Exploration of the Photodegradation of Naphtho[2,3-g] quinoxalines and Pyrazino[2,3-b]phenazines

Article abstract of DOI:10.1002/chem.201504453

Nitrogen-containing polycyclic aromatic hydrocarbons are very attractive compounds for organic electronics applications. Their low-lying LUMO energies points towards a potential use as n-type semiconductors. Furthermore, they are expected to be more stable under ambient conditions, which is very important for the formation of semiconducting films, where materials with high purity are needed. In this study, the syntheses of naphtho[2,3-g]quinoxalines and pyrazino[2,3-b]phenazines is presented by using reaction conditions, that provide the desired products in high yields, high purity and without time-consuming purification steps. The HOMO and LUMO energies of the compounds are investigated by cyclic voltammetry and UV/Vis spectroscopy and their dependency on the nitrogen content and the terminal substituents are examined. The photostability and the degradation pathways of the naphtho[2,3-g]quinoxalines and pyrazino[2,3-b]phenazines are explored by NMR spectroscopy of irradiated samples affirming the large influence of the nitrogen atoms in the acene core on the degradation process during the irradiation. Finally, by identifying the degradations products of 2,3-dimethylnaphtho[2,3-g]quinoxaline it is possible to track down the most reactive position in the compound and, by blocking this position with nitrogen, to strongly increase the photostability.

Full text of DOI:10.1002/chem.201504453

DOI: 10.1002/chem.201504453
Full Paper
&
Photodegradation
Exploration of the Photodegradation of Naphtho[2,3-g]
quinoxalines and Pyrazino[2,3-b]phenazines
Nicole Kolmer-Anderl,[a] Andreas Kolmer,[b] Christina M. Thiele,*[b] and Matthias Rehahn*[a]
Abstract: Nitrogen-containing polycyclic aromatic hydrocar-
bons are very attractive compounds for organic electronics
applications. Their low-lying LUMO energies points towards
a potential use as n-type semiconductors. Furthermore, they
are expected to be more stable under ambient conditions,
which is very important for the formation of semiconducting
films, where materials with high purity are needed. In this
study, the syntheses of naphtho[2,3-g]quinoxalines and pyra-
zino[2,3-b]phenazines is presented by using reaction condi-
tions, that provide the desired products in high yields, high
purity and without time-consuming purification steps. The
HOMO and LUMO energies of the compounds are investigat-
ed by cyclic voltammetry and UV/Vis spectroscopy and their
dependency on the nitrogen content and the terminal sub-
stituents are examined. The photostability and the degrada-
tion pathways of the naphtho[2,3-g]quinoxalines and pyrazi-
no[2,3-b]phenazines are explored by NMR spectroscopy of ir-
radiated samples affirming the large influence of the nitro-
gen atoms in the acene core on the degradation process
during the irradiation. Finally, by identifying the degrada-
tions products of 2,3-dimethylnaphtho[2,3-g]quinoxaline it is
possible to track down the most reactive position in the
compound and, by blocking this position with nitrogen, to
strongly increase the photostability.
merisation occurs during irradiation without oxygen.[5] Howev-
er, this reactivity especially of soluble pentacene and tetracene
derivatives is a disadvantage in terms of organic field effect
transistor applications were highly purified materials are
needed to achieve good performances. Hence, the sensibility
towards oxidation of these compounds is a disadvantage with
regard to the fabrication of stable organic field effect transis-
tors, especially because quinone compounds are known to act
as electronic traps.[6] That is why new organic semiconductors
with an increased durability towards photooxidation and low-
lying LUMO energies, that are suitable for n-type semiconduc-
tor applications, are needed. For this reason nitrogen-contain-
ing acenes are one focus of current research.[7]
Introduction
Polycyclic aromatic hydrocarbons, like tetracene and penta-
cene and their derivatives, are of special interest due to their
big potential as semiconductors in organic electronic applica-
tions, like organic field effect transistors (OFETs).[1] Especially
pentacene as excellent p-type semiconductor and its deriva-
tives are well studied also in terms of reducing LUMO energies
by adding electron-withdrawing substituents to generate n-
type semiconductors.[2] But compared to the number of known
organic p-type semiconductors only few organic semiconduc-
tors with n-type behaviour and good transistor properties are
known, although they are needed to create, for example, com-
plementary electronic circuits.[3]
Theoretical studies have already shown that increasing the
number of nitrogen atoms in the acene core is an efficient tool
to increase the electron affinity, which is advantageous for
stable n-type behaviour of organic semiconductors.[8] In addi-
tion, these materials exhibit low LUMO energies, which are
necessary for n-type semiconductors, too. Furthermore there
are several representatives that show good n-type properties
in OFET applications.[9] UV/Vis measurements indicate that the
nitrogen atoms increase the stability of acenes towards the un-
desirable oxidation process due to its higher electron affini-
ty.[9,10]
Unfortunately, polycyclic aromatic hydrocarbons are also
known for the photochemical reactions they undergo during
irradiation. On the one hand endoperoxide formation is report-
ed for pentacene and tetracene and several derivatives when
irradiated in the presence of oxygen.[4] On the other hand di-
[a] Dr. N. Kolmer-Anderl, Prof. Dr. M. Rehahn
Ernst-Berl-Institut für Technische und Makromolekulare Chemie
Technische Universität Darmstadt
Alarich-Weiss-Straße 4, 64287 Darmstadt (Germany)
E-mail: M.Rehahn@MC.tu-darmstadt.de
In this paper, we report the synthesis of naphtho[2,3-g]qui-
noxalines and pyrazino[2,3-b]phenazines and investigate the
influence of the nitrogen content in the tetracene core on the
LUMO energies and the stability towards photooxidation. For
this purpose the degradation process is studied by NMR spec-
troscopy and the chemical shifts of products and intermediates
are calculated to verify the proposed decomposition pathways.
[b] Dr. A. Kolmer, Prof. Dr. C. M. Thiele
Clemens-Schçpf-Institut für Organische Chemie und Biochemie
Technische Universität Darmstadt
Alarich-Weiss-Straße 4, 64287 Darmstadt (Germany)
E-mail: Cthiele@thielelab.de
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201504453.
Chem. Eur. J. 2016, 22, 5277 – 5287
5277
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Results and discussion
Table 1. Reaction conditions for the synthesis of the naphtho[2,3-g]qui-
noxaline derivatives 7a–7 f.
Synthesis
R
AcOH
T [8C]
t [min]
Yield [%]
The synthesis of the naphtho[2,3-g]quinoxalines 7a–7 f is
shown in Schemes 1 and 2. Starting from 2-aminoanthraqui-
none (1), 2,3-diaminoathraquinone (5) was prepared according
to the literature procedure of Schaarschmidt and Leu
(Scheme 1).[11] The reduction of 2,3-diaminoanthraquinone (5),
first reported by Clark,[12] was modified by using zinc powder,
which was activated with an aqueous copper sulphate solution
to ensure complete conversion of compound 5 (Scheme 2). Fi-
nally, the naphtho[2,3-g]quinoxalines 7a–7 f were synthesised
from 2,3-diaminoanthracene (6) and the corresponding 1,2-di-
ketones (Scheme 2). Except for 1,2-butandione, acetic acid was
used as catalyst in the final step of this reaction. Compared to
the reaction conditions described by Shitagaki et al.[13] for the
synthesis of phenyl-substituted naphtho[2,3-g]quinoxalines,
the selected conditions allow a drastic reduction of reaction
time from 24 h to 20–50 min. In addition the workup of the re-
action mixture is much easier as purification by column chro-
matography can be avoided, because the raw products can
easily be crystallised from toluene (compounds 7c–7 f) or pre-
cipitate directly from the reaction mixture (compounds 7a and
7b). With this method very good yields can be achieved. The
results for the derivatives 7a–7 f are listed in Table 1.
7a
7b
7c
7d
7e
7 f
CH3
CH2Br
C6H5
C6H4F
C6H4Br
C6H4OCH3
no
65
70
70
70
60
76
20
20
30
20
20
50
69
90
88
36
96
77
yes
yes
yes
yes
yes
Scheme 3. Synthesis of the pyrazino[2,3-b]phenazines 9a–9e. The reaction
conditions and yields are given in Table 2.
was added to catalyse the reaction leading to the advantage
that most of the products precipitate from the reaction mix-
ture and therefore, can be isolated easily with already high
purity. Table 2 lists the reaction conditions used and the yields
obtained.
The pyrazino[2,3-b]phenazines 9a–9e were synthesised in
analogy to the protocol reported by Armer et al.[14] 2,3-Diami-
nophenazine (8) was reacted with the appropriate 1,2-dike-
tones (see Scheme 3) but in contrast to the protocol of Armer
et al., tetrahydrofurane or methanol were used as solvent in-
stead of pure acetic acid. Only a small amount of acetic acid
Furthermore
2,3-dimethylnaphtho[2,3-g]quinoxaline-5,12-
dione (15a) and 2,3-dimethylnaphtho[2,3-g]quinoxaline-6,11-
dione (15b), which are supposed to be the degradation prod-
ucts of compound 7a, were synthesised, too (see Schemes 4
and 5). The synthesis of 2,3-dimethylnaphtho[2,3-g]quinoxa-
line-5,12-dione (15a) started from 1,4-dimethoxybenzene (10),
Scheme 1. Synthesis of 2,3-diaminoanthraquinone (5) according to Schaarschmidt and Leu.[11]
Scheme 2. Synthesis of the naphtho[2,3-g]quinoxalines 7a–7 f. The reaction conditions and yields for the last step are given in Table 1.
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5278
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 2. Reaction conditions for the synthesis of the pyrazino[2,3-b]phen-
azine derivatives 9a–9e.
R
Solvent
T [8C]
t [min]
Yield [%]
9a
9b
9c
9d
9e
CH3
CH2Br
C6H5
C6H5F
C6H5CF3
THF
70
70
70
70
70
120
45
360
47
62
18
71
72
53
methanol
methanol
methanol
methanol
Scheme 5. Synthesis of the quinone 15b, which is supposed to be one of
the degradation products of 2,3-dimethylnaphtho[2,3-g]quinoxaline (7a).
30
trolyte under a nitrogen atmosphere, by using the formula
from de Leeuw et al.[18] Ferrocene was used as internal stan-
dard. If the HOMO energies could not be obtained from cyclic
voltammetry measurements, they were also calculated from
the LUMO levels and the optical band gap measured by UV/Vis
spectroscopy (for more details see the Supporting Informa-
tion). Compared to compound 7a the LUMO levels of com-
pounds 7c–7 f are reduced by 0.15–0.24 eV and by 0.65 eV for
compound 7b (Figure 1).
which was converted into 2,3-dimethyl-5,8-dihydroxyquinoxa-
line (14) by a four-step procedure[15] (see Scheme 4).
