Unexpected photoluminescence of fluorinated naphthalene diimides

Article abstract of DOI:10.1002/chem.201405876

Two new amino core-substituted naphthalene diimides (cNDIs) bearing fluorinated side chains have been synthesised. Steady-state and time-resolved fluorescence spectroscopy reveals unprecedented optical properties for the cNDIs with high quantum yields of ~0.8 and fluorescence lifetimes of ~13 ns in a range of solvents. These properties are apparent at the level of single molecules, where the compounds also show exceptional photostability under pulsed-laser excitation. Photon emission is remarkably consistent with very few long timescale (millisecond or longer) interruptions with molecules regularly undergoing >10⁷ cycles of excitation and emission. Intermittencies owing to triplet-state formation occur on a sub-millisecond timescale with a low yield of 1-2%, indicating that the presence of the fluorine atoms does not lead to a significant triplet yield through the heavy-atom effect. These properties make the compounds excellent candidates for single-molecule labelling applications.

Full text of DOI:10.1002/chem.201405876

DOI: 10.1002/chem.201405876
Full Paper
&
Fluorescent Compounds
Unexpected Photoluminescence of Fluorinated
Naphthalene Diimides
Subashani Maniam, Rosalind P. Cox, Steven J. Langford,* and Toby D. M. Bell*[a]
Abstract: Two new amino core-substituted naphthalene di-
imides (cNDIs) bearing fluorinated side chains have been
synthesised. Steady-state and time-resolved fluorescence
spectroscopy reveals unprecedented optical properties for
the cNDIs with high quantum yields of ~0.8 and fluores-
cence lifetimes of ~13 ns in a range of solvents. These prop-
erties are apparent at the level of single molecules, where
the compounds also show exceptional photostability under
pulsed-laser excitation. Photon emission is remarkably con-
sistent with very few long timescale (millisecond or longer)
interruptions with molecules regularly undergoing >107
cycles of excitation and emission. Intermittencies owing to
triplet-state formation occur on a sub-millisecond timescale
with a low yield of 1–2%, indicating that the presence of
the fluorine atoms does not lead to a significant triplet yield
through the heavy-atom effect. These properties make the
compounds excellent candidates for single-molecule label-
ling applications.
Introduction
yield has been observed for a series of halogenated squaraine
dyes.[14] This investigation suggests that there is a redistribution
of electron density of the halogen atom from its HOMO to the
LUMO energy level of the conjugated system, therefore
strengthening the p-system upon excitation. Increases in quan-
tum yields and fluorescence lifetimes for series of fluorinated
boron-dipyrromethenes (BODIPYs) dyes has also been demon-
strated, as have the tunability of fluorescence in stilbene-based
molecules through different halogen- and hydrogen-bonding
interactions.[15,16] Further investigations have focused on the
effect of halogenation on the solid-state fluorescence and elec-
troluminescence properties of oligo(phenylene vinylene).[17] In
all of these examples, however, halogen atoms, and particularly
fluorine atoms, are directly attached to a conjugated system.
In our efforts to develop naphthalene diimides (NDIs) as ver-
satile aromatic compounds and realise some of their enormous
potential, we have developed synthetic strategies that have
led to the preparation of NDIs in single-molecule and biophysi-
cal applications,[18,19] as sensors,[20,21] and in supramolecular ar-
chitectures.[22,23] The potential for a chromophore to be ob-
served by fluorescence at a single-molecule level relies strongly
on its ‘molecular brightness’, which in its simplest form can be
approximated by the product of molar absorption coefficient
and fluorescence quantum yield.[24,25] Core-substituted NDIs
(cNDIs) perform quite well on the quantum yield side of the
equation, with a number of reported compounds displaying
quantum yields of 0.3<Ff <0.5.[26,27] On the absorption side of
molecular brightness, the issue with cNDIs is their relatively
low molar absorption coefficients, which typically fall in the
range of 10000–20000 cmꢀ1 mꢀ1.[21,27] Thus, the cNDIs reported
so far do not have the molecular brightness of the larger
rylene imide derivatives or other fluorophore families such as
the cyanines and rhodamines, for example. cNDIs have, howev-
er, been successfully imaged as single molecules[18] and are
Informed design of new compounds derived from an optically
active parent with the goal of obtaining improved spectro-
scopic properties is a long-standing and proven approach. Flu-
orination has proved to be one successful strategy to alter the
energy levels of molecular orbitals,[1,2] tune molecular pack-
ing,[3,4] and increase solubility within a family.[5,6] This method-
ology has been used as part of the molecular design strategy
for application in field-effect transistors (OFETs), dye-sensitised
solar cells (DSSCs) and photovoltaics (PVs).[7–9] One drawback,
however, is that luminescent properties are usually highly com-
promised in the presence of fluorine atoms due to the heavy-
atom effect (HAE). This effect is where the presence of ‘heavy’
atoms (e.g., substitution of hydrogen atoms by halogen
atoms) promotes singlet–triplet conversion by intersystem
crossing, thereby reducing the fluorescence quantum yield
and the excited-state lifetime of a molecule.[10–12] In addition to
the HAE, the high electronegativity of fluorine atoms can fur-
ther decrease fluorescence quantum yield by withdrawing
electron density away from the conjugated system of a chro-
mophore.[13] However, some recent investigations have shown
that halogens can have a positive effect on luminescence
properties. For example, an increase in fluorescence quantum
[a] Dr. S. Maniam,+ R. P. Cox,+ Prof. S. J. Langford, Dr. T. D. M. Bell
School of Chemistry
Monash University
Wellington Rd, Clayton 3800, Victoria (Australia)
Fax: (+613)9905 4597
E-mail: steven.langford@monash.edu
toby.bell@monash.edu
[+] Both authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201405876.
Chem. Eur. J. 2015, 21, 4133 – 4140
4133
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
good candidates in single-molecule labelling applications as
they undergo relatively few interruptions, or ‘blinks’ to their
fluorescence emission. This, plus their tunable optical proper-
ties and ready functionalisation, mean interest in cNDIs re-
mains high.
