An A
b
defect-free region (the clean-off area) was created by scanning across the
surface at þ3 V before exposing to water, as shown in Figure 8.4. According to the
assignment of Schaub et al. [15] exposure to water should have little effect in the clean-
off area as there would be no O
b
-vacs with which to react. However, exposing the
surface to water in fact replenishes the A
b
defects in the clean-off area and reduces the
number of A
d
defects (Figure 8.4c). If water dissociates in O
b
-vacs then the number of
O
b
-vacs is expected to decrease and the number of OH
b
species to increase, thus A
b
defects were reassigned as OH
b
and A
d
defects reassigned as O
b
-vacs. This also
means that approximately þ3 V scans and tip pulses remove hydrogen from OH
b
.
While the mechanism for this hydrogen desorption has not yet been elucidated, it
may be related to an empty state identified by Onda et al. [28] approximately 2.4 eV
above E
F
.
Returning to Figure 8.4 because the STM images were recorded from the same
area before and after exposure to water, it can be seen that some of the new OH
b
take
the positions of the reacting O
b
-vacs. Thus, Bikondoa et al. [16, 17] imaged water
dissociating in the O
b
-vacs forming one OH
b
in place of the O
b
-vacs and another OH
b
elsewhere, consistent with the mechanism depicted in Figure 8.1. STM measure-
ments at low temperature show that water dissociates at least down to approximately
187 K [19], a conclusion confirmed by high-resolution electron energy loss
spectroscopy [29].
Surprisingly, the number of OH
b
in positions previously taken by O
b
-vacs always
appears slightly higher than elsewhere. Accordingto the mechanism in Figure 8.1, an
equal number of OH
b
species should appear at the O
b
-vacs and at a distance from
them. Wendt et al. [18, 19] suggest that when the water molecules dissociate, the two
resulting OH
b
lie initially in pairs, adjacent in the [0 0 1] direction. As individual OH
b
from the pair are difficult to resolve with STM, each OH
b
pair therefore appears as a
single feature, which would account for the apparent anomaly. This interpretation is
supported by the observation of three apparent sizes of type-A defects in some STM
images [17– 20, 30]. The largest of these is assigned to OH
b
pairs, the next largest to
isolated OH
b
, and the smallest to O
b
-vacs, an assignment corroborated in recent STM
simulations of O
b
-vacs and OH
b
[30, 31].
Further evidence for the OH
b
pairs is given in an STM movie recorded at
approximately 187 K. The movie shows the OH
b
pairs separating across the O
b
rows
via proton exchange with water molecules [19]. Key frames from the movie are shown
in Figure 8.5. This water-assisted OH
b
diffusion mechanism is supported by
calculations, which show that the barrier to diffusion is lowered by the exchange
with water. The same diffusion mechanism is also observed for isolated OH
b
and
because of the misassignment of OH
b
and O
b
-vacs by Schaub et al. [15], this water-
mediated diffusion of OH
b
was incorrectly reported as oxygen-mediated diffusion of
O
b
-vacs [32, 33] via a mechanism inconsistent with subsequent isotope studies [34].
Figure 8.6 summarizes our current knowledge of the appearance of point defects
in STM images. The most prevalent point defects on sputtered/annealed TiO
2
(1 1 0)
1 1 surfaces have been identified as O
b
-vacs, OH
b
, and OH
b
pairs and these are
shown in a ball model together with an STM image decorated with a number of all
three types of defects.
224
j
8 Point Defects on Rutile TiO
2
(1 1 0): Reactivity, Dynamics, and Tunability