
react with CO up to 180 K. More interestingly, the removal of oxygen islands
accelerated when the surface coverage of (2 1)-O islands decreased to below
0.3 ML (Figure 3.15a). Below 0.3 ML, unlike the previous study on the Pt(1 1 1)
surface, the titration reaction rate with the (2 1)-O islands on Pd(1 1 1) shows a
linear relation with the surface area of oxygen, instead of the perimeter of oxygen
islands, as shown in Figure 3.15b and c. The authors speculated that a transient
occupation of CO on the oxygen island causes all O atoms to be accessible for the
reaction. There was no direct evidence for the existence of this kind of mixed O/CO
phase based on the STM images or other spectroscopic studies. Nonetheless, this
study unambiguously illustrated the superior reactivity of compressed oxygen
islands, especially when they become very small.
The adsorption of oxygen atoms often induces the reconstruction of metal surfaces
as is the case of (1 1 0) surfaces of fcc metals. It is expected that CO titration on such
surfaces would also involve the local transformation of metal substrates. Indeed,
accompanying the oxygen-induced reconstruction, the mobility of surface oxygen is
considerably reduced so that they can be resolved by STM at room temperature. For
this reason, the CO titration experiments using STM were initiated on (1 1 0) surfaces
of fcc metals in the early 1990s. CO oxidation was first visualized on a Rh(1 1 0)
surface by Leibsle et al. [60] where a pronounced reaction anisotropy was observed.
The experiments were carried out by titrating the oxygen precovered Rh(1 1 0)
surface with CO. Chemisorption of oxygen on Rh(1 1 0) forms several reconstructed
phases, which, in turn, were imaged as striped patterns along the ½1
10direction.
By monitoring the surface changes during CO exposure, STM images revealed
that oxygen was removed on the elongated stripes of the added rows in the
½1
10direction. Later, similar one-dimensional reactivity was also found on other
fcc(1 1 0) systems, such as Cu [61–63], Ni [64], and Ag [65, 66].
Figure 3.13 Reaction rates, determined from
the change in the size of the (2 2) area
between successive panels of the data of
Figure 3.12, normalized to (squares) the
length of the boundary between oxygen and
CO domains (the full line is a linear fit) and
(crosses) divided by q
O
(1 q
O
), which is equal
to q
O
q
CO
if q ¼1 implies maximum coverage
of the respective phase (the broken line is only
to guide the eye). (Reprinted with permission
from Ref. [58]. Copyright 1997, The American
Association for the Advancement of Science.)
3.3 Visualizing the Pathway of Catalytic Reactions
j
75