Room-temperature free-radical-induced polymerization of 1,1′-(methylenedi-1,4-phenylene)bismaleimide via a novel diphenylquinoxaline-containing hyperbranched aromatic polyamide

Article abstract of DOI:10.1021/ma030039z

Two new diphenylquinoxaline-containing AB₂ monomers, 2,3-bis(4-aminophenyl)quinoxaline-6-carboxylic acid, 5 and 2,3-bis[4-(4-aminophenoxy)phenyl]quinoxaline-6-carboxylic acid, 9 were prepared and polymerized via the Yamazaki reaction to form th

Full text of DOI:10.1021/ma030039z

Macromolecules 2003, 36, 4385-4396
4385
Room-Temperature Free-Radical-Induced Polymerization of
1,1′-(Methylenedi-1,4-phenylene)bismaleimide via a Novel
Diphenylquinoxaline-Containing Hyperbranched Aromatic Polyamide
J on g-Beom Ba ek ,† J oh n B. F er gu son ,‡ a n d Loon -Sen g Ta n *,‡
University of Dayton Research Institute, 300 College Park, Dayton, Ohio 45469, and Polymer Branch,
AFRL/ MLBP, Materials & Manufacturing Directorate, Air Force Research Laboratory,
Wright-Patterson Air Force Base, Ohio 45433-7750
Received J anuary 21, 2003; Revised Manuscript Received March 30, 2003
ABSTRACT: Two new diphenylquinoxaline-containing AB2 monomers, 2,3-bis(4-aminophenyl)quinoxaline-
6-carboxylic acid, 5, and 2,3-bis[4-(4-aminophenoxy)phenyl]quinoxaline-6-carboxylic acid, 9, were prepared
and polymerized via the Yamazaki reaction to form the hyperbranched aromatic polyamides (designated
as II and III, respectively) with -NH2 as the reactive chain-end groups. Although these AB2 monomers
and their respective hyperbranched polymers are structurally similar except for the presence of a
p-phenyloxy spacer between the quinoxaline and p-aminophenyl segments in 9 and III, the physical and
chemical properties of both monomers and hyperbranched polymers are distinctly different. It is believed
that the tautomerism in 5 and II is likely the basis for these differences. Since III was only marginally
soluble in polar aprotic solvents in which II readily dissolved, a known, soluble hyperbranched polyamide
(I) was prepared from 3,5-bis(4-aminophenyloxy)benzoic acid for comparison purposes in a subsequent
blends study. The curing behaviors and thermal properties of the hyperbranched polyamides I and II
blended in 0.75-3.75 wt % with a common bismaleimide [1,1′-(methylenedi-4,1-phenylene)bismaleimide,
BMI] resin were studied with differential scanning calorimetry (DSC) and Fourier transform infrared
(FT-IR) spectroscopy. Whereas the DSC results indicated that I reacted normally with BMI in a Michael-
addition fashion, followed by homopolymerization of the excess BMI, II appeared to be able to initiate
free radical polymerization of BMI at room temperature after co-dissolution with BMI in N-methyl-2-
pyrrolidinone. The DSC results of the BMI/II blends indicated that, at g1.5 wt % of II, no exotherm
attributable to the thermal curing of BMI was detected. Electron spin resonance (ESR) experiments
confirmed that the parmagnetic species present in II were more reactive toward BMI in solution at room
temperature than the radical detected in I. This unique property of II to initiate room-temperature radical
polymerization of BMI makes it important as a prototype for the development of low-temperature,
thermally curable thermosetting resin systems for high-temperature applications.
In tr od u ction
tooling can be fabricated easily from relatively inexpen-
sive materials such as wood, fiberglass, or foam. How-
ever, for these processing conditions, the material sys-
tems (prepreg, liquid resin, and adhesive) must possess
characteristics that are conducive to be processed at low
temperatures and pressures (e.g., ∼65 °C and 14 psi)
and after post-cure in free-standing fashion should pro-
vide structural performance equivalent to current aero-
space standards (e.g., epoxy-3501). This approach, nev-
ertheless, is hampered by the lack of suitable material
systems that can be cured at temperatures below 65 °C
to form structures with high-temperature properties.
Historically, fabrication of high performance, polymer
matrix composite (PMC) structures for aircraft and
space systems applications on a low volume basis is very
costly. This is because the nonrecurring costs such as
tooling and capital equipment are the major cost drivers
for the low volume production. Since the autoclaves and
hardened tooling, traditionally required for the fabrica-
tion of large composite-based structural components,
have the lion’s share in the fabrication cost, the afford-
ability issue can be logically addressed by developing
nonautoclave resins and processes. Toward this end,
electron beam (E-beam) curing1 is very attractive be-
cause of the following important advantages: (i) rapid
curing (seconds to minutes as opposed to hours for
thermal curing); (ii) good penetration depth up to an
inch, where curing of a large and thick component can
be performed on a “conveyor-belt” process with an over-
head horn providing an adjustable e-beam coverage; (iii)
low residual thermal stress. However, a high-energy
electron-beam source (<250 keV to >1 MeV) is required,
necessitating some measures of personnel protection.
Another approach is to drastically lower the curing
temperature and pressure for the composites so that
As a continued effort of our research program to
explore and develop niche applications for aromatic
hyperbranched polymers,2 we describe herein the re-
sults on the synthesis and characterization of two new
2,3-diphenylquinoxaline-containing AB2 monomers,
where A ) CO2H and B ) NH2, and the respective
hyperbranched aromatic polyamides with -NH2 as the
reactive chain-end groups. Also described are the un-
expected and interesting results from a comparative
study via thermal analysis, FT-IR, and ESR methods
on the reactivity of one of the new quinoxaline-contain-
ing hyperbranched polyamides (with electron-deficient
polymer backbone) and a known hyperbranched poly-
amide (with relatively electron-rich repeat units)3 when
blended in small amounts with a common bismaleimide
resin. These results are relevant to the development of
nonautoclave, thermally curable thermosetting systems.
* Corresponding author. E-mail address: Loon-Seng.Tan@
wpafb.af.mil. Telephone: (937) 255-9141. Fax: (937) 255-9157.
† University of Dayton Research Institute.
‡ Air Force Research Laboratory.
10.1021/ma030039z CCC: $25.00 © 2003 American Chemical Society
Published on Web 05/20/2003

4386 Baek et al.
Macromolecules, Vol. 36, No. 12, 2003
Sch em e 1. Syn th esis of 2,3-Bis(4-a m in op h en yl)qu in oxa lin e-6-ca r boxylic Acid (5)
Resu lts a n d Discu ssion
ture to form the benzoylated R-hydroxyketone, 2, which
under the strongly acidic, aqueous conditions of HBr/
DMSO oxidation14 was first converted to an R-hydroxy-
ketone and oxidized to the target benzil compound, 3.
Subsequently, (iii) tandem condensation-cyclodehydra-
tion reactions of 4,4′-dinitrobenzil with 3,4-diaminoben-
zoic acid in refluxing acetic acid resulted in excellent
yield of 2,3-bis(4-nitrophenyl)quinoxaline-6-carboxylic
acid, 4. Finally, catalytic hydrogenation of the dinitro
intermediate led to the formation of the desired AB2
monomer, 2,3-bis(4-aminophenyl)quinoxaline-6-carboxy-
lic acid, 5.
Poly(phenylquinoxalines) (PPQ’s) and aromatic poly-
amides are important classes of high performance
polymers.4 Although PPQ’s have several attractive
features such as high glass-transition temperatures,
heat-resistance, and amorphous character leading to
excellent processability,5 the utilization of PPQ’s, in
comparison with aromatic polyamides, is less wide-
spread mainly because of the high cost associated with
the preparations of their bis(dicarbonyl) and bis(o-
diaminophenyl) monomers as well as the choice of
m-cresol as the polymerization medium that is catalytic6
but relatively corrosive and toxic. Thus, the preformed
phenylquinoxaline monomers with suitable functions for
step-growth polymerizations provide an attractive ap-
proach to circumvent this problem.7,8,9 While there are
several examples of linear aromatic polyamides contain-
ing main-chain phenylquinoxanine10 or 2,3-bis(1,4-phen-
ylene)quinoxaline11 moieties described in the literature,
to our knowledge, analogous hyperbranched polyamides
have not been reported. However, a hyperbranched poly-
(phenylquinoxaline-arylene ether) was recently syn-
thesized, and its chain ends were appropriately func-
tionalized so as to control the morphological structures
of organic-inorganic hybrids.12
Mon om er Syn th esis. Since we are interested in
using an amide-forming reaction as the step-growth
process in conjunction with the facts that 3,4-diami-
nobenzoic acid is commercially available and can easily
react with suitable benzil derivatives to form the qui-
noxaline compounds, we designed our diphenylquinoxa-
line-containing AB2 monomers accordingly.
Thus, the AB2 monomer, 2,3-bis(4-aminophenyl)-
quinoxaline-6-carboxylic acid, was synthesized accord-
ing to Scheme 1. Initially, we attempted the synthesis
of an R-hydroxyketone dinitro compound, a key inter-
mediate to 4,4′-dinitrobenzil from the benzoin condensa-
tion of 4-nitrobenzaldehyde in the presence of KCN.
Unfortunately, the effort resulted in poor yield of the
corresponding R-hydroxyketone. Thus, the synthesis of
4,4′-dinitrobenzil was prepared in three steps. The first
two are (i) in situ benzoylation of the nitrobenzaldehyde
cyanohydrin anion13 and isolation of 2-benzoyloxy-2-(4-
nitrophenyl)acetonitrile, 1, (ii) condensation of 1 and
4-nitrobenzaldehyde in a 10% NaOH(aq)/benzene mix-
The AB2 monomer, 2,3-bis[4-(4-aminophenoxy)phenyl]-
quinoxaline-6-carboxylic acid, was synthesized accord-
ing to Scheme 2. Double condensation reaction of
commercially available 4,4′-dimethoxybenzil with 3,4-
diaminobenzoic acid in refluxing acetic acid resulted in
excellent yield of 2,3-bis(4-dimethoxyphenyl)quinoxa-
line-6-carboxylic acid, 6. Demethylation of 6 in refluxing
48% hydrobromic acid-acetic acid mixture afforded 2,3-
bis(4-hydroxyphenyl)quinoxaline-6-carboxylic acid, 7,
which was subsequently subjected to an aromatic nul-
ceophilic substitution reaction with 4-fluoronitrobenzene
in the presence of potassium carbonate and 18-crown-6
under Dean-Stark conditions to result in the isolation
of 2,3-bis(4-nitrophenyloxyphenyl)quinoxaline-6-carbox-
ylic acid, 8. Finally, catalytic hydrogenation of the
dinitro intermediate led to the formation of the desired
AB2 monomer, 2,3-bis[4-(4-aminophenoxy)phenyl]qui-
noxaline-6-carboxylic acid, 9.
The AB2 monomer (11) for hyperbranched polymers
I was prepared according to the reported procedures
(Scheme 3).3
1H NMR a n d Ta u tom er ism of Mon om er 5. Sur-
prisingly, monomer 9 turned out to be less soluble in
the solvents that easily dissolved monomer 5 even
though the presence of two extra phenyl ether groups
should render 9 structurally more flexible and in turn,
more soluble. Furthermore, when monomer 5 (orange
solid) dissolved in NMP, a red solution resulted whereas
the NMP solution of monomer 9 (light yellow solid) was
yellow, suggesting greater conjugation length in the
molecular structure of 5. The tautomerism of monomer
5 in polar aprotic solvent would be a plausible explana-
tion for the observed disparity in solubility because

Macromolecules, Vol. 36, No. 12, 2003
Free-Radical-Induced Polymerization 4387
Sch em e 2. Syn th esis of 2,3-Bis[4-(4-a m in op h en oxy)p h en yl]qu in oxa lin e-6-ca r boxylic Acid (9)
Sch em e 3. Syn th esis of 3,5-Bis(4-a m in op h en oxy)ben zoic Acid (11)
Sch em e 4. Syn th esis of
2,3-Dip h en ylqu in oxa lin e-6-ca r boxylic Acid (12)
precursor to those of monomer 9 and its nitro precursor
reveals that changing the distal nitro group to amino
group only slightly shifted Ha, Hb, and Hc signals
upfield, consistent with the inductive effect exerted by
p-aminophenoxy group. However, this is not the case
for monomer 5 and its nitro precursor; a much larger
upfield shift was observed (e.g., ∼1.5 ppm for Ha; see
Figure 1).
Since there are theoretically three tautomers for
monomer 5, namely the bis(4-aminophenyl)quinoxaline
form (5), the bis(quinoneimine)piperazine form (5b), and
there are more intermolecular hydrogen-bonding pos-
sibilities with polar aprotic solvents in 9 than 5. To
support the hypothesis, we examined closely the proton
nuclear magnetic resonance (1H NMR) spectra of the
diphenylquinoxaline compounds prepared in our labora-
tory and the 1H NMR spectral data reported for a
number of similar 6-substituted 2,3-diarylquinoxaline
compounds (where 6-substituents are H, Cl, F, CF3, and
3-aminophenoxy).15,16 It appears that Ha, Hb, and Hc of
the benzene component of quinoxaline ring have char-
acteristically their chemical shift (δ) values in the range
of 7.6-8.5 ppm. The predominant deshielding effect
(downfield shift) on these protons is due to the strongly
electron-withdrawing nature of the pyrazine component.
For 2,3-diphenylquinoxaline-6-carboxylic acid, 12 (see
Scheme 4), the range is 8.2-8.6 ppm, and for monomer
9, 8.3-8.6 ppm. As depicted in Figure 1, the chemical
shifts of Ha, Hb, and Hc of monomer 5 are in the range
of 6.6-7.2 ppm, which is shifted upfield in comparison
with the aforementioned 6-substituted 2,3-diarylqui-
noxaline compounds (∆ ∼ 1.0-1.3 ppm), diphenylqui-
noxaline-6-carboxylic acid (∆ ∼ 1.6 ppm) and monomer
9 (∆ ∼ 1.4-1.7 ppm). Such large upfield shifts can only
be effected by the dramatic change in the electronic
character of the C4N2 fused ring from strongly electron-
withdrawing (pyrazine ring) to strongly electron-donat-
ing (piperazine ring).17 In addition, a comparison be-
1
the intermediate form, 5a , we used a H NMR predic-
tion/simulation software program18 to generate the
simulated proton NMR spectra in the aromatic proton
region (δ range ) 6.00-9.00 ppm) for the these tau-
tomers. The results are shown in Figure 2. In compari-
son with the experimental 1H NMR spectrum for mono-
mer 5 (Figure 1a), it is quite clear that the experimental
spectrum is more closely matched with the simulated
spectrum corresponding to the bis(quinoneimine)pip-
erazine form (Figure 2, bottom) than the other two
spectra. Thus, we tentatively conclude that only bis-
(quinoneimine)piperazine form (5b) was present in
solution. This is very surprising because from the
enthalpy consideration, the bis(4-aminophenyl)quinoxa-
line form (5) should be more stable than the other two
tautomers since there is a loss of the resonance energies
in benzene ring and to lesser extent, piperazine ring in
going from the 5 form to the 5a and 5b forms. As there
are more intermolecular hydrogen bonding sites in 5b
than 5 and 5a 19 for the interactions with polar aprotic
solvents (H-bonding acceptors), the heat of solvation
may be able to compensate for the loss in resonance
energy and drive the tautomerization process. However,
the existence of 5 and 5a in solution cannot be ruled
out completely because tautomerization processes20 can
also be influenced by the nature of the solvent.21
1
tween the H NMR spectra of monomer 5 and its nitro

4388 Baek et al.
Macromolecules, Vol. 36, No. 12, 2003
1
F igu r e 2. Computer-simulated H NMR spectra for the three
tautomers for monomer 5: the bis(4-aminophenyl)quinoxaline
form, 5 (top), the intermediate form, 5a (middle), and the bis-
(quinoneimine)piperazine form, 5b (bottom). Note that the
ACD software program was unable to distinguish the labile
protons such as OH and NH.
F igu r e 1. 1H NMR spectra: (a) monomer 5 and its nitro
precursor; (b) monomer 9 and its nitro precursor.
Hyp er br a n ch ed P olya m id e/BMI Blen d s. Our ini-
tial objective of preparing the hyperbranched poly-
amides was to investigate their use as reactive additives
for high performance thermoset resins such as bisma-
leimides (BMI) to improve the fracture toughness
properties based on the work by Månson et al. that the
addition of aliphatic hyperbranched polyesters increased
significantly the toughness of epoxy resins.24 Since the
end groups in the ether-amide (I) and diphenylqui-
noxaline-amide (II) hyperbranched polymers are amine
functions, which can thermally react with maleimide via
Michael addition without the generation of volatile
byproducts, they were ideal for blends studies. This
effort was carried out in parallel with our recently
reported work on the blends of a BMI resin with an
allyl-terminated hyperbranched poly(arylene-ether-
ketone-imide) (AT-P AEKI).25
We selected 1,1′-(methylenedi-4,1-phenylene)bismale-
imide as the BMI component in this study mainly be-
cause it is the most common component used in com-
mercially available BMI formulations. Although both
hyperbranched polymers I and II were found to be able
to form homogeneous and transparent melts with this
BMI at temperatures 160-165 °C, it was more conve-
nient to prepare the blends via solution mixing. Thus,
the mixtures of BMI and I or II were first completely
dissolved in NMP and the solvent was then removed
under the reduced pressure (1 mmHg) at 100 °C for
200 h. The DSC thermograms were run on powder
samples after they had been heated to 200 °C, cooled to
20 °C, heated to 320 °C, cooled to 20 °C, and heated
to 320 °C again with heating and cooling rates of 10
Ar om a t ic H yp er b r a n ch ed P olym er s, I, II, a n d
III. The direct polycondensation of the three monomers
via the Yamazaki reaction22 was conducted following
literature procedures.3,23 The hyperbranched aromatic
polyamides with the structures for the repeat units of
I-III as shown in Scheme 5, were obtained as white
(I) or light brown (II and III) powders from monomers
11, 5, and 9, in that order. It is noteworthy that only
polymers I and II are soluble in polar aprotic solvents,
such as DMF, DMAc, NMP, and DMSO, and polymer
III is not. This precludes the use of III in the BMI
blends study (vide infra). Therefore, hyperbranched
polymer I, a known polymer,3 was prepared for com-
parison purposes.
As expected (see Table 1), the intrinsic viscosity
values for both hyperbranched polymers I and II are
low and fortuitously the same (0.19 dL/g). From dif-
ferential scanning calorimetry (DSC) experiments, the
glass-transition temperature (Tg) of II (252 °C) is found
to be 35 °C higher than that of I. However, no Tg was
detected for III from room temperature to 450 °C via
either DSC or thermomechnical analysis (TMA). The
thermal and thermooxidative stabilities of both het-
eroaromatic hyperbranched polymers (powder samples)
were considerably higher (∼142-174 °C in air, and
∼174 °C in helium) than I as indicated by temperatures
at which a 5% weight losses from thermogravimetric
analysis (TGA) experiments. This is consistent with the
high thermal stability observed for linear aromatic poly-
(quinoxaline-amides).10

Macromolecules, Vol. 36, No. 12, 2003
Free-Radical-Induced Polymerization 4389
Sch em e 5. Rep ea tin g Un its of Am in e-Ter m in a ted Hyp er br a n ch ed P olym er s
Ta ble 1. In tr in sic Viscosities a n d Th er m a l P r op er ties of
Am in e-Ter m in a ted Hyp er br a n ch ed P olym er s]
polymer II content increased (Table 3 and Figure 4).
We were surprised to observe that there was a dramatic
reduction in the intensity of the cure exotherm (∼31%
of ∆Hexo BMI) when only 0.75 wt % II was present in
the blend. Furthermore, at all other compositions, no
cure exotherm was detected by DSC as the curing had
already taken placed prior to DSC runs. The results
were intriguing, as we had expected that the amine
functions of II to be much less nucleophilic than those
of II because of the π-accepting nature of quinoxaline.
The greatly reduced nucleophilicities of the amino
functions of the quinoxaline derivatives, 2,3-bis(4-ami-
nophenyl)quinoxaline and 2,3-bis(4-aminophenyl)-6-
methylquinoxaline, were noted when they were inves-
tigated as curing agents of epoxy resins. The exotherms
on DSC thermograms of the mixtures of the diglycidyl
ether of bisphenol A (DGEBA) with them were observed
at higher temperatures than that of the mixture of
DGEBA with a common aromatic diamine such as 4,4-
diaminodiphenyl sulfone.26
Td5% (°C)c
entry
[η] (dL/g)a
Tg (°C)b
in air
in helium
I
II
III
0.19
217
252
e
360
502
523
359
532
532
0.19
insol.d
a
b
Intrinsic viscosity determined in NMP at 30 ( 0.1 °C. Glass
transition temperature (Tg) determined by DSC with a heating
rate of 10 °C/min. c The temperature at which 5% weight loss
occurred on TGA thermogram obtained with a heating rate of 10
°C/min. Insoluble in NMP. e Not observed.
d
Ta ble 2. Th er m a l P r op er ties of BMI a n d I Blen d sa
I content
(wt %)
mp (°C)
∆Hf (J /g)
Texo (°C)
∆Hexo (J /g)
0.00
0.75
1.50
2.25
3.00
3.75
161.7
154.0
152.7
150.4
151.7
152.7
105.1
78.1
83.9
69.6
64.7
64.1
252.7
261.9
260.4
253.6
248.0
247.3
-130
-160
-205
-241
-185
-157
The aforementioned DSC samples were also subjected
to an FT-IR study. The representative infrared spec-
tra are shown in Figure 5. Three important deductions
can be drawn from the comparison of these spectra: (i)
the spectra of the BMI/II (3 wt %) blend taken before
and after DSC curing (Figure 5, parts e and f, respec-
tively) are practically identical, confirming the DSC
result that this blend sample contained cured BMI
without any prior heat treatment; (ii) the spectra of BMI
(Figure 5a) and the BMI/I(3 wt %) blend (Figure 5c)
before thermal cure are practically identical, revealing
the absence of any chemical reactions (Michael addition
and/or radical polymerization) between BMI and II at
room temperature; (iii) the spectra of BMI and the
BMI/I blend that were similarly cured under DSC
conditions (Figure 5, parts b and d, respectively) as well
as the spectra of the BMI/II blend (Figure 5, parts e
and f) are basically the same, implicating that the
chemical structures of the cured BMI in all three
samples should resemble one another and have arisen
from the same curing mechanism. All these four spectra
depicted a characteristic strong broad band centered at
1167-1168 cm-1. This band was assigned to the C-N-C
(succinimide ring) stretch; the corresponding C-N-C
(maleimide ring) stretch is present as a strong band at
1149 cm-1 in the spectrum of uncured BMI.27 To sum
up, the IR results clearly pointed out that (a) the
diametrically opposite reactivities of I and II toward
BMI at room temperature, and (b) Michael addition
reaction did not occur in both BMI/I and BMI/II blends
at room temperature, strongly implicating the radical
polymerization of BMI at room temperature in the
presence of II.
a
Ta ble 3. Th er m a l P r op er ties of BMI a n d II Blen d s
II content
(wt %)
mp (°C)
∆Hf (J /g)
Texo (°C)
∆Hexo (J /g)
0.00
0.75
1.50
2.25
3.00
3.75
161.7
139.4
138.4
133.0
128.4
a
105.1
34.0
2.4
1.3
1.3
a
252.7
224.0
a
a
a
a
-130
-42
a
a
a
a
a
Not observed.
°C/min. The thermal analysis data are summarized in
Tables 2 and 3. In the case of I/BMI blends (Table 2
and Figure 3), the melting points (mp) of BMI compo-
nent decreased approximately 10-12 °C as compared
to pure BMI. The heat of fusion (∆Hf) decreased as
content of I increased. We observed that the peak
temperature (Texo) during curing increased to 262 °C (∼9
°C > pure BMI) and then decreased to 247 °C (∼5 °C <
pure BMI) as the hyperbranched polymer I content
varied from 0.75% to 3.75%. The heats of curing (∆Hexo
)
were higher at all compositions, indicating that the large
numbers of free amine at chain ends indeed reacted with
BMI. However, the amount of enthalpy detected during
curing also showed a maximum value at relatively low
polymer I content, 2.25 wt %.
In the case of II/BMI blends, both melting points and
heat of fusion (∆Hf) of BMI decreased as hyperbranched
Finally, for the sake of reproducibility, three vial
samples of NMP (1 mL) solutions were prepared con-

4390 Baek et al.
Macromolecules, Vol. 36, No. 12, 2003
F igu r e 3. DSC thermograms of a I/BMI blend with heating and cooling rate of 10 °C/min: (a) first heating scan, (b) second
heating scan, (c) second cooling scan, and (d) third heating scan.
F igu r e 4. DSC thermograms of a II/BMI blend with heating and cooling rate of 10 °C/min: (a) first heating scan, (b) second
heating scan, (c) second cooling scan, and (d) third heating scan.
taining (a) BMI/II (200 mg/1.5 mg), (b) BMI/II (200
mg/3.0 mg), and (c) II (200 mg). The resulting solutions
were all clear and (a) orange, (b) yellow-orange, and (c)
pink, in that order. They were allowed to stand at room
temperature, and several hours later, the solutions a
and b were cloudy and insoluble gels started to precipi-
tate from the mixture, indicative of some polymerization
reaction under ambient conditions. After 24 h, at the
bottom of the vials, both samples a and b formed clear,
red gels that remained immobile when the vials were

Macromolecules, Vol. 36, No. 12, 2003
Free-Radical-Induced Polymerization 4391
ent from that of II alone. The existence of a multiplet
in the ESR spectrum for II is an indication that the
radical species in II is more mobile than I and, there-
fore, also has a greater probability to hop from II to BMI
monomer and initiate the polymerization. Unfortu-
nately, the overlap of individual structure elements is
too great to allow an exact determination of multiplet
detail.30 A number of possible paramagnetic species
present in II are shown in Scheme 6.
Con clu sion
In summary, we have prepared two new diphenylqui-
noxaline-containing AB2 monomers, 5 and 9 and their
hyperbranched polyamides, II and III. Although these
AB2 monomers and their respective hyperbranched
polymers are structurally similar except for the presence
of a p-phenyloxy spacer between the quinoxaline and
p-aminophenyl groups in 9 and III, the physical and
chemical properties of both monomers and hyper-
branched polymers are noticeably different. In the case
of the AB2 monomers, the observations on their solution
behaviors in polar aprotic solvents; i.e., their relative
1
solubilities, colors and H NMR spectra suggested that
their molecular structures must be distinctly different.
Thus, we believe that the tautomerism that exists only
for the AB2 monomer 9 (without the p-phenyloxy spac-
ers) is a reasonable explanation. Whereas hyper-
branched polymer II remained in solution (NMP)
throughout the polymerization process, III precipitated
out of the solution after substantial viscosity buildup
and would not redissolve in polar aprotic solvents after
the usual workup. This is contrary to our expectation
since the presence of p-phenyloxy units should make III
more flexible and soluble than II. With the understand-
ing that the nature of myriad end groups collectively
control the solubility of the hyperbranched polymer,
among other physical properties, it is possible that the
intermolecular hydrogen-bonding between the terminal
amino group and the quinoxaline units (see Scheme 7)
provides the driving force for a large-scale aggregation,
which eventually led to the observed precipitation
during the polymerization process of III. The inability
of polar aprotic solvents to penetrate such deeply
imbeded H-bonded structures resulted in the observed
marginal solubility of III31 In the case of II, the
predominant bis(iminophenyl)piperazine form of the end
groups can interact with polar aprotic solvents which
are strong H-bond acceptors. Finally, the results from
the blends study indicate that the amine function in
conjugation with a phenyloxy group behaves like normal
nucleophile and reacts with maleimide via a Michael
addition reaction whereas the amine in conjugation with
a quinoxaline ring is more prone to be engaged in radical
processes. Since tautomerization is an equilibrium
process that not only involves the rapid hopping of a
proton from one heteroatom to another, the electrons
are also moving back and forth. Therefore, we believe
that the electrons in the tautomeric segments of hyper-
branched polymer II is more mobile and susceptible to
facile air oxidation to generate stable nitrogen-centered
radical-cations. We propose that such radical species are
likely to be responsible for the observed room-temper-
ature reactivity toward BMI. Since it is known that
quinoxaline derivatives in conjunction with UV light
and a tertiary amine could initiate free-radical polym-
erization of methyl methacrylate,32 we have yet to
consider the photochemical effect in our future work in
addition to the unequivocal identification of the respon-
F igu r e 5. FT-IR (KBr pellet) spectra: (a) pure BMI before
thermal cure, (b) BMI after thermal cure, (c) BMI/I (3 wt %)
before thermal cure, (d) BMI/I (3 wt %) after thermal cure, (e)
BMI/II (3 wt %) before thermal cure, and (f) BMI/II (3 wt %)
after thermal cure.
inverted. Sample c, however, was still a pink and clear
solution.28 These observations in conjunction with the
results from the DSC and FT-IR studies clearly ruled
out Michael addition reaction at room temperature,29
reaffirming the possibility of free-radical polymerization
of BMI promoted by II. Indeed, the electron spin
resonance (ESR) experiments of II and a BMI/II blend
indicated that active paramagnetic species did appear
in II and reacted with BMI in solution at room temper-
ature.
To probe the active species that were responsible for
initiating the curing of a BMI under ambient conditions,
we prepared the following four samples for an electron
spin resonance (ESR) study: I, II, BMI/I (3.75 wt %)
blend, and BMI/II (3.75 wt %) blend. As shown in Figure
6, parts b and d, the sample containing only hyper-
branched polymer I and the BMI/I (3.75 wt %) blend
showed almost identical ESR behavior. Since pure BMI
is diamagnetic and shows no ESR signal, the signal for
both samples must be due to some radical species
associated with the hyperbranched polymer I. As there
are numerous aromatic amine groups present as the end
groups of the hyperbranched polymer, the ESR signal
is most likely due to the oxidized amines that formed
radical cations as in the case of aromatic amine serving
as antioxidants. Ostensibly, these radical species are not
reactive enough to initiate room-temperature polymer-
ization of BMI. On the other hand, the ESR spectrum
for hyperbranched polymer II (Figure 6a) suggests the
presence of a structurally different radical species in the
sample. For the sample BMI/II (3.75 wt %) blend, the
resulting ESR spectrum (Figure 6c) is distinctly differ-

4392 Baek et al.
Macromolecules, Vol. 36, No. 12, 2003
F igu r e 6. ESR spectra: (a) II, (b) I, (c) II (3.75 wt %)/BMI, and (d) I (3.75 wt %)/BMI.
Sch em e 6. Molecu la r Str u ctu r es of P r op osed
Ca tion -Ra d ica l (Left Colu m n ) a n d Dica tion -Ra d ica ls
(Righ t Colu m n ) a n d Th eir Reson a n ce Str u ctu r es
Sch em e 7. Id ea lized Str u ctu r e Sh ow in g th e
In ter d igita tion of Hyp er br a n ch ed Ma cr om olecu les of
III via Hyd r ogen Bon d in g
and provides an important step in the design and
development of nonautoclave, thermally curable ther-
mosetting systems.
Exp er im en ta l Section
Ma ter ia ls. All reagents and solvents were purchased from
Aldrich Chemical Inc. and used as received, unless otherwise
specified. N-Methyl-2-pyrrolidinone (NMP) was distilled under
reduced pressure over phosphorus pentoxide. The commer-
cially available AB2 monomers and 3,5-diaminobenzoic acid
(Aldrich Chemical Inc.) were recrystallized from deoxygenated
water after treating with charcoal to give white needles for
the first two and white flakes for the latter, respectively (mp
238 °C, dec, mp >300 °C, respectively). 1,1′-(Methylenedi-4,1-
phenylene)bismaleimide (BMI, Aldrich Chemical Co.) was first
purified via column chromatography as described before,34
followed by recrystallization from ethanol to give yellow
crystals (mp 161.7 °C and Texo 252.7 °C, from DSC).
sible paramagnetic species. Nonetheless, the unique
property of hyperbranched polymer II to initiate radical
polymerization of BMI33 has never been observed before
In str u m en ta tion s. Proton and carbon nuclear magnetic
resonance (1H NMR and 13C NMR, 270 and 50 MHz, respec-

Macromolecules, Vol. 36, No. 12, 2003
Free-Radical-Induced Polymerization 4393
tively) spectra for the intermediates, monomers, and polymers
were obtained on a J EOL-270 spectrometer. Electron spinning
resonance (ESR) spectra were obtained from a Bruker EMX
instrument. Infrared (FT-IR) spectra were recorded on a
Bruker IFS 28 Equinox Fourier transform spectrophotometer.
Elemental analysis and mass spectral analysis were performed
by the System Supports Branch, Air Force Research Lab,
Dayton, OH. The melting points (mp) of all the compounds
were determined on a Mel-Temp melting point apparatus and
are uncorrected. Intrinsic viscosities were determined with
Cannon-Ubbelohde No. 150 viscometers. Flow times were
recorded for N-methyl-2-pyrrolidinone (NMP) solution contain-
ing 1% lithium bromide and polymer solutions with concentra-
tions of approximately 0.5-0.10 g/dL at 30.0 ( 0.1 °C.
Differential scanning calorimetry (DSC) analyses were per-
formed in nitrogen at a heating rate of 10 °C/min using a
Perkin-Elmer model 2000 thermal analyzer equipped with
differential scanning calorimetry cell. Themogravimetric analy-
ses (TGA) were obtained in helium and air atmospheres with
a heating rate of 10 °C/min using a TA Hi-Res TGA 2950
themogravimetric analyzer. Thermomechamical analysis (TMA)
was conducted in helium with heating rate of 4 °C/min in
nitrogen. Sample III, which was not consolidated at oven
temperatures up to 240 °C, was compressed with a hydraulic
press at a ram force of 40 000 lbs.
2-Ben zoyloxy-2-(4-n itr op h en yl)a ceton itr ile (Cya n o(4-
n itr op h en yl)m eth yl ben zoa te, 1). In a 500 mL three-necked
round-bottomed flask equipped with a magnetic stirbar and
nitrogen inlet were dissolved 4-nitrobenzaldehyde (50.0 g, 331
mmol) and benzoyl chloride (50 mL) in dichloromethane (50
mL). Potassium cyanide (33.0 g, 507 mmol) in water (100 mL)
was added dropwise at an ice-bath temperature. Triethylben-
zylammonium chloride (TEBA, 2.5 g) was added, and this two-
phase system was stirred for 24 h at room temperature. The
organic layer was diluted with dichloromethane (100 mL),
separated, washed with aqueous sodium bicarbonate, and
dried over magnesium sulfate. Solvent was completely re-
moved from the CH2Cl2 extract, and the resulting light orange
residue was dissolved in warm ethanol and cooled to room
temperature to give 65.4 g (70% yield) of off-white crystals:
mp 114-116 °C (lit. mp 116-117.5 °C). Anal. Calcd for
yellow crystals: mp 211-212 °C. Anal. Calcd for C21H12N4O6:
C, 56.01; H, 2.69; N, 9.33; O, 31.97. Found: C, 55.75; H, 2.50;
N, 9.30; O, 33.98. FT-IR (KBr, cm-1): 1347, 1527, 1682. Mass
spectrum (m/e): 282 (M+, 100% relative abundance). 1H NMR
(DMSO-d6, ppm): δ 8.27-8.30 (dd, 4H, Ar), 8.41-8.44 (dd, 4H,
Ar). 13C NMR (DMSO-d6, ppm): δ 123.94, 131.66, 136.81,
150.75, 190.06.
2,3-Bis(4-n it r op h en yl)q u in oxa lin e-6-ca r b oxylic a cid
(4). In a 250 mL three-necked round-bottomed flask equipped
with a magnetic stirbar, a condenser, and nitrogen inlet were
dissolved 4,4′-dinitrobenzil (8.2 g, 27,3 mmol) and 3,4-diami-
nobenzoic acid (4.4 g, 28.9 mmol) in acetic acid (100 mL) and
heated under reflux for 12 h. After being cooled to room
temperature, the dark red solution was filtered, and the filtrate
was poured into 5% hydrochloric acid. The resulting precipitate
was collected by suction filtration and then air-dried overnight
to give 11.1 g (98% yield) of an off-white powder: mp 287-
289 °C. Anal. Calcd for C21H12N4O6: C, 60.58; H, 2.91; N,
13.46; O, 23.06. Found: C, 60.25; H, 3.10; N, 13.12; O, 24.84.
FT-IR (KBr, cm-1): 1521, 1670. Mass spectrum (m/e): 419 (M+,
100% relative abundance). 1H NMR (DMSO-d6, ppm): δ 7.78-
7.81 (d, 4H, Ar), 8.24-8.27 (d, 4H, Ar), 8.30 (s, 1H, Ar), 8.36-
8.37 (d, 1H, Ar), 8.39-8.40 (d, 1H, Ar), 8.68-8.69 (d, 1H, Ar).
13C NMR (DMSO-d6, ppm): δ 123.34, 129.44, 130.31, 130.74,
131.29, 139.81, 142.31, 144.04, 147.61, 147.67, 152.16, 152.80,
166.27, 171.89.
2,3-Bis(4-a m in op h en yl)qu in oxa lin e-6-ca r boxylic a cid
(5). Into a 500 mL high-pressure bottle were charged 2,3-bis-
(4-nitrophenyl)quinoxaline-6-carboxylic acid (8.0 g, 19 mmol),
palladium on activated carbon (10%, 0.5 g), and mixture of
ethanol (100 mL) and acetone (50 mL). The bottle was placed
on a hydrogenator. It was then subjected to five cycles of
charging and discharging hydrogen gas and agitated at 60-
65 psi for 24 h. After the resulting mixture had been filtered
through Celite 545 to remove the catalyst, the solvent was
removed on a rotary evaporator. The gray solid was slurred
in deoxygenated 2-propanol and collected by suction filtration
to give 5.2 g (76.5% yield) of orange powder: mp ) 187-195
°C. DSC showed three broad melting endotherms centered at
181, 199, and 216 °C and a strong endotherm centered at 240
°C). The polymorphoism of quinoxaline derivatives with
multiple substitutents is not uncommon.8b Anal. Calcd for
C
15H10N2O4: C, 63.83; H, 3.57; N, 9.92; O, 22.67. Found: C,
63.96; H, 3.62; N, 9.42; O, 23.71: FT-IR (KBr, cm-1): 1733,
C21H16N4O2: C, 70.78; H, 4.53; N, 15.72; O, 8.98. Found: C,
2923. Mass spectrum (m/e): 282 (M+, 100% relative abun-
dance), 255, 225, 105. H NMR (CDCl3, ppm): δ 6.79 (s, 1H,
69.26; H, 5.78; N, 13.86; O, 10.50. FT-IR (KBr, cm-1): 1670,
1
3356. Mass spectrum (m/e): 358, 360 (M+, 100% relative
1
CH), 7.47-7.53 (t, 2H, Ar), 7.63-7.66 (t, 1H, Ar), 7.82-7.75
(dd, 2H, Ar), 8.06-8.09 (dd, 2H, Ar), 8.33-8.36 (dd, 2H, Ar).
13C NMR (CDCl3, ppm): δ 62.23, 115.25, 124.12, 127.46,
128.87, 130.16, 134.57, 138.28, 149.03, 164.32.
2-Ben zoyloxy-1,2-bis(4-n itr op h en yl)eth a n on e (2). In a
1000 mL three-necked, round-bottomed flask equipped with
abundance). H NMR (DMSO-d6, ppm): δ 6.29-6.33 (dd, 4H,
Ar), 6.48-6.56 (m, 4H, Ar), 6.62-6.76 (m, 1H, Ar), 7.10-7.18
(m, 2H, Ar). 13C NMR (DMSO-d6, ppm): δ 57.19, 58.23, 110.87,
112.83, 112.97, 113.78, 115.10, 117.41, 118.67, 120.20, 128.20,
128.38, 128.72, 129.24, 132.70, 138.51, 146.72, 146.81, 157.17,
167.92.
a
magnetic stirbar and nitrogen inlet, 2-benzoyloxy-2-(4-
2,3-Bis(4-m eth oxyph en yl)qu in oxalin e-6-car boxylic acid
(6). In a 500 mL three-necked, round-bottomed flask equipped
with a magnetic stirbar, nitrogen inlet, and a condenser was
completely dissolved 3,4-diaminobenzoic acid (14.2 g, 93.4
mmol) in acetic acid (300 mL). 4,4′-Dimethoxybenzol (25.0 g,
92. mmol) was then added, and the reaction mixture was
heated under reflux for 8 h. While the red-brown mixture was
allowed to cool on its own, light orange needles separated from
the mother liquid; 34.8 g (97% crude yield, mp 295-297 °C.
Anal. Calcd for C23H18N2O4: C, 71.49; H, 4.70; N, 7.25.
Found: C, 71.51; H, 4.45; N, 6.96. FT-IR (KBr, cm-1): 1693.
Mass spectrum (m/e): 386 (M+, 100% relative abundance). 1H
NMR (DMSO-d6, ppm): δ 3.80 (s, 6H, CH3), 6.93-6.96 (d, 4H,
Ar), 7.45-7.49 (dd, 4H, Ar), 8.12-8.15 (d, 1H, Ar), 8.22-8.26
(dd, 1H, Ar), 8.58-8.59 (d, 1H, Ar). 13C NMR (DMSO-d6,
ppm): δ 55.12, 113.55, 128.92, 130.48, 130.74, 131.08, 131.17,
131.52, 139.41, 142.11, 153.49, 154.09, 159.88, 159.97, 166.59.
2,3-Bis(4-h ydr oxyph en yl)qu in oxalin e-6-car boxylic acid
(7). In a 1000 mL three-necked round-bottomed flask equipped
with a magnetic stirbar, nitrogen inlet, and a condenser, was
dissolved 2,3-bis(4-methoxyphenyl)quinoxaline-6-carboxylic acid
(34.7 g, 89.8 mmol) in acetic acid (260 mL). Hydrobromic acid
(500 mL) was then added to the clear, yellow mixture at room
temperature. The reaction mixture was heated under reflux
nitrophenyl)acetonitrile (42.0 g, 149 mmol) in benzene (450
mL) were stirred 10% sodium hydroxide (30 mL) and TEBA
(2.2 mg) for 10 min under the nitrogen. Then, 4-nitrobenzal-
dehyde (22.52 g, 149 mmol) in benzene (100 mL) was added
at ice-bath temperature and the resulting mixture was stirred
for 4 h at room temperature. The organic layer was separated,
washed with water, and dried over magnesium sulfate. Solvent
was removed to dryness and the residue dissolved in ethanol
and cooled to room temperature to give 69.6 g (77% yield) of
off-white crystals: mp 131-133 °C (lit. mp 132-134 °C). Anal.
Calcd for C21H14N2O7: C, 62.07; H, 3.47; N, 6.89. Found: C,
62.26; H, 3.36; N, 6.77. FT-IR (KBr, cm-1): 1701, 1720. Mass
spectrum (m/e): 406, 150 (M+, 1% relative abundance).
4,4′-Din itr oben zil (1,2-Bis(4-d in itr op h en yl)eth a n e-1,2-
d ion e, 3). In a 500 mL three-necked, round-bottomed flask
equipped with a magnetic stirbar, dropping funnel, and ni-
trogen inlet was dissolved 2-benzoyloxy-1,2-bis(4-nitrophenyl)-
ethanone (15.0 g, 36.9 mmol) in DMSO (150 mL) and stirred
until the solution become homogeneous. Then, hydrobromic
acid (48%, 50 mL) was added through dropping funnel and
the reaction stirred at 60 °C. During this process the light
yellow crystals were isolated. After cooled, the crystals were
collected by suction filtration to give 10.1 g (91% yield) of

4394 Baek et al.
Macromolecules, Vol. 36, No. 12, 2003
with vigorous stirring until the solution become homogeneous.
It took about 6 h. After the red-brown mixture was allowed to
cool on its own, it was poured into distilled water. The resulting
light brown precipitate was collected by suction filtration and
dried under reduced pressure to give 31.9 g (99% crude
yield) of yellow solid: mp 319-320 °C dec. Anal. Calcd for
abundance). 1H NMR (DMSO-d6, ppm): δ 7.29-7.34 (d, 4H,
Ar), 7.38-7.40 (t, 1H, Ar), 7.50-7.51 (s, 2H, Ar), and 8.26-
8.32 (d, 4H, Ar). 13C NMR (DMSO-d6, ppm): δ 116.67, 118.38,
126.19, 134.77, 142.98, 156.22, 161.67, 165.58.
3,5-Bis(4-a m in op h en yloxy)ben zoic a cid (11). Into a 500
mL high-pressure bottle were added 3,5-bis(4-nitrophenyloxy)-
benzoic acid (15.0 g, 38 mmol), 10% palladium on activated
carbon (0.5 g), and a mixture of ethanol (150 mL) and acetone
(50 mL). The bottle was placed on the hydrogenation vessel.
It was then subjected to five cycles of charging and discharging
hydrogen gas and agitated at 60-65 psi for 24 h. After the
resulting mixture had been filtered through Celite 545 to
remove the catalyst, the solvent was removed on a rotary
evaporator. The orange residue was recrystallized from deoxy-
genated 90% 2-propanol to give 10.65 g (83.7% yield) of brown
crystals: mp 228-230 °C (free amine, dec), 245 °C (dihydro-
chloric acid salt, dec). Anal. Calcd for C19H16N2O4: C, 67.85;
H, 4.79; N, 8.33; O, 19.03. Found: C, 66.44; H, 6.25; N, 6.97;
O, 19.70. FT-IR (KBr, cm-1): 1594, 3412. Mass spectrum
C
21H14N2O4: C, 70.38; H, 3.94; N, 7.82. Found: C, 66.70; H,
3.98; N, 7.20. FT-IR (KBr, cm-1): 1698, 3396. Mass spectrum
1
(m/e): 358 (M+, 100% relative abundance). H NMR (DMSO-
d6, ppm): δ 6.76-6.79 (d, 4H, Ar), 7.36-7.40 (dd, 4H, Ar),
8.09-8.12 (d, 1H, Ar), 8.21-8.25 (dd, 1H, Ar), 8.57 (d, 1H, Ar).
13C NMR (DMSO-d6, ppm): δ 114.93, 128.84, 129.21, 130.34,
131.11, 131.52, 139.32, 142.03, 153.75, 154.35, 158.44, 166.68.
2,3-Bis(4-n itr op h en yloxyp h en yl)qu in oxa lin e-6-ca r box-
ylic a cid (8). Into a 250 mL three-necked round-bottomed
flask equipped with a magnetic stirbar, nitrogen inlet, and a
condenser were placed 4-fluoronitrobenzene (8.3 g, 57 mmol),
2,3-bis(4-hydroxyphenyl)quinoxaline-6-carboxylic acid (10.0 g,
28 mmol), potassium carbonate (14.0 g, 100 mmol), and a
mixture of NMP (110 mL) and toluene (60 mL). The reaction
mixture was then heated and maintained around 140 °C for 8
h with vigorous nitrogen flow. The dark solution was filtered
while it was warm, and the filtrate was poured into distilled
water containing 5% hydrochloric acid. The resulting precipi-
tates were collected by suction filtration and then air-dried
overnight. Bright yellow crude product was boiled in acetic
acid and filtered while hot to afford 16.4 g (98% yield) of light
yellow powder: mp 252-254 °C. Anal. Calcd for C33H20N4O8:
C, 66.00; H, 3.36; N, 9.33; O, 21.31. Found: C, 65.72; H, 3.55;
N, 9.24; O, 21.37. FT-IR (KBr, cm-1): 1586, 1701. Mass
spectrum (m/e): 600 (M+, 100% relative abundance). 1H NMR
(DMSO-d6, ppm): δ 7.13-7.18 (d, 4H, Ar), 7.26-7.29 (d, 4H,
Ar), 7.62-7.65 (d, 4H, Ar), 8.21-8.24 (dd, 4H, Ar), 8.28 (s, 1H,
Ar), 8.32-8.36 (dd, 1H, Ar), 8.68-8.69 (d, 1H, Ar). 13C NMR
(DMSO-d6, ppm): δ 117.58, 120.11, 126.13, 129.21, 132.12,
132.24, 135.29, 142.34, 153.40, 154.04, 154.96, 155.04, 162.39,
166.45.
2,3-Bis(4-a m in op h en yloxyp h en yl)q u in oxa lin e-6-ca r -
boxylic a cid (9). Into a 500 mL high-pressure bottle were
added 2,3-bis(4-nitrophenoxyphenyl)quinoxaline-6-carboxylic
acid (13.3 g, 22.1 mmol), 10% palladium on activated carbon
(0.5 g), and a mixture of ethanol (150 mL) and acetone (150
mL). The bottle was placed on a hydrogenator. It was subjected
to five cycles of charging and discharging with hydrogen gas.
The reaction mixture was then agitated mechanically at
60-65 psi for 24 h. After the resulting mixture had been
filtered through Celite 545 to remove the catalyst, the solvent
was removed on a rotary evaporator. The yellow solid was
poured into deoxygenated water. The resulting precipitate was
collected to give 10.65 g (83.7% yield) of light yellow solids:
mp 184.5 °C dec. Anal. Calcd for C33H24N4O4: C, 73.32; H, 4.47;
N, 10.36; O, 11.84. Found: C, 72.98; H, 4.60; N, 9.52; O, 12.05.
FT-IR (KBr, cm-1): 1705, 3373. Mass spectrum (m/e): 540 (M+,
100% relative abundance). 1H NMR (DMSO-d6, ppm): δ 6.77-
6.80 (d, 4H, Ar), 6.88-6.91 (d, 8H, Ar), 7.48-7.51 (d, 4H, Ar),
8.25-8.28 (d, 1H, Ar), 8.29 (s, 1H, Ar), 8.60 (s, 1H, Ar). 13C
NMR (DMSO-d6, ppm): δ 48.38, 116.11, 116.22, 116.60,
116.71, 116.86, 116.94, 118.85, 120.80, 120.89, 121.03, 129.24,
131.43, 131.75, 131.92, 142.14, 166.50.
1
(m/e): 336 (M+, 100% relative abundance). H NMR (DMSO-
d6, ppm): δ 6.63-6.67 (d, 4H, Ar), 6.69-6.71 (t, 1H, Ar), 6.81-
6.87 (d, 4H, Ar), 6.97-6.98 (d, 2H, Ar). 13C NMR (DMSO-d6,
ppm): δ 108.88, 109.54, 114.90, 121.29, 133.04, 144.59, 145.91,
160.43, 166.45.
2,3-Dip h en ylqu in oxa lin e-6-ca r boxylic a cid (12). In a
500 mL three-necked round-bottomed flask equipped with a
magnetic stirrer, a condenser, and a nitrogen inlet was
dissolved 3,4-diaminobenzoic acid (16.0 g, 105 mmol) in
deoxygenated acetic acid (250 mL). Benzil (21.0 g, 100 mmol)
was then added in one portion. The mixture was heated under
reflux for 12 h. During this time, an off-white precipitate fell
out of the solution. After the reaction mixture had been allowed
to cool, the precipitate was collected to give 32.2 g (99% yield)
of the crude product, mp 290.5-292 °C. Recrystallization of
the crude product from DMF afforded 29.6 g (91% yield) of
pink crystals, mp 291-292.5 °C. Anal. Calcd for C24H14N2O2:
C, 77.29; H, 4.32; N, 8.57; O, 9.80. Found: C, 76.93; H, 4.77;
N, 8.52; O, 9.63. FT-IR (KBr, cm-1): 696, 1691, 2922, 2958,
3400. Mass spectrum (m/e): 326 (M+, 100% relative abun-
1
dance). H NMR (DMSO-d6, ppm): δ 7.35-7.43 (m, 6H, Ar),
7.48-7.51 (d, 4H, Ar), 8.18-8.21 (d, 1H, Ar), 8.28-8.32 (dd,
1H, Ar), 8.65 (s, 1H, Ar), 13.51 (s, 1H, COOH). 13C NMR
(DMSO-d6, ppm): δ 128.00, 128.20, 128.95, 129.04, 129.15,
129.41, 129.67, 130.65, 132.01, 138.31, 139.61, 142.23, 153.98,
154.61, 166.50.
Hyp er br a n ch ed P olya m id e Der ived fr om 3,5-Bis(4-
a m in op h en yloxy)ben zoic Acid (I). Into a 100 mL three-
necked round-bottomed flask equipped with a magnetic stir-
bar, nitrogen inlet, and a reflux condenser were placed 3,5-
bis(4-aminophenyloxy)benzoic acid (3.0 g., 8.9 mmol), triphenyl
phosphite (7 mL), pyridine (5 mL), and freshly distilled
N-methyl-2-pyrrolidinone (40 mL), and the reaction mixture
was heated to 80 °C and maintained at this temperature for
12 h. After the dark brown, homogeneous mixture had been
allowed to cool to room temperature, it was poured into a
deoxygenated mixture of methanol and acetic acid (1/1, v/v).
The resulting white precipitate was collected by suction
filtration and dried under reduced pressure (1 mmHg) at 100
°C for 200 h.
3,5-Bis(4-n itr op h en yloxy)ben zoic a cid (10). Into a 500
mL three-necked round-bottomed flask equipped with a mag-
netic stirbar, nitrogen inlet, and a condenser were placed
4-fluoronitrobenzene (30.0 g, 0.21 mol), 3,5-dihydroxybenzoic
acid (15.4 g, 0.10 mol), potassium carbonate (50.0 g, 0.36 mol),
and NMP (250 mL). The reaction mixture was then heated
and maintained between 120 and 140 °C for 8 h. The dark
solution was filtered while it was still warm, and the fil-
trate was poured into distilled water containing 5% hydro-
chloric acid. The resulting precipitate was collected by suc-
tion filtration and then air-dried overnight. The crude product
was recrystallized from hot acetic acid to afford 39.6 g (91%
yield) of brown crystals: mp 228-229 °C. Anal. Calcd for
Hyp er br a n ch ed P olya m id e Der ived fr om 2,3-Bis(4-
a m in op h en yl)qu in oxa lin e-6-ca r boxylic Acid (II). This
was prepared and worked up similarly as above from a mixture
of 2,3-bis(4-aminophenyl)quinoxaline-6-carboxylic acid (3.0 g,
8.4 mmol), triphenyl phosphite (7 mL), pyridine (5 mL), and
freshly distilled N-methyl-2-pyrrolidinone (40 mL).
Hyp er br a n ch ed P olya m id e d er ived fr om 2,3-Bis(4-
a m in op h en yloxyp h en yl)qu in oxa lin e-6-ca r boxylic Acid
(III). This was prepared and worked up similarly as above
from a mixture of 2,3-bis(4-aminophenyloxyphenyl)quinoxa-
line-6-carboxylic acid (4.5 g, 8.3 mmol), triphenyl phosphite
(7 mL), pyridine (5 mL), and freshly distilled N-methyl-2-
pyrrolidinone (40 mL). However, 2.5 h after the reaction
mixture was heated at 80 °C, the solution viscosity built up
drastically, and the mixture became heterogeneous due to
some precipitation. Although stirring such a viscous reaction
C
19H12N2O8: C, 57.58; H, 3.05; N, 7.07; O, 32.30. Found: C,
57.45; H, 2.74; N, 6.96; O, 33.46. FT-IR (KBr, cm-1): 1105,
1350, 1454, 1625. Mass spectrum (m/e): 396 (M+, 100% relative

Macromolecules, Vol. 36, No. 12, 2003
Free-Radical-Induced Polymerization 4395
A.; Harris, F. W. Macromolecules 2001, 34, 2427. (c) Klein,
D. J .; Korleski, J . E.; Harris, F. W. J . Polym. Sci., Part A:
Polym. Chem. 2001, 39, 2037. (d) Ooi, I. H.; Hergenrother,
P. M.; Harris, F. W. Polymer 2000, 41, 5095.
mixture had become inefficient, the heterogeneous mixture
was maintained at 80 °C overnight (∼12 h). After the very
viscous mixture had been allowed to cool to room temperature,
it was poured into a deoxygenated mixture of methanol and
acetic acid (1/1, v/v). The resulting white precipitate was
collected by suction filtration and dried under reduced pressure
(1 mmHg) at 100 °C for 200 h. The insolubility of this
hyperbranched polymer in common organic solvents and
strong acid such as methanesulfonic acid (MSA) precluded the
viscosity determination.
Gen er al P r ocedu r e for th e P r epar ation of BMI Blen ds.
The hyperbranched polyamides (I or II) was dissolved in NMP
(1 wt %) and filtered. Exact weight of the filtrate (stock
solution) was added into sample vials containing preweighted
BMI, and the resulting mixtures were agitated until the
solutions became homogeneous. The content of I or II in the
blends were varied from 0 to 3.75 wt % in both series of the
blends. The solvent from each of the blend/NMP solutions was
removed under reduced pressure (1 mmHg) for 1 week at 100
°C. The solvent residue was monitored by FT-IR to ensure that
none was present in the samples before thermal characteriza-
tion (see Tables 2 and 3). This was confirmed by thermograve-
metric analysis results.
(9) (a) Hedrick, J . L.; Labadie, J . W. Macromolecules 1988, 21,
1883. (b) Hedrick, J . L.; Labadie, J . W. Macromolecules 1990,
23, 1561. (c) Hedrick, J . L.; Labadie, J . W.; Russell, T. P.
Macromolecules 1991, 24, 4559. (d) Hedrick, J .; Twieg, R.;
Matray, T.; Carter, K. Macromolecules 1993, 26, 4833. (e)
Hedrick, J . L.; Twieg, R. Polymer 1997, 38, 995.
(10) (a) Bruma, M.; Schulz, B.; Ko¨pnick, T.; Stiller, B.; Mercer, F.
Mater. Sci. Eng.: C 1999, 8-9, 361. (b) Bruma, M.; Schulz,
B.; Ko¨pnick, T.; Stiller, B.; Belomoina, N.; Mercer, F. Eur.
Polym. J . 1999, 35, 1253. (c) Hamciuc, E.; Hamciuc, C.; Sava,
I.; Bruma, M. Eur. Polym. J . 2001, 37, 287.
(11) (a) Akutsu, F.; Matsuo, K.; Naruchi, K.; Miura, M. Polym.
Commun. 1989, 30, 182. (b) Akutsu, F.; Hirata, H.; Hayashi,
H.; Miura, M.; Nagakubo, K. Makromol. Chem. Rapid Com-
mun. 1985, 6, 215.
(12) Hedrick, J . L.; Hawker, C. J .; Miller, R. D.; Twieg, R.;
Srinivasan, S. A.; Trollsås, M. Macromolecules 1997, 30, 7607.
(13) Rozwadowska, M. D. Tetrahedron 1985, 41, 3135.
(14) (a) Elce, E.; Hay, A. S. Polymer 1996, 37, 1745. (b) Floyd, M.
B.; Du, M. T.; Fabio, P. F.; J acob, L. A.; J ohnson, B. D. J .
Org. Chem. 1985, 50, 5022.
(15) (a) Hedrick, J .; Twieg, R.; Matray, T.; Carter, K. Macromol-
ecules 1993, 26, 4833. (b) Hedrick, J . L.; Labadie, J . W.
Macromolecules 1990, 23, 1561.
P r ep a r a tion of ESR Sa m p les. The blend samples (from
the above preparation) were ball-milled as fine mesh as
possible and packed into ESR glass sample tubes and dried
under reduced pressure (1 mmHg) at 100 °C for 24 h. The
tubes were filled with nitrogen and then sealed with a torch.
(16) J andke, M.; Strohriegl, P.; Berleb, S.; Werner, E.; Brutting,
W. Macromolecules 1998, 31, 6434.
(17) For example, changing the substituent from NO2 to NHMe
results in the following changes in chemical shifts for o-H
(1.75 ppm), m-H (0.48 ppm) and p-H (1.08 ppm); see:
Memory, J . D.; Wilson, N. K. “NMR of Aromatic Compounds,”
Wiley: New York, 1982; p 64-67.
(18) ACD/HNMR version 1.0, Advanced Chemistry Development
Inc., 141 Adelaide St., West, Suite 1501, Toronto, Canada.
(19) In going from 5 to 5b, the H-bond accepting sites (quinoxaline
nitrogens) are effectively converted to H-donating sites
(piperazine NH’s).
(20) (a) Katritzky, A. R.; Karelson, M.; Harris, P. A. Heterocycles
1991, 32, 329-369. (b) Elguero, J .; Marzin, C.; Katritzky, A.
R.; Linda, P. The Tautomerization of Heterocycles. In Ad-
vances in Heterocyclic Chemistry, Supplement 1; Katrizky,
A. R., Boulton, A. J ., Eds.; Academic Press: New York, 1976.
(c) Katritzky, A. R.; Lagowski, J . M. Prototropic Tautomerism
of Heteroaromatic Compounds: I. General Discussion and
Methods of Study. In Advances in Heterocyclic Chemistry;
Katrizky, A. R., Ed.; Academic Press: London, 1963; Vol. 1.
(d) Katritzky, A. R.; Lagowski, J . M. Prototropic Tautomerism
of Heteroaromatic Compounds: II. Six-Membered Rings. In
Advances in Heterocyclic Chemistry; Katrizky, A. R., Ed.;
Academic Press: London, 1963; Vol. 1.
Ack n ow led gm en t. We are grateful to Marlene
Houtz (University of Dayton Research Institute) for the
TGA and TMA data and Dr. Kevin Church (Department
of Chemistry, University of Dayton) for the NMR
spectra.
Su p p or tin g In for m a tion Ava ila ble: A color photograph
showing the three sample vials containing NMP (1 mL)
solutions, (a) II/BMI (1.5 mg/200 mg), (b) II/BMI (3.0 mg/200
mg), and (c) II (200 mg), after they had been allowed to stand
under room conditions for 24 h. This material is available free
of charge via the Internet at http://pubs.acs.org.
Refer en ces a n d Notes
(1) (a) Crivello, J . V.; Fan, M.; Bi, D. J . Appl. Polym. Sci. 1992,
44, 9. (b) Sutcliffe, J . In Electron Beam Curing (EBC);
Randell, D. R., Ed.; The Royal Society of Chemistry: Cam-
bridge, U.K., 1991; Vol. Special Publ. No. 89, p 22. (c) Walton,
T. C.; Crivello, J . V. Proc., 41st Int.SAMPE Symp. Exhib.
1996, 41, 1760. (d) Walton, T. C.; Crivello, J . V. Proc., 41st
Int. SAMPE Symp. Exhib. 1996, 40, 1266.
(2) (a) Baek, J .-B.; Tan, L.-S. Polym. Prepr. 2001, 42 (2), 468.
(b) Baek, J .-B.; Mather, P. T.; Tan, L.-S. Polym. Mater. Sci.
Eng. 2001, 84, 742. (c) Baek, J .-B.; Ferguson, J . B.; Mather,
P. T.; Tan, L.-S. Polym. Mater. Sci. Eng. 2001, 84, 724. (d)
Baek, J .-B.; Simko, S. R.; Tan, L.-S. Polym. Prepr. 2001, 42
(2), 502. (e) Baek, J .-B.; Tan, L.-S. Polym. Prepr. 2002, 43
(1), 514. (f) Baek, J .-B.; Tan, L.-S. Polym. Prepr. 2002, 43
(1), 533.
(3) (a) Yang, G.; J ikei, M.; Kakimoto, M. Macromolecules 1998,
31, 5964. (b) Yang, G.; J ikei, M.; Kakimoto, M. Macromol-
ecules 1999, 32, 2215. (c) Ishida, Y.; Sun, A. C. F.; J ikei, M.;
Kakimoto, M. Macromolecules 2000, 33, 2838. (d) J ikei, M.;
Fujii, K.; Yang, G.; Kakimoto, M. Macromolecules 2000, 33,
6228.
(4) Cassidy, P. E. Thermally Stable Polymers, Marcel Dekker:
New York, 1980.
(5) Hergenrother, P. M. In Encyclopedia of Polymer Science and
Engineering, 2nd ed.; Mark, H. F., Bikales, N. M., Overberger,
C. G., Menges, G., Eds.; J ohn Wiley and Sons: New York,
1988; Vol. 13, p 55.
(6) Hagnauer, G. L.; Mulligan, G. D. Macromolecules 1973, 6,
477.
(7) (a) Connell, J . W.; Hergenrother, P. M. Polym. Prepr. 1988,
29 (1), 172. (b) Connell, J . W.; Hergenrother, P. M. Polymer
1992, 33, 3739.
(21) Beak, P. Acc. Chem. Res. 1977, 10, 186.
(22) Yamazaki, N.; Matsumoto, M.; Higashi, F. J . Polym. Sci.,
Polym. Chem. Ed. 1975, 13, 1373.
(23) Russo, S.; Boulares, A.; Mariani, A. Macromol. Sym. 1998,
128.
(24) (a) Boogh, L.; Pattersson, B.; Kaiser, P.; Månson, J .-A.
SAMPE J . 1997, 33, 45. (b) Boogh, L.; Pettersson, B.; Månson,
J . A. E. Polymer 1999, 40, 2249.
(25) Baek, J .-B.; Qin, H.; Mather, P. T.; Tan, L.-S. Macromolecules
2002, 35, 4951. Further work on AT-PAEKI/BMI indicates
that AT-PAEKI can serve as a processing aid to greatly
reduce the melt viscosity of the BMI resin, and potentially a
toughening additive as well; for a preliminary account, see:
Qin, H.; Mather, P. T.; Baek, J .-B.; Tan, L.-S. Polym. Prepr.
2003, 44 (1), 132.
(26) Akutsu, F.; Inoki, M.; Daicho, N.; Kasashima, Y.; Shiraishi,
N.; Marushima, K. J . Appl. Polym. Sci. 1998, 69, 1737.
(27) For the assignment of the IR vibrational bands of BMI and
related model compounds, see: Giulio, C. D.; Gauthier, M.;
J asse, B. J . Appl. Polym. Sci. 1984, 29, 1771.
(28) A color photograph showing the three vials is provided as
Supporting Information.
(29) Michael addition reaction of an aromatic diamine to a
bismaleimide compound occurs typically around 120-145 °C,
see: Fasce, D. P.; Williams, R. J . J . Polym. Bull. (Berlin)
1995, 34, 515 and references therein.
(30) For hyperbranched polyamide I, both pure and mixed with
BMI, a g value of 2.0038 is observed, indicating no reaction
has occurred. However, for the pure II, the initial g value is
2.0034. This value is within the range of g values reported
(8) (a) Kim, B.-S.; Korleski, J . E.; Zhang, Y.; Klein, D. J .; Harris,
F. W. Polymer 1999, 40, 4553. (b) Klein, D. J .; Modarelli, D.

4396 Baek et al.
Macromolecules, Vol. 36, No. 12, 2003
in the literature for various quinoxaline-based cation radicals;
see: (a) Kato, T.; Shida, T. J . Am. Chem. Soc. 1979, 101, 6869.
(b) Braun, D. Angew. Makrom. Chem. 1991, 185/ 186, 217.
(c) Carre, C.; Courtieu, J . J .; Stahl-Lariviere, H. Spectrochim.
Acta 1986, 42A, 1201.
(33) Under similar conditions, an NMP solution of the hyper-
branched polymer II and another BMI (2,2-bis[4-(4-maleimi-
dophenoxy)phenyl]propane or BPA-BMI) in the same com-
position also underwent similar free-radical induced polymeri-
zation. II has been stored under room conditions for over a
year and took a longer time (3-4 days vs about a day) to
effect a similar gelation.
(31) After about a month in methanesulfonic acid at room tem-
perature and on a mechanical shaker, some of the hyper-
branched polymer III dissolved to form a dark purple solution
but most of it formed microgels. Thus, the insufficient
solubility of III even in a strong acid practically precluded
any solution characterization effort.
(34) Tan, L.-S.; Arnold, F. E.; Soloski, E. J . J . Polym. Sci., Part
A: Polym. Chem. 1988, 26, 3103.
(32) Arsu, N.; Aydin, M. Angew. Makromol. Chem. 1999, 270, 1.
MA030039Z