As the first step, nitration of 1,4-dimethoxybenzene (10) was
performed in acetic acid with concentrated nitric acid.[16] Re-
duction with H2/Pd/C in ethyl acetate gave a mixture of com-
pounds 12a and 12b.[17] Separation of the regioisomers was
done in the next reaction step. Only compound 12a gave the
quinoxaline 13 during the reaction with 2,3-butanedione,
which then could be separated easily from the byproduct 12b.
Thus, the mixture of compounds 12a and 12b was reacted
with 2,3-butanedione in acetic acid affording compound 13,
which was transferred into compound 14 by reaction with
AlCl3. Condensation of 2,3-dimethyl-5,8-dihydroxyquinoxaline
(14) and o-phthaldialdehyde gave 2,3-dimethylnaphtho[2,3-g]
quinoxaline-5,12-dione (15a) in 70% yield. 2,3-Dimethylnaph-
tho[2,3-g]quinoxaline-6,11-dione (15b) was prepared from 2,3-
diaminoanthraquinone (5) and 2,3-butanedione in THF and
acetic acid in 80% yield.
Additionally, phenyl substituents increase the solubility of
the products, which is advantageous regarding solution-based
applications of the semiconducting film.
Comparing the naphtho[2,3-g]quinoxalines 7 and the pyrazi-
no[2,3-b]phenazines 9 with the same terminal substituent, the
LUMO energies of the pyrazino[2,3-b]phenazines are lowered
by 0.53 eV for the methyl derivative (i.e., compounds 7a and
9a), 0.56 eV for the phenyl derivatives (i.e., compounds 7c and
9c) and by 0.61 eV for the fluorophenyl derivatives (i.e., com-
pounds 7d and 9d) (Figure 2). From this it follows that the
phenyl substituents used affect the energy levels to a lesser
extent than the number of nitrogen atoms in the acene core.
Orbital energies
In order to investigate the influence of the substituents and
the nitrogen content on the HOMO and LUMO energies, cyclic
voltammetry (CV) and UV/Vis measurements were performed.
The HOMO and LUMO levels were obtained from cyclic voltam-
metry experiments (fully reversible), in dry THF with tetra-n-bu-
tylammonium hexafluorophosphate (0.1m) as supporting elec-
Possible degradation pathways and NMR spectroscopic
probes thereof
In principle, two degradation pathways for the naphtho[2,3-g]-
quinoxalines and the pyrazino[2,3-b]phenazines are expected.
On the one hand, photooxidation of one of the non-nitrogen-
Scheme 4. Synthesis of the quinone 15a, which is supposed to be one of the degradation products of 2,3-dimethylnaphtho[2,3-g]quinoxaline (7a).
Chem. Eur. J. 2016, 22, 5277 – 5287 www.chemeurj.org ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5279

Full Paper
of this [4+4] cycloaddition occurring, this leads to the forma-
tion of the 2N dimers 17a–17l (see Figure 3) and to the 4N
dimers 20a–20 f (see Figure 4).
The first question that needs to be addressed is which of
the two pathways of degradation is predominant. Thus, to
study the degradation of the naphtho[2,3-g]quinoxalines 7 and
the pyrazino[2,3-b]phenazines 9 NMR measurements were per-
formed. Therefore, saturated solutions (3–4 mg) of the methyl
derivatives of the naphtho[2,3-g]quinoxaline 7a and the pyrazi-
no[2,3-b]phenazine 9a in [D8]THF (0.5 mL, euriso-top 99.5% D)
were irradiated at room temperature and under air in an NMR
tube (Norell XR 55) with a 100 W light bulb. After certain time
intervals 1D (1H, 13C) and 2D (HSQC, HMBC) NMR spectra were
recorded on a Bruker AV-III 600 MHz NMR spectrometer (see
the Supporting Information for details) to identify the reactive
positions of the molecules 7a and 9a and thus gaining insight
on the decomposition pathways of the analysed compounds.
To differentiate between the two pathways, it is necessary to
identify the decomposition intermediates and products whose
signals appear in the recorded NMR spectra. There are two as-
pects: First, a principal distinction between the appearances of
either peroxide, quinone or dimer compounds is needed. Af-
terwards, in each of those compounds, the reactive position
has to be identified.
Figure 1. HOMO and LUMO energies of the naphtho[2,3-g]quinoxalines 7a–
7 f obtained from cyclic voltammetry measurements in a 0.1m solution of
tetra-n-butylammonium hexafluorophosphate in dry THF. Ferrocene was
used as internal standard. The value marked with an asterisk is estimated
from the LUMO energy obtained from cyclic voltammetry and the optical
band gap predicted from UV/Vis measurements (see the Supporting Infor-
mation). This has been necessary because decomposition of the compound
occurred during the cyclic voltammetry measurement.
For the first aspect—the differentiation of the decomposi-
tion pathways—the chemical shift of the protons and carbon
atoms at the bridgehead position and the expected multiplici-
1
ty (in the H spectra) of the endo-peroxides or dimers in each
case should be indicative. In terms of chemical shifts one
would expect chemical shifts of approximately d=80 ppm in
the 13C spectra for the endo-peroxides 16a–16c and 18a–18b,
whereas for the dimers 17a–17l and 20a–20 f chemical shifts
of approximately d=50 ppm in the carbon spectra are expect-
ed. Furthermore, it is expected that the peroxide compounds
16 convert into the corresponding 2N-quinones 15 in the
course of the photodegradation. The carbonyl 13C chemical
shifts of these are known to be d=180.1 ppm for compound
15a and d=181.1 ppm for compound 15b as they have been
obtained from synthesis for comparison purposes (see above).