Continued progress in the application of cNDIs[27,28] requires
further development of the synthetic methodologies that
access systems with novel properties. Fluorinated moieties are
easily introduced at the N-imide positions of NDIs;[29–33] howev-
er, thus far, there are few examples of fluorinated cNDIs.[34,35]
One successful approach has been to append fluoroalkyl moi-
eties directly onto the core, which causes strong negative
spectroscopic changes.[17] As part of our continuing develop-
ment of cNDIs with new and improved optical properties, we
have synthesised new cNDIs 4 and 5 bearing fluorinated alkyl
side groups, which are typically used to impart solubility. We
observed an unusual increase in the fluorescence quantum
yield and lifetime, to around 0.8 and 13 ns, respectively, in
both cNDIs 4 and 5 with only small variation with solvent. This
is carried through to the single-molecule level, where mole-
cules of 4 show a significant increase in molecular brightness
compared with previous cNDIs, exceptional photostability, and
almost uninterrupted photon
Figure 1. a) Molecular structures of 4 and 5, and b) X-ray crystallographic
structure of 5 showing canted vertical stacking.
emission intensity.
Results and Discussion
Compounds 4 and 5 (Figure 1a),
bearing non-fluorinated and flu-
orinated alkyl substituents at the
N-imides, respectively, and core
di-substituted with amino-meth-
ylene-perfluoroalkyl
groups,
were synthesised by using Hart- Scheme 1. Synthesis of compounds 4 and 5. dba=dibenzylidenacetone; dppf=1,1’-bis(diphenylphosphine)ferro-
cene.
wig–Buchwald coupling condi-
tions (Scheme 1). This core-sub-
stitution synthetic methodology
is not commonly utilised for NDIs;[28] however, was necessary
as the known electron-withdrawing nature of the fluorinated
moiety decreased the nucleophilicity of the amine. Full syn-
thetic characterisation of 4 and 5 can be found in the Support-
ing Information along with synthetic details for the intermedi-
ate for 4 (Figures S1–S11 in the Supporting Information).
Figure 2 shows the normalised steady-state absorption and
emission spectra of 4 and 5 in a CH2Cl2 solution measured at
room temperature. The absorption spectra display characteris-
tic bands in the near-UV range, at 350 nm and 370 nm, which
are assigned to p–p* transitions, along with a broad band cen-
Deep-red crystals of 5 suitable for X-ray analysis[36] were
grown from CHCl3 solution and the resultant structure is
shown in Figure 1b. Single-crystal analysis of compound 5
shows that it is planar along the NDI core. Neighbouring mole-
cules of 5 display a canted vertical stacking arrangement (Fig-
ure 1b) with the nitrogen atoms of the N-imide centrally locat-
ed along the ring system of the adjacent NDI group. This ori-
entation leads to p–p stabilisation of the NDI ring with the
imide nitrogen giving an average plane separation of 3.4 ꢁ.
Within the crystal lattice, columns of 5 extend along the crys-
tallographic a-axis, whereas in the b,c-plane, 5 adopts a step-
stacking approach with the neighbouring molecules, an ar-
rangement that maximises the London dispersive forces within
the alkyl chains (Figure S12 in the Supporting Information).
Figure 2. Normalised steady-state absorption and emission spectra
(lex =520 nm) of 4 and 5 in CH2Cl2 (1ꢂ10ꢀ5 m).
Chem. Eur. J. 2015, 21, 4133 – 4140
www.chemeurj.org
4134
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
tered at 560–565 nm, attributed to an n–p* transition. The
molar absorption coefficients of 4 and 5 in CH2Cl2 are deter-
mined to be 15400 and 11500mꢀ1 cmꢀ1, respectively (Fig-
ure S13 in the Supporting Information) and these spectral fea-
tures are consistent with previously reported di-amino-substi-
tuted NDIs.[14,20,37–40] Spectra are very similar for both com-
pounds, indicating that the nature of the imide-substituted
group has only a minimal effect on the optical properties ob-
served.
Interestingly, emission spectra for both compounds are cen-
tered at 590 nm with favourably high fluorescence quantum
yields (Ff) of 0.85 and 0.75 for 4 and 5, respectively, in CH2Cl2.
The emission maximum of 4 redshifts from 570 nm in n-
hexane to 594 nm in benzonitrile (PhCN) and the emission
maximum of 5 shows a similar trend (see Table 1, and Figur-
es S14 and S15 in the Supporting Information). This observa-
tion is likely due to the symmetry of the molecules, which bal-
Figure 3. TCSPC decay histograms and fitted single-exponential functions
(black) for 4 in n-hexane and PhCN. IRF is the instrument response function.
higher than other known NDIs mainly owing to the non-radia-
tive decay being much less efficient than for non-fluorinated
cNDIs, presumably due to a decrease in vibrationally
mediated non-radiative loss as a result of the F
Table 1. Steady-state and time-resolved optical data for 4 and 5 in various solvents.
atoms adding mass to the core substituents. knr in-
creases slightly with polarity (Table S1 in the Support-
ing Information), thus, the lifetimes marginally de-
crease; this is consistent with the small reduction in
HOMO–LUMO band gap as shown by the redshift of
the absorption and emission maxima with increasing
polarity. The presence of the F atoms might be ex-
pected to increase the rate constant and yield of in-
tersystem crossing to the triplet state through the
heavy atom effect, however, the bulk measurements
show that this is not occurring to any significant
extent. This is likely attributable to the F atoms being
located on the substituents and not adjacent to the aromatic
core where they would exert more influence. Furthermore, all
fluorinated substituents, whether on the core or at the imides,
first contain an alkyl carbon (ꢀCH2) before the fluorinated car-
bons, heading outwards from the point of attachment.
