Predicting the Outcome of Photocyclisation Reactions: A Joint Experimental and Computational Investigation

Article abstract of DOI:10.1002/asia.201801761

Photochemical oxidative cyclodehydrogenation reactions are a versatile class of aromatic ring-forming reactions. They are tolerant to functional group substitution and heteroatom inclusion, so can be used to form a diverse range of extended polyaromatic systems by fusing existing ring substituents. However, despite their undoubted synthetic utility, there are no existing models—computational or heuristic—that predict the outcome of photocyclisation reactions across all possible classes of reactants. This can be traced back to the fact that “negative” results are rarely published in the synthetic literature and the lack of a general conceptual framework for understanding how photoexcitation affects reactivity. In this work, we address both of these issues. We present experimental data for a series of aromatically substituted pyrroles and indoles, and show that quantifying induced atomic forces upon photoexcitation provides a powerful predictive model for determining whether a given reactant will photoplanarise and hence proceed to photocyclised product under appropriate reaction conditions. The propensity of a molecule to photoplanarise is related to localised changes in charge distribution around the putative forming ring upon photoexcitation. This is promoted by asymmetry in molecular structures and/or charge distributions, inclusion of heteroatoms and ethylene bridging and well-separated or isolated photocyclisation sites.

Full text of DOI:10.1002/asia.201801761

CHEMISTRY
AN ASIAN JOURNAL
www.chemasianj.org
Accepted Article
Title: Predicting the outcome of photocyclisation reactions: a joint
experimental and computational investigation
Authors: Akassong Dominique Dinga Wonanke, Jayne L Gulbrandsen,
Chris M Fitchett, and Deborah Lee Crittenden
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Asian J. 10.1002/asia.201801761
Link to VoR: http://dx.doi.org/10.1002/asia.201801761
A Journal of
A sister journal of Angewandte Chemie
and Chemistry – A European Journal

10.1002/asia.201801761
Chemistry - An Asian Journal
Predicting the outcome of photocyclisation reactions: a joint
experimental and computational investigation
A. D. Dinga Wonanke, Jayne L. Ferguson,
Christopher M. Fitchett, Deborah L. Crittenden∗
School of Physical and Chemical Sciences, University of Canterbury,
Christchurch, New Zealand
∗
Corresponding author
Abstract
Photochemical oxidative cyclodehydrogenation reactions are a versatile class
of aromatic ring-forming reactions. They are tolerant to functional group
substitution and heteroatom inclusion, so can be used to form a diverse
range of extended polyaromatic systems by fusing existing ring substituents.
However, despite their undoubted synthetic utility, there are no existing
models – computational or heuristic – that predict the outcome of photocy-
clisation reactions across all possible classes of reactants. This can be traced
back to the fact that “negative” results are rarely published in the synthetic
literature and the lack of a general conceptual framework for understand-
ing how photoexcitation affects reactivity. In this work, we address both of
these issues. We present new experimental data for a series of aromatically
substituted pyrroles and indoles, and show that quantifying induced atomic
forces upon photoexcitation provides a powerful predictive model for deter-
mining whether a given reactant will photoplanarise and hence proceed to
photocyclised product under appropriate reaction conditions. The propen-
sity of a molecule to photoplanarise is related to localised changes in charge
distribution around the putative forming ring upon photoexcitation. This
is promoted by asymmetry in molecular structures and/or charge distribu-
tions, inclusion of heteroatoms and ethylene bridging and well-separated or
isolated photocyclisation sites.
1
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
1 Introduction
Since photochemical reactions were discovered over 100 years ago, photochemical
synthesis has become an important branch of organic chemistry. [1,2] Photochemi-
cal syntheses involve the absorption of light by a molecule, causing it to transition
to an electronically excited state. This results in redistribution of electrons, chang-
ing the chemical properties of the molecule and broadening its reactivity spectrum
by opening reaction pathways that may not be accessible by conventional thermal
procedures. [3] This often leads to novel photo-products. [4]
From the plethora of chemical reactions facilitated by electronic excitation, elimi-
native photochemical cyclisation has become one of the most important and widely
used, because it provides an efficient route to the synthesis of polycyclic aromatic
and heteroaromatic macromolecules. These molecules function as organic semi-
conductors, making them well suited for use in technological applications such as
in organic light emitting diodes, [5–11] organic solar cells, [12–16] organic field
effect transistors, [17–21] liquid crystals [22–25] and sensors. [26–28]
Eliminative photocyclisation is an intramolecular light-activated aromatic ring
forming reaction. Experimental evidence indicates that the reaction proceeds in
two steps; planarisation followed by elimination. [29] The photo-planarisation step
involves a 6-π electrocyclisation of the reactant molecule into a cyclised intermedi-
ate, while the elimination step leads to ring aromatisation through the elimination
of two atoms or functional groups adjacent to the newly formed bond. A simple
illustration of this two-step mechanism is provided in Scheme 1.
