Bowker M., Davies P.R. (Eds.) Scanning Tunneling Microscopy in Surface Science, Nanoscience and Catalysis

Подождите немного. Документ загружается.

also found, in the presence of 2 10

7

Torr H

at 65 K, vacancy aggregates with four

or more vacancies are also ﬁlled by H atoms from the dissociation of H

within the

aggregates and transformed into a single vacancy or totally annihilated. Through a

series of STM movies on the diffusion of surface hydrogen vacancies, the authors

found that only an aggregate with three or more vacancies could be annihilated by H

dissociation. The dimers or isolated vacancies are never occupied by H atoms.

Instead, the dimers always dissociate creating isolated vacancies with an average

lifetime of 10 min at 65 K. From these data, the authors concluded that three or more

empty palladium sites are necessary for the dissociation of H

molecules. This

ﬁnding is rather surprising since it has traditionally been assumed that two

neighboring empty sites are sufﬁcient for the dissociation of a diatomic molecule.

The discovery of the active sites for H

dissociation on Pd(1 1 1) illustrates the power

of in situ STM in addressing the elementary steps of surface reactions and in testing

the conventional assumptions in catalysis.

3.3.3

Determining the Sites for Chemisorption on Oxide Surfaces

On reduced oxide surfaces, the diffusion of an adsorbate is often limited by the

localized bonding, either ionic or covalent, between the adsorbate and the oxide

substrate. The relatively slow diffusion of adsorbates allows chemisorption and

diffusion on oxides to be studied by STMat elevated temperatures. For example, TiO

is an excellent photocatalyst for dissociation of water and decomposition of organic

molecules, critical to pollution control and the hydrogen economy. Studies of the

adsorption and diffusion of water, oxygen, and organic molecules on TiO

are of

primary importance to our understanding of photocatalysis by TiO

Being the most stable phase of TiO

, the rutile TiO

(1 1 0) crystal has been studied

most extensively by STM and other surface science techniques [37]. Figure 3.6 shows

a structural model of the rutile TiO

(1 1 0)-(1 1) surface. The surface contains two

types of titanium atoms that form rows along the [0 0 1] direction. Rows of six-

coordinated Ti atoms alternate with ﬁve-coordinated terminal Ti atoms, which miss a

single O atom perpendicular to the surface. The surface also contains two kinds of

oxygen atoms, that is, three-coordinated oxygen atoms, sitting in the surface plane,

and bridging oxygen atoms, sitting above the surface plane and bonded to two six-

coordinated Ti atoms. Undersaturated bridging oxygen atoms can be easily removed

from the surface by annealing, electron bombardment, or ion sputtering to form

bridging oxygen vacancies. Bridging oxygen vacancies are the most common and

well-deﬁned defects on the TiO

(1 1 0) surface.

STM studies on the adsorption and diffusion of small molecules on the TiO

(1 1 0)

surface began in the late 1990s [37–39] and have provided a general understanding of

the important role of bridging oxygen vacancies in the dissociation of water, oxygen,

and small organic molecules. However, due to the difﬁculties in distinguishing active

surface sites and dissociation products, only recently have the fundamental steps

of adsorption and dissociation processes been understood with the help of

in situ STM.

3 In Situ STM Studies of Model Catalysts

Wendt et al. [40] and Bikondoa et al. [41] studied the adsorption and dissociation of

water and O

on TiO

(1 1 0). In situ STM studies, in parallel with the use of DFT

methods, allow surface features such as bridging oxygen vacancies, adsorbed oxygen

atoms, surface hydroxyls, and adsorbed water to be distinguished. Figure 3.7a

illustrates the difference between a bridging oxygen vacancy, a surface hydroxyl,

and adsorbed water in their appearance in STM images. Due to electronic effects,

STM resolves the ﬁve-coordinated terminal Ti rows as bright rows whereas the

bridging oxygen rows are imaged as dark rows. The nuances in the appearance of

oxygen vacancies and hydroxyls in STM images are distinguished by visualizing

oxygen vacancies being transformed into OH species in situ. The authors have also

found that applying voltage pulse (3 V) over the hydroxyls desorbs individual H

atoms from the hydroxyl groups while leaving the oxygen vacancies intact. Using in

situ STM, Wendt et al. [42] demonstrated that water dissociation takes place at

bridging oxygen vacancies of the TiO

(1 1 0) surface at 187 K and form paired

hydroxyl groups, with one positioned at the oxygen vacancy site and the other formed

at the nearest bridging oxygen site. The diffusion of these pairs is inhibited at 187 K

but can be initiated in the presence of neighboring water molecules adsorbed in the

ﬁve-coordinated Ti trough. Through the interaction with neighboring water mole-

cules, hydrogen atoms from the paired hydroxyl groups are transferred to adjacent

bridging oxygen rows, causing the cross-row diffusion of hydroxyl groups.

Dohnalek and coworkers [43, 44] have further measured the diffusion kinetics of

hydroxyl groups (or H atom) at room temperature and above. In situ STM

images (Figure 3.8) have conﬁrmed that H

O dissociates at the bridging oxygen

Figure 3.6 The ball model of the TiO

(1 1 0) surface. Large gray

(red) balls represent O atoms, small light gray (gray) balls five-

coordinated surface Ti atoms (5f-Ti), and small black balls six-

coordinated Ti atoms (6f-Ti). The bridging oxygen atoms (O

single oxygen vacancies (V

), and O atoms adsorbed on top of the

5f-Ti row (O

) are indicated. (Reprinted with permission from

3.3 Visualizing the Pathway of Catalytic Reactions

Figure 3.7 STM images (16 nm 16 nm) of clean, reduced

TiO

(1 1 0) samples showing the difference between bridging

oxygen vacancies, surface hydroxyls, and adsorbed water. The

sample in (a) was less reduced than the sample in (b). (c) STM

height profiles along the ½1



10 direction of species indicated in

Elsevier.)

Figure 3.8 STM images of the same area on

TiO

(1 1 0) at 357 K (V

¼1.5 V, I

¼0.1 nA)

as a function of time (Dt ¼60 s): (a) clean

TiO

(1 1 0) with bridging oxygen (BBO)

vacancies; (b) TiO

(1 1 0) with a geminate

hydroxyl pair formed by adsorption and dis-

sociation of a water molecule. H

marks the

OH hydrogen and H

the hydrogen that split off

from the OH; (c) same area after a single hop

of H

; and (d) after subsequent hop of H

Insets exhibit the ball models illustrating the

corresponding processes. (Reprinted with

The American Chemical Society.)

3 In Situ STM Studies of Model Catalysts

vacancies, producing paired hydroxyl groups. Hydrogen atoms readily diffuse at

room temperature along the bridging oxygen row. However, the two hydrogen atoms

in the paired hydroxyl groups exhibit inequivalent diffusivity. The hydrogen atom

positioned at the healed oxygen vacancy site (H

) diffuses much slower than the

one adsorbed at neighbor bridging oxygen sites (H

). The different diffusion rates of

and H

were measured between 300 and 410K and the activation barrier of H

estimated to be approximately 0.22eV lower than H

. The diffusion barrier of

these hydrogen atoms increases with the separation between hydroxyl groups,suggest-

inga repulsive OHOHinteraction.Theauthors speculatedthat a long-livedpolaronic

state is responsible for the inequivalent diffusion rates of H

and H

.However,the

measured kinetic parameters (prefactors and diffusion barriers) could not be repro-

duced by DFT calculations, suggesting a rather complex diffusion mechanism.

The pathway of dissociation and diffusion discovered in theabove water adsorption

experiment also applies to the adsorption of alcohols on TiO

(1 1 0). In situ STM has

been used to study the adsorption of methanol [45] and butanol on TiO

(1 1 0) [46].

Figure 3.9 shows a series of STM images obtained on the same area following

exposure to methanol. Figure 3.9a depicts the clean surface before exposure, with

bridging oxygen vacancies marked as yellow circles. The exposure of 0.06 ML

methanol led to the dissociative adsorption of methanol at the bridging oxygen

vacancies of TiO

(1 1 0) (Figure 3.9b). The dissociation of methanol forms a methoxy

group at the bridging oxygen vacancy, resolved as bright features in Figure 3.9b, and a

hydroxyl group at the nearest neighbor. The methoxy group has a similar appearance

to the bridging oxygen vacancy in STM images, except that the methoxy group is 0.8 Å

higher than the bridging oxygen vacancy. The hydroxyl group could not be differ-

entiated from the bright features of neighboring methoxy groups in Figure 3.9b.