In terms of the multiplicity the expected patterns for the
bridgehead proton NMR signals are shown in Figures 3 and 4
and Schemes 6 and 7 as denoted by the usual abbreviations (s
for singlet, d for doublet, dd for doublet of doublet). As can be
seen for both decomposition pathways the expected multiplic-
ities of the bridgehead protons do not allow a distinction be-
tween the peroxide and the dimerisation products. In addition,
the appearance of various signals for the various possible com-
pounds might lead to an overlap and further complicate the
identification and assignment of multiplet structures. Due to
the symmetry of most of the compounds, HMBC spectra are
not helpful in discriminating between the two pathways.
Figure 2. Comparison of the HOMO and LUMO energies of the methyl (i.e.,
compounds 7a and 9a), phenyl (i.e., compounds 7c and 9c) and fluoro-
phenyl derivatives (i.e., compounds 7d and 9d). The values marked with an
asterisk are estimated from the LUMO energy obtained from cyclic voltam-
metry measurements and the optical band gap predicted from UV/Vis meas-
urements (see the Supporting Information).
containing rings could occur, leading to the 2N-quinones 15a,
15b or 15c (see Scheme 6) and to the 4N-quinones 19a and
19b (see Scheme 7).
On the other hand, a [4+4] cycloaddition is also possible as
a degradation pathway, as this type of reaction is also known
from polycyclic aromatic hydrocarbons like tetracene.[6] In case
Thus, to conclude on the possible NMR spectroscopic pa-
rameters for the distinction of the two pathways: The only ac-
cessible information is the chemical shift of the bridgehead
(proton and) carbon atoms. The expected values for the chemi-
cal shifts (d=50 ppm for dimers, d=80 ppm for endo-perox-
ides) are thus backed up with the calculated chemical shifts
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5280
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Scheme 6. Photooxidation processes of compound 7a with the proposed nomenclature and the expected multiplicity of the bridgehead protons.
Scheme 7. Photooxidation processes of compound 9a with the proposed nomenclature and the expected multiplicity for the bridgehead protons.
(see Table 3 in which experimental and calculated chemical
shifts for all compounds observed in the following sections are
compared). For details on the calculation by using ORCA[19]
and CHESHIRE CCAT[20] see the Supporting Information. It
stable conditions for the irradiation process. After every hour
1H and 13C NMR spectra were recorded. In addition, HSQC and
HMBC spectra were measured after four and eight hours irradi-
ation time and an additional HMBC spectrum was recorded
after ten hours (see Figure 5). Irradiation is continued after-
wards. As no decomposition without irradiation has been ob-
served, the time for the NMR experiment does not need to be
taken into account.
1
needs to be pointed out that H chemical shift predictions are
not conclusive and are shown here only for comparison with
the experimental chemical shifts. From the test set shown in
the Supporting Information we can estimate an accuracy of
2 ppm for the calculated 13C chemical shifts. Differences of
more than 5 ppm are considered interpretable.
1
Figure 5a shows the measured H NMR spectra in which the
signal for compound 7a decreases (green signals) and several
additional signals appear during the irradiation process. The
most informative signals appearing are the ones of 1H chemical
shifts at d=6.15 (red signals) and 6.18 ppm (purple signals).
These are singlets with the corresponding 13C chemical shifts
of d=80.1 and 78.1 ppm, respectively, extracted from the
HSQC spectra (Figure 5b). Thus, these signals must belong to
the 2N-centre-peroxides 16a and 16b (for assignment, which
signal set belongs to which compound, see below). Based on
the multiplicity, compound 16c can be excluded (d expected);
Irradiation of 2,3-dimethylnaphtho[2,3-g]quinoxaline (7a) in
the presence of oxygen
First, the investigation of the photodegradation of compound
7a in the presence of oxygen is studied. A saturated solution
of 2,3-dimethylnaphtho[2,3-g]quinoxaline (7a) (3–4 mg of com-
pound 7a in 0.5 mL [D8]THF) was irradiated in one hour steps
in the presence of oxygen by a 100 W light bulb to generate
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5281
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 3. Possible products of the [4+4] photodimerisation of compound 7a with the expected multiplicity for the bridgehead protons. Each dimer that is
formed when one of the outer rings is involved is called 2N-outer-dimer. The ones where only centre rings are involved in the photodimerisation are called
2N-centre-dimers.
this is further confirmed by the calculated chemical shift for
the bridgehead carbon atom of compound 16c, which devi-
ates approximately 6 ppm from the experimentally observed
chemical shifts. Also, the dimers can be excluded based on
chemical shifts (dꢀ50 ppm expected). As expected for com-
When further irradiating the sample it can be observed, that
the signals of compounds 16a and 16b start to fade again,
whereas new signals appear. These can be assigned to the 2N-
quinones 15a and 15b (the spectra and assignments of which
are known because they were synthesised as compounds for
comparison, see above and Table 3). This is also confirmed by
the appearance of HMBC correlations in the carbonyl region
(see Figure 5d–e). From this it follows that the oxidation of
compound 7a leads to the 2N-centre-peroxides 16a and 16b
as intermediates and continues to the 2N-centre-quinones
15a and 15b.
As can be seen in the spectra, the signals for the endo-per-
oxides 16a and 16b do not appear in equal amounts and
thus, it would be intriguing to find out, which position is the
more reactive one. This is the one that is more prone to oxida-
tion.
