen atom transfer to CO for anionic cobalt, nickel, and copper oxides. For both
cobalt and nickel the same stoichiometry observed in iron oxides of (x, y = x+1) is the
most active. However, in copper oxides, a stoichiometry with the same number of metal
atoms and oxygen atoms is the most reactive. For cationic clusters of nickel and cobalt
oxides non-equivalent numbers of metal and oxygen atoms were the most active toward
oxidation. Shown in Figure 8-4 for NiO3+ and CoO2
+, the major reaction channel was O2
replacement by CO. Detailed below are the individual metal oxides and the observed
products for anion and cation reactions with CO.
Cobalt. Oxygen atom transfer was observed for all anionic monomer through
trimer cobalt clusters. CoO3- exhibited a weakly bound oxygen atom as shown in N2
studies at high pressures, while Co2O4- and Co3O4
- had only a slight reaction with less
than 1% relative product intensity. Molecular oxygen loss was another product shown in
reactions with CO especially in larger oxygen rich clusters, Co4O6- and Co4O7
-. It is
interesting that Co3O5- and Co3O6
-, which are also oxygen rich species, underwent
oxygen atom transfer as the major reaction channel instead of molecular oxygen loss.
For cationic monomer cobalt clusters, O2 replacement by CO was the major
product followed by O2 loss. In the dimers, O2 loss was the dominant pathway with
attachment of CO to the metal cluster following. For Co2O2+, however, CoO2 and Co
replacement by a CO molecule was detected. Also multiple replacement reactions were
observed for Co2O4+ and Co2O6
+ with two CO molecules attached to the metal cluster.
Nickel. Nickel oxide anions underwent mostly oxygen atom transfer to CO
forming neutral CO2.
147
Figure 8-3: Branching ratios for a) Co2O3- b) Ni2O3
-and c) Cu3O3- reacted with carbon
monoxide. Each cluster shows an active reaction pathway for oxygen atom transfer.
Rel
ativ
e In
tens
ity R
eact
ant
0.7
0.8
0.9
1
0 5 10 15
CO (mTorr)
0
0.1
0.2
0.3
Ni2O3-
Ni2O2-
Rel
ativ
e In
tens
ity
Rea
ctan
t Relative Intensity Product
0.7
0.8
0.9
1
0 5 10 15CO (mTorr)
0
0.1
0.2
0.3
Cu3O3-
Cu3O2-
Cu3O-
Rel
ativ
e In
tens
ity
Rea
ctan
t Relative Intensity Products
(b)
(c)
0.7
0.8
0.9
1
0 4 8 12 16CO (mTorr)
0
0.1
0.2
0.3
Co2O3-
Co2O2-
Relative Intensity Product
(a)
Rel
ativ
e In
tens
ity R
eact
ant
0.7
0.8
0.9
1
0 5 10 15
CO (mTorr)
0
0.1
0.2
0.3
Ni2O3-
Ni2O2-
Rel
ativ
e In
tens
ity
Rea
ctan
t Relative Intensity Product
0.7
0.8
0.9
1
0 5 10 15CO (mTorr)
0
0.1
0.2
0.3
Cu3O3-
Cu3O2-
Cu3O-
Rel
ativ
e In
tens
ity
Rea
ctan
t Relative Intensity Products
(b)
(c)
0.7
0.8
0.9
1
0 4 8 12 16CO (mTorr)
0
0.1
0.2
0.3
Co2O3-
Co2O2-
Relative Intensity Product
(a)
0.7
0.8
0.9
1
0 5 10 15
CO (mTorr)
0
0.1
0.2
0.3
Ni2O3-
Ni2O2-
Rel
ativ
e In
tens
ity
Rea
ctan
t Relative Intensity Product
0.7
0.8
0.9
1
0 5 10 15CO (mTorr)
0
0.1
0.2
0.3
Cu3O3-
Cu3O2-
Cu3O-
Rel
ativ
e In
tens
ity
Rea
ctan
t Relative Intensity Products
(b)
(c)
0.7
0.8
0.9
1
0 4 8 12 16CO (mTorr)
0
0.1
0.2
0.3
Co2O3-
Co2O2-
Relative Intensity Product
(a)
148
Figure 8-4: Branching ratios for a) CoO2+ and b) NiO3
+ reacted with carbon monoxide. The dominant reaction pathway for both clusters is molecular oxygen replacement by CO.
0
0.25
0.5
0.75
1
0 2 4 6 8CO (mTorr)
0
0.25
0.5
0.75
1
0 2.5 5 7.5 10CO (mTorr)
NiO3+
NiOCO+
Ni+
NiO2+R
elat
ive
Inte
nsit
y
CoO2+
CoCO+
Co+
Rel
ativ
e In
tens
ity
(b)
(a)
0
0.25
0.5
0.75
1
0 2 4 6 8CO (mTorr)
0
0.25
0.5
0.75
1
0 2.5 5 7.5 10CO (mTorr)
NiO3+
NiOCO+
Ni+
NiO2+R
elat
ive
Inte
nsit
y
CoO2+
CoCO+
Co+
Rel
ativ
e In
tens
ity
(b)
(a)
149
Ni2O4- had a weakly bound oxygen atom as nitrogen studies also showed oxygen atom
loss with high collision pressures. For larger metal clusters, only Ni3O4- and Ni4O5
-
showed oxygen atom transfer, evidence that clusters with one more oxygen atom than
metal atom have a particularly active stoichiometry. For Ni3O5-, molecular oxygen loss
was observed followed by oxygen atom loss. No reactions were observed with clusters
containing the same number of metal and oxygen atoms, Ni3O3- and Ni4O4
-.
In cationic nickel oxides, oxygen atom transfer was observed for NiO+, NiO2+,
and Ni2O3+; however, it was the major reaction channel only for NiO+. Molecular
oxygen loss was a minor reaction channel observed for select clusters including, NiO3+
and NiO4+. The major product channel detected was O2 replacement by a CO molecule.
Dimer species also underwent Ni, NiO, or NiO2 replacement by CO and in some cases
two CO molecules were bound to the metal cluster. For Ni2O2+, all the products had
metal atom loss and CO attached to either Ni+ or NiO2+. In studies by Parks and
coworkers on CO binding to nickel, it was observed that adsorption of the CO molecule
weakened the Ni-Ni bonding (18). This may explain why metal atom loss was prevalent
when carbon monoxide was attached to the product species.
Copper. In all the anionic copper oxides studied the major product channel was
oxygen atom transfer, with Cu4O4- being only slightly reactive. This reaction pathway
was followed by O2 loss and metal atom loss for Cu2O3- and Cu4O4
-. For Cu2O5- oxygen
atom transfer was observed with CO; however, this was a very weakly bound oxygen
atom since experiments with N2 showed the same results.
150
The mass degeneracy in oxygen rich cationic copper oxide clusters made the
study of these species difficult due to the resolution of our mass spectrometer and the
results are not reported here.
Reactivity Trends. The reactivity of transition metal oxide clusters was studied by
investigating the effects of charge state, cluster size, composition, and stoichiometry on
the observed reaction pathways. The results for nickel, cobalt, and copper oxides show
that the reaction pathways were similar for each charge state. However, unique
characteristics for each metal composition depending on the stoichiometry and metal to
oxygen ratio were observed. Oxygen atom transfer was the major pathway for negatively
charged clusters and molecular oxygen replacement by CO was the major reaction
channel for positively charged metal oxide clusters.
The reactions of positively charged transition metal oxides are different from the
corresponding anionic clusters, since CO is attached to the metal oxide center for most
cationic clusters. Figure 8-5 shows NiO2-,+ clusters reacted with CO. From these
branching ratios, it can be seen that oxygen atom transfer occurs for both charge states,
however cationic NiO2+ has a competing O2 replacement by CO reaction pathway.
Differences in the reactive behavior of these small transition metal oxides may be
explained by the cluster charge state and effects of electron density towards bonding. For
cationic transition metals, the positive metal charge induces electrostatic effects which
create a strong bond between the charge on the metal and CO (36). The general
mechanism for CO adsorption onto a metal surface is described by the Blyholder model,
which involves σ donation and π backdonation (61).
151
Figure 8-5: Branching ratios for a) NiO2- and b) NiO2
+ reacted with carbon monoxide. The dominant reaction pathway for anions is oxygen atom transfer while cations have two competing reactions, oxygen atom transfer and molecular oxygen replacement by CO.
0
0.25
0.5
0.75
1
0 5 10 15
CO (mTorr)
NiO2-
NiO-
0
0.25
0.5
0.75
1
0 2.5 5 7.5 10
CO (mTorr)
Rel
ativ
e In
tens
ity
Rel
ativ
e In
tens
ity
NiO2+
NiCO+
NiO+
(b)
(a)
0
0.25
0.5
0.75
1
0 5 10 15
CO (mTorr)
NiO2-
NiO-
0
0.25
0.5
0.75
1
0 2.5 5 7.5 10
CO (mTorr)
Rel
ativ
e In
tens
ity
Rel
ativ
e In
tens
ity
NiO2+
NiCO+
NiO+
(b)
(a)
152
Essentially the lone pair electron of CO in the occupied 5σ antibonding orbital donates
into the metal σ valence orbital, while the unoccupied 2π antibonding orbital in CO
accepts charge from the d orbitals of the metal. Therefore electrostatic effects are used to
generate stronger metal-CO bonds. In cationic nickel and cobalt oxide clusters the
positive metal center disperses its charge into the metal-carbon (M-C) bond which creates
more covalent bond character (37). In free CO, electron density is polarized toward the
more electronegative O atom. When CO binds to M+, the positive charge at the metal
center draws electron density away from the O atom and into the bond, increasing
covalency. Also, the M+-C-O- attraction increases the bond strength. Investigations of
late d-block metal cations have also suggested stronger M-CO bonds due to the effect of
coulombic attraction between positive metal centers and CO (62).