The HOMO and LUMO energy levels of 4 and 5 were mea-
sured by using cyclic voltammetry (Figure 4 and Table S2 in
the Supporting Information). Compound 4 displays two well-
defined reversible one-electron redox processes, with the first
reduction and oxidation E1/2 values of ꢀ1.19 V and 1.08 V (vs
Fc/Fc+). Thus, for compound 4, the HOMO energy level is
Solvent
lmax(abs)
[nm]
5
lmax(em)
[nm]
5
Ff
tflu
[ns]
5
4
4
4
5
4
n-hexane 550 555 570 574 0.89ꢁ0.02[a] 0.83ꢁ0.03 13.8ꢁ0.2 13.8ꢁ0.1
toluene
CH2Cl2
PhCN
560 565 587 592 0.85ꢁ0.02
560 565 590 593 0.85ꢁ0.03
564 569 594 597 0.75ꢁ0.04
0.77ꢁ0.03 12.7ꢁ0.1 13.0ꢁ0.1
0.75ꢁ0.04 12.9ꢁ0.2 12.8ꢁ0.2
0.73ꢁ0.04 12.3ꢁ0.1 12.2ꢁ0.1
[a] Errors are ꢁ1 standard deviation from triplicate measurements.
ances the charge distribution, leading to only a weak depend-
ence on solvent polarity. Table 1 also indicates that both 4 and
5 show high Ff values across a range of solvent polarities,
from completely non-polar n-hexane to the quite polar solvent
PhCN. The compounds, however, are not soluble in water,
where they aggregate. The highest Ff value of 0.89 is achieved
with 4 in n-hexane and Ff remains greater than 0.7 for both 4
and 5 in PhCN. These Ff values are significantly higher than
those reported thus far for core-substituted NDIs[27] and en-
courages the development of water-soluble derivatives with
the hope that the high Ff carries through into this solvent,
thus opening the way for use of these compounds in biologi-
cal-labelling applications. We have recently synthesised and
characterised some water-soluble cNDIs with substituents con-
taining a number of oxygen atoms[19] and derivatives of 4 with
such groups at the imide positions are current targets of ours.
The kinetics of the fluorescence emission of 4 and 5 were
explored by using time-correlated single-photon counting
(TCSPC). A single exponential decay function adequately de-
scribed the emission in all solvents for 4 and 5 (Figure S16 in
the Supporting Information) with observed fluorescence life-
times of 12.7 and 12.6 ns, respectively, in CH2Cl2 (Figure 3). On
the basis of the Ff values and observed lifetimes, the radiative
lifetimes for 4 and 5 were deduced to be 15.3 and 16.1 ns, re-
spectively, in CH2Cl2. In addition, their non-radiative decay rate
constants, knr, were calculated to be 1.3ꢂ107 sꢀ1 for 4 and 1.8ꢂ
107 sꢀ1 for 5. This suggests that the Ff values for 4 and 5 are
Figure 4. Cyclic voltammograms of 4 and 5 in CH2Cl2 (0.5 mm) solutions con-
taining 0.1m Bu4NPF6 as the supporting electrolyte at a scan rate of
100 mVsꢀ1
.
Chem. Eur. J. 2015, 21, 4133 – 4140
www.chemeurj.org
4135
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ꢀ5.88 eV and the LUMO is ꢀ3.62 eV. By introducing fluorinated
groups at the N-imide positions, compound 5 has deeper
HOMO and LUMO energy levels of ꢀ5.99 eV and ꢀ3.82 eV, re-
spectively. These HOMO–LUMO electrochemical band gaps
agree well with the S1–S0 energy difference (optical band gaps)
indicated by the spectral maxima of these compounds. The
presence of the electron-withdrawing groups has reduced the
HOMO–LUMO band gap, from 2.26 eV for compound 4 to
2.17 eV for compound 5, which is consistent with the slight
redshift (~5 nm, 0.05 eV) recorded in the optical spectra.
A detailed investigation of 4 as single molecules was per-
formed to see if its favourable optical properties were carried
through to the single-molecule level. 4 was chosen for this
analysis as it outperforms 5 in both quantum yield and molar
absorption coefficient and is thus expected to show greater
molecular brightness. Figure 5a shows a wide-field fluores-
cence image of 4 embedded in a poly(methyl methacrylate)
(PMMA) film. It can be seen that single molecules manifest as
bright diffraction-limited point-spread functions when excited
with 532 nm light. Under quite gentle illumination conditions
(~0.1 kWcmꢀ2), the molecules typically showed a signal-to-
noise value of ~4.
Figure 6. Emission trajectories of four single molecules of 4 in a PMMA film
in ambient atmosphere. Excitation at 560 nm.
cies or transitions to dim states were very infrequent and ob-
served in less than 10% of the total single molecules studied.
Factoring in the detection efficiency of the single molecule
(SM) set-up, where ~5% of emitted photons are detected by
the avalanche photodiode (APD), it can be estimated that over
50% of single molecules of 4 emit more than 1ꢂ107 photons.
Some molecules showed remarkable longevity, for example,
molecule d emitted an almost continuous stream of photons
for longer than 10 min, undergoing more than 1ꢂ108 excita-
tion–emission cycles before photobleaching.
Single molecules of 4 were also studied by using confocal
fluorescence microscopy. 10ꢂ10 mm areas (Figure 5b) were
The fluorescence lifetimes of 4 varied from molecule to mol-
ecule over a range from ~6–22 ns, however, they were centred
at approximately 13 ns, which is similar to the bulk values
measured in solvent. Figure 7 shows an example of lifetimes
calculated along the trajectory of a single molecule, an exam-
ple fluorescence-decay profile, and the lifetime distribution for
all molecules analysed. Decay lifetimes were determined from
histograms created by binning consecutive amounts of
10000 photons. Individual molecules showed stable lifetimes—
the variation of <1 ns in the decay lifetime of the molecule in
Figure 5. Single molecules of 4 embedded in a PMMA film. a) Wide-field
fluorescence microscopy image. Scale bar: 10 mm. Excitation at 532 nm.
b) Confocal scan of a 10ꢂ10 mm area, 1 ms/pixel dwell time. Scale bar:
2 mm. Excitation at 560 nm.
raster scanned to identify single molecules and their time-
tagged photon streams and spectral data were collected. In
total, emission from 150 individual molecules was studied.