However, with time, the hydrogen from the hydroxyl group (red dots in Figure 3.9c

and d) diffuses along the bridging oxygen row and across the bridging oxygen rows

through interactions with methanol molecules weakly bounded to the Ti trough. The

diffusing hydrogen atoms were identiﬁed by their apparent height in STM images

and by a tip desorption experiment (Figure 3.9e) proposed by Bikondoa et al. [41].

Figure 3.9f gives a graphic illustration of the dissociation and diffusion pathway of

methanol on TiO

(1 1 0), which was also observed in the adsorption experiment of

2-butanol (CH

CH(OH)CH

)onTiO

(1 1 0) at room temperature [45].

The adsorption and dissociation of O

on TiO

(1 1 0) have been investigated by

Wendt et al. [40] and Du et al. [47]. Both observed that O

molecules adsorb and

dissociate at the bridging oxygen vacancies, with one O adatom healing the vacancy

and the other O adatom bounded to the neighboring ﬁve-coordinated Ti site. Du et al.

also analyzed the lateral distribution of the O adatoms upon dissociation and

discovered a transient mobility of O adatoms along the Ti trough in the [0 0 1]

direction. Unlike the dissociative adsorption of O

on metal surfaces where both

adatoms have equal diffusivity, the diffusivity of O adatoms on TiO

(1 1 0) was found

to be inequivalent. While the O adatoms ﬁlling the vacancy are locked in the bridging

oxygen row, O adatoms in the Ti trough are relatively free to move. A majority of O

adatoms on the Ti trough (81%) were found separated from the O adatoms in the

previous vacancy sites by two lattice constants.

3.3 Visualizing the Pathway of Catalytic Reactions

Figure 3.9 STM images of same area before

and after adsorption of methanol on reduced

TiO

(1 1 0) at 300 K (V

¼1.0 0.3 V and

¼<0.1 nA): (a) bare surface; (b) after 80 s

exposure to methanol; (c) after 110 s exposure

to methanol; (d) taken on (c) after spontaneous

tip change; (e) after high bias (3.0 V) sweep of

(c); (f) schematic model of the adsorption

process. Insets show magnified areas marked

by squares. Yellow circles show the position of

bridging oxygen vacancies. Blue circles show the

methoxy groups on oxygen vacancies. Red

squares show H atoms diffusing on bridging

oxygen rows. (Reprinted with permission from

Chemical Society.)

3 In Situ STM Studies of Model Catalysts

Zhang et al. [48] have recently measured the stability of bridging oxygen vacancies

on TiO

(1 1 0) using in situ STM. Sequences of STM images between 340 and 420 K

suggest that bridging oxygen vacancies migrate along the bridging oxygen row via the

slow diffusion of bridging oxygen atoms with a diffusion barrier of 1.15 eV, in

agreement with DFT calculations. All the above studies suggest that the surface

chemistry of TiO

(1 1 0) is dictated by bridging oxygen vacancies, which can account

for approximately10% of the bridging oxygen sites.

However, there are disagreements. Lyubinetsky et al. [49] studied the adsorption

of trimethylacetic a cid ((CH

)

CCOOH, TMAA), a photoreact ive molecul e, on

TiO

(110)atroomtemperature.In situ STM found that the deprotonation of

TMAA to form TMA does not necessarily occur at bridging oxygen vacancies. None

of the hydroxyl groups was found during the adsorption of TMAA. Instead, the

hydrogen a tom was bound to a pair of bridging oxygen atoms and stabilized by

the adjacent TMA groups sitting on the ﬁve -coordinated Ti trough. At saturation

coverage, TMAA formed a (2 1) overlayer on the TiO

(1 1 0) surface.

Wendt et al. [50] recently studied the interaction between O

and TiO

(1 1 0) surface

in detail and suggested that bridging oxygen vacancies are only the minor sites that

account for O

dissociation. Even though bridging oxygen vacancies account only for

approximately 10% of surface bridging oxygen sites, exposing the clean TiO

(1 1 0) to

a few Langmuirs ofO

could not fully remove all bridging oxygen vacancies. To isolate

the inﬂuence of bridging oxygen vacancies in O

dissociation, the authors created a

perfect TiO

(1 1 0) surface by exposing the TiO

(1 1 0) surface to water at room

temperature. Hydroxyl groups, formed via water dissociation, covered all bridging

oxygen vacancies, yielding a vacancy-free TiO

(1 1 0) surface (Figure 3.10a).

Figure 3.10c and d illustrates that O

exposure can fully remove surface hydroxyl

groups and create a TiO

(1 1 0) surface with perfect bridging oxygen rows, as

previously suggested in TPD studies [51]. With the titration of hydroxyl groups,

oxygen adatoms on the ﬁve-coordinated Ti row (O

) also increase (Figure 3.10b).

However, the increase in oxygen adatoms does not seem to stop even after all the

hydroxyl groups have been replaced with oxygen (Figure 3.10c and d). Paired O

atoms start to appear on the ﬁve-coordinated Tirow during extended O

exposure. On

the basis of these observations, the authors showed that a second and primary O

dissociation channel is operative on the ﬁve-coordinated Ti row.

STM results combined with photoelectron spectroscopy (PES) on the valence

state of TiO

(1 1 0) further show that the removal of all hydroxyl groups by oxygen,

leading to a perfect TiO

(1 1 0) surface, only slightly attenuates the Ti 3d defect

state (Figure3.10e). The full attenuation of Ti3d state requires 420 L of O

. Figure 3.9f

plots the evolutions of the Ti 3d defect state and the OH 3s state over O

exposure

and suggests the Ti 3d defect state is not mainly caused by bridging oxygen vacancies.

The authors suggest that other types of defects, Ti

3 þ

interstitials that form during

the reduction of TiO

(1 1 0) and are hidden beneath the surface, are primarily

responsible for the formation of the Ti 3d defect state and the dissociative adsorption

of O

Indeed, the importance of Ti

3 þ

interstitials has also been realized in previous in

situ STM studies of the reoxidation of TiO

(1 1 0) [52–54]. It is noted that Ti

3 þ

3.3 Visualizing the Pathway of Catalytic Reactions

interstitials diffuse to the surface in the presence of O

and form TiO

species, which

serve as the building blocks for the regrowth of TiO

(1 1 0) plane. Using in situ STM,

Bowker and coworkers [52] measured the reoxidation kinetics of reduced TiO

(1 1 0)

surface at temperatures between 573 and 1000 K and oxygen pressures of 5 10

8

2 10

6

mbar. By monitoring the TiO

species that diffused and coalesced on the

TiO

(1 1 0) surface, the surface growth rate could be measured through the change of

island morphology. This growth rate was found to be linear with respect to the oxygen

partial pressure. The regrowth of TiO

(1 1 0) has a low activation energy of

approximately 25 kJ/mol, suggesting a high mobility of Ti

3 þ

interstitials in the

presence of oxygen. Previous studies on the reduction of TiO

(1 1 0) [37] have shown

the rutile TiO

(1 1 0) bulk serves as a huge reservoir for Ti

3 þ

interstitials during

reduction, so that the surface is maintained near-stoichiometry and is thermody-

namically stable.

Figure 3.10 (a–d) STM images (105 Å 105 Å)

of the TiO

(1 1 0) surface covered with

hydroxyls [h-TiO

(1 1 0)] and then exposed to

increasing amounts of O

at room temperature.

(e) Selected PES valence-band spectra recorded

on an h-TiO

(1 1 0) surface that was exposed to

at RT. Arrows indicate the representative

STM images. (f) Normalized integrated

intensities of the OH 3s (red) and Ti 3d (blue)

features for O

exposures up to 420 L from PES

spectra; circles indicate intensity values that

were obtained from the spectra shown in (e).

(Reprinted with permission from Ref. [50].

the Advancement of Science.)

3 In Situ STM Studies of Model Catalysts

3 þ

interstitials are essentially oxygen vacancies within the bulk whereas bridging

oxygen vacancies are basically undercoordinated Ti ions at the surface. It is not

surprising to expect that Ti

3 þ

interstitials in the subsurface play a role in the

dissociation of adsorbed molecules. In the above STM study by Wendt et al.,itis

worth noting that subsurface Ti

3 þ

interstitials were neither visualized nor seen to

diffuse to the TiO

(1 1 0) surface during O

adsorption at room temperature.

Considering PES usually probes the top few layers at the surface, the complete

attenuation of the Ti 3d defect state suggests that Ti

3 þ

interstitials within those layers

have been oxidized and therefore quenched. Details of how the excess charge of Ti

3 þ

interstitials involves in the bond breaking of O

molecules remain to be elucidated.

Nevertheless, the above studies demonstrate the power of in situ STM in tracing the

active sites, hidden or unhidden. The ﬁnding of mobile defects in a rigid TiO

(1 1 0)

will stimulate more investigations on other reducible oxides and encourage revisiting

their surface chemistry.

3.3.4

Visualizing Reaction Intermediates and the Mechanism of Hydrogen Oxidation

In situ STM has also been used to study the pathway of surface reactions and to

measure their kinetics. The hydrogen oxidation reaction, catalyzed by Pt group

metals to produce water, was the ﬁrst catalytic reaction discovered in 1823. This

reaction is still a core catalytic reaction at the heart of fuel cell technologies. Although

the oldest of catalytic reactions, the mechanism of catalytic hydrogen oxidation is still

unclear, especially for its surprising reactivity at or below the water desorption

temperature (170 K). It has been proposed that the reaction proceeds via the

combination of dissociatively adsorbed O atoms (O

) and H atoms (H

), forming

hydroxyl groups that subsequently bind with another H

to form water. The

formation of hydroxyl groups has been postulated as the rate-limiting step. However,

the hydroxyl group, as a reaction intermediate, could not be conﬁrmed in early

surface spectroscopic studies. It has not been until the past decade, with the help of in

situ STM, that the reaction mechanism has become apparent for the hydrogen

oxidation reaction on metal surfaces.

Wintterlin and coworkers [55, 56] studied the hydrogen oxidation reaction on

Pt(1 1 1) using in situ STM. The study was conducted as a titration experiment, in

which the surface was precovered with O

and then the O-terminated surface

subsequently exposed to H

molecules. The O-terminated Pt(1 1 1) surface was

prepared by exposing the clean Pt(1 1 1) surface to 10 L of O

, followed by annealing at

225 K to dissociate O

. The surface was then exposed to 8 10

9

mbar H

at 131 K

and monitored by STM, as shown in Figure3.11. The Pt(1 1 1) surface was precovered

by a (2 2)-O

layer, where O

atoms were imaged as dark dots (Figure 3.11a). Upon

exposure to H

, a few bright islands form on top of the (2 2)-O

layer

(Figure 3.11b). These islands continue to grow and eventually develop into an

ordered layer with hexagonal and honeycomb phases (Figure 3.11c). These two

phases are characterized as a surface hydroxyl (OH

) overlayer, formed by hydrogen

bonding. Additional diffusing islands seen in Figure 3.11c are attributed to adsorbed

3.3 Visualizing the Pathway of Catalytic Reactions

O islands (H

). Figure 3.11a–c shows the formation of hydroxyl group as the

reaction intermediate and reveal the atomic details of hydrogen oxidation catalyzed by

Pt(1 1 1).

Figure 3.11d–f presents snapshots of STM images of the O

-terminated Pt(1 1 1)

surface exposed to 2 10

8

mbar H

at 112 K. The imaged area includes several

surface terraces covered with numerous small bright dots and a bright ring that

expands with time. These small bright dots have been assigned to small OH

islands

that appear after H

exposure. The white ring, termed as the reaction front, is

concentrated with small OH

islands and continues to grow as the reaction

progresses, suggesting a fast reaction at the boundaries between O

atoms and

the diffusing H

. The fast reaction between O

and H

produces two OH

which in turn forms H

through a rapid reaction with H

atom. Since the

combination of O

and H

to form OH

is the rate-limiting step, the presence of

removes this kinetic limit and promotes an autocatalytic cycle until depletion

of O

atoms. STM images thus illustrate an autocatalytic reaction mechanism that

accounts for the low activation barrier and high reactivity of hydrogen oxidation at or

Figure 3.11 Series of successive STM images,

recorded during dosing of the O-covered

Pt(1 1 1) surface with H

.(a–c) Frames

(17 nm 17 nm) from an experiment at 131 K

[P(H

) ¼8 10

9

mbar]. The hexagonal pattern

in (a) is the (2 2)-O structure; O atoms appear

as dark dots and bright features are the initial

OH islands. In (c), the area is mostly covered by

OH, which forms ordered structures. The white,

fuzzy features are H

O-covered areas.

(d–f) Frames (220 nm 220 nm) from an

experiment at 112 K [P(H

) ¼2 10

8

mbar].

In (d), the surface is mostly O-covered

(not resolved). The bright dots are small OH

islands, most of which are concentrated in the

expanding ring. H

O in the interior of the ring is

not resolved here. Thin, mostly vertical lines are

atomic steps. (Reprinted with permission from

Association for the Advancement of Science.)

3 In Situ STM Studies of Model Catalysts

below the water desorption temperature. At temperatures higher than 170 K, where

O starts to desorb, the shortened lifetime of H

breaks down the autocatalytic

cycle by stopping the fast reaction between H

and O

to form OH

The autocatalytic reaction mechanism apparent at low temperatures is expected to

apply to catalytic hydrogen oxidation at high pressures. In addition, the above study is

the ﬁrst to use STM to observe the formation of dynamic surface patterns at the

mesoscopic level, which had previously been observed by other imaging techniques

in surface reactions with nonlinear kinetics [57]. This study illustrates the ability of in

situ STM to visualize reaction intermediates and to reveal the reaction pathway with

atomic resolution.

3.3.5

Measuring the Reaction Kinetics of CO Oxidation

Another important catalytic reaction that has been most extensively studied is CO

oxidation catalyzed by noble metals. In situ STM studies of CO oxidation have focused

on measuring the kinetic parameters of this surface reaction. Similar to the above

study of hydrogen oxidation, in situ STM studies of CO oxidation are often conducted

as a titration experiment. Metal surfaces are precovered with oxygen atoms that are

then removed by exposure to a constant CO pressure. In the titration experiment, the

kinetics of surface reaction can be simpliﬁed and the reaction rate directly measured

from STM images.

Wintterlin et al. [58] investigated the catalytic oxidation of carbon monoxide on

Pt(1 1 1) using in situ STM. Oxygen atoms were preadsorbed on the Pt(1 1 1) surface

by exposing the surface to 3 L O

at 96 K, followed by a short anneal at 293 K to

dissociate O

. The oxygen-covered Pt(1 1 1) surface was then cooled to 247 K and

exposed to 5 10

8

Torr CO. STM was used to follow the change of surface adsorbate

structures as a function of the CO exposure time (Figure 3.12). At 247 K, O

atoms

form an ordered (2 2) overlayer, imaged as dark dots. At t ¼0, the Pt(1 1 1) surface is

mainly covered by the (2 2)-O layer together with empty Pt sites, imaged as bright

islands, scattered on the surface. The addition of CO molecules lowers the mobility of

surface oxygen atoms and slowly compresses the (2 2)-O layer into large islands.

The adsorbed CO molecules form ordered c(4 2) domains on the Pt(1 1 1) surface.

As time progresses, the areas of c(4 2) CO domains continue to grow at the expense

of the (2 2)-O islands. From the series of in situ STM images, the rate of CO

oxidation can be estimated based on the reduction rate of the surface areas of the

(2 2)-O islands.

Figure 3.13 plots the dependence of the reaction rate on the surface area or

perimeter of oxygen domains, as a function of time. Approximately, the reaction rate

is linear with respect to the perimeter of the surface oxygen domains, suggesting CO

oxidation mainly occurs along the boundary between the oxygen and the CO domains

on the Pt(1 1 1) surface. The titration experiments were repeated at various

temperatures between 237 and 274 K. An Arrhenius plot gives an activation energy

of 0.49 eV and a prefactor of 3 10

2

1

, in good agreement with the kinetic

parameters obtained from macroscopic measurements.

3.3 Visualizing the Pathway of Catalytic Reactions