The anionic counterparts show weaker M-C bonds for several reasons including
differences in the electronic structure compared to cations. For anions, a non-optimal
overlap exists due to the small 3d metal orbitals and the 5σ orbital of CO. The presence
of the extra electron in anionic systems greatly affects its behavior (17). For example,
CuO- has its extra electron occupying the 2pπ orbital resulting in a closed shell electronic
ground state (22). This anionic configuration does not accept the charge from the lone
pair of CO and therefore we do not see CO attached to the anionic cluster.
Another significant effect that charge state has on the reactivity of transition metal
oxides is on the qualitative reaction rate. From Figure 8-5 it can be seen that the
reactions of cations are more active at lower CO pressures (10 mTorr versus 15 mTorr in
anions). This may be a consequence of the electron deficient cation exhibiting higher
activity as it will be able to accept σ electron density from CO forming a bond. In studies
153
by Russell et al., smaller ionic clusters and clusters with open coordination sites
displayed an increased relative reaction rate (63).
The metal to oxygen ratio is observed to affect the reactivity in the present
studies. Stable cobalt oxides contain a different number of oxygen atoms than nickel and
copper oxides, with more oxygen atoms surrounding cobalt. Oxygen rich species with a
stoichiometry of M3O6- and M4O6,7
- were only detected in the cobalt mass distribution
(Figure 8-1b). It should also be noted that no anionic cobalt or nickel oxides with the
same number of metal and oxygen atoms were reactive in the presence of CO. Only
copper oxides, Cu3O3- and Cu4O4
-, were active towards oxygen atom transfer followed by
O2 loss. In contrast, studies by Huang et al. showed that non-stoichiometric copper oxide
species (ie. Cu3O4- and Cu4O5
-) were more active toward CO oxidation because these
oxygen rich species were able to transport surface lattice oxygen more efficiently (40).
Indeed in oxygen deficient environments one can see the usefulness of oxygen rich
species in order to supply oxygen to the reaction. However for copper oxides, in terms of
active sites, there is a special reactivity associated with metal and oxygen equivalent
species in the present studies.
Cations with the same number of metal atoms and oxygen atoms readily undergo
reactions with CO compared to the anions. From Table 8-4, the results of one oxygen
atom on nickel or cobalt show oxygen atom transfer. Ni2O2+ and Co2O2
+ both show MO2
(M=Ni, Co) replacement by CO followed by M (M=Ni, Co) replacement by a CO
molecule. These clusters are the only species to contain two CO molecules as the major
reaction channel and it is interesting that these species readily lose a metal atom. This
suggests that the CO binding energy is greater than the energy required to break a M-M
154
or M-O bond. For small clusters the energy released in binding CO produces competitive
reaction pathways between unimolecular decomposition and collisional stabilization. In
this case, the charge donation from CO binding causes destabilization in the metal (24)
resulting in weaker M-M bonds then M-C bonds (18). Cox et al. have shown that cobalt
and nickel have the strongest M-C bond strengths, followed by iron and then copper (12).
Size dependence is another important aspect in studying the reactivity of small
transition metal oxide clusters. For the monomer and dimer clusters, all the metal anions
exhibit oxygen atom transfer. However, the major reaction pathways begin to differ
starting with the trimer depending on the metal composition. Tetramer oxide clusters,
Co4O6- and Co4O7
-, do not readily promote the oxidation of CO through the oxygen atom
transfer pathway. In fact, only Ni4O5-, Cu4O4
- and Cu4O5- promote the oxidation of CO.
Size may also play a role in the cationic adsorption mechanism of oxygen. The amount
of oxygen adsorbed onto the metal clusters differs with size. For example, the cobalt
monomer oxide contained dissociated oxygen; however, the dimer species absorbed an
even number of oxygen atoms. This provides evidence that oxygen adsorbs molecularly
to cobalt dimers and may explain the dominant oxygen molecule replacement by CO
channel in these clusters.
Periodic Trends. Previous studies of metal oxide clusters reacted with CO allow
us to make other periodic comparisons (43, 64). All the metals discussed in the present
study are heavier 3d transition metals, which show electron delocalization only for the 4s
electrons (65). This means that the lightest elements, like titanium have 3d electrons that
contribute to delocalization and participate in valence orbitals and bonding. Although not
directly involved in the bonding, metal d orbital contributions strengthen bonds. This
155
effect becomes less critical moving from left to right on the periodic table because the
radial extent of the d orbitals decreases relative to the s orbital (35). In fact, other studies
have shown that the metal d contribution to bonding is higher for CO adsorption on Ni
than on Cu surfaces (11). Iron oxides belonging to the first row transition metals exhibit
a behavior depending on the metal to oxygen ratio similar to nickel and cobalt for the
most reactive stoichiometries toward CO oxidation (51). These metals have open d shells
and may allow more coordination of oxygen resulting in the observed active
stoichiometries. However, copper oxides displayed an equivalent number of metal and
oxygen atoms for the most active species in oxidation. The reason copper may differ
from the other metal oxide is that copper possesses a full d orbital shell and bonding
occurs primarily through s interactions. Copper oxides may become more stable with
less oxygen coordinated.
Differences in the electronic structure and bonding energies of small anionic and
cationic metal oxides are indeed evident from our reactivity studies. The significance of
additional electron density in anionic clusters is also exhibited in the present studies.
Closed shell electron configurations produce more stable clusters, and studies show that
the lone pair electrons on CO readily donate charge to clusters needing additional
electrons to form a closed shell (16). Cations were shown to have more CO binding in
order to gain electron density and better cluster stability.
The nickel group metals include some of the most reactive metals in catalysis
besides the commonly employed platinum and palladium nanoparticles. Studies
involving Pt, Pd, and Ni with CO were conducted and reveal fundamental differences in
the electronic structure of the different metals (24). Compared to Pt and Pd, nickel
156
showed the highest uptake of CO. Also Ni-CO bonds tend to be stronger with more
contributions from π backbonding (13,17). However, studies have not been conducted in
our laboratory on other metal oxides in this group with CO and therefore it is hard make
comparisons based on the reactivity of these metal oxides with CO.
The coinage metals have been studied for their reactivity towards CO. Sueyoshi
and coworker found that the low activation energy for the reaction between adsorbed CO
and oxygen toward the oxidation of CO on Cu(110) surfaces makes this surface more
active than standard Pd, Pt, or Rh catalytic materials (31). They extended their low
temperature studies to include Ag and Au surfaces and suggested a trend for low
activation energy on these surfaces that allows for more energetically favorable CO
oxidation than currently used metal surfaces. Our studies on copper, silver, and gold
oxides show mostly anionic species promote the oxidation of CO. It is interesting that no
anionic trimer gold oxides (64) or dimer silver oxides show oxygen atom transfer
products, even though they are in the same group of the periodic table. These reactions
are highly dependent on the structure of the metal oxide as pointed out in previous studies
(64). However, copper seems to be the most active for all sizes studied and we provide
further evidence that charge transfer between the negative metal center and oxygen
enhances the oxidation of CO.
8.4 Comparison to Bulk
Low temperature oxidation of carbon monoxide has been carried out on mixed
transition metals (66) and on transition metal oxide supports (67, 68). Also, X-ray
157
photoelectron spectroscopy has been used to study the specific composition of these
species. Studies involving cobalt oxide catalytic supports showed these materials to be
composed of Co3O4 and CoO species in the near-surface region (68). Isotope studies of
oxygen found that CO2 was formed over preoxidized cobalt oxide surfaces and that
surface oxygen participated in the reaction (41). There have been numerous other studies
that now show that surface oxygen from metal oxide supports is responsible for
promoting the oxidation of CO to CO2. It has also been shown previously that diffusion
of lattice oxygen enhances the activity of CO oxidation. Thus, the ability of transition
metal oxides to adsorb and release oxygen greatly affects the activity. Oxygen rich
anionic clusters with at least one more oxygen atom than metal atom are shown to
promote the oxidation of CO in the present experiments. This supports the fact that
transition metal oxides composed of iron, cobalt, nickel, or copper do promote CO
oxidation. With further studies on mixed metal systems, the full potential of these
reactive species may be realized.
8.5 Conclusion
The dissociation patterns and reactivity in the presence of CO of several first row
transition metal oxides were investigated as a function of size, stoichiometry, and ionic
charge state. Numerous factors affected the reaction selectivity including oxygen
stoichiometry and metal composition. This comprehensive study of the electron density
changes in charge state moving across the periodic table has revealed similarities and
differences in the observed reaction products. Small anionic metal oxides are more
158
selective toward CO oxidation. However cationic metal oxides are more efficient in their
observed reaction pathways. This study has shed light on the active species for CO
oxidation and provided evidence that transition metal oxides, commonly used in catalytic
supports, may contribute to the overall reactivity. The ability of transition metal oxide
clusters to transfer oxygen to CO forming CO2 potentially mimics the mechanism that
occurs on condensed phase catalyst supports. Further research into the nature of mixed
metal cluster systems would be beneficial to understand the interactions of supported
catalysts at the molecular level.
159
References
1. Gu, X.; Hohn, K. L. Ind. Eng. Chem. Res. 2004, 43, 30.
2. Heiz, U. Appl. Phys. 1998, 67, 621.
3. Guo, B. C.; Kerns, K. P.; Castleman, A. W., Jr. J. Phys. Chem. 1992, 96, 6931.
4. Kapiloff, E.; Ervin, K. M. J. Phys. Chem. A 1997, 101, 8460.
5. Parks, E. K.; Zhu, L.; Ho, J.; Riley, S. J. J. Chem. Phys. 1994, 100, 7206.
6. Morse, M. D.; Geusic, M. E.; Heath, J. R.; Smalley, R. E. J. Chem. Phys. 1985, 83, 2293.
7. Hintz, P. A.; Ervin, K. M. J. Chem. Phys. 1995, 103, 7897.
8. Guo, B. C.; Kerns, K. P.; Castleman, A. W., Jr. J. Chem. Phys. 1992, 96, 8177.
9. Leuchtner, R. E.; Harms, A. C.; Castleman, A. W., Jr. J. Chem. Phys. 1990, 92, 6527.
10. Blitz, M. A.; Mitchell, S. A.; Hackett, P. A. J. Phys. Chem. 1991, 95, 8719.
11. Föhlisch, A.; Nyberg, M.; Bennich, P.; Triguero, L.; Hasselström, J.; Karis, O.; Pettersson, L. G. M.; Nilsson, A. J. Chem. Phys. 2000, 112, 1946.
12. Cox, D. M.; Reichmann, K. C.; Trevor, D. J.; Kaldor, A. J. Chem. Phys. 1988, 88, 111.
13. Hintz, P. A.; Ervin, K. M. J. Chem. Phys. 1994, 100, 5715.
14. Vajda, Š.; Wolf, S.; Leisner, T.; Busolt, U.; Wöste, L. H.; Wales, D. J. J. Chem. Phys. 1997, 107, 3492.
15. Fayet, R.; McGlinchey, M. J.; Wöste, L. H. J. Am. Chem. Soc. 1987, 109, 1733.
16. Nygren, M. A.; Siegbahn, P. E. M. J. Phys. Chem. 1992, 96, 7579.
17. Rohlfing, C. M.; Hay, P. J. J. Chem. Phys. 1985, 83, 4641.
18. Parks, E. K.; Kerns, K. P.; Riley, S. J. J. Chem. Phys. 2000, 112, 3384.
160
19. Reddy, B. V.; Nayak, S. K.; Khanna, S. N.; Rao, B. K.; Jena, P. J. Phys. Chem. A 1998, 102, 1748.
20. Schulze Icking-Konert, G.; Handschuh, H. Ganteför, G.; Eberhardt, W. Phys. Rev.
Lett. 1996, 76, 1047.
21. Kerns, K. P.; Parks, E. K.; Riley, S. J. J. Chem. Phys. 2000, 112, 3394.
22. Wang, L.-S. Photoionization and Photodetachment: Part II, Adv. Series in Phys. Chem. Vol 10B (C.-Y. Ng, Ed.) World Scientific: River Edge, NJ 2000, 854-957.
23. Wu, H.; Desai, S. R.; Wang, L.-S. J. Chem. Phys. 1995, 103, 4363.
24. Ganteför, G.; Schelze Icking-Konert, G.; Handschuh, H.; Eberhardt, W. Int. J.
Mass Spec. Ion Proc. 1996, 159, 81.
25. Gord, J. R.; Bemish, R. J.; Freisher, B. S. Int. J. Mass. Spec. Ion Proc. 1990, 102, 115.
26. Spasov, V. A.; Lee, T.-H.; Ervin, K. M. J. Chem. Phys. 2000, 112, 1713.
27. Krückeberg, S.; Schweikhard, L.; Ziegler, J.; Dietrich, G.; Lützenkirchen, K.; Walther, C. J. Chem. Phys. 2001, 114, 2955.
28. Lian, L.; Su, C.-X.; Armentrout, P. B. J. Chem. Phys. 1992, 96, 7542.
29. Armentrout, P. B.; Halle, L. F.; Beachamp, J. L. J. Chem. Phys. 1982, 76, 2449.
30. Konova, P.; Stoyanova, M.; Naydenov, A.; Christoskova, S.; Mehandjiev, D. Appl. Catal. A 2006, 298, 109.
31. Sueyoshi, T.; Sasaki, T.; Iwasawa, Y. J. Phys. Chem. 1996, 100, 1048.
32. Sueyoshi, T.; Sasaki, T.; Iwasawa, Y. Chem. Phys. Lett. 1995, 241, 189.
33. Coulman, D. J.; Wintterlin, J.; Behm, R. J.; Ertl, G. Phys. Rev. Lett. 1990, 64, 1761.
34. Nayak, S. K.; Khanna, S. N.; Rao, B. K.; Jena, P. J. Phys. Chem. A 1997, 101, 1072.
35. Russon, L. M.; Heidecke, S. A.; Birke, M. K.; Conceicao, J.; Morse, M. D.;
Armentrout, P. B. J. Chem. Phys. 1994, 100, 4747.
161
36. Wittborn, A. M. C.; Costas, M.; Blomberg, M. R. A.; Siegbahn, P. E. M. J. Chem. Phys. 1997, 107, 4318.
37. Fielicke, A.; von Helden, G.; Meijer, G.; Pedersen, D. B.; Simard, B.; Rayner, D. M. J. Chem. Phys. 2006, 124, 194305.
38. Nagase, K.; Zheng, Y.; Kodama, Y.; Kakuta, J. J. Catal. 1999, 187, 123.
39. Jernigan, G. G.; Somorjai, G. A. J. Catal. 1994, 147, 567.
40. Huang, T.-J.; Tsai, D.-H. Catal. Lett. 2003, 87, 173.
41. Jansson, J. J. Catal. 2000, 194, 55.
42. Grillo, F.; Natile, M. M.; Glisenti, A. Appl. Catal. B 2004, 48, 267. 43. Reilly, N. M.; Johnson, G. E.; Reveles, J. U.; Khanna, S. N.; Castleman, A. W.,
Jr. Chem. Phys. Lett. 2007, 435, 295.
44. Negishi, Y.; Nakamura, Y.; Nakajima, A.; Kaya, K. J. Chem. Phys. 2001, 115, 3657.
45. Arrii, A.; Morfin, F.; Renouprez, A. J.; Rousset, J. L. J. Am. Chem. Soc. 2004,
126, 1199.
46. Comotti, M.; Li, W. C.; Spliethoff, B.; Schüth, F. J. Am. Chem. Soc. 2006, 128, 917.
47. Chen, M.; Cai, Y.; Yan, Z.; Goodman, D. W. J. Am. Chem. Soc. 2006, 128, 6341.
48. Kozlova, A. P.; Sugiyama, S.; Kozlov, A. I.; Asakura, K.; Iwasawa, Y. J. Catal.
1998 176, 426.
49. Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M. J.; Delmon, B. J. Catal. 1993, 144, 175.
50. Bell, R. C.; Zemski, K. A.; Justes, D. R.; Castleman, A. W., Jr. J. Chem. Phys.
2001, 114, 798.
51. Reilly, N. M.; Reveles, J. U.; Johnson, G. E.; Khanna, S. N.; Castleman, A. W., Jr. J. Phys. Chem. A 2007, Accepted.
52. Bell, R. C.; Zemski, K. A.; Kerns, K. P.; Deng, H. T.; Castleman, A. W., Jr. J.
Phys. Chem. A 1998, 102, 1733.
162
53. Uzunova, E. L.; Nikolov, G. St.; Mikosch, H. J. Phys. Chem. A 2002, 106, 4104.
54. Pramann, A.; Koyasu, K.; Nakajima, A.; Kaya, K. J. Phys. Chem. A 2002, 106, 4891.
55. Aristov, N.; Armentrout, P. B. J. Phys. Chem. 1986, 90, 5135.
56. Vann, W. D.; Castleman, A. W., Jr. J Phys. Chem. A 1999, 103, 847.
57. Vardhan, D.; Liyanage, R.; Armentrout, P. B. J. Chem. Phys. 2003, 119, 4166.
58. Wang, L.-S.; Wu, H.; Desai, S. R.; Lou, L. Phys. Rev. B 1996, 53, 8028.
59. Tan, L.; Liu, F.; Armentrout, P. B. J. Chem. Phys. 2006, 124, 084302.
60. Justes, D. R.; Mitrić, R.; Moore, N. A.; Bonačić-Koutecký, V.; Castleman, A. W., Jr. J. Am. Chem. Soc. 2003, 125, 6289.
61. Blyholder, G. J. Phys. Chem. 1964, 68, 2772.
62. Lupinetti, A. J.; Jonas, V.; Thiel, W.; Strauss, S. H.; Frenking, G. Chem. Eur. J. 1999, 5, 2573.
63. Anderson Fredeen, D.; Russell, D. H. J. Am. Chem. Soc. 1985, 107, 3762.
64. Kimble, M. L.; Moore, N. A.; Johnson, G. E.; Castleman, A. W., Jr.; Bürgel, C.; Mitrić, R.; Bonačić-Koutecký, V. J. Chem. Phys. 2006, 125, 204311.
65. Neukermans, S.; Janssens, E.; Tanaka, H.; Silverans, R. E.; Lievens, P. Phys. Rev.
Lett. 2003, 90, 033401.
66. Teruuchi, K.; Habazaki, H.; Kawashima, A.; Asami, K.; Hashimoto, K. Appl. Catal. 1991, 76, 79.
67. Saito, N.; Sakamoto, M.; Nishiyama, H.; Inoue, Y. Chem. Phys. Lett. 2001, 341,
232.
68. Epling, W. S.; Hoflund, G. B.; Weaver, J. F.; Tsubota, S.; Haruta, M. J. Phys. Chem. 1996, 100, 9929.
Chapter 9
General Conclusions
The field of heterogeneous catalysis is ever-evolving with new technologies and
better methods for energy-efficient conversion of fossil fuels. One major concern that
arises from the incomplete combustion of fuel is the environmental pollutant, carbon
monoxide. With increasingly strict standards for automotive emissions and the push for
cleaner burning fuels, improved methods for converting fuel are needed. The studies in
this thesis were aimed at providing a fundamental picture of the oxidation of CO and the
mechanisms that govern the conversion of CO to CO2 in the presence of alternative
metals. Current catalysts employed in industry to reduce pollution consist of costly
platinum or palladium metals which require high working temperatures. Metal catalysts
capable of low temperature conversion are ideal in order to avoid unnecessary energy
consumption for non-viable products. With a better understanding of the mechanisms
governing potentially active alternative metals, the ability to tailor design catalysts to
effect specific oxidation reactions could be realized. There were two main alternative
catalytic materials investigated in this thesis, noble metal oxides in Chapters 3 and 4, and
transition metal oxides in Chapters 5-8.
The combined studies examined in this thesis reveal new information on the
mechanisms of different catalytic compositions toward the oxidation of CO. Also, these
studies provide corroborative evidence of recent findings with a proposed mechanism to
explain observed products, such as the existence of intermediate species with dissociated
164
oxygen (1, 2, 3). These initial studies uncover information based mostly on individual
active sites in catalysis, gleaned from studies based on size and stoichiometry. For
instance transition metal oxides, often employed in catalytic supports, were found to
promote the direct oxidation of CO. This is evidence that support materials actively
participate in the catalytic conversion of CO to CO2, an issue which has stirred much
debate in the literature (4).
9.1 Major Findings
The major theme throughout these studies was the investigation of the ionic
charge state and its effect on the electron density and charge transfer to different metal
atoms at the nano-scale. It was shown that especially in the catalytic oxidation of CO in
noble metal systems that negative charge was more effective in promoting oxidation. It
has been suggested in the literature that negatively charged metal centers donate charge
to O2 thereby elongating the O-O bond and further activating the dissociation (5).
It is interesting that calculations by our collaborators at the Humboldt University
and VCU predicted that positively charged metal centers also would promote the
oxidation of CO. The reason was that metal-oxygen bonds were found to be weaker in
cationic systems compared to anionic systems due to charge transfer effects. Indeed, a
kinetic analysis of the rate constants for the direct oxidation of CO with Fe2O3+ and
Fe2O3- showed the cationic rate constant was 8.5x10-13 ± 0.8 cm3s-1 compared to the
anionic rate contant of 2.9x10-13 ± 0.2 cm3s-1. However, cationic gold oxide clusters of
similar stoichiomerty to anionic gold oxide clusters, found to undergo oxygen atom
165
transfer previously, did not undergo CO oxidation. This is an indication of the different
mechanisms occurring on gold centers compared to transition metal oxide centers which
promoted CO oxidation.
For transition metal oxide systems, oxygen atoms are bound directly to the metal
atom, which enables direct oxidation with the use of oxygen atoms coming from the
metal. In contrast, the Langmuir-Hinshelwood type mechanism is the operative
mechanism for most catalysts (6). A reaction proceeding through this mechanism would
adsorb CO and react with adsorbed oxygen, not oxygen that is already bound to the metal
catalyst. Thus, oxygen dissociation is a barrier needed to be overcome for the reaction to
proceed. This step is eliminated in transition metals that are not saturated with oxygen
and do not contain O-O bonds. Therefore, transition metals tend to be reduced as CO is
oxidized and the cycle continues based on a redox cycle mechanism. The nature of
transition metals is to be stable in several oxidation states, therefore making them
particularly suitable for direct oxidation applications and catalytic supports.
The stoichiometry and oxygen to metal ratio was also an important factor in the
observed reaction pathways. It is not surprising that our findings show oxygen rich
transition metal oxide clusters were more active toward CO oxidation since they employ
their own oxygen atoms in the reaction. Interestingly, AuO- and Au2O-, both oxidize CO
whereas no negatively charged transition metal oxides were even formed with the same
number of oxygen and metal atoms. In comparison, positively charged transition metal
oxides were formed and underwent CO oxidation for clusters containing the same
number of metal and oxygen atoms. This demonstrates the formation mechanisms of
metal oxides and the chemistry of oxygen bonding to different charged metal species.
166
The studies in this thesis were based on dissociated as well as molecularly adsorbed
oxygen since oxygen was added directly in the source region where cluster formation
occurs. This allowed the investigation of potential barriers to CO oxidation after O-O
bond dissociation. Indeed, recent studies conducted in our laboratory found that
dissociated oxygen is necessary but not sufficient for the reaction to proceed on anionic
gold oxide clusters (7). The reaction rates for CO oxidation in the presence of Au2O3-
and Au2O4-, which both contained dissociated oxygen, were on the order of 10-13 cm3s-1.
Size was found to be very important in the overall reaction pathways observed. In
the nano-scale regime, there is much controversy over the smallest active size of metal
nanoparticles that are able to oxidize CO (8). Our studies reveal only larger cationic gold
oxides (clusters with at least four gold atoms) undergo CO oxidation. Whereas, previous
investigations show that only select monomer and dimer anionic gold oxide clusters were
effective in promoting the oxidation of CO (7). Also, cationic gold oxides and larger
anionic gold oxides had other reaction products that decreased the overall efficiency and
selectivity of the oxidation reaction channel.
Although CO oxidation is the most viable reaction channel to understand for use
in industrial catalysts, it is important to identify the mechanisms of other reaction
channels observed. Oxygen replacement by CO was a common reaction channel observed
in cationic noble metal and transition metal oxides. The positively charge metal centers
were more efficient in this reaction channel than CO oxidation was in anionic metal
centers due to weaker oxygen bonding. In a kinetic analysis of the rate constants, the
replacement rate constant for Au2O+ was an order of magnitude greater, 10-12 cm3s-1, then
the oxidation rate constant for Au2O-, 10-13 cm3s-1. It is possible that the replacement
167
reaction pathway may have prevented the slower CO oxidation pathway from occurring
or being the major reaction channel. Also, the strength of M-CO bonds formed between
cationic metal clusters showed that CO was bound strongly and desorption of CO2 was
limited. Therefore, potential intermediates in the oxidation of CO were too stable to react
further.
Thus many factors contribute to the overall reactivity of metal oxides and it is
important to study the role of ionic charge, size, stoichiometry, and composition in order
to understand the governing active sites and mechanisms in catalysis. From the studies
presented in the previous chapters of this thesis, there remain several questions to be
investigated as a result of the findings.
9.2 Future Studies
Further research on bimetallic systems would be beneficial to providing a
complete picture of the operative mechanisms of the oxidation of CO. The production of
larger clusters with different stoichiometries of silver and gold atoms would reveal
interesting information on trends and active sites for CO oxidation. Although the dual
rod source described in Chapter 2 produced mixed cluster species of Ag-Au oxides,
increased signal intensity is needed to further this research. Method development on a
new source may be the most promising for larger bimetallic clusters. Different mixed
metal systems would also be of interest as electron transfer differs in different metal
centers shown in condensed phase studies of increased reactivity for only certain metal
oxide supports.
168
It would be interesting to study the transition metal systems with oxygen added
downstream from the source. The 5-component guided ion beam mass spectrometer
could be employed to study this reaction by adding another reaction cell octopole and
third quadrupole to the chamber. Mass selection would occur in the 1st quadrupole, and
CO could then be added to the first octopole. Investigations with molecular oxygen added
into the first octopole could determine how oxygen is adsorbed to the ionic metal centers.
Next CO added to the second octopole would provide conditions close to condensed
phase adsorption of both oxygen and CO. Product species detected from these studies
would be very interesting to compare to those predicted by the Langmuir-Hinshelwood
catalytic mechanism especially for transition metal oxide clusters. The transition metal
oxide clusters show promise in becoming more active alternatives to costly precious
metal catalysts and more insight into their reaction mechanisms would be beneficial.
169
References
1. Socaciu, L. D.; Hagen, J.; Heiz, U.; Bernhardt, T. M.; Leisner, T.; Wöste, L. Chem. Phys. Lett. 2001, 340, 282.
2. Socaciu-Siebert, L. D.; Hagen, J.; Le Roux, J.; Popolan, D.; Vaida, M.; Vajda, Š.;
Bernhardt, T. M.; Wöste, L. Phys. Chem. Chem. Phys. 2005, 7, 2706.
3. Henao, J. D.; Caputo, T.; Yang, J. H.; Kung, M. C; Kung, H. H. J. Phys. Chem. B 2006, 110, 8689.
4. Negishi, Y.; Nakamura, Y.; Nakajima, A.; Kaya, K. J. Chem. Phys. 2001, 115,
3657.; Arrii, A.; Morfin, F.; Renouprez, A. J.; Rousset, J. L. J. Am. Chem. Soc. 2004, 126, 1199.; Comotti, M.; Li, W. C.; Spliethoff, B.; Schüth, F. J. Am. Chem. Soc. 2006, 128, 917.
5. Böhme, D. K.; Schwarz, H. Angew. Chem. Int. Ed. 2005, 44, 2336.; Yoon, B.;
Häkkinen, H.; Landman, U. J. Phys. Chem. A 2003, 107, 4066.; Stolcic, D.; Fischer, M.; Ganteför, G.; Kim, Y. D.; Sun, Q.; Jena, P. J. Am. Chem. Soc. 2003, 125, 2848.; Chen, M. S.; Goodman, D. W. Science 2004, 306, 252.
6. Baxter, R. J.; Hu, P. J. Chem. Phys. 2002, 116, 4379.
7. Kimble, M. L.; Moore, N. A.; Johnson, G. E.; Castleman, A. W., Jr.; Bürgel, C.;
Mitrić, R.; Bonaćič-Koutecký, V. J. Chem. Phys. 2006, 125, 204311.
8. Lee, S.; Fan, C.; Wu, T.; Anderson, S. L. J. Am. Chem. Soc. 2004, 126, 5682.; Sanchez, A.; Abbet, S.; Heiz, U.; Schneider, W.-D.; Häkkinen, H.; Barnett, R. N.; Landman, U. J. Phys. Chem. A 1999, 103, 9573.; Wallace, W. T.; Whetten, R. L. J. Am. Chem. Soc. 2002, 124, 7499.
Appendix
Kinetic Analysis of the Reaction between (V2O5)1,2+ and Ethylene
A.1 Abstract
A systematic experimental and theoretical investigation of the influence of
reactant energy on the reactivity of (V2O5)n=1,2+ clusters with ethylene provided evidence
of the rate controlling steps in the reaction (1). Herein, we present further experimental
and theoretical evidence for our recently proposed radical cation mechanism for oxygen
atom transfer from (V2O5)n=1,2+ clusters to ethylene. In particular the results of ab initio
molecular dynamics simulations are found to further support the radical cation
mechanism. Experimental reaction cross sections at the zero pressure limit and rate
coefficients show that the energy dependence of the reaction cross section is in accord
with the Langevin formula. Evidence is presented that ion-molecule association is the
rate determining step, whereas subsequent hydrogen transfer and formation of
acetaldehyde proceed without significant barriers. We propose a kinetic model for the
reaction cross section which fully accounts for the experimental findings. The model
offers the prospect of elucidating the details of the general reaction mechanisms through
a study of the energy dependence of the reaction cross sections.
(Reproduced in part with permission from J. Phys. Chem. B 2006, 110, 3015-3022. Copyright 2006 American Chemical Society.)
171
A.2 Introduction
There is growing interest in the use of metal oxide clusters as model systems for
gaining insight into the mechanisms of various reactions of importance in the field of
catalysis (2-4). Gas phase cluster research is proving to be very valuable in determining
the properties and behavior of transition metal oxides, yielding information on
thermochemistry (5), structure (6-13), and reactivity (14-16). Recently, we have
undertaken a comprehensive joint theoretical and experimental effort to reveal the
reactive behavior of vanadium oxides of various sizes and stoichiometries with a variety
of small organic molecules. Our findings show that among vanadium oxides of widely
differing stoichiometries and sizes, only V2O5+ and V4O10
+ effect oxygen transfer
reactions with ethylene (1, 17-18). In particular, these results have prompted us to devote
considerable attention to a study of these species. Prior evidence from experimental
investigations obtained in our laboratory (1) and theoretical findings revealed evidence of
a facile oxygen transfer mechanism in direct analogy to ones believed to occur on
vanadia surfaces at reaction sites of similar stoichiometry (15, 18). Our earlier theoretical
work, in combination with the experimental studies, proposed the involvement of an
oxygen centered radical reaction, proceeding subsequent to formation of a metal oxide
cluster-molecule complex (1).
Specifically, we have obtained evidence of a general radical cation mechanism for
the oxidation of ethylene on cationic vanadium oxide clusters with the composition
(V2O5)n=1,2+, leading to the formation of acetaldehyde as a neutral product according to
the reactions (1):
172
V2O5+ + C2H4 → V2O4
+ + C2H4O
V4O10+ + C2H4 → V4O9
+ + C2H4O
(I)
(II)
The key point is a radical center located at a peripheral oxygen atom on the V2O5+
and V4O10+ clusters. The reaction is considered to proceed through a generic reaction
profile based on energetic data obtained from ab initio calculations as shown in Figure 1-
1. The initial reaction step is believed to proceed via the formation of a complex arising
from an ion-molecule collision of A and B (Figure 1-1). The structural transformation
along the reaction pathway indicates that a single bond is ultimately formed between the
first carbon atom of the ethylene and the oxygen atom (complex C). This leads to the
shift in the radical center, initially located at the oxygen atom, to the second carbon atom
of the ethylene molecule. In the second reaction step, a hydrogen atom is transferred
from the first to the second carbon atom, forming a complex in which the acetaldehyde
molecule is bound to the vanadium atom (complex D). Finally, in the third step, the V-O
bond is broken and an acetaldehyde molecule is released, resulting in an overall oxygen
transfer process between the two reactants, leading to (E+F).
To provide a more complete understanding of the factors influencing the reaction
dynamics of this observed oxygen atom transfer reaction, we have undertaken further
experimental and theoretical studies of the energetics of the reaction. The findings,
which are reported herein, demonstrate consistency with a proposed mechanism of this
reaction that is of considerable interest in the area of fundamental catalysis, as well as in
the industrial production of acetaldehyde (19).
173
Figure 1-1: General scheme of the energetic profile and structural transformation of the mechanism for the reaction between (V2O5)n=1,2
+ and C2H4. Source: Figure courtesy of the group of Professor V. Bonačić-Koutecký at the Humboldt Unisversity in Berlin.
A+B
C
D
E+F
A+B
C
D
E+F
174
A.3 Experimental and Computational Methods
The experiments performed for this study were conducted using a guided ion
beam mass spectrometer coupled to a laser vaporization source, described in detail
previously (20) and briefly explained here. Metal oxide species were generated using a
laser vaporization source, whereby a vanadium rod was ablated using the second
harmonic of a Nd:YAG laser while rotating and translating to continually expose a fresh
surface. At a predetermined time, a small amount of oxygen seeded in helium, ~8%, was
pulsed over the surface of the rod, creating a plasma in which the vanadium oxide
clusters were formed. The clusters were then cooled in a region of supersonic expansion
and guided into the first quadrupole through a skimmer and a set of electrostatic lenses.
The ion of interest was then mass selected and focused through a second set of
electrostatic lenses into a reaction cell incorporated within an octopole guide. The
effective path length of the octopole reaction cell was calculated to be 12.9 cm using the
trapezoidal pressure fall-off approximation (20). The pressure of the ethylene in the
reaction cell was kept constant while the energy of the octopole was increased from 0 to
20 volts (laboratory frame) in single volt increments. The lab-frame energies (Elab) were
converted to center-of-mass frame energy (ECM), ECM=Elab[M/(M+m)], in order to enable
a comparison of calculated and experimental values. Here, M is the mass of neutral
target gas and m is the mass of the selected ion. The described procedure was repeated
for pressures ranging from 0.1 to 0.6 mTorr of ethylene, chosen to effect single as well as
multiple collision reaction conditions. The products generated in the octopole reaction
175
cell were focused into the second quadrupole through a third set of electrostatic lenses,
mass analyzed, and detected using a channel electron multiplier.
The calculations performed by the Bonačić-Koutecký group and presented here
were carried out at the density functional level of theory using the Becke’s hybrid three
parameter non-local exchange functional combined with the Lee-Yang-Parr gradient
corrected correlation functional (B3LYP) (21) and employing the all-electron triple-ζ
valence plus polarization basis set (TZVP) developed by Ahlrichs and co-workers (22).
As shown in our previous work, this method allows for the accurate description of
structural properties of cationic vanadium oxide clusters and their interaction with
ethylene (1). The reaction pathways were studied employing ab initio molecular
dynamics (MD) with forces calculated by using density functional theory (23). For the
solution of Newton’s equations of motion, the Verlet algorithm was used. The MD
simulations were initiated from the activated stable complexes of (V2O5)n=1,2+ with
ethylene by randomly distributing the energy among all degrees of freedom. This
allowed us to verify reaction steps deduced from the calculation of the stationary points
and obtain support for the proposed mechanism.
A.4 Results and Discussion
Ab initio MD simulations at constant energy, corresponding to the stability of the
initial complex with respect to the reactants, were carried out to gain further evidence for
the operative reaction mechanism. This corresponds to the limiting case of a single
176
collision condition in the experiment (1) in which the energy of the initial complex is not
released by collisions with gas molecules.
The snapshots of the MD trajectory for the reaction between V2O5+ and C2H4 are
shown in Figure 1-2. The formation of complex C is a barrierless process. The
simulation beginning with this complex shows that the association of ethylene to V2O5+ is
followed by a very rapid (20 fs) hydrogen transfer from the carbon atom bound to oxygen
toward the terminal carbon atom. After the hydrogen transfer has taken place (complex
D), the V-O bond is broken within the next 40 fs, leading to acetaldehyde and V2O4+ as
the final products (E and F noted in Figure 1-1). Analogous snapshots for the reaction
between V4O10+ and C2H4 are shown in Figure 1-2. Similar to the case of V2O5
+, the
reaction with V4O10+ also involves very rapid hydrogen transfer (~30 fs) followed by the
formation of acetaldehyde which occurs typically within 100 fs.
In addition to ab initio MD simulations, experiments were performed by
systematically raising the energy of the reaction under conditions of constant selected
pressure of ethylene within the reaction cell. Figures 1-3 and 1-4 show the branching
ratios of the reactions of V2O5+ and V4O10
+ with ethylene, respectively. The relative
intensities of the ions are plotted as a function of the energy added to the octopole
reaction cell. Minor molecular oxygen loss products, V2O3+ and V4O8
+, are observed and
are insensitive to the energy addition. The decrease in reactivity with increasing energy
is evident from the data in Figure 1-3, which show that there is comparatively less
conversion of V2O5+ reactant ion to the major product, V2O4
+, at higher collision energies.
A similar situation arises for V4O10+ and V4O9
+ as seen in Figure 1-4.
177
Figure 1-2: Snapshots of the ab initio MD trajectory calculated using DFT methods for the reaction of a) V2O5
+ with C2H4 and b) V4O10+ with C2H4 leading to the formation of
acetaldehyde. Source: Figure courtesy of the group of Professor V. Bonačić-Koutecký at the Humboldt Unisversity in Berlin.
178
Figure 1-3: Branching ratios for the energetic analysis of V2O5+ with ethylene conducted
under the following pressures: a) 0.1 mTorr, b) 0.2 mTorr, c) 0.3 mTorr, d) 0.4 mTorr,e) 0.5 mTorr and f) 0.6 mTorr of ethylene. These are defined by the ratio of each product ion divided by the total ion signal. Notice that the minor contribution of V2O3
+ is essentially independent of the center-of-mass reaction energy.
179
Figure 1-4: Branching ratios for the energetic analysis of V4O10+ with ethylene
conducted under the following pressures: a) 0.1 mTorr, b) 0.2 mTorr, c) 0.3 mTorr, d) 0.4 mTorr, e) 0.5 mTorr and f) 0.6 mTorr of ethylene.
180
In both cases, as the pressure of the ethylene is increased from 0.1 to 0.6 mTorr,
the energy under which equal amounts of the parent and the oxygen transfer product ions
are attained is increased. This mimics a negative temperature dependence and is not the
behavior expected in the case of collision induced dissociation reactions. In fact,
collision induced dissociation experiments have shown that the observed oxygen
deficient products, V2O4+ and V4O9
+, were not the result of a collisional process because
they were not observed in separate experiments using an inert gas at near thermal
energies (6). Accordingly, the pressure corresponding to the crossing point of the
branching ratios can be calculated as follows:
),)(
exp(0 Tk
pEII
B
CMp
lσ−= (A-1)
where I0 is the sum of the reactant and all product ion intensities and Ip is the final
intensity of the product ion, in this case VxOy-1+. The total reaction cross section is
σ(ECM) and kB is the Boltzmann constant. The effective path length of the collision cell is
ℓ, while T and p are the temperature and the pressure of the reactant gas, respectively.
Cross Section Anaylsis. The experimental cross sections of both V2O5+ and
V4O10+ clusters with ethylene are calculated according to eq A-1. In order to account for
the pressure dependence of the cross section measurements, data from several pressures
are plotted and then extrapolated to zero pressure in the customary way (24). At the low
pressure limit, eq A-1 becomes
181
Tk
pI
I
E
B
p
CM l0)( =σ
(A-2)
yielding the cross section from the slope of the line in Figure 1-5 for the intensity ratio
Ip/I0 as a function of pressure. These plots are then repeated for every energy input into
the reaction cell.
For V4O10+, charge transfer and associated ion reaction products, C2H4
+ and
C3H6+, can be seen in the branching ratios of Figure 1-4. These products are accounted
for in the cross section calculations through the sum of product intensities via the analysis
using eq A-2. The secondary reaction products, which are a consequence of multiple
collisions, become significant only at 0.6 mTorr. The experimental cross sections are
presented and discussed below with the kinetic model.
Reaction Rates. The reaction rate coefficients, which are dependent on the
lifetime of the ion-molecule complex, were also calculated for both oxygen transfer
reactions. The experimental cross section data are converted to an expression for the
phenomenological rate coefficient (24) by eq A-3
)()( 00 CMEvvk σ= (A-3)
where the relative velocity (ν0) of the reactant ion is
2/1
0
2⎟⎟⎠
⎞⎜⎜⎝
⎛=μ
CMEv
(A-4)
182
It should be emphasized that this equation is valid only in the case that the energy
distribution of the reactant ions is very narrow (monoenergetic ions). The energy input
into the reaction cell (ECM) is in the center-of-mass frame, and μ is the reduced mass.
Figure 1-6 shows the phenomenological rate constants calculated from zero
pressure cross section data and therefore does not account for multiple collision
conditions in our higher pressure experiments. Another approach to calculate the rate is
via the usual method given by eq A-5
[ ][ ],BAkdt
dA −= (A-5)
and employed in analyzing data acquired in a flow tube reactor operated at low pressures
and low electric drift fields. Use of eq A-5 requires knowledge of the interaction time in
the reaction cell. This time is determined by means of calculation in the present case.
The slope of Figure 1-7, which is a plot of ln[Ip/I0] versus the concentration of the
reactant ion, is equal to –kt. A series of graphs, similar to Figure 1-7, are plotted for each
energy value.
The time is found by estimating the sum of the velocity due to the supersonic
nozzle expansion (25) and the velocity acceleration from the voltage field. The
supersonic expansion velocity is given by eq A-6
2/1
2
11 ⎥
⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
⎟⎠⎞
⎜⎝⎛ −+
=T
Ts
Mm
RTMv
γγ
(A-6)
where MT is the mach number, m is the mass of the ion, and γ is the specific heat capacity
ratio.
183
Figure 1-5: a) Ratio of product ion intensity, Ip, to total ion intensity, I0, versus the reaction cell gas pressure for reactions of V2O5
+ with ethylene at ECM= 0 eV and b) at ECM= 2.0 eV.
Figure 1-6: a) The rate constant versus energy for V2O5
+ with ethylene calculated from cross section data. b) The rate constant versus energy for V4O10
+ with ethylene calculated from cross section data. The error bars represent the relative uncertainty in the cross section measurements.
y = 1.77x
R 2 = 0.95 0
0.2 0.4 0.6 0.8 1.0 1.2
0 0.2 0.4 0.6 0.8 Pressure (mTorr)
a) y = 0.70x
R 2 = 0.93 0
0.1 0.2 0.3 0.4 0.5 0.6
0 0.2 0.4 0.6 0.8
b)
I p/I
0
Pressure (mTorr)
2.0x10-10
4.0x10-10
6.0x10-10
8.0x10-10
1.0x10-9
1.2x10-9
0.0 1.0 2.0 3.0 EnergyCM (eV)
2.0x10-10
4.0x10-10
6.0x10-10
8.0x10-10
1.0x10-9
1.2x10-9
0.0 0.3 0.6 0.9 1.2 1.5 EnergyCM (eV)
Rat
e C
onst
ant
(cm
3 s-1
) a) b)
184
In this case, the heat capacity was calculated for atoms because the carrier gas was mostly
comprised of helium being pulsed through the nozzle. The velocity due to the voltage
applied to the octopole in the reactant gas cell is deduced from eq A-7
,2
m
qEv LAB=
(A-7)
where the charge of the ion is q, and ELAB is the electric field voltage applied to the
reaction cell in laboratory frame. The electrostatic lens immediately before the reaction
cell is set at 0 volts, creating a voltage field difference when energy is added to the
octopole. The effective path length of the reaction cell divided by the sum of the
velocities provides reaction times ranging from approximately 0.3x10-4 to 1x10-4 seconds
for V2O5+, over the energy range studied, and 0.4x10-4 to 1.4x10-4 seconds for V4O10
+.
Compared to the effect of the electric field, the contribution due to supersonic expansion
is minimal. As the energy in the reaction cell is increased the time decreases because the
ions are moving faster through the cell. These times are within the experimental time
frame used in collision induced dissociation experiments and are due to the short distance
between the reaction collision and the quadrupole entrance (20). Finally, the negative
slope of Figure 1-7 divided by the time corresponding to the energy input yields the rate
coefficient.
Figure 1-8 shows the rate coefficients plotted versus energy. These values were
calculated from the experimental data, which include effects due to the pressure and
concomitant multiple collision conditions. It can be seen from Figure 1-8 that the rate of
V2O5+ oxygen transfer is on the order of high-10-10 cm3s-1 and is slightly higher than the
rates for V4O10+ oxygen transfer which are in the mid-10-10 cm3s-1 range.
185
Figure 1-8: a) The rate constant versus center-of-mass energy for the reaction of V2O5+
with ethylene calculated from the velocity of the ions and reaction time and b) The rate constant versus center-of-mass energy for the reaction of V4O10
+ with ethylene calculated from the velocity of the ions and reaction time. The error bars are the relative uncertainties of measurements considered in the calculation.
Figure 1-7: a) Logarithm of the intensity ratio [V2O5
+/V2O5++ΣIp] versus the
concentration of C2H4 calculated as an ideal gas at ECM=0 eV and b) at ECM= 2 eV.
Rat
e C
onst
ant
(cm
3 s-1
)
5.0x10-11
3.5x10-10
6.5x10-10
9.5x10-10
1.3x10-9
0.0 1.0 2.0 3.0 EnergyCM (eV)
1.0x10-11
3.1x10-10
6.1x10-10
9.1x10-10
0.0 0.3 0.6 0.9 1.2 1.5 EnergyCM (eV)
a) b)
y = - 6E - 14x - 0.11 R 2
- 1.4 - 1.2
- 1 - 0.8 - 0.6 - 0.4 - 0.2
0 0 1E+13 2E+13
[C 2 H 4 ] mc/cc
Ln r /I ]
y = - 2.8E - 14x + 0.02 R 2 = 0.94
- 0.6 - 0.5 - 0.4 - 0.3 - 0.2 - 0.1
0 0 1E+13 2E+13 3E+13
[C 2 H 4 ] mc/cc -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2
0 1x1013 2x1013 3x1013
Conc. [C2H4] mc/cc -0.6 -0.5 -0.4 -0.3 -0.2 -0.1
0 0 1x1013 2x1013 3x1013
Conc. [C2H4] mc/cc
0
R2=0.92 R2=0.94
a) b)
y=-6x10-14x-0.11 y=-2.8x10-14x+0.02
ln[I
r/I0]
186
The rate constant for V2O5+ at lower energies first increases, and then slowly
decreases at values above ~1 eV center-of-mass energy. This decrease in the rate
constant is what is expected for an exothermic reaction with negative temperature
dependence and no potential barriers. The trend in V4O10+ is similar with the rate
constant increasing and then leveling off at higher energies. It is interesting to note that
there is reasonably good agreement in the rate constants derived by the two methods, and
presented in Figures 1-6 and 1-8, respectively. The data in Figure 1-6 have been obtained
by analyzing data based on single collision conditions to acquire cross sections, and then
treated on the basis of a narrow velocity distribution to acquire rate constants.
Conversely, the data in Figure 1-8 are acquired through calculations that describe our
experimental conditions of higher pressures in a manner typically employed in flow tube
kinetics. Although branching ratios are taken into account, the latter method involves an
average rate constant under multiple collision conditions.
Kinetic Model. A kinetic model for the reaction cross section for the oxygen
transfer rate follows from eq A-1. Combining with eq A-3, and as kt→0 for short times
and low pressure, it follows that
tTk
pk
I
I
B
R )(1 0
0
ν−= (A-8)
where IR is the intensity of the reactant ion and t is time. This is rearranged in terms of
the rate constant and gives eq A-9
tTk
pI
I
k
B
p
00 )( =ν .
(A-9)
187
Employing eq A-3 and the relationship, l=t0ν , along with the crossing points, Ip/I0 =
0.5, shown in Figures 1-3 and 1-4, it follows that
l)(
2ln2/1
CM
B
E
Tkp
σ=
(A-10)
and the crossing point for a given energy is a direct measure of the reaction cross section.
If we assume that the initial step involves formation of an ion-molecule association
between ethylene and the mass selected reactant oxide ion, understanding the reaction
cross section requires consideration of the kinetics of several potentially operative
reaction steps in the overall conversion mechanism.
In order to derive the expression for the reaction cross section as a function of
energy we write the reaction symbolically in accordance with the mechanism depicted in
Figure 1-1 as
FEDCBAkk
k
k
k+→⇔⇔+
53
4
1
2
. (III)
By using the steady state approximation for the intermediates C and D, shown in Figure
1-1, the change in C and D with respect to time are obtained by the following equations
54
3 ][][
kk
CkD
+=
(A-11)
and
54
4332
1 ]][[][
kk
kkkk
BAkC
+−+
= (A-12)
The individual rate constants shown in the equations are also defined in Figure 1-1. It
readily follows that the total reaction rate d[E]/dt, is given by eq A-13,
188
== ][][
5 Dkdt
Ed]][[
))(( 435432
531 BAkkkkkk
kkk
−++
and the overall rate constant by eq 14
(A-13)
435432
531
))(( kkkkkk
kkkk
−++=
(A-14)
The first step of the reaction can be modeled after a long-range ion-molecule association,
and the reaction cross section can be calculated from the Langevin cross section, σL,
2/1
0
2
4 ⎟⎟⎠
⎞⎜⎜⎝
⎛=
CML E
e αε
σ (A-15)
where the collisional energy of the reactant gas is ECM, α is the polarizability of ethylene,
and ε0 is the vacuum permittivity. In principle, the rate constants k and k1 can be
obtained from the energy dependent reaction cross sections by averaging over the
velocity distribution. However, because the energy distribution of the ions under the
considered experimental conditions is narrow, it can be assumed that the velocity
distribution is also narrow. In this limiting case, as an approximation, the rate constants k
and k1 are simply products of the velocity of the ions and the corresponding cross
sections. This yields a total reaction cross section as
435432
53
))(()(
kkkkkk
kkE LCM −++
= σσ (A-16)
The rate constants for the elementary reaction steps can be approximated using the RRK
expression, yielding the following four equations for k2, k3, k4 and k5.
1
122 )( −
Δ+= N
EE
EAk
(A-17)
189
1
1
#1
33 )( −
Δ+−Δ+= N
EE
EEEAk
(A-18)
1
2
#1
44 )( −
Δ+−Δ+= N
EE
EEEAk
(A-19)
1
2
3255 )( −
Δ+Δ−Δ+= N
EE
EEEAk
(A-20)
In the above equations, N is taken to be the number of vibrational degrees of freedom.
The parameters ΔE1, ΔE2, ΔE3, and E# are defined in Figure 1-1 and A2, A3, A4, and A5
represent the respective frequency factors. The frequency factor A2 is approximated by
the frequency of the normal mode which leads to the formation of complexes V2O5+-
C2H4 and V4O10+-C2H4 (Figures 1-2(a) and 1-2(b), respectively, at t=0). This results in
values of ν = 992 cm-1 for V2O5+ and ν = 502.0 cm-1 for V4O10
+. Frequency factors A3
and A4 are calculated using theoretical normal mode frequencies for the complexes C and
D. Complex D is related to the structures shown in Figure 1-2(a) (t=20 fs) for V2O5+-
C2H4 and Figure 1-2(b) (t=50 fs) for V4O10+-C2H4 and the transition state (1) according
to:
#1)4(3i
Ni
iNiA
νν
−ΠΠ= .
(A-21)
A5 is approximated by the frequency of the V-O stretching vibration which leads to the
products and has a value of ν = 484 cm-1 for V2O5+-C2H4 and ν = 397.3 cm-1 for V4O10
+-
C2H4 complexes.
To apply RRKM theory to the reaction mechanism given in Figure 1-1, it is
necessary to identify the transition states of the association, hydrogen transfer, and
190
acetaldehyde formation. No transition states are available because the association step
engages an ion-molecule reaction without a barrier and the formation of acetaldehyde
involves a barrier-less V-O bond dissociation. Therefore, RRK theory is used to
calculate the rate constants for these two reaction steps using approximate frequency
factors of the normal modes which lead to the association and emanation of acetaldehyde.
For the hydrogen transfer step, RRK theory is also used because no low frequency modes
are present for which the harmonic approximation underlying RRK theory fails.
Moreover, because the barrier for the hydrogen transfer is significantly lower than the
total internal energy acquired in the association step, quantitative differences between the
RRK and RRKM theory are not expected to influence the final reaction cross section.
Figure 1-9 compares the experimental cross sections at the zero pressure limit
and the cross sections calculated from the Langevin model corresponding to the center-
of-mass energy. The theoretical cross sections for the reaction of V2O5+ with ethylene in
Figure 1-9(a) employ eqs A-16 to A-20 with the following values for parameters defined
in Figure 1-1: ΔE1 = 3.85 eV, ΔE2 = 5.19 eV, ΔE3 = 2.66 eV, and E# = 1.45 eV obtained
from DFT calculations (1,26). Figure 1-9(b) is a plot for the reaction of V4O10+ and
ethylene calculated in the same way with parameter values as follows: ΔE1 = 2.18 eV,
ΔE2 = 4.17 eV, ΔE3 = 2.16 eV, and E# = 0.13 eV.1 As can be seen from eq A-16, the total
reaction cross section consists of the product of two terms. The first term is the pure
Langevin cross section for the association of ethylene with the metal oxide, and the
second term is dependent on the rate constants for the subsequent steps (A-16 to A-20).
191
The second term involving the rate constants is found through calculations to be unity for
these particular parameter values due to the low reaction barriers. Therefore, based on
the above assumptions the total reaction cross section is predicted to be accounted for
exclusively by the Langevin cross section. It is interesting to note that Hanmura et al.
have found that the reaction of benzene with nickel cluster ions is also governed by
Langevin kinetics (27).
Comparing the cross sections obtained theoretically using eq A-16 with those
from experiment, we find good agreement between experiment and theory for both
systems as shown in Figure 1-9. It is reasonable that the experimental cross sections are
systematically lower, showing that not every collision is an effective collision. Both
V2O5+ and V4O10
+ with ethylene have decreasing reaction cross sections with increasing
energy, characteristic of an exothermic reaction process with a negative temperature
dependence.
This provides consistent evidence that subsequent oxygen transfer and formation
of acetaldehyde proceed through a mechanism where the barriers are significantly lower
than the energy of the initial association complex. On the basis of the theoretical model
presented here, together with the experimental findings, we show that the reaction
kinetics are comparatively insensitive to the detailed energetics of the involved steps.
This applies as long as the barrier for the hydrogen transfer is significantly lower than the
energy of the initial complex. This is in contrast to what has been recently claimed for
the energy profile differences between the 2A'' ground state and the 2A' ground state in
V2O5+ (9). Although the reaction energies for V2O5
+ + C2H4 change for the 2A'' ground
192
state, it is insignificant. The barriers for the reaction are lower in energy than those for
the initial rate-limiting association step, showing that the reaction is governed by
Langevin kinetics. Hence subsequent reaction steps readily proceed.