Figure 6 shows example emission trajectories of four individual
molecules. Most molecules produced intensity traces with little
variation in photon count rate other than statistical noise. En-
couragingly, single molecules showed little to no blinking and
most emitted for longer than 1 min, indicating excellent pho-
tostability.
Molecules a and b in Figure 6 show typical trajectories
where the emission has a stable count rate before irreversibly
photobleaching in a single step. Molecule c displays a clear
blinking event at 55 s, which lasts less than 500 ms before par-
tial recovery of intensity that lasts for a further 1.2 s. This is fol-
lowed by a transition to a long-lived (at least 4 s) dim state of
a similar intensity to the blinking event. Long-lived intermitten-
Figure 7. a) Emission trajectory and decay times for a single molecule of 4
embedded in a PMMA film. Lifetimes were calculated from decay profiles
created by binning successive packets of 10000 photons. b) A typical decay
profile and fit for 4. c) A fluorescence lifetime distribution of all the single
molecules analysed.
Chem. Eur. J. 2015, 21, 4133 – 4140
www.chemeurj.org
4136
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 7 is typical—making assignment of each molecule into
1 ns bins for the histogram straightforward.
are often observed in single molecules and can arise from
mechanisms such as electron trapping and oxidation.[41,42] NDI
4 appears to not suffer any appreciable amount of these blinks
and single molecules of 4 emit an uninterrupted and reliable
stream of photons. Another intrinsic blinking mechanism avail-
able to many fluorophores is triplet formation. Triplet states
are usually much shorter lived (microseconds), especially in
ambient conditions where collision with oxygen will return the
molecule to its ground state whereupon emission can resume.
Triplet formation does not present a significant problem when
measuring the fluorescence of a compound under bulk con-
centrations, where a proportion of triplet formation will result
in the same proportion decrease in the fluorescent output.
However, when studying fluorescent single molecules, the life-
time of the triplet state will have an effect on the photon effi-
ciency. Whereas the triplet yield remains the same, the overall
fluorescence intensity over time will be lower as the molecule
will not be excited for the duration of the ‘off’ period. When
triplet lifetimes are too short to be seen by simple intensity
binning, their presence can be revealed by using autocorrela-
tion analysis of the emission intensity from a single molecule.
Autocorrelation functions were generated for the single mol-
ecules studied. Figure 9 shows an example molecule emission
trajectory with an autocorrelation curve and fit. Intensity auto-
correlation functions were generated by calculating the corre-
lation of photon output, I, at any given time, t, with the
photon output at a time (t+t), shifted by the lag time t, by
using Equation (1):
Emission spectra were measured for single molecules em-
bedded in PMMA. Example emission spectra are shown in
Figure 8, with single molecules emitting sufficient photons for
successive spectra to be taken every second. The fluorescence
is very stable, with the emission maxima centred at 578 nm
and varying only ꢁ3 nm across 60 spectra. This molecule is
Figure 8. Emission spectra from a single molecule of 4 embedded in
a PMMA film. Successive spectra were recorded every second. Excitation at
532 nm.
representative of the molecules studied, with minimal spectral
shifts across the time observed. Emission maxima varied slight-
ly between molecules, ranging between 576–581 nm. This is
consistent with the emission maxima measured in the bulk
properties.
< dIðtÞdIðt þ tÞ >
ð1Þ
GðtÞ ¼
< I >2
Interruptions or ‘blinks’ to a single molecule’s photon emis-
sion can occur through a variety of mechanisms. Inspection of
single-molecule trajectories of 4 reveals that blinks of sufficient
duration to be visible in the trajectories when binned in 10 ms
bins (e.g., the event at 14 s in Figure 9a, which lasts 20 ms) are
exceedingly rare and when they do occur are not particularly
long-lasting. Long duration interruptions to photon emission
Autocorrelation traces were fit by an exponential decay
function, G(t)=AACexp[ꢀt/tAC] with the decay time, tAC, and
the pre-exponential factor, AAC, used to determine ‘on’ and ‘off’
times from Equations (2) and (3):
1
tAC toff ton
1
1
¼
þ
ð2Þ
ð3Þ
toff
ffi
ton
AAC
These equations are based on the analysis by Weston
et al.,[43] and are applicable to a two-state system where the in-
tensity of one state approaches zero, that is, ‘on–off’ behaviour
such as that generated by a single fluorescent molecule under-
going triplet-state excursions. For the molecule in Figure 9, tAC
is 97 ms and AAC is 0.750, yielding a triplet lifetime (toff) of
170 ms and ton of 226 ms. The triplet yield, or yield of intersys-
tem crossing (ISC), can be calculated by using Equation (4):
Figure 9. a) Emission trajectory and the calculated triplet lifetime for a single
molecule of 4 embedded in a PMMA film. Triplet lifetimes were calculated
from autocorrelation curves created by binning successive packets of 50000
photons. b) Autocorrelation curve and fit for 4 in ambient atmosphere.
c) The triplet lifetime distribution of all the single molecules analysed.
ꢀfꢀdet
ð4Þ
ꢀISC
¼
tonIf
where Fdet is the detection efficiency (~5%) and If is the emis-
sion intensity in counts per second (for this molecule,
Chem. Eur. J. 2015, 21, 4133 – 4140
www.chemeurj.org
4137
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
cedures.[18] Acetone, acetic acid, ethanol, dichloromethane, and di-
methylformamide (DMF) were purchased from commercial suppli-
ers. Benzonitrile, DMF, toluene and poly(methyl methacrylate)
(PMMA) used for spectroscopic analysis were of the highest purity
available from Aldrich or Merck.
~15000 countssꢀ1); thus, FISC is 1.2%. The triplet lifetime was
measured as a function of time from autocorrelation curves
created by binning successive packets of 5000 photons. There
was little fluctuation across the lifetime duration of the mole-
cule, with triplet lifetimes between 150 and 200 ms. A similar
lack of variation in triplet lifetime over time was seen in all the
molecules studied under ambient atmosphere, with the triplet
lifetimes falling in the range of 100–300 ms and yields of 1–2%.
These values are of a similar magnitude to that of other
cNDIs;[44] however, the triplet yield is slightly increased, sug-
gesting there is a small HAE from the F atoms, but not suffi-
cient to lead to significant increases in the ISC rate constant
and yield. The triplet lifetime of 4 increased to 500–600 ms
when measured under N2 (Figure S17 in the Supporting Infor-
mation), yet triplet yields remained at ~1–2%, confirming trip-
let formation as the origin of these short-lived intermittencies.
Despite the introduction of fluorine atoms, the triplet yield re-
mains quite low and with short, sub-millisecond lifetimes, en-
suring a constant fluorescent output from single molecules.
These properties point to the suitability of these cNDIs for use
in single-molecule labelling applications.
Characterisation
Low-resolution electron-impact mass spectrometry (LR-MS) was
performed on a Micromass Platform II API quadrupole electrospray
mass spectrometer. High-resolution electron-impact mass spec-
trometry (HR-ESI) was performed on an Agilent Technologies 6220
1
Accurate-Mass Time-of-Flight LC/MS. H and 13C NMR spectra were
recorded by using a Bruker DRX 400 MHz NMR spectrometer
(400 MHz for 1H NMR, 100 MHz for 13C NMR) and a Bruker AV
600 MHz NMR spectrometer (600 MHz for 1H NMR, 150 MHz for
13C NMR), using CDCl3 unless otherwise stated. Chemical shifts are
reported relative to the resonances of residual CHCl3 at d=
7.26 ppm (H) and d=77.2 ppm (C). Fluorine NMR spectra were re-
corded on a Bruker AV400 spectrometer at 400 MHz using CFCl3 as
the external reference at d=0 ppm. NMR spectra for 2–5 are given
in the Supporting Information, Figures S1–S11.
Synthesis
Synthesis of 4: N,N’-Bis(n-hexyl)-2,6-dibromo-1,4,5,8-naphthalene-
tetracarboxy diimide (2) (see the Supporting Information) (50 mg,
0.08 mmol), 2,2,3,3,4,4,5,5,6,6,6-undecafluorohexan-1-amine (0.10 g,
Conclusion
The synthesis and structural characterisation of two new fluori-
nated cNDIs have been presented and a comprehensive inves-
tigation into their optical properties in bulk solution and at the
single-molecule level has been conducted. Both compounds
show high quantum yields (~0.8) and long single-exponential
fluorescence lifetimes (~13 ns) in a range solvents; values un-
precedented in cNDIs. As single molecules, the compound
with fluorinated core side chains showed excellent photon-
emission properties with consistent photon output over ~107–
108 cycles of excitation and emission. Interruptions or blinks in
the photon stream lasting longer than milliseconds were ex-
ceptionally rare and shorter intermittencies due to triplet-state
formation were also infrequent and had durations of ~100 ms
in ambient conditions and yields of only 1–2%. These proper-
ties arise from fluorination of the core substituents, which is
contrary to a common perception about heavy atom quench-
ing effects and shows that fluorination can, in the right circum-
stances, lead to improved optical properties and an increase in
molecular brightness without a significant increase in intersys-
tem crossing yield. The high photostability and consistent
photon emission from single molecules demonstrates that
these compounds have high potential for single-molecule
imaging and labelling applications.
0.34 mmol),
2.53 mmol),
tris(dibenzylidenacetone)dipalladium(0)
1,1’-bis(diphenylphosphine)ferrocene
(2.3 mg,
(0.5 mg,
0.84 mmol) and sodium tert-butoxide (9.7 mg, 1.08 mmol) were
heated in dry toluene (50 mL) at 908C for 12 h. After this period,
the reaction was cooled and the solvent removed under reduced
pressure. The crude material was dissolved in 20% n-hexane in
CH2Cl2 (50 mL) and passed through a pad of silica gel. The filtrate
was concentrated and purified by column chromatography (20%
n-hexane in CH2Cl2) to give the title compound 4 as a bright-pink
solid (43 mg, 50%). M.p.>2008C; 1H NMR (400 MHz): d=9.88 (t,
J=6.8 Hz, 2H, NH), 8.20 (s, 2H, NDI), 4.31–4.23 (m, 4H, NCH2CF2),
4.18 (dd, J=7.6, 7.6 Hz, 4H, NCH2), 1.76–1.68 (m, 4H, CH2), 1.45–
1.31 (m, 12H, CH2), 0.90 ppm (t, J=6.8 Hz, 6H, CH3); 19F NMR
(400 MHz): d=ꢀ80.72–ꢀ80.77 (m, 6F), ꢀ117.32–ꢀ117.40 (m, 4F),
ꢀ122.58–ꢀ122.60
(m,
4F),
ꢀ122.61–ꢀ122.16
(m,
4F),
ꢀ126.09–ꢀ126.17 ppm (m, 4F); 13C NMR (100 MHz, partial): d=
166.4, 162.5, 149.0, 126.3, 121.8, 118.1, 104.0, 43.1, 40.9, 31.6, 28.1,
27.0, 22.7, 14.1 ppm; HRMS (TQ-MS-ESI): m/z calcd for
C35H38F22N4O4: 1029.2302 [M+H]+; found: 1029.2312 [M+H]+.
Synthesis of 3: 2,2,3,3,4,4,5,5,6,6,6-Undecafluorohexan-1-amine
(1.05 g, 3.52 mmol) was added to a suspension of 2,6-dibromo-
1,4,5,8-naphthalenetetracarboxylic
dianhydride
(1)
(0.50 g,
1.17 mmol) in acetic acid (80 mL), and the reaction heated at
1308C for 4 h. The formed precipitate was filtered and recrystal-
lised with CH2Cl2/methanol to give the title compound as a pale-
orange powder 3 (0.32 g, 28%). The compound was found to be
rather insoluble in many organic solvents. M.p.>2508C; 1H NMR
(400 MHz, [D6]DMSO): d=9.43 (s, 2H, NDI), 5.56 ppm (t, J=16.0 Hz,
4H, NCH2); 19F NMR (400 MHz, [D6]DMSO): d=ꢀ80.72–ꢀ80.77 (m,
3F), ꢀ114.38–ꢀ114.46 (m, 2F), ꢀ122.18–ꢀ122.23 (m, 2F),
ꢀ123.23–ꢀ123.37 (m, 2F), ꢀ125.87–ꢀ125.94 ppm (m, 2F); 13C NMR
(150 MHz, partial, [D6]DMSO): d=155.7, 134.6, 128.8, 124.9 ppm;
HRMS (TQ-MS-ESI): m/z calcd for C26H6Br2N2F22O4: 987.8323 [M]ꢀ;
found: 987.8335 [M]ꢀ.
Experimental Section
Materials
1,4,5,8-Naphthalenetetracarboxylic dianhydride and 1-hexanamine
were obtained from Sigma–Aldrich and 2,2,3,3,4,4,5,5,6,6,6-undeca-
fluorohexan-1-amine was purchased from Tokyo Chemical Industry
(TCI). All chemicals were used as received. 2,6-Dibromo-naphtha-
lene dianhydride was prepared following established literature pro-
Synthesis of 5: N,N’-Bis(2,2,3,3,4,4,5,5,6,6,6-undecafluorohexyl)-2,6-
dibromo-1,4,5,8-naphthalenetetracarboxy diimide (3) (50 mg,
Chem. Eur. J. 2015, 21, 4133 – 4140
www.chemeurj.org
4138
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
0.05 mmol), 2,2,3,3,4,4,5,5,6,6,6-undecafluorohexan-1-amine (60 mg,
FWHM). Goodness-of-fit of the data by the fitting function was de-
termined by the c2 fitting parameter and the distribution of the fit
residuals.
0.20 mmol),
tris(dibenzylidenacetone)dipalladium(0)
(1.4 mg,
1.52 mmol),
1,1’-bis(diphenylphosphine)ferrocene
(0.06 mg,
0.51 mmol) and sodium tert-butoxide (2.8 mg, 0.05 mmol) were
heated in dry toluene (50 mL) at 908C for 12 h. After this period,
the reaction was cooled and the solvent removed under reduced
pressure. The crude material was dissolved in 20% n-hexane in
CH2Cl2 (50 mL) and passed through a pad of silica gel. The filtrate
was concentrated and purified by column chromatography (20%
n-hexane in CH2Cl2) to give the title compound 5 as a bright-pink
Single-molecule samples of 1 were prepared by serial dilution into
1% solution of PMMA until a final concentration of ~10ꢀ10 m. Sam-
ples were spin-cast onto thoroughly cleaned glass cover slips at
2000 rpm for 60 s, yielding a film thickness of ~100–200 nm. Wide-
field imaging was achieved by using a previously described set
up,[47] equipped with a 100ꢂ 1.3 N.A. oil immersion objective
(Olympus). Confocal microscopy measurements were performed by
using a pulse-picked supercontinuum fiber laser (Fianium, SC 400–
4-pp) with pulses of ~40 ps across the visible and near-IR range.
The excitation wavelength was selected by using a 10 nm band
pass filter (Chroma). The light was then passed through a focusing
lens through a 10 mm pinhole (Thorlabs) and then collimated. The
light was directed through an inverted Olympus IX71 microscope
and focused onto the sample using a 100ꢂ 1.4 N.A. oil immersion
objective (Olympus). Emission from the sample was collected
through the same objective, separated from excitation light by a di-
chroic mirror (Chroma), passed through a long pass filter (Chroma)
and was either sent directly though to an avalanche photodiode
(APD, PicoQuant, t-SPAD) or was split using a 50:50 beamsplitting
prism (Edmund Optics) between the APD and through a spectro-
graph (Princeton Instruments) to an EMCCD camera (Princeton In-
struments, ProEM512). Molecules were selected by scanning a 10ꢂ
10 mm area by using a Piezo stage (Physik Instrumente, P-733.2CL)
and controller (Physik Instrumente, E-710). Emission times were re-
corded by using a photon counting device (Picoquant, PicoHarp
300) through a router (Picoquant, PHR 800) with a trigger diode as-
sembly (PicoQuant, TDA 200) to provide the start signal. Data col-
lection was performed by using the SymPhoTime (Picoquant) and
WinSpec (Princeton) software suites. The photon stream from the
APD was collected in t3r mode where both macro times (time
from start of measurement) and micro times (time from previous
laser pulse) were recorded for every photon. Single-molecule tra-
jectories (from macro times) and decay profiles (from micro times)
were constructed and analysed by using the BIFL software package
(Scientific Software Technologies Center). Single-molecule decay
profiles were fit by an exponential decay convolved with an IRF re-
corded from excitation light scattered by a blank coverslip (IRF typ-
ically ~700–800 ps, FWHM for the confocal set-up). Goodness-of-fit
of the data by the fitting function was determined by the reduced
c2 fitting parameter returned by the BIFL software and the distribu-
tion of the fit residuals.
1
solid (20 mg, 28%). M.p.>2008C; H NMR (400 MHz): d=9.75 (br t,
2H, NH), 8.33 (s, 2H, NDI), 5.07–4.50 (br t, 4H, NCH2CF2), 4.30 ppm
(br t, 4H, NCH2); 19F NMR (400 MHz): d=ꢀ80.74–ꢀ80.79 (m, 12F),
ꢀ114.84–ꢀ114.90
ꢀ122.56–ꢀ122.59
(m,
(m,
4F),
8F),
ꢀ117.14–ꢀ117.21
ꢀ123.10–ꢀ123.13
(m,
(m,
4F),
4F),
ꢀ123.63–ꢀ126.65 (m, 4F), ꢀ126.16–ꢀ126.18 ppm (m, 8F); 13C NMR
(100 MHz, partial): d=165.5, 162.3, 149.5, 126.1, 122.2, 119.1, 103.7,
43.4, 38.7 ppm; HRMS (TQ-MS-ESI): m/z calcd for C38H11F44N4O4:
1423.0078 [MꢀH]ꢀ; found: 1423.0088 [MꢀH]ꢀ.
Electrochemical Methods
Electrochemical experiments were carried out in a standard three-
electrode arrangement with a Bioanalytical Systems BAS Model
100B electrochemical workstation (Bioanalytical Systems, West La-
fayette, IN, USA). A Pt wire was used as the auxiliary electrode,
while another Pt wire in a frit (CH2Cl2, 0.1m [Bu4N][PF6]) separated
from bulk solution was used as a quasi-reference electrode.
The potentials are quoted relative to the ferrocene/ferrocenium
(Fc/Fc+) potential scale, as an internal reference. A glassy carbon
(GC) working electrode (ALS, Japan) with a diameter (d=0.10 cm)
was used for cyclic voltammetry experiments. Prior to each voltam-
metry experiment, the working electrode was polished with 0.3 mm
alumina on a clean polishing cloth (Buehler, USA), rinsed with
water, then acetone and finally dried under nitrogen gas. The solu-
tions used for voltammetry measurements were purged with sol-
vent-saturated nitrogen gas prior to commencement of an experi-
ment and blanketed with nitrogen during the course of the experi-
ment. All voltammetry studies were carried out at room tempera-
ture (20ꢁ28C). The supporting electrolyte [Bu4N][PF6] (GFS Chemi-
cals, 98%) was recrystallised twice from ethanol.[45]
Spectroscopic Methods
Absorbance spectra were run on a Varian model Cary 60 UV/Visible
spectrophotometer in 1 cm quartz cuvettes. Steady-state fluores-
cence measurements were taken on a Varian model Cary Eclipse
fluorescence spectrophotometer with intensity corrected for detec-
tor efficiency. Fluorescence quantum yields of compounds 4 and 5
were determined by using the comparative method with Rhoda-
mine 6G (Ff =0.95 in ethanol)[46] as the reference standard. Solu-
tions were prepared in 1 cm quartz cuvettes, with absorbance
maxima less than 0.10 in order to prevent inner filter effects. Sam-
ples were deoxygenated by bubbling with N2 for 20 min immedi-
ately prior to measurement of fluorescence emission spectra. Time-
resolved fluorescence decays were measured on a previously de-
scribed home-built set-up:[20] Briefly, excitation was provided by
the supercontinuum fiber laser described below, emission passed
through a monochromator (CVI, dk480) and detected by an avalan-
che photodiode (APD: Id-Quantique, Id-100). Photon timing was
achieved by using the PicoQuant hardware and software described
below. Data were fit by an exponential decay convolved with the
instrument response function (IRF) recorded from light scattered
by a solution of milk powder in water (IRF typically ~90–100 ps,
Supporting Information available: Synthesis of 2, NMR spectra for
2–5, electrochemical data for 4 and 5, X-ray crystallography data
for 5, molar absorption coefficients for 4 and 5, optical spectra for
4 and 5, time-resolved fluorescence data and calculation of radia-
tive and non-radiative rate constants, triplet lifetimes of single mol-
ecules of 4.
Acknowledgements
Financial support from the Australian Research Council
through the Discovery Grant Scheme (DP130101861) is grate-
fully acknowledged. We thank Craig Forsyth for technical assis-
tance with X-ray crystallography.
Keywords: fluorescence quantum yield
·
naphthalene
diimide · photostability · single-molecule studies · time-
resolved spectroscopy
Chem. Eur. J. 2015, 21, 4133 – 4140
www.chemeurj.org
4139
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[1] H. Brinkmann, C. Kelting, S. Makarov, O. Tsaryova, G. Schnurpfeil, D.
Woehrle, D. Schlettwein, Phys. Status Solidi A 2008, 205, 409–420.
[2] F. Babudri, G. M. Farinola, F. Naso, R. Ragni, Chem. Commun. 2007,
1003–1022.
[29] T. He, M. Stolte, F. Wꢃrthner, Adv. Mater. 2013, 25, 6951–6955.
[30] M. Stolte, S. L. Suraru, F. Wꢃrthner, J. H. Oh, Z. Bao, J. Brill, M. Kçne-
mann, J. Qu, U. Zschieschang, H. Klauck, Proc. of SPIE 2010, 7778,
777801–777808.
[3] T. Okamoto, K. Nakahara, A. Saeki, S. Seki, J. H. Oh, H. B. Akkerman, Z.
Bao, Y. Matsuo, Chem. Mater. 2011, 23, 1646–1649.
[31] B. J. Jung, J. Sun, T. Lee, A. Sarjeant, H. E. Katz, Chem. Mater. 2009, 21,
94–101.
[4] M. Gliboff, H. Li, K. M. Knesting, A. J. Giordano, D. Nordlund, G. T. Seidler,
J.-L. Bredas, S. R. Marder, D. S. Ginger, J. Phys. Chem. C 2013, 117,
15139–15147.
[5] M. G. Dhara, S. Banerjee, Prog. Polym. Sci. 2010, 35, 1022–1077.
[6] J. A. Gladysz, M. Jurisch, Top. Curr. Chem. 2012, 308, 1–24.
[7] M.-L. Tang, Z.-N. Bao, Chem. Mater. 2011, 23, 446–455.
[8] S. Maniam, A. B. Holmes, J. Krstina, G. A. Leeke, G. E. Collis, Green Chem.
2011, 13, 3329–3332.
[32] H. E. Katz, J. Johnson, A. J. Lovinger, W. Li, J. Am. Chem. Soc. 2000, 122,
7787–7792.
[33] B. A. Jones, A. Facchetti, T. J. Marks, M. R. Wasielewski, Chem. Mater.
2007, 19, 2703–2705.
[34] Z. Yuan, J. Li, Y. Xiao, Z. Li, X. Qian, J. Org. Chem. 2010, 75, 3007–3016.
[35] Y. Li, C. Li, W. Yue, W. Jiang, R. Kopecek, J. Qu, Z. Wang, Org. Lett. 2010,
12, 2374–2377.
[36] Crystal data for 5: [C41H19F44N4O4]; 0.25ꢂ0.12ꢂ0.05 mm3; a=5.0868(9),
¯
[9] H. E. Katz, A. J. Lovinger, J. Johnson, C. Kloc, T. Siegrist, W. Li, Y. Y. Lin, A.
Dodabalapur, Nature 2000, 404, 478–481.
b=15.041(2), c=16.0457(18) ;, a=92, b=90, g=988; triclinic; P1; Z=2,
V=1213.2(3) A3, 1calc =2.009 Mgmꢀ3
F(000)=721; l=0.71073 ꢁ; T=
;
[10] A. R. Gutierrez, D. G. Whitten, J. Am. Chem. Soc. 1974, 96, 7128–7130.
[11] T. C. Werner, Fluoresc. News 1976, 9, 1–4.
123(2) K; m=0.244 mmꢀ1; Nonius Kappa CCD diffractometer, f and w
scan data, 10079 data collected, 5458 unique reflections (Rint =0.0342)
and 10079 observed [I>2s(I)], 437 refined parameters; R1=0.0486;
Rw2=0.0950. Crystallographic data (excluding structure factors) for 5
have been deposited with the Cambridge Crystallographic Data Centre.
CCDC 1026275 contains the supplementary crystallographic data for
this paper. These data can be obtained free of charge from The Cam-
bridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_
request/cif.
[12] M. Rae, F. Perez-Balderas, C. Baleizao, A. Fedorov, J. A. S. Cavaleiro, A. C.
Tome, M. N. Berberan-Santos, J. Phys. Chem. B 2006, 110, 12809–12814.
[13] D. F. Eaton, J. Am. Chem. Soc. 1981, 103, 7235–7239.
[14] F. Wꢃrthner, S. Ahmed, C. Thalacker, T. Debaerdemaeker, Chem. Eur. J.
2002, 8, 4742–4750.
[15] M. A. H. Alamiry, A. C. Benniston, J. Hagon, T. P. L. Winstanley, H. Lemme-
tyinen, N. V. Tkachenko, RSC Adv. 2012, 2, 4944–4950.
[16] D. Yan, D.-K. Bucar, A. Delori, B. Patel, G. O. Lloyd, W. Jones, X. Duan,
Chem. Eur. J. 2013, 19, 8213–8219.
[37] C. Rçger, Y. Miloslavina, D. Brunner, A. R. Holzwarth, F. Wꢃrthner, J. Am.
Chem. Soc. 2008, 130, 5929–5939.
[17] C.-L. Sun, J. Li, H.-W. Geng, H. Li, Y. Ai, Q. Wang, S.-L. Pan, H.-L. Zhang,
Chem. Asian J. 2013, 8, 3091–3100.
[38] F. Doria, I. Manet, V. Grande, S. Monti, M. Freccero, J. Org. Chem. 2013,
78, 8065–8073.
[18] T. D. M. Bell, S. Yap, C. H. Jani, S. V. Bhosale, J. Hofkens, F. C. De Schryver,
S. J. Langford, K. P. Ghiggino, Chem. Asian J. 2009, 4, 1542–1550.
[19] H. F. Higginbotham, S. Maniam, S. J. Langford, T. D. M. Bell, Dyes Pigm.
2015, 112, 290–297.
[39] S. Guo, J. Sun, L. Ma, W. You, P. Yang, J. Zhao, Dyes Pigm. 2013, 96, 449–
458.
[40] Y. V. Suseela, M. Sasikumar, T. Govindaraju, Tetrahedron Lett. 2013, 54,
6314–6318.
[20] R. P. Cox, H. F. Higginbotham, B. A. Graystone, S. Sandanayake, S. J.
Langford, T. D. M. Bell, Chem. Phys. Lett. 2012, 521, 59–63.
[21] H. F. Higginbotham, R. P. Cox, S. Sandanayake, B. A. Graystone, S. J.
Langford, T. D. M. Bell, Chem. Commun. 2013, 49, 5061–5063.
[22] S. V. Bhosale, C. H. Jani, C. H. Lalander, S. J. Langford, I. Nerush, J. G.
Shapter, D. Villamaina, E. Vauthey, Chem. Commun. 2011, 47, 8226–
8228.
[41] M. Orrit, Photochem. Photobiol. Sci. 2010, 9, 637–642.
[42] M. Haase, C. G. Huebner, F. Nolde, K. Mꢃllen, T. Basche, Phys. Chem.
Chem. Phys. 2011, 13, 1776–1785.
[43] K. D. Weston, P. J. Carson, H. Metiu, S. K. Buratto, J. Chem. Phys. 1998,
109, 7474–7485.
[44] T. D. M. Bell, S. Yap, C. Jani, S. J. Langford, J. Hofkens, F. De Schryver, K. P.
Ghiggino, Proc. of SPIE 2007, 6444, 644404–1.
[23] K. A. Lee, V. Lozan, S. J. Langford, B. Kersting, Dalton Trans. 2009, 7481–
7485.
[45] Electrochemistry for Chemists, 2nd ed. (Eds.: D. T. Sawyer, A. Sobkowiak,
J. J. L. Roberts) Wiley, Hoboken, 1995.
[24] K. A. Willets, O. Ostroverkhova, M. He, R. J. Twieg, W. E. Moerner, J. Am.
Chem. Soc. 2003, 125, 1174–1175.
[25] W. E. Moerner, J. Phys. Chem. B 2002, 106, 910–927.
[26] S. V. Bhosale, C. H. Jani, S. J. Langford, Chem. Soc. Rev. 2008, 37, 331–
342.
[46] R. F. Kubin, A. N. Fletcher, J. Lumin. 1982, 27, 455–462.
[47] D. R. Whelan, T. Holm, M. Sauer, T. D. M. Bell, Aust. J. Chem. 2014, 67,
179–183.
[27] N. Sakai, J. Mareda, E. Vauthey, S. Matile, Chem. Commun. 2010, 46,
4225–4237.
Received: October 29, 2014
[28] S.-L. Suraru, F. Wꢃrthner, Angew. Chem. Int. Ed. 2014, 53, 7428–7448;
Angew. Chem. 2014, 126, 7558–7578.
Published online on January 29, 2015
Chem. Eur. J. 2015, 21, 4133 – 4140
www.chemeurj.org
4140
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim