Page 54 - Photoreactive Organic Thin Films

PHOTOiSOMERIZATIONOFAZOBENZENES 33

tion activation volumes in their experiments under pressure, a dependence
of the mechanism on solvent and even parallel reaction paths along these
mechanistic coordinates. A close proximity of the calculated energies of
the rotation and inversion transition states was pointed out by Cimiraglia
176
and Hofmann. They draw attention to the strong lowering of the energy
barriers of rotation on substitution.
Isomerization needs some extra sweep volume. The volumes for the two
3
mechanisms of azobenzene should be quite different—ca. 0.25 nm for rota-
3
177 178
tion and ca. 0.12 nm for inversion. ' This bears out in restricted spaces.
In some zeolites azobenzene can isomerize whereas stilbene does not.
Kuriyama and Oishi found that there are two separate AH* versus AS* lines
for azobenzenes isomerizing by inversion (azobenzene type) and rotation
(pseudo-stilbene type). 179

i .6.I.2 The Photoisomerization Mechanism
5
In their 1971 review, Ross and Blanc expressed doubts as to the opera-
tion of the inversion mechanism in the excited states. This opened another
round of heated discussion. The rotation/inversion controversy invoked much
theoretical and experimental work.
180
The use of Walsh diagrams, based on one-electron molecular orbitals,
shows that on n -» n* excitation the azobenzene molecule is stretched, which
is the beginning of inversion. All calculations and suggestions for an inversion
mechanism agree that the potential energy curve for inversion has a relatively
steep slope at the E- and the Z- geometries. This is corroborated by the
experimental evidence of a continuous n —> n* absorption band in both
isomers. In fact, a structured n —> it* band in an azo compound that can
isomerize has never been observed.
The crucial finding was that in the case of azobenzene, the isomerization
l
quantum yields for excitation of the higher (n,n*) state are about one-half
1
those on excitation to the lower (n,xc*) state (Table 1.1). Similar facts hold
for aminoazobenzene 1.1 x 2 type molecules with low-lying ^n,^*) state
(Table 1.3). On the other hand, for azobenzene-type molecules whose struc-
ture inhibits rotation, the quantum yields become equal. To rationalize their
52
results, Rau and Liiddecke suggested that two different isomerization path-
ways in the excited states in azobenzene-type molecules should be active:
1
rotation in the high energy (n,n*) state and inversion in the low energy
l 34
(ti,n*) state. This concept was extended to a potential energy diagram
1
(Figure 1.15B). At the geometry of E-azobenzene the (n,jc*) state has a
relatively steep slope in the direction of the inversion coordinate leading to
a minimum from which the deactivation to the ground state occurs in the
vicinity of a ground state maximum providing for 25% of isomerization. In
l
contrast, in the (n,n*) state, the slope of inversion is uphill, rotation is
favored and leads to a "bottleneck" state near the geometry of the maximum
of the twisted ground state, and the partition of the population toward the
1
Z-isomer is different from that of (n,n*) inversion. The crucial point in
I
this model is the vanishing internal conversion (TC,JC*) —> ^n,^*) near the
E-geometry. This channel, however, seems to be disfavored by the large
energy gap if the rotation channel is open.