Subsequent to the above study, we conducted an investigation to ascertain under
which conditions deviations from the Langevin cross sections would be expected to
occur. In Figure 1-10 the reaction cross sections are plotted for four different values of
the hydrogen transfer barrier, 1.45 eV, 3.75 eV, 3.85 eV, and 3.88 eV. Whereas the cross
section for the situation of a barrier of 1.45 eV is indistinguishable from the Langevin
cross sections, deviations start to occur when the barrier reaches a value of 3.85 eV which
corresponds to the energy of the initial complex. For barriers higher than 3.85 eV, the
behavior is qualitatively different, giving rise to a cross section which initially increases
with energy, reaches a maximum, and then decreases. This is illustrated in Figure 1-10
for only a slightly increased barrier with respect to 3.85 eV. These findings will allow for
insight into mechanisms of related reactions through a study of the energetics of the
reactive cross sections.
A.5 Conclusion
A theoretical and experimental analysis of the reaction of V2O5+ and V4O10
+ with
ethylene provides additional information that demonstrates consistency with the
previously proposed energetic profile and mechanism of the oxygen transfer radical
cation reaction.
193
Figure 1-9: a) Energy dependence of the total reaction cross sections for V2O5+ with
ethylene for the theoretical model (solid line) and experimental measurements (diamonds). b) Energy dependence of the total reaction cross sections for V4O10
+ with ethylene. The error bars represent one standard deviation of the experimental cross section calculated from the root mean square value of the least squares analysis.
Figure 1-10: Dependence of the reaction cross section (σ(E)) on the energy for different values of the barrier for hydrogen transfer: 1.45 eV (black line), 3.75 eV (blue line), 3.85 eV (green line), 3.88 eV (red line). Source: Figure courtesy of the group of Professor V. Bonačić-Koutecký at the Humboldt Unisversity in Berlin.
0 10
20
30 40 50 60
0 0.5 1.0 1.5 2.0 2.5 3.0 0
10
20
30 40 50 60
0 0.5 1.0 1.5 2.0 2.5 3.0 0
10
20
30 40 50 60
0 0.5 1.0 1.5 2.0 2.5 3.0 0
10
20
30
40
50
60
0 0.5 1.0 1.5 2.0 2.5 3.0 0
10
20
30
40
50
60
70
0 0.5 1.0 1.5 2.0
a) b)
EnergyCM (eV) EnergyCM (eV)
σ[Å
2 ]
194
Comparison between experimental and theoretical results has given evidence that all
steps following the formation of the initial complex as well as the hydrogen transfer
reaction proceed with barriers which are significantly lower than that of the initial
association complex. This finding supports our earlier proposed reaction mechanism,
furthermore showing that the reactions effectively proceed according to the Langevin
model. This established that the reaction is driven by the initial rate constant and
associated cross section that governs the encounter of the reactants and leads to the
formation of an intermediate collision complex. Moreover, the kinetic model proposed
allows one, in principle, to estimate the barriers from experimental data on the energy
dependence of the reaction cross section, providing important information about cluster
reactivity.
Acknowledgements
NAM, DRJ, and AWC, Jr. gratefully acknowledge the U. S. Department of
Energy, Grant No. DE-FG02-92ER14258, for financial support of the experimental work
reported herein, and RM and VBK acknowledge support from Deutsche
Forschungsgemeinschaft (DFG). We thank Dr. Michele Kimble for helpful discussions
during the course of the study.
195
References
1. Justes, D. R.; Mitrić, R.; Moore, N. A.; Bonačić-Koutecký, V.; Castleman, A. W., Jr. J. Am. Chem. Soc. 2003, 125, 6289.
2. Muetterties, E. L. Science 1977, 196, 839. 3. Fialko, E. F.; Kikhtenko, A. V.; Goncharov, V. B.; Zamaraev, K. I. J. Phys. Chem. B
1997, 101, 5772. 4. Waters, T.; O’Hair, R. A. J.; Wedd, A. G. J. Am. Chem. Soc. 2003, 125, 3384. 5. Armentrout, P. B. Int. J. Mass Spectrom. 2000, 200, 219.; Xu, J.; Rodgers, M. T.;
Griffin, J. B.; Armentrout, P. B. J. Chem. Phys. 1998, 108, 9339. 6. Bell, R. C.; Zemski, K. A.; Kerns, K. P.; Deng, H. T.; Castleman, A. W., Jr. J. Phys.
Chem. A 1998, 102, 1733. 7. Oliveira, M. C.; Marçalo, J.; Vieira, M. C.; Almoster Ferreiraac, M. A. Int. J. Mass
Spectrom. 1999, 185-187, 825. 8. Asmis, K. R.; Brümmer, M.; Kaposta, C.; Santambrogio, G.; von Helden, G.; Meijer,
G.; Rademann, K.; Wöste, L. Phys. Chem. Chem. Phys. 2002, 4, 1101. 9. Asmis, K. R.; Meijer, G.; Brümmer, M.; Kaposta, C.; Santambrogio, G.; Wöste, L.;
Sauer, J. J. Chem. Phys. 2004, 120, 6461. 10. Nalewajski, R. F.; Korchowiec, J. Computers Chem 1995, 19, 217.
11. Pacchioni, G.; Ferrari, A. M.; Giamello, E. Chem. Phys. Lett. 1996, 255, 58.
12. Vyboishchikov, S. F.; Sauer, J. J. Phys. Chem. A 2000, 104, 10913.
13. Fielicke, A.; Mitrić, R.; Meijer, G.; Bonačić-Koutecký, V.; von Helden, G. J. Am. Chem. Soc. 2003, 125, 15716.
14. Kimble, M. L.; Castleman, A. W., Jr.; Mitrić, R.; Bürgel, C.; Bonačić-Koutecký, V. J.
Am. Chem. Soc. 2004, 126, 2526. 15. Zemski, K. A.; Justes, D. R.; Castleman, A. W., Jr. J. Phys. Chem. B 2002, 106, 6136.
16. Berg, C.; Beyer, M.; Achatz, U.; Joos, S.; Niedner-Schatteburg, G.; Bondybey, V. E. J. Chem. Phys. 1998, 108, 5398.
196
17. Justes, D. R.; Mitrić, R.; Bonačić-Koutecký, V.; Castleman, A. W., Jr. Eur. Phys. J. D 2003, 24, 331.
18. Zemski, K. A.; Justes, D. R.; Castleman, A. W., Jr. J. Phys. Chem. A 2001, 105,
10237. 19. Oyama, S. T.; Middlebrook, A. M.; Somorjai, G. A. J. Phys. Chem. 1990, 94, 5029. 20. Bell, R. C.; Zemski, K. A.; Justes, D. R.; Castleman, A. W., Jr. J. Chem. Phys. 2001,
114, 798. 21. Becke, A. D. Phys. Rev. A 1988, 98, 3098.; Becke, A. D. J. Chem. Phys. 1993, 98,
5648.; Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1998, 37, 785. 22. Schäfer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829.
23. Mitrić, R. Ph.D. Thesis, Humboldt University of Berlin, 2003.
24. Ervin, K. M.; Armentrout, P. B. J. Chem. Phys. 1985, 83, 166.
25. Anderson, J. B.; Fenn, J. B. Phys. Fluids 1965, 8, 780.
26. Note: ∆E1 = 3.85 eV has been obtained based on the 2A′ ground state of V2O5+ (ref.
9). 27. Hanmura, T.; Ichihashi, M.; Kondow, T. J. Phys. Chem. A 2002, 106, 4525.
VITA
Nelly Moore Reilly
Education The Pennsylvania State University, University Park, Pennsylvania Doctorate of Philosophy in Chemistry, May 2007.
Ursinus College, Collegeville, Pennsylvania
Bachelor of Science in Chemistry, May 2000 Experience 2002-2007 Graduate Research Assistant Department of Chemistry The Pennsylvania State University, University Park, PA 2000-2002 Associate Scientist
Nutritional Research Wyeth Pharmaceuticals, Collegeville, PA
Awards 2006 Geiger Fellowship 2006 Graduate Student Travel Award for APS Conference 2005 Braucher Fellowship 2004 Graduate Student Travel Award to Humboldt University 2002 Roberts Fellowship Professional Affiliations 2000-present American Chemical Society, Analytical Division 2006-present American Physics Society Select Publications
1. Reilly, N. M., Johnson, G. E., Reveles, J. U., Khanna, S. N., and Castleman, A. W., Jr. “Experimental and Theoretical Study of the Structure and Reactivity of Fe1-2O≤6
- Clusters with CO.” J. Phys. Chem. A 2007, Accepted.
2. Reilly, N. M., Johnson, G. E., Reveles, J. U., Khanna, S. N., and Castleman, A. W., Jr. Chem. Phys. Lett. 2007, 435, 295.
3. Moore, N. A., Mitrić, R., Justes, D. R., Bonačić-Koutecký, V., and Castleman, A. W., Jr. J. Phys. Chem. B 2006, 110, 3015-3022. Select Presentation
1. Moore, N. A., Kimble, M. L., Johnson, G. E., and Castleman, A. W., Jr. “Investigation of Gas Phase Gold Oxide Cations towards the Oxidation of Carbon Monoxide” American Physics Society, Baltimore, MD (March 2006).
top related