1
pounds 16a and 16b the H NMR spectrum shows two sin-
glets at d=6.15 and 6.18 ppm for the bridgehead protons and
two singlets and four doublet doublets (of which the two at
d=7.38 ppm overlap) in the aromatic region of the spectra.
Moreover, the measured chemical shifts of the bridgehead pro-
tons are in a very good agreement with the chemical shifts
measured for the endo-peroxide of tetracene (d=6.13 ppm) by
Bjarneson and Petersen.[5]
Additionally, it is seen in the HMBC spectrum after 4 h (Fig-
ure 5c) that no correlation to a carbonyl carbon atom is ob-
servable yet.
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5282
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 4. Possible products of the [4+4] photo dimerisation of compound 9a with the expected multiplicity for the bridgehead protons. Each dimer that is
formed when one of the outer rings is involved is called 4N-outer-dimer. The ones where only centre rings are involved in the photodimerisation are called
4N-centre-dimers.
Table 3. A summary of the chemical shifts assigned from the decomposition spectra as shown in Figures 5, 6 and 7 (exptl), calculated chemical shifts
(calcd), and experimental chemical shifts of the quinones 15a and 15b as synthesised according to Schemes 4 and 5 (synthesised exptl) of the bridge
head atoms. For the unsymmetrical dimers 17c, 17d, 17e, 17 f, 17i, 17j, 20e and 20 f, the first value is the one for the “upper” half of the molecule as dis-
played in the Figures 3 and 4.
1H [ppm]
exptl
13C [ppm]
exptl
synthesised
exptl
1H [ppm]
exptl
13C [ppm]
exptl
calcd
calcd
calcd
calcd
16a
16b
16c
15a
15b
15c
17a
17b
17c
17d
17e
17 f
17g
17h
17i
6.15
6.18
5.33
5.28
4.93
–
–
–
4.38
4.33
4.93/4.58
5.08/4.58
4.63/4.53
4.68/4.53
5.12
80.1
78.1
79.6
77.8
74.6
177.0
178.5
180.7
49.7
51.3
55.3/49.9
55.4/50.1
54.7/49.3
55.1/48.8
54.9
–
–
–
18a
18b
6.36
5.08
4.78
81.5
81.5
76.8
–
–
–
180.2
181.0
180.1
181.1
19a
19b
–
–
–
–
178.6
174.9
178.7
-
–
–
–
–
–
–
–
–
–
–
–
–
20a
20b
20c
20d
20e
20 f
4.23
4.31
4.88
4.88
4.53/4.68
4.50/4.68
48.7
51.0
54.0
54.1
49.8/54.0
49.3/53.7
5.35
5.36
51.8
51.8
4.83–4.95
4.83–4.95
4.83–4.95
4.83–4.95
4.83–4.95
4.83–4.95
52.0–53.5
52.0–53.5
52.0–53.5
52.0–53.5
52.0–53.5
52.0–53.5
5.03
55.1
4.78/5.03
4.83/5.03
4.92
55.0/54.2
55.0/54.2
54.1
17j
17k
17l
4.83
54.3
One could speculate that the endo-peroxide 16a or 16b,
which is formed in a higher amount (purple signals in Figure 5)
would also lead to more quinone 15a or 15b. This, however,
would require that the two respective reaction rates are com-
parable. With this assumption and the known chemical shifts
of compounds 15a and 15b one could assign the more abun-
dant quinone to be compound 15b and thus, the more abun-
dant endo-peroxide to be compound 16b (purple signals in
Figure 5). We, however, prefer to have additional spectroscopic
evidence for the assignment, which is provided based on cor-
relations in the HMBC spectra, as will be discussed below.
The purple signals in Figure 5 can be related to the 2N-
centre-peroxide 16b because there are two HMBC correlations
from the bridgehead carbon atom at d=78.1 ppm (1H chemi-
cal shift d=6.18 ppm from HSQC) to proton signals at d=7.38
(dd) and 7.81 ppm (s) (data not shown). Thus, the bridgehead
must be “seeing” two neighbouring (3J(C,H)) protonated
carbon atoms with singlet and doublet of doublet multiplicity
of the attached protons.
This is expected only for compound 16b. For compound
16a only one correlation to a singlet is expected in the HMBC
spectrum. This is actually observed: one HMBC correlation
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5283
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 5. a) 1H NMR spectra of the irradiation of compound 7a with light in the presence of oxygen after 0, 1, 2, …, 10 h in [D8]THF. In addition, several
2D NMR experiments were recorded, a selection is shown. The green signals correspond to compound 7a, the red signals correspond to the 2N-centre-perox-
ide 16a, the purple signals correspond to the 2N-centre-peroxide 16b and the blue signals correspond to the 2N-centre-quinones 15a and 15b. b) HSQC
spectrum recorded after 4 h of irradiation. The peak at d=78.1 ppm, assigned to the 2N-centre-peroxide 16b, is larger than the one at d=80.1 ppm, as-
signed to the other 2N-centre-peroxide 16a. c) HMBC spectrum recorded after 4 h of irradiation. There are no peaks in the carbonyl region yet. d) HMBC re-
corded after 8 h of irradiation. There is one peak in the carbonyl region, which we assigned to the 2N-centre-quinone 15b. e) HMBC spectrum recorded after
10 h of irradiation. A second peak appears in the carbonyl region, which we assigned to the other 2N-centre-quinone 15a.
from the bridgehead proton signal (singlet) at d=6.15 ppm
(1H, d=80.1 ppm for 13C, red signals in Figure 5) to only one
other CH carbon atom, which is related through HSQC to a sin-
glet at d=7.93 ppm is observed (data not shown).
Hence, the red signals in Figure 5 must belong to the 2N-
centre-peroxide 16a.
As the intensity of the signals of the 2N-centre-peroxide
16b is larger than the one of the 2N-centre-peroxide 16a, the
corresponding position therefore, should be the most reactive
one. Furthermore, there are no signals in the NMR spectra that
can be assigned to the 2N-outer-peroxide 16c. Consequently,
this position should be the least reactive.
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5284
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Irradiation of 2,3-dimethylnaphtho[2,3-g]quinoxaline (7a) in
the absence of oxygen
of compound 7a (approximately 80%) is still left in the
sample. In the case of oxygen being present during the irradia-
tion, the signals of compound 7a disappear already after 5 h
of irradiation, showing that oxidation of compound 7a is faster
than dimerisation.
If oxygen is excluded during the irradiation process, the only
degradation pathway left is the dimerisation leading to the
compounds 17a–17l. In the proton spectra of solutions of
compound 7a irradiated in the absence of oxygen, which are
shown in Figure 6, several new signals appear at approximately
d=5 ppm. Their corresponding carbon chemical shifts with
d=52.0–53.5 ppm are significantly different from the carbon
chemical shifts of the peroxide compounds 16a–16c. We
assume that compounds 17a–17 f are not formed, as the
outer ring should also be less reactive in dimerisation reac-
tions. Furthermore, the calculated chemical shifts for the
bridgehead carbon atoms in compounds 17a–17 f always in-
clude a carbon atom at d=48–50 ppm (not observed) addi-
tionally to the one at approximately d=54 ppm. The observed
13C chemical shifts of d=52–53.5 ppm are closer to d=
54 ppm than to d=48 ppm, but this observation can only be
taken as an indication (as we estimate the accuracy of the cal-
culations to be 2 ppm). As many different dimerisation prod-
ucts are formed and spectra are severely overlapped, multiplic-
ities can only provide a rough guidance. However, if we
assume that compounds 17a–17 f are not formed due to the
reduced reactivity of the outer ring, one would expect four sin-
glets for the compounds 17g, 17h, 17k and 17l (symmetric,
see Figure 3) and four doublets for the dimers 17i and 17j
(not symmetric) in the region of d=5 ppm. This would lead to
twelve lines in this region, which is in good agreement with
what is actually observed.
Irradiation of 2,3-dimethylpyrazino[2,3-b]phenazine (9a) in
the presence of oxygen
Once the most reactive position of compound 7a is known, it
can be blocked by two nitrogen atoms to avoid degradation
through oxidation of this position, as is the case in 2,3-dimeth-
ylpyrazino[2,3-b]phenazine (9a). In order to prove that this
leads to a more stable compound, we tried to prepare a solu-
tion of 2,3-dimethylpyrazino[2,3-b]phenazine (9a) in [D8]THF.
The solubility of compound 9a in THF is very limited though
(we estimate the concentration of the sample to be 1–2 mg
per 0.5 mL). We have nevertheless irradiated the sample in
time steps of one hour under the same conditions as used for
the solution of 2,3-dimethylnaphtho[2,3-g]quinoxaline (7a), up
to a total irradiation time of 22 h (see the Supporting Informa-
tion). It was observed that the signals assigned to 2,3-dimeth-
ylpyrazino[2,3-b]phenazine (9a) disappear much slower com-
pared to 2,3-dimethylnaphtho[2,3-g]quinoxaline (7a) leading
to the assumption that 2,3-dimethylpyrazino[2,3-b]phenazine
(9a) is the more stable compound. The novel signals that ap-
peared in the spectrum might be assigned to the dimers 20c
and 20d because the observed multiplicity is in accordance
with the expectation (singlet) and the chemical shifts of these
signals are in agreement with what would be expected for
1
Combining these three indications it can be assumed that
the compounds 17g–17l are the main degradation products
of compound 7a, if oxygen is excluded during the irradiation.
Analysing the degradation spectra after 8 h of irradiation of
the oxygen-free sample, it was found that a significant amount
these two structures (see Table 3), when assuming that the H
chemical shifts of the bridgehead protons would appear at d=
4–5 ppm as is the case for the corresponding compounds 17.
However, neither 13C nor HSQC or HMBC spectra can be mea-
sured properly due to a low signal intensity.
Figure 6. 1H NMR spectra of the irradiation of compound 7a with light with exclusion of oxygen after 0, 1, 2, …, 8 h in [D8]THF. The grey signals correspond
to compound 7a. In addition, a HSQC spectrum recorded after 8 h of irradiation is shown, the proton trace of which shows eleven lines. The corresponding
carbon atoms possess chemical shifts of d=52.0–53.5 ppm, which indicates that they belong to the 2N-centre-dimers 17g–17l. No signals can be found in
the chemical shift regions where the 2N-peroxides appeared in Figure 5 (13C: d=75–80 ppm).
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5285
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
To gain better information about the degradation process,
a saturated solution of 2,3-dimethylpyrazino[2,3-b]phenazine
(9a) in CDCl3 (3–4 mg of compound 9a in 0.5 mL CDCl3) is irra-
diated under the conditions described above. Here, chloroform
dimers 20c and 20d. Looking at the integrals of the appearing
signals it can be assumed that the singlet at d=5.35 ppm and
the multiplets at d=7.59 and 7.88 ppm belong to the same
molecule. These values match to the predicted values for com-
pound 20c.
1
is a better solvent for this compound. The recorded H NMR
spectra are shown in Figure 7.
By using the same arguments, the singlet at d=5.36 ppm is
related to the multiplets at d=7.42 and 7.72 ppm, which
match to the predicted values for compound 20d.
Apparently, degradation in CDCl3 is even slower than in
[D8]THF (despite the higher concentration). Even after a total ir-
radiation time of 61 h approximately 60% of 2,3-dimethylpyra-
zino[2,3-b]phenazine (9a) are still present in solution. However,
it should be mentioned that during the irradiation process
a black precipitate has appeared, which could not be analysed.
It is nevertheless possible to measure 13C, HSQC and HMBC
spectra.
The singlet at d=6.36 ppm (brown signal in Figure 7) can
be correlated to the carbon signal at d=81.5 ppm through the
HSQC spectrum. Its existence and multiplicity shows that the
4N-centre-peroxide 18a is present. As additional evidence, the
experimental and calculated values for the bridgehead proton
and the bridgehead carbon atom are compared, which are in
good agreement. There should be more 1H signals present
that belong to compound 18a, but unfortunately, due to
a very low signal intensity and the many overlapping signals
the expected multiplets cannot be found/assigned safely.
At longer irradiation times, it is possible to observe signals
from the 4N-centre-quinone 19a (dark red signal in Figure 7).
In addition to the measured values for the four aromatic pro-
tons (multiplets at d=8.01 and 8.48 ppm) and the aromatic
CH carbon atoms (d=133.4 and 130.4 ppm) being in very
good agreement with the calculated values of this compound,
a carbon signal appears at d=178.6 ppm, which is assigned to
the carbonyl group.
We have conducted the assignment of compounds follow-
ing the same train of thought as described above in detail for
the degradation of compound 7a.
The situation is different to the one of compound 7a
1
though: Signals of H chemical shifts at d=5.35/5.36 ppm (in-
dicative of dimerisation products) and d=6.36 ppm (indicative
of endo-peroxides) start to appear at approximately the same
time (first signs can be seen after 5 h of irradiation time). For-
tunately, the expected number and multiplicity of signals is
significantly reduced due to the nitrogen substitution.
From the HSQC spectra, the singlets at d=5.35 (pink signal
in Figure 7) and d=5.36 ppm (dark pink signal in Figure 7) can
be correlated to the carbon signal(s) at d=51.8 ppm. Due to
the observed multiplicity (singlet) and the comparison of the
measured and calculated chemical shifts (as listed in Table 3) it
seems reasonable to assign these signals to the 4N-centre-
From this it follows that the degradation of 2,3-dimethylpyr-
azino[2,3-b]phenazine (9a) proceeds through the 4N-centre-
peroxide 18a resulting in the formation of the 4N-centre-qui-
none 19a, but is much slower than the oxidation of 2,3-di-
Figure 7. 1H NMR spectra of the irradiation of compound 9a with light in the presence of oxygen after 0, 1, 3, 5, 20, 22, 29, 42, 48 and 61 h in CDCl3. The
light green signals correspond to compound 9a, the brown signals correspond to compound 18a, the dark red signals correspond to compound 19a and
the pink and dark pink signals correspond to compounds 20c and 20d, respectively.
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5286
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
methylnaphtho[2,3-g]quinoxaline (7a). At the same time the
formation of the 4N-centre-dimers 20c and 20d in the pres-
ence of oxygen is also possible. Furthermore, the degradation
in general is much slower, as approximately 60% of compound
18a are still present after 61 h at a concentration comparable
to the ones used before.
These results confirm that by adding nitrogen atoms to the
acene core it is possible to influence the orbital energies and
the compound stability. Moreover, the decomposition to qui-
none compounds can be drastically reduced by either exclud-
ing oxygen or adding nitrogen atoms to the core, thus, reduc-
ing the number of potential electronic traps in the material.
This opens the opportunity to tailor the properties of organic
semiconductors with regard to their application.
Thus, compound 9a is significantly more stable towards
photodegradation as compared to compound 7a.
Conclusion
Acknowledgements
Naphtho[2,3-g]quinoxalines and pyrazino[2,3-b]phenazines
have been synthesised in very good yields by the condensa-
tion of 2,3-diaminoanthracene or 2,3-diaminophenazine with
the appropriate diketones, respectively. The proposed synthet-
ic protocol yields the desired products in high purity so that
time-consuming workup like column chromatography and
working under inert conditions can be avoided.
Generous support from the European Research Council
(through 2010-SG_257041), the MerckLab and the Merck KGaA
is gratefully acknowledged.
Keywords: degradation pathways
· NMR spectroscopy ·
OFETs · photochemistry · polycyclic aromatic hydrocarbons
This work shows that inserting nitrogen atoms into polycyc-
lic aromatic hydrocarbons is an efficient tool to tune the LUMO
energies and to increase the photostability of the products.
Cyclic voltammetry measurements suggest that the LUMO
energies are lowered by an increasing nitrogen content inde-
pendent from the terminal substituents. The terminal substitu-
ent does affect the HOMO and LUMO energies as well, but to
a minor extent.
[1] H. Dong, X. Fu, J. Liu, Z. Wang, W. Hu, Adv. Mater. 2013, 25, 6158–6183.
[2] Y. Sakamoto, T. Suzuki, M. Kobayashi, Y. Gao, Y. Fukai, Y. Inoue, F. Sato, S.
