10.1002/chem.201702209
Chemistry - A European Journal
FULL PAPER
dye absorption onsets ~750 nm) through the use of indolizine
donors and retained high molar absorptivities, we next evaluated
the emissive properties of the dyes in solution. The fluorescence
spectra of the compounds were recorded in toluene and DCM. A
large Stokes shift, the difference in energy between the
absorption maximum and fluorescence maximum, is desirable
for applications requiring significant separation between input
and output energy, such as live fluorescence imaging. For this
application specifically, molecules should also emit within the
therapeutic window (~700-1400 nm). The emission maxima for
these dyes were found to range from 756-777 nm with the
exception of MesIn2SQ, which emits at noticeably higher energy
(735 nm). This gives Stokes shifts of around 30-50 nm or 0.063-
0.119 eV for In2SQ series (except MesIn2SQ 11 nm, 0.026 eV),
which is a significantly larger separation of input and output
energies than observed for the benchmark 1 (12 nm, 0.034 eV)
(Table 1). The absorption and emission of the most water
soluble derivative, MeIn2SQ, was also investigated by dissolving
the dye in minimal DMSO then diluting with water (Figure S9). A
blue shift of the absorption spectrum is noted; however, no
evidence of aggregation was observed based on absorption or
emission curve shapes since there are no noticeable changes
comparing water with organic solvents.
The excited-state lifetimes for the dyes in this series
ranged from 0.30 ns to 0.79 ns with the longest excited-state
lifetime from bis-CF3PhIn2SQ (0.79 ns) and the shortest excited-
state lifetimes from MesIn2SQ (0.30 ns) and MeOPhIn2SQ (0.33
ns). It is noteworthy that the resonance withdrawing substituted
dyes CNPhIn2SQ and NO2PhIn2SQ have relatively long-lived
excited-states (both at 0.65 ns) and the resonance donating
substituted MeOPhIn2SQ has a relatively short-lived excited-
state. While all of the dyes show shorter excited-state lifetimes
than the benchmark squaraine 1 (2.40 ns), this is expected as
dyes emitting at lower energies typically have reduced excited-
state lifetimes due to the Energy Gap Law.
The fluorescence quantum yield is one of the most
important metrics for evaluating fluorescent materials, especially
in the NIR range where Φ is typically very low, often <1%.7 For
a number of applications, Φ dictates the amount of dye needed
for good image resolution and device performance. Interestingly,
PyIn2SQ was found to have a fluorescence quantum yield of
>12%. This is a dramatic enhancement over what is commonly
observed for materials emitting photons at energies lower than
800 nm, with notable exceptions8. The next highest Φ values
were observed for bis-CF3PhIn2SQ (12%) and NO2PhIn2SQ
(11%), with the remaining dyes emitting between 3.7%-8.9%.
Even the lowest observed Φ for this series is a significant
enhancement over that of many materials currently being
evaluated for biological imaging purposes. Although
substantially larger Stokes-shifts are observed compared to
benchmark, 1, with less absorption and emission overlap, we do
note that considerable overlap of absorption and emission
curves is observed for the indolizine squaraine dyes, which can
lead to re-absorption and diminished quantum yields. It is
interesting that of the dyes evaluated computationally, PyIn2SQ
had a substantial influence of the HOMO-1 orbital (which is
heavily influenced by the π-system of pyrene) on the vertical
transition (see below). Since the PyIn2SQ dye has the highest
quantum yield in the series, the modulation of the HOMO-1 to
have a significant contribution to vertical transition upon transfer
of charge to the LUMO may be a novel strategy for enhancing Φ
values for NIR emissive materials. This suggests tuning lower
level orbitals could have substantial influences on Φ values.
The solid-state absorption of all the dyes was also
measured by diffuse reflectance spectroscopy. All of the dyes
showed absorption in the 600-800 nm range in the solid state
with several of the dyes’ having absorption extending well into
the 800-900 nm range (Figure S7). Solid-state emission of dyes
in this range is useful for potential applications in secure display
technologies (NIR OLEDs). As a demonstration for the use of
these materials in as solid emitters, PhIn2SQ powder was
arranged in the shape of an “M”, then covered with a white filter.
As shown in Figure 8, under visible light irradiation only the filter
is visible. However, under NIR irradiation at 785 nm through the
white filter, an image of the PhIn2SQ powder is clearly visible via
NIR photon detection at 850 nm. The emission spectra of each
dye in the solid state can be found in Figure S8.
Substituent selection has relatively minor effects on
absorption and emission energies. Electrochemical analysis was
used to find if the substituents are effectively isolated from the
In2SQ π-system responsible for the observed absorption and
emission energies. The electrochemical properties of the dyes
are also critical for applications involving solar cells, display
applications, photochemical reactions and for assessing of the
stability of the dyes under atmosphere. The dye ground-state
oxidation potentials (E(S+/S)) ranged from 1.26 to 1.44 V versus
ferrocene and the excited-state oxidation potentials (E(S+/S*)),
varied from -0.88 V to -1.22 V. The ground-state reduction
potentials (E(S/S-)) show a similar range of potentials (-1.37 to -
1.55 V). Interestingly, this indicates the substituents have
tunable control of E(S+/S), E(S/S-) and E(S+/S*) energy levels without
significantly altering the optical band gap energy (Egopt) which
only varies by 0.04 eV between dyes. In general, the dye
substituents give the expected changes in energy levels where
added π-electron density from aryl groups raise energy levels to
more destabilized values and substituents withdrawing π-
electron density lead to more stabilized energy levels (Table S2).
The observed Stokes shifts are hypothesized to be the
result of a geometric reorganization in the excited-state to allow
for increased planarization of the π-system. To evaluate this
hypothesis, computational analysis was performed on PhIn2SQ
with DFT at the B3LYP/6-311g(d,p) level. Ground-state and
excited-state geometries were first optimized, then TD-DFT at
the B3LYP/6-311g(d,p) level was carried out to predict the
vertical excitation (ground-state geometry) and relaxation
energies (excited-state geometry), where the difference of these
corresponds to a Stokes-shift. Provided reasonable agreement
is observed between theory and experiment, the geometry
changes from the ground to the excited-state can be analyzed to
better understand the origin of the Stokes-shift observed.
In the ground state, the HOMO and LUMO were found to
be delocalized throughout the indolizine-squaraine π-system
with little orbital contribution from the indolizine substituents
(Figure 9). The HOMO-1 orbital was found to be the first orbital
with substantial contribution of the indolizine-aryl substituent to
the π-system (Figure S10). The indolizine-squaraine backbone
is not fully planar and the PhIn2SQtrans derivative was found to
have a twist angle of 10 degrees between the indolizine plane
This article is protected by copyright. All rights reserved.