2
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
R2
R2
X
R2
Planarisation
R3 R1
Elimination
R1
X
R1
R3
R3
H
H
Scheme 1 General reaction scheme for eliminative cyclisation, where R1, R2,
R3 denote different ring systems. X represents a ring substituent which could be
hydrogen or any suitable leaving group.
Reactions in which X is a hydrogen atom require the presence of an oxidising agent
to drive the elimination step. Such reactions are referred to as oxidative eliminative
photocyclisations. On the other hand, reactions in which X is a suitable leaving
group (alkoxy groups or halogens) are referred to as non-oxidative eliminative cy-
clisation since the elimination step can proceed spontaneously without an oxidising
agent. [30] In this paper, our focus will be on oxidative eliminative cyclisation re-
actions because they do not require substituted starting materials to be prepared
and so are synthetically more convenient. Further, the use of oxidising agent to
promote elimination allows an additional degree of synthetic control and further
potential for optimising reaction conditions to promote product formation. For
brevity, photo-induced oxidative 6π electrocyclisation reactions will henceforth be
referred to simply as photocyclisations.
Although photocyclisation reactions are widely used synthetically, [31–35] there
is still a fundamental lack of understanding as to why certain molecules fail to
yield products. Most experimental studies focus on optimising reaction conditions
- solvent, oxidant, temperature - in an attempt to identify experimental condi-
3
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
tions that will lead to product formation. However, recent computational studies
have shown that the elimination step is thermodynamically favourable and can
be kinetically enhanced by optimising reaction conditions, regardless of whether
elimination proceeds in the excited state or following internal conversion back to
the ground state. [30] Further, if internal conversion is coupled to elimination, the
whole process will proceed spontaneously and exothermically following photoexci-
tation and photoplanarisation. [30]
In either case, photoplanarisation is a necessary pre-condition for elimination, so
it is important to be able to predict and/or control the process. Controlling the
outcome of the photoplanarisation step is far harder than for elimination, because
both the thermodynamic and kinetic favourability are determined primarily by
changes in the molecule’s electronic structure upon photo-excitation. In other
words, photoplanarisability is an inherent property of a molecule that cannot be
easily controlled by modifying reaction conditions.
In this joint experimental and computational study, our aim is to understand
the factors that determine a molecule’s ability to photoplanarise, and therefore
photocyclise under appropriate reaction conditions. Experimentally, we will test
a diverse set of polycyclic aromatic and heteroaromatic hydrocarbons for their
ability to undergo photocyclisation under suitable oxidising conditions. Compu-
tationally, we will investigate how changes in molecular electronic structure upon
photo-excitation affect a molecule’s ability to photoplanarise and therefore photo-
cyclise if reaction conditions are appropriately optimized to promote elimination.
Changes in electronic structure will be characterised by analysing the forces that
the changing electronic distribution induces at the atomic centres. Finally, we will
assess computational predictions against experimental observations.
4
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
2 Theory
2.1 Theoretical background
Photoplanarisation is the first step in the photocyclisation process. Electronically,
it can be viewed as a reorganisation of the π-electrons within the putative forming
ring, with concomitant formation of a new σ bond, creating a proto-aromatic sys-
tem. Upon subsequent elimination of leaving groups adjacent to the newly formed
ring, full aromaticity is realized. Geometrically, it can be viewed as a sterically-
strained bond rotation, coupled to bond length changes indicative of aromatic ring
formation.
For photoplanarisation to proceed, photo-induced changes in molecular electronic
and geometric structure must be consistent with progress along this 6-π electro-
cyclisation reaction pathway. Previous studies have focussed on directly captur-
ing changes in electron distribution upon photoexcitation, monitoring changes in
quantities computed via Hückel molecular orbital theory such as free-valence in-
dices, [36, 37] localization energies, [36–38] electronic overlap populations, [39, 40]
and bond orders. [41]
Although these reactivity predictors have been employed to good effect to pre-
dict regioselectivity and overall reactivity for polyaromatic hydrocarbons, applying
them to heteroatom-containing systems becomes problematic. Hückel molecular
orbital theory can be generalised to describe heteroaromatic systems, but exten-
sive parameterisation is required. [42] Separate parameters are required for every
different atom and bond type, and atom type parameters can also vary according
to the local bonding environment. [42] Further, there is no single unique set of
optimal parameters. [42] As a consequence, no software exists to perform calcula-
5
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
tions of this type.
This suggests that a more sensible strategy would be to proceed directly to us-
ing modern electronic structure methods in combination with atomic population
analysis schemes, to develop “modern-day” replacements. Unfortunately, atomic
populations are not quantum observables, so all population analysis schemes are
arbitrary. [43] This makes it impossible to construct uniquely defined reactivity
predictors. Further, any reactivity predictor will inherit the limitations and weak-
nesses of the underlying atomic population analysis scheme and, in some cases,
may not even be physically meaningful. [43]
Here, we contend that geometrically and energetically based indicators will provide
a more robust, more generally applicable and less method-dependent way of as-
sessing whether photo-induced changes in electronic structure are consistent with
promoting planarisation, or not. Our approach relies on the fact that instanta-
neous changes in electron distributions upon photo-excitation at a given molecular
conformation will induce forces on the atoms within the molecule, driving them
to reorganise from their ground state conformation towards an excited state equi-
librium structure. If these forces are consistent with the atoms moving to adopt
a more planar and/or more aromatic ring-like system, then it is likely that photo-
planarisation will proceed.
This suggests two possible metrics; the projection of the induced atomic forces
along the torsional coordinate about which bond rotation must occur for the puta-
tive forming ring to planarise, or the projection of the induced atomic forces along
bond vectors within the putative forming ring. The former is the most intuitive,
as it directly answers the question - will the induced forces act to decrease the
6
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
torsion angle and so planarise the molecule? However, this may be complicated
by equal and/or opposing forces arising from steric strain. Analysing whether the
bonds in the proposed forming ring become more aromatic is likely to be more
computationally robust, because bond stretching vibrations tend to occur at high
frequencies and low amplitudes, i.e. are quite rigid motions due to high energy
penalties for displacement from equilibrium. Therefore, instantaneous forces along
bond vectors are likely to provide an accurate representation of how the structure
will immediately adapt, uncomplicated by competing forces that may arise during
longer time scale, large-amplitude motions.
Consider any two atoms, i and j, within a putative forming ring whose positions
are represented by the position vectors ri and rj respectively, with instantaneous
photo-induced atomic forces of fi and fj computed at the molecule’s ground state
equilibrium geometry. The bond vector between these two atoms is ri,j = ri − rj
and the corresponding force vector is fi,j = fi − fj.
Then the component of the force that acts directly along the bond is simply:
fi,j · ri,j
proj
i,j
F
=
(1)
||ri,j||
Finally, it remains to consider how the photo-induced forces affect all bonds within
the photocyclising ring system, excluding the forming bond. To be consistent with
aromatization, atom-pairs that are σ-bonded in the ground state would acquire
negative projected forces in the excited state, indicating a driving force towards
these bonds becoming shorter. Conversely, π-bonded atom-pairs would acquire
positive projected forces, indicative of bond lengthening.
If these conditions are fulfilled, then the total projected excited state “bond arom-
7
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
atization force” is computed as:
Fπ∗,σ
n−1
n
X X
proj
||F ||
=
(2)
i,j
i=1 j=i+1
where the sums run over all unique atom-pairs in the putative forming ring that
are bonded in the ground state.
2.2 Computational details
Ground state geometries are optimised for isolated, gas phase species at B3LYP/6-
31G*. [44–48] The B3LYP functional is chosen because it is well-parameterised for
organic systems, and is one of the few functionals for which analytic gradients
and Hessians are available in both ground and excited states. The 6-31G* basis
set is used because it is the basis in which the B3LYP functional was parameter-
ized and so affords near optimal error cancellation between BSSE and dispersion
contributions to the energy that are not explicitly accounted for. [49] All excited
state gradients are computed for the first singlet excited state at TD-B3LYP/6-
31G*, unless otherwise specified. All calculations are performed using the Q-
Chem5.0 software package. Our force projection code can be freely downloaded
from http://github.com/dlc62/force-projection.
3 Experimental Results and Discussion
To properly benchmark computational predictions against experiment, a compre-
hensive data set of photocyclisation reaction conditions and outcomes is required.
Photocyclisation reaction conditions and outcomes have been compiled from the
literature in Table 1 for a series of compounds based upon (1) stilbenyl, [29,31,50]
(2) phenanthryl, [50,51] (3) acrylamide, [52] (4) furyl, [50] (5) thiophenyl [50], (6)
pyrrolyl [53,54] and (9) benzyl [50,53] scaffolds illustrated in Fig. 1.
8
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
To extend the breadth and scope of our reference data set, we have also experimen-
tally investigated photocyclisation of a novel set of 2,3-diaryl substituted indoles,
7. First, a series of substituted N-ethyl or N-benzyl indoles were synthesised by
alkylation of the appropriate N-H indole to give 7a – 7h (see ESI for details).
Scaffolds of these molecules are illustrated in Fig. 1 and their substituents listed
in Table 1.
9
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
R1
R2
X
R3
R4
R1
A
B
R3
C
N
O
R2
A
H
1. Stilbenyl
R5
R4
3. Acrylamide
2. Phenanthryl
X
N
O
S
R4
R4
R1
R1
R4
R1
R3
R3
R2
R2
R3
R2
4. Furyl
5. Thiophenyl
6. Pyrrolyl
X
N
Y
R1
X
N
R
1 R6
R2
R3
R1
R2
R5
R2
R4
Y
7. Indolyl
9. Benzyl
8. Dibenzo[e,g]indolyl
Fig. 1 Scaffold of molecules present in our dataset where A, B, and C represent
ring constituents, R represent ring substituents that are potentially involved in
photocyclisation, X and Y represent electronically active substituents, and the
identity of these groups are defined in Table 1. For each molecule, photocyclisation
can occur between ring substituents, between a ring substituent and the scaffold
or between the scaffold atoms (. . . ).
10
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Photocyclisation reactions for 7a – 7h were carried out in a Rayonet photore-
actor irradiating with 300nm light over 12 hours. All reactions were performed
in a quartz tube with toluene as the solvent, 1.1 molar equivalents of I2 and 5
mL of propylene oxide. [33] A constant stream of argon was bubbled through
the reaction solution for the duration of the reaction to remove oxygen. The
dibenzo[a,c]carbazole products, 10a – 10h (Fig. 2), were isolated in moderate to
good yields after column chromatography (see Table 1), and have been fully char-
acterised using NMR spectroscopy and electrospray mass spectrometry (ES-MS).
X
(a) X = Et, Y = H
Y
(b) X = Et, Y = OMe
N
(c) X = Et, Y = Me
(d) X = Et, Y = tBu
(e) X = Bz, Y = H
(f) X = Bz, Y = OMe
(g) X = Bz, Y = Me
(h) X = Bz, Y = tBu
Y
Fig. 2 Dibenzo[a,c]carbazoles (10) formed by oxidative photocyclisation of 2,3-
diaryl substituted indoles (7)
Cyclisation was confirmed by examination of 1H-NMR spectra, which showed con-
siderable desymmetrisation due to aryl ring rotation (see ESI for spectra). This was
also associated with a general shift of the resonances to more downfield positions,
with the most downfield protons in 10 ranging from 8.21 – 8.94ppm (compared
to 7.76 – 7.85 ppm for 7). The formation of two singlets for the protons adjacent
to the new bond for 10b–d and 10f–h is indicative of the loss of these protons
during the cyclodehydrogenation reaction. The protons of the N-alkylating group
were also affected, with the protons of the attaching methylene shifted to ≈ 4.9
11
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
ppm (for 10a – 10d) and ≈ 6.0 ppm (for 10e – 10h) from ≈ 4.1 ppm and ≈ 5.3
ppm, respectively. These changes are indicative of the increased aromatic area of
10, and the associated increase in the ring anisotropy. [54]
We were able to grow single crystals of 10c, 10f and 10g, and these were charac-
terised using single crystal X-ray diffraction, along with the structures of 7a, 7d,
and 7h (see ESI). The molecules all crystallised with one molecule in the asym-
metric unit, as exemplified by the structure of 10f shown in Fig. 3 (see ESI for
structures of 10c and 10g).
Fig. 3 Crystallographic molecular structure of 10f, showing the newly formed
ring system. Thermal ellipsoids are shown at the 50% level. Selected bond lengths
(Å): C11-C14 1.417(2), C3-C4 1.441(2), C3-C7 1.392(2), C4-C8 1.420(2), C7-C11
1.438(2), C8-C14 1.467(2).
12
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
The structures show the cyclised products, with newly formed C-C bond lengths
of 1.465(2) Å(10c), 1.467(2) Å(10f) and 1.462(2) Å(10g). The enlarged aromatic
dibenzo[a,c]carbazole ring systems are remarkably planar, with RMS deviations of
0.045 Å(10c), 0.066 Å(10f) and 0.092 Å(10g). This planarity forces the N-ethyl
and N-benzyl groups to be almost perpendicular to the plane, with torsion angles
of 97.7(2)◦(10c), 104.5(2)◦(10f) and 90.8(2)◦(10g). Due to the large area of the
dibenzo[a,c]carbazole ring, the structures all exhibit face-to-face interactions in
the solid state, with the closest approach being 3.475(1) Å(10c), 3.379(1) Å(10f)
and 3.551(1) Å(10g).
We have also investigated the propensity of the structurally-related dibenzo[e,g]indoles,
8, to undergo cyclization under similar reaction conditions. However, no evidence
of cyclised product was observed via either NMR or ES-MS. However, we have
previously shown that dibenzo[e,g]indoles can themselves be formed by photocy-
clisation from appropriately substituted pyrroles, 6. [54]
In total, we report reaction conditions and outcomes for 40 different potentially
photocyclisable systems in Table 1, representing a diverse range of molecular scaf-
folds and substitution patterns.
13
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Table 1 Table of photocyclisation yields and experimental conditions of data compiled from this studies and
published literatures. The molecular scaffolds and substituent patterns for these molecules are illustrated in Fig. 1
and the identities of functional groups are defined below.
Molecules
Solvent
1.Stilbenyl
Oxidant Forming ring Yields %
Ref
a R1 = R2 = 2-furyl, R3 = R4 = H
b R1 = R2 = 2-thiophenyl, R3 = R4 = H
c R1 = R2 = phenyl, R3 = R4 = H
d R1 = R2 = R3 = R4 = phenyl
ethanol
CuCl2/I2
I2/O2
I2/O2
I2/O2
I2
R1–R2
R1–R2
R1–R2
R1–R2
R1–R2
R1–R2
24
90
82
88
14
68
[50]
[31]
[29]
[50]
[51]
[51]
cyclohexane
cyclohexane
cyclohexane
toluene
e R1 = R2 = R3 = R4 = 2-thiophenyl
f
R1 = R2 = 2-thiophenyl, R3 = R4 = phenyl
toluene
I2
2. Phenanthryl, A = CH
cyclohexane
toluene
a R1 = R2 = phenyl, R3 = R4 = R5 = H
I2
–
–
[50]
[51]
b R1 = R2 = 2-thiophenyl, R3 = R4 = R5 = H
I2/O2
R1–R2
69
Continues on next page
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Table 1 – Continued from previous page
Molecules
Solvent
Oxidant Forming ring Yields %
Ref
c R1 = R2 = R5 = H, R3 = R4 = phenyl
d R1 = R2 = R3 = H, R4 = R5 = phenyl
e R1 = R2 = R3 = R4 = H, R5 = phenyl
benzene
benzene
I2
I2
I2
I2
–
–
–
[50]
[50]
[50]
[50]
–
benzene
A–R5
A–R5
46
35
f
R1 = R2 = R3 = R4 = H, R5 = 2-naphthyl
benzene
3. Acrylamides
benzene/acetic acid
benzene/acetic acid
benzene/acetic acid
benzene/acetic acid
benzene/acetic acid
4. Furyl
a A = B = CH, C = N, X = H
b A = B = CH, C = N, X = Me
c A = C = CH, B = N, X = Me
d A = CH, B = N, C = CCl, X = Me
e A = N, B = C = CH, X = Me
SeO2
SeO2
SeO2
SeO2
SeO2
. . .
. . .
. . .
. . .
. . .
17
78
22
25
72
[52]
[52]
[52]
[52]
[52]
a R1 = R2 = phenyl, R3 = R4 = H
b R1 = R2 = R3 = R4 = phenyl
benzene
O2
O2
R1–R2
–
40
p
[50]
[50]
benzene
Continues on next page
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Table 1 – Continued from previous page
Molecules
Solvent
5. Thiophenyl
benzene
Oxidant Forming ring Yields %
Ref
a R1 = R2 = phenyl, R3 = R4 = H
b R1 = R2 = R3 = R4 = phenyl
O2
O2
R1–R2
–
70
p
[50]
[50]
benzene
6. Pyrrolyl
benzene
a R1 = R2 = R3 = R4 = phenyl, X = H
b R1 = R2 = R3 = R4 = phenyl, X = Et
O2
–
–
[53]
[54]
toluene
I2/PrO
R1–R2
73
7. Indolyl
toluene
a R1 = R2 = phenyl, X = Et
I2/PrO
I2/PrO
I2/PrO
I2/PrO
I2/PrO
I2/PrO
R1–R2
R1–R2
R1–R2
R1–R2
R1–R2
R1–R2
64
73
60
89
84
62
This work
This work
This work
This work
This work
This work
b R1 = R2 = 4-methoxyphenyl, X = Et
c R1 = R2 = 4-methylphenyl, X = Et
d R1 = R2 = 4-tert-butylphenyl, X = Et
e R1 = R2 = phenyl, X = Bz
toluene
toluene
toluene
toluene
f
R1 = R2 = 4-methoxyphenyl, X = Bz
toluene
Continues on next page
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Table 1 – Continued from previous page
Molecules
Solvent
Oxidant Forming ring Yields %
Ref
g R1 = R2 = 4-methylphenyl, X = Bz
toluene
toluene
I2/PrO
I2/PrO
R1–R2
R1–R2
56
78
This work
This work
h R1 = R2 = 4-tert-butylphenyl, X = Bz
8. Dibenzo[e,g]indole
toluene
a R1 = R2 = phenyl, X = Et, Y = H
I2/PrO
I2/PrO
I2/PrO
–
–
–
–
–
–
This work
This work
This work
b R1 = R2 = 4-methylphenyl, X = Et, Y = Me
c R1 = R2 = 4-tert-butylphenyl, X = Et, Y = t-Bu
toluene
toluene
9. Benzyl
benzene
a R1 = R2 = phenyl, R3 = R4 = R5 = R6 = H
b R1 = R2 = R3 = phenyl, R4 = R5 = R6 = H
c R1 = R4 = H, R2 = R3 = R5 = R6 = phenyl
d R1 = R2 = R3 = R4 = phenyl, R5 = R6 = H
e R1 = R2 = R3 = R4= R5 = phenyl, R6 = H
I2/N2
I2/N2
I2
R1–R2
88
21
p
[50]
[53]
[53]
[53]
[53]
[53]
benzene
benzene
benzene
benzene
benzene
R1-R2-R3
R2–R3
I2
–
–
–
–
I2
–
f
R1 = R2 = R3 = R4 = R5 = R6 = phenyl
I2
–
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Abbreviations
p : Product identified but not purified to determine yield
PrO = propylene oxide
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
4 Computational Results and Discussion
Total photo-induced bond aromatisation forces around all possible rings that may
form within reactants are reported in Table 2. The complete set of molecular
structures and first singlet excited state transition energies are available in the
Electronic Supporting Information. Most transition energies are predicted to fall
in the near-UV range (300 – 400 nm, 3.1 – 4.1 eV), although some fall in the
experimentally accessible 250 – 300 nm mid-UV range (4.1 – 4.9 eV), primarily
molecules in the benzyl and acrylamide classes.
Table 2 Total photo-induced bond aromatisation forces, F∗π,σ, along all pos-
sible photocyclisation pathways for all molecules in our data set. For clarity,
experimentally-determined yields are reproduced from Table 1.
Molecules
Forming ring
F∗π,σ (Eh/a0) Yield %
1. Stilbenyl
a R1 = R2 = 2-furyl, R3 = R4 = H
R1–R2
R1–R2
0.49
0.49
0.47
0.32
0.30
0.34
0.25
24
90
82
88
14
68
–
b R1 = R2 = 2-thiophenyl, R3 = R4 = H
c R1 = R2 = phenyl, R3 = R4 = H
d R1 = R2 = R3 = R4 = phenyl
R1–R2
R1–R2 = R3–R4
R1–R2 = R3–R4
R1–R2
e R1 = R2 = R3 = R4 = 2-thiophenyl
f
R1 = R2 = 2-thiophenyl,
R3 = R4 = phenyl
R3–R4
2. Phenanthryl, A = CH
a R1 = R2 = phenyl, R3 = R4 = R5 = H
b R1 = R2 = 2-thiophenyl, R3 = R4 = R5 = H
c R1 = R2 = R5 = H, R3 = R4 = phenyl
d R1 = R2 = R3 = H, R4 = R5 = phenyl
R1–R2
R1–R2
R3–R5
R3–R4
0.12
0.30
0.09
0.09
–
69
–
–
Continues on next page
19
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Table 2 – Continued from previous page
Molecules
Forming ring
F∗π,σ (Eh/a0) Yield %
e R1 = R2 = R3 = R4 = H, R5 = phenyl
R1 = R2 = R3 = R4 = H, R5 = 2-naphthyl
A–R5
A–R5
0.11
0.11
46
35
f
Continues on next page
20
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Table 2 – Continued from previous page
Forming ring
Molecules
F∗π,σ (Eh/a0) Yield %
3. Acrylamide
a A = B = CH, C = N, X = H
b A = B = CH, C = N, X = Me
c A = C = CH, B = N, X = Me
d A = CH, B = N, C = CCl, X = Me
e A = N, B = C = CH, X = Me
. . .
. . .
. . .
. . .
. . .
0.29
0.29
0.28
0.27
0.26
17
78
53
25
72
4. Furyl
5. Thiophenyl
6. Pyrrolyl
a R1 = R2 = phenyl, R3 = R4 = H
b R1 = R2 = R3 = R4 = phenyl
R1–R2
R1–R2
R2–R3
R3–R4
0.40
0.21
0.11
0.21
40
p
a R1 = R2 = phenyl, R3 = R4 = H
b R1 = R2 = R3 = R4 = phenyl
R1–R2
R1–R2
R2–R3
R3–R4
0.39
0.21
0.11
0.21
70
p
R1–R2
R2–R3
R3–R4
R1–R2
R2–R3
R3–R4
0.17
0.09
0.17
0.21
0.09
0.15
a R1 = R2 = R3 = R4 = phenyl, X = H
b R1 = R2 = R3 = R4 = phenyl, X = Et
p
73
Continues on next page
21
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Table 2 – Continued from previous page
Forming ring
Molecules
F∗π,σ (Eh/a0) Yield %
7. Indolyl
a R1 = R2 = phenyl, X = Et
R1–R2
R1–R2
R1–R2
R1–R2
R1–R2
R1–R2
R1–R2
R1–R2
0.30
0.35
0.31
0.33
0.30
0.33
0.32
0.33
64
60
87
89
84
62
56
78
b R1 = R2 = 4-methoxyphenyl, X = Et
c R1 = R2 = 4-methylphenyl, X = Et
d R1 = R2 = 4-tert-butylphenyl, X = Et
e R1 = R2 = phenyl, X = Bz
f
R1 = R2 = 4-methoxyphenyl, X = Bz
g R1 = R2 = 4-methylphenyl, X = Bz
h R1 = R2 = 4-tert-butylphenyl, X = Bz
8. Dibenzo[e,g]indole
a R1 = R2 = phenyl, X = Et, Y = H
R1–R2
R1–R2
R1–R2
0.04
0.04
0.04
–
–
–
b R1 = R2 = 4-methylphenyl, X = Et, Y = H
c R1 = R2 = 4-tert-butylphenyl, X = Et, Y = H
Continues on next page
22
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Table 2 – Continued from previous page
Forming ring
Molecules
F∗π,σ (Eh/a0) Yield %
9. Benzyl
a R1 = R2 = phenyl, R3= R4 = R5 = R6 = H
b R1 = R2 = R3 = phenyl, R4 = R5 = R6 = H
c R2 = R3 = R5 = R6 = phenyl, R1 = R4 = H
R1–R2
R1–R2 = R2–R3
R2–R3 = R5–R6
R1–R2 = R3–R4
R2–R3
0.36
0.20
88
21
p
0.25
R1 = R2 = R3 = R4 = phenyl
0.175
0.162
0.149
0.144
d
–
–
R5 = R6 = H
R1 = R2 = R3 = R4 = R5 = phenyl
R1–R2 = R4–R5
R2–R3 = R3–R4
R1–R2
e
R6 = H
= R2–R3
= R3–R4
f
R1 = R2 = R3 = R4 = R5 = R6 = phenyl
0.125
–
= R4–R5
= R5–R6
= R6–R1
4.1 Correlating predicted photoplanarisation with observed
photocyclisation
There are 6 classes of molecules within our data set for which photocyclised prod-
ucts are consistently formed, regardless of substitution pattern: stilbenyl, acry-
lamide, furyl, thiophenyl, pyrrolyl and indolyl. In all cases, the total bond aroma-
tisation force is larger than 0.15 Eh/a0 around the photocyclising ring system that
leads to the observed isomeric product. In the cases where separation and charac-
terisation of products was not attempted, at least one had a corresponding value
23
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
of F∗π,σ above 0.15. Fig. 4 illustrates photo-induced atomic forces for a prototypical
photocyclisable system, cis-stilbene (1c). They clearly act to elongate π bonds
and shorten σ bonds, i.e. make the photocyclising ring system more aromatic.
Fig. 4 Molecular structure of cis-stilbene (left) and photo-induced atomic forces
(right)
Despite trialling a range of experimental conditions for a series of molecules with
different substitution patterns, no photocyclised product could be obtained from
dibenzo[e,g]indoles. Computationally, these molecules are found to have low bond
aromatisation forces (0.04 Eh/a0). Photo-induced atomic forces for a prototypical
dibenzo[e,g]indole system (8a) are illustrated in Fig. 5. This figure shows that the
electronic transition occurs solely on the dibenzo[e,g]indole scaffold, and so does
not promote aromatization or drive photoplanarisation. If the photoplanarised
intermediate cannot be formed, then the elimination step is unlikely to proceed.
This is consistent with experimental observations that no product can be formed.
24
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Et
N
Fig. 5 Molecular structure of N-ethyl-2,3-diaryl-dibenzo[e,g]indole (left) and
photo-induced atomic forces (right)
For substituted benzyl and phenanthryl systems, experimental outcomes are de-
pendent on substitution pattern. Heavily substituted and sterically crowded hexaphenyl-
benzene (9d) does not photocyclise while 1,2-diphenylbenzene (9a) and 1,2,3-
triphenylbenzene (9b) do. This too can be explained by inspection of the photo-
induced atomic forces for each molecule, as illustrated in Fig. 6.
25
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
Fig. 6 Photo-induced atomic forces within 1,2-diphenylbenzene (left) and
hexaphenylbenzene (right)
In contradistinction to the dibenzo[e,g]indole case, bond aromatisation forces in
hexaphenylbenzene are small not because the electronic transition is localized to
an isolated and rigid molecular subfragment but rather because it is delocalized
across the whole system.
Overall, there is a clear qualitative difference between molecules with strong bond
aromatisation forces (F∗π,σ > 0.15 Eh/a0) that generally proceed to form planarised
intermediates and photocyclised products, and those with weaker bond aromati-
sation forces (F∗π,σ < 0.15 Eh/a0) that do not.
Therefore, we propose the following categorisation scheme for predicting whether
a polycyclic molecule will photoplanarise or not:
26
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
• F∗π,σ > 0.15 - molecule will photoplanarise
• F∗π,σ ≈ 0.15 - molecule may photoplanarise
• F∗π,σ < 0.15 molecule will not photoplanarise
Applying this categorization scheme to our complete data set, we obtain the re-
sults shown in Table 3.
Table 3 Cross-correlation matrix of predicted planarisation against observed
cyclization for all molecules in our data set
Observed cyclization
yes
no
0
Predicted
yes 29
planarisation no
2
9
Our model has high predictive power, with a strong correlation between predicted
ability to photocyclise and observed product formation in experiments. Once
again, this supports the hypothesis that the elimination step can be controlled
by changing reaction conditions, but the overall outcome will be critically depen-
dent on a molecule’s intrinsic propensity to become aromatic in the excited state
and hence undergo photoplanarisation.
However, there are two cases in which we predict that a molecule will not photopla-
narise, yet it nonetheless is experimentally observed to photocyclise. Clearly our
predictions are wrong, which warrants further more detailed investigation. Chemi-
cal structures and photo-induced atomic forces for 1-phenylphenanthrene (2e) and
1-(2-naphthyl)phenanthrene (2f) are illustrated in Fig. 7 and Fig. 8, respectively.
27
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
From these figures, it is firstly clear that these structures are closely related, com-
prising two rigid aromatic units connected by a single rotatable bond. In both
cases, the cyclization reaction occurs between the substituent and the scaffold.
Close inspection of Fig. 8 reveals that the photo-induced forces are predicted to
largely localise on the phenanthrene unit, and do not correspond to aromatisation
of the putative forming ring.
Fig. 7 Molecular structure of 1-phenylphenanthrene (left) and 1-(2-
naphthyl)phenanthrene (right)
Fig. 8 Photo-induced atomic forces within 1-phenylphenanthrene (left) and 1-
(2-naphthyl)phenanthrene (right)
28
This article is protected by copyright. All rights reserved.

10.1002/asia.201801761
Chemistry - An Asian Journal
One possible explanation is that rotation about the single bond connecting the
two aromatic units breaks π conjugation and prevents aromatisation. However,
the equilibrium torsion angles in these systems do not differ substantially from
other molecules that are predicted to photoplanarise, which makes this explana-
tion unlikely. Further, accounting for thermally accessible bond rotations would
substantially increase the computational cost and complexity of our model, and
prevent it from being routinely applied.
Alternatively, it is plausible that the electronic structure model we have chosen,
TD-B3LYP/6-31G*, does not adequately model the excited state electronic struc-
ture of these particular systems. To test this hypothesis, we repeated our analysis
using the Coulomb-attentuated version of B3LYP (CAM-B3LYP) functional de-
signed to more accurately describe long-range interactions in electronically excited
states. [55] For both systems, we find that the excited state becomes more localized
to the phenanthrene ring, and F∗π,σ drops to from 0.11 to 0.09 Eh/a0.
Overall, it is clear that photo-induced atomic forces computed at TD-B3LYP (with
or without Coulomb attenuation) computed at the ground-state optimised geom-
etry are not consistent with aromatisation of putative forming rings within these
systems. This is a fundamental limitation of the approximations we have made
in our model overall - in our choice of electronic structure model, and relatively
simple computation and analysis procedure.
Nonetheless, we contend that our model still provides useful qualitative insights
for very modest computational cost across a diverse set of molecules, even if it
should be interpreted with caution for molecules of this type.
29
This article is protected by copyright. All rights reserved.