Tokito, J. Am. Chem. Soc. 2004, 126, 8138–8140.
[3] H. Usta, A. Fachetti, T. J. Marks, Acc. Chem. Res. 2011, 44, 501–510.
[4] J. Reichwagen, H. Hopf, A. Del Guerzo, J.-P. Desvergne, H. Bouas-Lau-
rent, Org. Lett. 2004, 6, 1899–1902; A. G. L. Olive, A. Del Guerzo, J.-L.
Pozzo, J.-P. Desvergne, J. Reichwagen, H. Hopf, J. Phys. Org. Chem. 2007,
20, 838–844.
By NMR measurements of irradiated samples of 2,3-dimeth-
ylnaphtho[2,3-g]quinoxaline (7a) and 2,3-dimethylpyrazino[2,3-
b]phenazine (9a), it has not only been possible to analyse the
dependency of the photostability on the nitrogen content, but
also the degradation pathway of compounds 7a and 9a has
been explored, showing that oxidation is the main degradation
pathway.
[5] D. W. Bjarneson, N. O. Petersen, J. Photochem. Photobiol. A 1992, 63,
327–335.
[6] J. E. Anthony, Angew. Chem. Int. Ed. 2008, 47, 452–483; Angew. Chem.
2008, 120, 460–492.
[7] U. H. F. Bunz, J. U. Engelhardt, B. D. Lindner, M. Schaffroth, Angew. Chem.
Int. Ed. 2013, 52, 3810–3821; Angew. Chem. 2013, 125, 3898–3910.
[8] M. Winkler, K. N. Houk, J. Am. Chem. Soc. 2007, 129, 1805–1815.
[9] Z. Liang, Q. Tang, J. Xu, Q. Miao, Adv. Mater. 2011, 23, 1535–1539.
[10] G. J. Richards, J. P. Hill, T. Mori, K. Ariga, Org. Biomol. Chem. 2011, 9,
5005–5017.
[11] A. Schaarschmidt, H. Leu, Justus Liebigs Ann. Chem. 1915, 407, 176–194.
[12] M. Clark, J. Chem. Soc. C 1966, 277–283.
[13] S. Shitagaki, H. Yamazaki, S. Seo, 2004, US2005/0003232A1.
[14] A. M. Amer, A. A. El-Bahnasawi, M. R. H. Mahran, M. Lapib, Monatsh.
Chem. 1999, 130, 1217–1225.
[15] W. F. Gum, M. M. JoulliØ, J. Org. Chem. 1967, 32, 53–59.
[16] F. E. King, N. G. Clark, P. M. H. Davis, J. Chem. Soc. 1949, 3012–3016.
[17] P. Hammershøj, T. K. Reenberg, M. Pittelkow, C. B. Nielsen, O. Hammer-
ich, J. B. Christensen, Eur. J. Org. Chem. 2006, 2786–2794.
[18] D. M. de Leeuw, M. M. J. Simenon, A. R. Brown, R. E. F. Einerhand, Synth.
Met. 1997, 87, 53–59; Y. Li, Y. Cao, J. Gao, D. Wang, G. Yu, A. J. Heeger,
Synth. Met. 1999, 99, 243–248; H. Krüger, S. Janietz, D. Sainova, A.
Wedel, Macromol. Chem. Phys. 2003, 204, 1607–1615.
[19] ORCA—an ab initio, density functional and semiempirical program
package, Version 2.8, F. Neese, Max Planck Institut für Bioanorganische
Chemie, Mühlheim an der Ruhr (Germany), 2011; F. Neese, WIREs
Comput. Mol. Sci. 2012, 2, 73–78.
Elucidation of the degradation products has been used to
identify the most reactive position in compound 7a. Chemical
shifts of the expected intermediates and products of the deg-
radation have been predicted, showing a very good agreement
with the measured data. In addition the quinones 15a and
15b have been synthesised and analysed by NMR spectrosco-
py. The measured NMR spectra show that these two com-
pounds are the degradation products of compound 7a. If
oxygen is excluded during the irradiation process of com-
pound 7a, degradation proceeds much slower. The recorded
NMR spectra of the oxygen-free samples of 2,3-dimethylnaph-
tho[2,3-g]quinoxaline (7a) suggest that in this case dimerisa-
tion occurs leading to the compounds 17g–17l.
Subsequently, the CH groups of the most reactive positions
in compound 7a have been substituted by nitrogen atoms
leading to compound 9a, which shows an increased stability
compared to compound 7a. Furthermore, the NMR studies un-
derline that the nitrogen atoms efficiently block the position
most susceptible to oxidation of 2,3-dimethylnaphtho[2,3-g]-
quinoxaline. Moreover the studies of compound 9a indicate
that through the decrease in sensitivity towards oxygen the di-
merisation process becomes more important.
[20] http://chenshirenmr.info/; M. W. Lodewyk, M. R. Siebert, D. J. Tantillo,
Chem. Rev. 2012, 112, 1839; M. W. Lodewyk, D. J. Tantillo, J. Nat. Prod.
2011, 1339, 1343; R. J. Jain, T. Bally, P. R. Rablen, J. Org. Chem. 2009, 74,
4017–4023; P. R. Rablen, S. A. Pearlman, J. Finkbiner, J. Phys. Chem. A
1999, 103, 7357–7363.
Received: November 5, 2015
Published online on February 25, 2016
Chem. Eur. J. 2016, 22, 5277 – 5287
www.chemeurj.org
5287
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim