The relationship between the chemical structure and thermal conversion temperatures of thermally rearranged (TR) polymers

Article abstract of DOI:10.1016/j.polymer.2012.04.032

An attempt was made to elucidate the relationship between chemical structure, in terms of glass transition temperature (T_g), and conversion temperature (T_TR) of thermally rearranged polybenzoxazole membranes, by analyzing DSC and TGA data for a total of 15 sets of o-hydroxypolyimides and copolyimides, derived from two experimental and four commercial structurally different diamines, as well as four different dianhydride monomers. Our research revealed that T_TR was influenced by chain ridigity of o-hydroxypolyimides, and exhibited a linear relationship with T_g. Therefore, structure and thermal property of o-hydroxypolyimide should be considered when studying TR polymers to determine thermal conversion to polybenzoxazole.

Full text of DOI:10.1016/j.polymer.2012.04.032

Polymer 53 (2012) 2783e2791
Contents lists available at SciVerse ScienceDirect
Polymer
journal homepage: www.elsevier.com/locate/polymer
The relationship between the chemical structure and thermal conversion
temperatures of thermally rearranged (TR) polymers
,b,
Mariola Calle a, Yishee Chan b, Hye Jin Jo a, Young Moo Lee a
*
a WCU Department of Energy Engineering, Hanyang University, Seoul 133-791, Republic of Korea
b School of Chemical Engineering, Hanyang University, Seoul 133-791, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history:
An attempt was made to elucidate the relationship between chemical structure, in terms of glass tran-
sition temperature (Tg), and conversion temperature (TTR) of thermally rearranged polybenzoxazole
membranes, by analyzing DSC and TGA data for a total of 15 sets of o-hydroxypolyimides and copolyi-
mides, derived from two experimental and four commercial structurally different diamines, as well as
four different dianhydride monomers. Our research revealed that TTR was influenced by chain ridigity of
o-hydroxypolyimides, and exhibited a linear relationship with Tg. Therefore, structure and thermal
property of o-hydroxypolyimide should be considered when studying TR polymers to determine thermal
conversion to polybenzoxazole.
Received 1 December 2011
Received in revised form
29 March 2012
Accepted 17 April 2012
Available online 22 April 2012
Keywords:
Thermal conversion temperature
Glass transition temperature
Thermal rearrangement
Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction
a function of increasing flexibility in the dianhydride moiety.
Subsequently, Tullos and his colleagues carried out a more in depth
In-situ thermal conversions of ortho-hydroxy-containing poly-
imides (HPIs) to polybenzoxazoles (PBOs) have been widely reported.
Likhatchev and co-workers studied for the first time the thermal
behavior of soluble aromatic polyimides based on 2,20-bis(3-amino-4-
hydroxyphenyl)hexafluoropropane (APAF) diamine and several
aromaticdianhydrides(i.e. pyromelliticdianhydride(PMDA), 3,30,4,40-
benzophenonetetracarboxylic acid dianhydride (BTDA) and 4,40-oxy-
diphthalic anhydride (ODPA)) [1]. They established that the thermal
conversion of hydroxypolyimides to polybenzoxazoles proceeds
through a carboxybenzoxazole intermediate, followed by decarbox-
ylation to give the fully aromatic final benzoxazole (Scheme 1).
The carbon dioxide evolution from the decarboxylation step was
easily detected by thermogravimetric analysis (TGA) as a well-
defined, weight-loss step prior to the generalized decomposition of
the in-situ formed polybenzoxazole. Thus, they claimed that the
starting temperature of weight loss and the temperature at the peak
(maximum amount of CO2 evolution temperature) in the derivative
thermogravimetric curves (DTG) were controlled to some extent by
the structure of dianhydride moiety. In fact, the imide-to-
benzoxazole rearrangement shifted to lower temperatures as
study about the thermal rearrangement of several structurally
different o-hydroxypolyimides derived from different bis(o-amino-
phenol)s such APAF and 3,30-dihydroxybenzidine (HAB), and several
dianhydrides such as 3,30,4,40-bisphenyltetracarboxylic dianhydride
(BPDA), 4,40-(hexafluoroisopropylidene)diphthalic anhydride 6FDA,
4,40-(4,40-isopropylidenediphenoxy)bis(phthalicanhydride) BPADA,
BTDA, and ODPA, in an attempt to confirm the final rearranged
polymer structure and the thermal conversion mechanism [2,3].
In fact, they reasserted the previously established imide-to-
polybenzoxazole rearrangement mechanism, and found that the
more flexible hydroxy-containing polyimides, with lower glass
transition temperatures, underwent thermal conversion at a faster
rate and at a lower temperature. However, Tullos and his colleagues
did not establish a clear and direct relationship between the thermal
rearrangement or conversion temperature (TTR) and glass transition
temperature (Tg) of the precursor polyimides. After their report,
several studies have described the thermal conversion of HPIs to
PBOs [4e7], but none of them have centered on claryfing the possible
connection between Tg and TTR
.
Recently we found that thermally rearranged (TR) poly-
benzoxazole membranes show unusual microporous characteris-
tics resulting from a significant increase of free volume elements
during the thermal rearrangement process in the solid state [8e12].
These TR-polymer membranes are promising materials for gas
separation applications, such as CO2 separation for carbon capture,
* Corresponding author. Hanyang University, School of Chemical Engineering,
Seoul 133-791, Republic of Korea. Tel.: þ82 2 2220 0525.
E-mail address: ymlee@hanyang.ac.kr (Y.M. Lee).
0032-3861/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymer.2012.04.032

2784
M. Calle et al. / Polymer 53 (2012) 2783e2791
O
HO
2.2.2. Synthesis of 1,4-Bis(3-methoxy-4-nitrophenoxy)2,5-di-tert-
butylbenzene (1)
O
N
O
-CO2
N
Ar
N
The dinitro dimethoxylated intermediate 1 was synthesized by
the reaction of 2,5-di-tert-butylhydroquinone (11.12 g, 50 mmol)
and 5-fluoro-2-nitroanisole (18.82 g, 110 mmol) in the presence of
potassium carbonate (15.48 g, 112 mol) and DMF (100 mL) at 160 ꢀC
for 18 h. The mixture was then cooled and poured into distilled
water, filtered, and washed again with water. The crude product
was recrystallized from DMF to provide a yellow solid. The yield
was 85%. mp 243 ꢀC. Elemental Anal. Calcd. For C28H32N2O8: C,
64.11; H, 6.15; N, 5.34; Found: C, 63.95; H, 6.10; N, 5.20. 1H NMR
(300 MHz, DMSO-d6): 8.00 (d, 2H, J ¼ 9.1 Hz), 7.00 (s, 2H), 6.96 (d,
2H, J ¼ 2.2 Hz), 6.51 (dd, 2H, J ¼ 2.2 Hz, J ¼ 9.1 Hz), 3.39 (s, 6H).
OH
O
O
Scheme 1. Proposed reaction for thermal conversion of hydroxy-imide to benzoxazole.
since they show outstanding gas performance, particularly for the
gas pairs CO2/CH4 and CO2/N2 [13]. One of the crucial factors to
design a cost effective thermal treatment process for TR-polymers,
is the thermal rearrangement or conversion temperature (TTR). This
temperature, as pointed out above, can be predicted to be greatly
influenced by the polymer morphology and chemical structure.
Therefore, it is our objective to explore the relationship between
the chain mobility, Tg and TTR for this family of o-hydrox-
ypolyimides. As a continuation in our studies about TRePBO
polymer membranes and for further understanding on how TTR is
affected by the chemical structure of o-hydroxypolyimides, herein
we have examined in detail the differential scanning calorimetry
(DSC) and TGA thermograms, accompanied by DTG curves for
a large set of o-hydroxypolyimides and copolyimides. A great
variety of different chemical structures have been synthesized,
from two commercially available bis(o-aminophenol)s, (APAF and
HAB), and two experimental ones, 2,2-Bis(4-(4-amino-3-
hydroxyphenoxy)phenyl)hexafluoropropane (6FBAHPP) and 1,4-
Bis(4-amino-3-hydroxyphenoxy)2,5-di-tert-butylbenzene
2.2.3. Synthesis of 1,4-bis(3-hydroxy-4-nitrophenoxy)2,5-di-tert-
butylbenzene (2)
A mixture of 9.00 g (11.44 mmol) of 1 and 54.0 g of pyridine
hydrochloride was heated at 160 ꢀC for 24 h under nitrogen. The
reaction mixture was then poured into distilled water. Then the
precipitate was collected by filtration, and the crude product was
washed with water and dried. The product was recrystallized from
penthanol to afford a brown solid. The yield was 65%. mp 248 ꢀC.
Elemental Anal. Calcd. For C26H28N2O8: C, 62.89; H, 5.68; N, 5.64;
Found: C, 62.65; H, 5.45; N, 5.50. 1H NMR (300 MHz, DMSO-d6):
11.19 (s, 2H), 8.05 (d, 2H, J ¼ 8.6 Hz), 7.00 (s, 2H), 6.59 (d, 2H,
J ¼ 1.9 Hz), 6.50 (dd, 2H, J ¼ 1.9 Hz, J ¼ 8.6 Hz).
2.2.4. Synthesis of 1,4-bis(4-amino-3-hydroxyphenoxy)2,5-di-tert-
butylbenzene (TBAHPB)
(TBAHPB), incorporating flexible connecting linkages, together
with three commercial, commonly used dianhydrides, such as,
BPDA, 6FDA and BPADA.
A flask was charged with 2 (5 g, 10.07 mmol), 25.0 ml of
hydrazine monohydrate, 40 ml of ethanol, and 0.100 g of 10%
palladium on carbon (PdeC). The mixture was heated to reflux
(80 ꢀC) for 20 h. After this time, the reaction suspension was poured
into distilled water. The precipitate was collected by filtration, and
the crude solid was recrystallized using a mixed solution of DMF
(DMF:water ¼ 2:1, v/v) under a nitrogen atmosphere. The removing
of the PdeC catalyst was carried in the recrystallization step by
filtering through Celite. The yield was 70%. mp 335 ꢀC. Elemental
Anal. Calcd. For C26H32N2O4: C, 71.53; H, 7.39; N, 6.42; Found: C,
71.35; H, 7.10; N, 6.20. 1H NMR (300 MHz, DMSO-d6): 9.20 (s
(broad), 2H), 6.69 (s, 2H), 6.55 (d, 2H, J ¼ 8.6 Hz), 6.33 (d, 2H,
J ¼ 2.5 Hz), 6.22 (dd, 2H, J ¼ 2.5 Hz, J ¼ 8.6 Hz), 4.30 (s (broad), 4H).
2. Experimental section
2.1. Materials
Solvents and reactants were of reagent-grade quality and used
without further purification. 5-Fluoro-2-nitrophenol, hydrazine
monohydrate and palladium 10 wt% on activated carbon were
purchased from Aldrich, 4,40-(hexafluoroisopropylidene)diphenol
and 2,5-di-tert-butylhydroquinone to Alfa Aesar, and 5-fluoro-2-
nitroanisole from Apollo (U.K).
The dianhydride 3,30,4,40-bisphenyltetracarboxylic dianhydride
(BPDA), was purchased from Shanghai Resin Factory Co., Ltd. (China),
4,40-(4,40-isopropylidenediphenoxy)bis(phthalicanhydride) (BPADA)
was purchased from Aldrich and the 4,40-(hexafluoroisopropylidene)
diphthalic anhydride (6FDA) from Daikin Industries, Ltd. (Osaka,
Japan).
2.3. Poly(o-hydroxyimide)s synthesis
A three-necked flask, equipped with a mechanical stirrer and
gas inlet and outlet, was charged with 10.0 mmol of diamine and
10.0 mL of NMP. The mixture was stirred at room temperature
under nitrogen atmosphere until the solid was entirely dissolved.
Then, the solution was cooled to 0 ꢀC, had dianhydride (10.0 mmol)
added to it along with 10.0 ml of NMP. The reaction mixture was
stirred for 15 min at 0 ꢀC. Then, the temperature was raised to room
temperature and left overnight. o-Xylene (30 mL) as an azeotropic
agent was then added to the solution, which was stirred vigorously
and heated for 6 h at 180 ꢀC to promote imidization. During this
step, the water released by the ring-closure reaction was separated
as an o-xylene azeotrope. The resulting brown-colored solution was
cooled to room temperature, precipitated in distilled water, washed
several times with water and dried in a convection oven at 120 ꢀC
for 12 h.
The diamines, 2,20-bis(3-amino-4-hydroxyphenyl) hexa-
fluoropropane (APAF) was purchased from Central Glass Co. Ltd
(Tokyo, Japan) and the 3,30-dihydroxybenzidine (HAB) from Tokyo
Chemical Industry (TCI) Co., Ltd. (Tokyo, Japan).
2.2. Monomers synthesis
2.2.1. Synthesis of 2,2-Bis(4-(4-amino-3-hydroxyphenoxy)phenyl)
hexafluoropropane (6FBAHPP)
It was synthesized in two steps, according to the previously
reported method [14,15], from 4,40-(hexafluoroisopropylidene)
diphenol and 5-fluoro-2-nitrophenol by nucleophilic aromatic
substitution in the presence of potassium carbonate (K2CO3) and
DMF as solvent, followed by catalytic reduction with hydrazine
hydrate and Pd/C as a catalyst. Elemental analyses and 1H NMR data
of intermediate and final monomers have been recently reported
elsewhere [14].
2.4. Polyimide film formation
The casting of the polyimide was done from a 15 wt% filtered
solution in NMP onto a clean glass plate. Cast film was placed in

M. Calle et al. / Polymer 53 (2012) 2783e2791
2785
a vacuum oven and heated slowly to 250 ꢀC with holds for 1 h at
100 ꢀC, 150 ꢀC and 200 ꢀC to evaporate the solvent under high
vacuum. The solid film was taken off from the glass plate, rinsed
with deionized water, and dried at 120 ꢀC until the water was
removed. The defect-free and clean membranes were cut into small
sized strips, placed between quartz plates and further heated in
a muffle furnace up to 300 ꢀC, under a high-purity argon atmo-
sphere. It was then held for 1 h to eliminate residual solvent.
Membranes with glass transition temperatures well below 300 ꢀC,
were only heated up to 250 ꢀC to complete solvent removal. The
cooling rate of the cast membranes after annealing was 10 ꢀC/min.
hydrochloride in solvent-free conditions, and final conventional
catalytic reduction with hydrazine/PdeC (Scheme 2). Elemental
analysis and 1H NMR spectroscopic techniques were used to identify
the structures of the intermediate compounds and the final hydroxyl
diamine monomer.
3.2. Synthesis of hydroxyl-containing precursor polyimides and
copolyimides
A series of polyimides containing hydroxy groups ortho to the
imide group (HPIs) were synthesized from two commercially
available bis(o-aminophenol)s monomers: the stiff biphenyl
diamine 3,30-dihydroxybenzidine (HAB), and the semi flexible
diamine (APAF) containing the bulky [C(CF3)2] linkage. Three
commercial dianhydrides with different degrees of flexibility were
also used: the rigid biphenyl BPDA dianhydride, the 6FDA dianhy-
dride incorporating the eC(CF3)2e central link, and the very flex-
ible BPADA monomer containing ether and isopropylidene groups.
Moreover, new ether-containing HPIs, with a higher degree of
chain mobility, were prepared based on two experimental bis(o-
aminophenol)s incorporating aryl ether linkages, namely the large
and highly flexible 2,2-bis(4-(4-amino-3-hydroxyphenoxy)phenyl)
hexafluoropropane (6FBAHPP) that included the [C(CF3)2] group as
central linkage. The novel 1,4-bis(4-amino-3-hydroxyphenoxy)2,5-
di-tert-butylbenzene (TBAHPB) contained bulky tert-butyl side
groups in ortho positions in the central ring resulting in poly(ether-
imide)s with a more contorted, rotation restricted and stiff
conformation [19]. Inclusion of poly(ether-imide)s will contribute
to a more in depth study of the polymer structure-thermal
conversion temperature relationship.
2.5. Measurements
1H spectra were recorded on a Murcury Plus 300 MHz spec-
trometer (Varian, Inc., CA, USA). Elemental analyses were per-
formed with a Thermofinnigan EA1108 (Fisions Instrument Co.,
Italy) elemental analyzer. Molecular weights of precursor poly-
imides were measured by gel permeation chromatography (GPC,
Tosoh HLC-8320 GPC, Tokyo, Japan) with a TSKÔ SuperMultipore
HZ-M column, and a refractive index (RI) detector in THF based on
standard polystyrenes. Thermogravimetric analyses (TGAs) were
performed on a TA Q-500 thermobalance (TA Instruements, DE,
USA) and coupled with mass spectroscopy (MS) ThermoStarÔ GSD
301T (Pfeiffer Vacuum GmbH, Asslar, Germany).
Glass transition temperatures (Tg) of HPIs films were measured
by differential scanning calorimetry (DSC) analyses on a TA Instru-
ments Q-20 calorimeter. A total of two heatingecooling cycles, at
a heating rate of 20 ꢀC/min, were conducted and Tg was obtained
from the second heating cycle. Testing samples were heated to
a temperature usually below the starting temperature of conversion
to PBO for each HPI film during the first heating, quenched at room
temperature and reheated up to 475 ꢀC in the second scan. Heat of
decarboxylation reaction or rearrangement reaction occurred
around 350e450 ꢀC for HPIs can be detected by DSC.
Alternatively, for further understanding of how thermal rear-
rangement process is affected by the chain composition and flexi-
bility of HPI, a series of copolyimides in different ratios were also
synthesized based on 6FDA and APAF comonomers including no-
hydroxylated diamine moieties, such as the flexible oxydianiline
(ODA) or the very rigid 2,4,6-trimethyl-1,3-phenylenediamine
(DAM) in different contents.
3. Results and discussion
All the polymers were prepared by a two step polyimidization
method using a poly(amic acid) intermediate. In the second stage,
o-xylene as an azeotropic agent was added to the polymer solution,
stirred vigorously, and heated for 6 h at 180 ꢀC to promote imid-
ization. During this step, the water released by the ring-closure
reaction was separated as an o-xylene azeotrope. Structures and
codes of the monomers and polymers are summarized in Table 1.
The polymer nomenclature is defined by the used monomers. For
example, APAFe6FDA describes a polyimide obtained by a reaction
of APAF diamine with 6FDA dianhydride. In addition, APAF/DAM-
6FDA (2:8) described a copolyimide that was prepared from 6FDA
dianhydride and a mixture of APAF/DAM diamines in a ratio of 2:8.
3.1. Monomer synthesis
Two ether-containing non-commercially available bis(o-ami-
nophenol)s were synthesized in order to obtain a wider and clearer
picture of how different degrees of flexibility in the nucleophilic
aromatic diamine monomer can influence the thermal conversion
temperature (TTR).
The synthesis of bis(o-aminophenol)s including flexible ether
linkages has previously been reported [14]. The synthetic strategy
usually employed in these studies involves the base mediated
aromatic nucleophilic substitution of 2-hydroxy-4-fluoronitro-
benzene with diverse bisphenols. This is done to provide readily
the corresponding ether-containing bisnitrobenzene compound,
which is further reduced to the final bis(o-aminophenol) by
a conventional catalytic reduction in the presence of a palladium
catalyst on carbon. According to this procedure, 2,2-bis(4-(4-amino-
3-hydroxyphenoxy)phenyl)hexafluoropropane (6FBAHPP) was
synthesized successfully as reported elsewhere [14,15]. Nevertheless,
in some cases, protection of the hydroxy group in the fluoronitro
derivative, usually as a benzyloxy group, was proved to enhance the
efficiency in the condensation step [16e18]. Thus, as a modification of
this approach, the novel 1,4-bis(4-amino-3-hydroxyphenoxy)2,5-di-
tert-butylbenzene (TBAHPB) monomer, including bulky di-tert-butyl
side groups, was efficiently prepared from the commerciallyavailable
5-fluoronitroanisole as the starting material in the condensation
reaction with 5-di-tert-butylhydroquinone. This was followed by
demethylation of the methoxy protecting groups using pyridine
K2CO3
HO
OH
+
O2N
O
O
NO2
F
NO2
DMF
H3CO
OCH3
OCH3
1
1. Py . HCl
H2N
HO
O
O
NH2
2. NH2NH2.H2O
Pd/C
OH
TBAHPB
Scheme 2. Synthesis of 1,4-bis(4-amino-3-hydroxyphenoxy)2,5-di-tert-butylbenzene
(TBAHPB).

2786
M. Calle et al. / Polymer 53 (2012) 2783e2791
Table 1
Thermal property and Mw of hydroxyl polyimides (HPIs).
a
Polymer code
Diamine
Dianhydride
Mw
Tg
TTR1 TTR2 TTR3
(ꢀC) (ꢀC) (ꢀC) (ꢀC)
O
F3C CF3
O
O
O
H2N
HO
NH2
OH
O
HABe6FDA
170,860 331 347 452 490
O
HAB
6FDA
O
O
O
O
H3C
CH3
O
HABeBPADA
64,700 286 315 401 439
O
O
O
BPADA
O
O
O
O
H2N
HO
O
O
NH2
OH
O
O
TBAHPBeBPDA
28,900 310 337 426 456
BPDA
TBAHPB
O
O
O
F3C CF3
O
O
O
TBAHPBe6FDA
70,200 304 321 424 450
6FDA
O
O
O
O
H3C
CH3
O
O
TBAHPBeBPADA
41,700 245 316 396 434
O
O
BPADA
O
O
O
O
O
O
F3C
CF3
H2N
HO
NH2
OH
6FBAHPPeBPDA
6FBAHPPe6FDA
6FBAHPPeBPADA
185,200 295 300 382 427
97,200 280 300 411 469
85,200 234 290 384 436
O
O
6FBAHPP
BPDA
O
F3C CF3
O
O
O
O
O
6FDA
O
O
O
O
H3C
CH3
O
O
O
O
BPADA
O
O
O
O
F3C CF3
H2N
HO
NH2
OH
O
O
APAFeBPDA
73,200 322 330 452 508
APAF
BPDA
O
O
O
F3C CF3
O
O
O
APAFe6FDA
102,700 313 328 430 482
6FDA
O
O
O
O
H3C
CH3
O
O
APAFeBPADA
59,400 252 324 408 473
O
O
BPADA

M. Calle et al. / Polymer 53 (2012) 2783e2791
2787
Table 1 (continued )
a
Polymer code
Diamine
Dianhydride
Mw
Tg
TTR1 TTR2 TTR3
(ꢀC) (ꢀC) (ꢀC) (ꢀC)
CH3
O
F3C CF3
F3C CF3
O
O
O
H2N
H3C
NH2
CH3
H2N
HO
NH2
OH
O
APAF/DAMe
6FDA (5:5)
86,000 326 335 432 477
31,200 358 373 454 eb
285,900 306 335 428 473
413,900 304 332 430 469
O
APAF
6FDA
DAM
APAF/DAMe
6FDA (2:8)
O
F3C CF3
F3C CF3
O
O
O
H2N
HO
NH2
OH
H2N
O
NH2
O
O
APAF/ODAe
6FDA(5:5)
ODA
APAF
6FDA
APAF/ODAe
6FDA(2:8)
a
From the second trace of DSC measurements conducted with a heating rate of 20 ꢀC/min under nitrogen atmosphere.
Not detected.
b
From gel permeation chromatography (GPC), it was confirmed that
most of the polymers had high molecular weights (weight-average
molecular weights, Mw) as shown in Table 1. Therefore creasable
films could be prepared in every case by casting from polymer
solutions. Polymer structures were confirmed by 1H NMR. As an
example, the 1H NMR spectra of poly(ether-imide)s derived from
the newly synthesized TBAHPB bis(o-aminophenol), are compiled
in Fig. 1.
3.3. Thermal rearrangement temperature (TTR
)
As mentioned earlier, thermal rearrangement or conversion
temperature (TTR), is one of the crucial factors to design a cost-
effective thermal treatment process for TRePBO polymer
membranes. It has been widely reported that the thermal behavior
of HPI by TGA usually shows two distinct weight losses. The first
one appeared in the range of 300e500 ꢀC corresponding to the CO2
l
a
l
i
HO
O
h
O
O
k
g
e
c
N
N
O
OH
CH
j
f
O
O
O
O
CH
k
b
d
c, d, e
g
a
i
h
b
BPADA
f
j
i
a
g
e
i
HO
O
f
O
O
b
d
N
N
O
OH
h
f
O
O
g
b
c
a
e
d
c
h
BPDA
i
a
i
g
HO
O
f
O
O
d
h
N
N
O
OH
CF
e
O
f
O
CF
g
c
b
d
a
e
b
h
c
6FDA
8
6
2
1
10
(ppm)
δ
Fig. 1. 1H NMR (DMSO-d6, 300 MHz) spectra of TBAHPB bis(o-aminophenol) containing poly(ether-imide)s.

2788
M. Calle et al. / Polymer 53 (2012) 2783e2791
evolved in the rearrangement to PBO. The second one showed the
generalized decomposition of the polymer backbone at around
500e600 ꢀC. Furthermore, thermogravimetric analysis coupled
with mass spectroscopy (TGA-MS) provided evidence for the CO2
evolution by detection of the mass weight of 44 [11,12]. Based on
these observations, we concluded that it was possible to monitor, to
some extent, the thermal conversion by TGA. Thus, we have
recorded TGA data for all HPIs synthesized in this work, and
analyzed the thermograms in an attempt to explore the relation-
to combine with three aromatic dianhydrides (BPDA, 6FDA and
BPADA) to obtain a clear picture on how different chemical struc-
tures and rigidities in the nucleophilic o-hydroxy-diamine as well as
in the dianhydride monomers affect Tg and TTRs in the final precursor
polyimides. The highly rigid HABeBPDA polymer could not be
synthesized by solution thermal imidization (azeotropic imidiza-
tion), as were the rest of polyimides, because of premature precip-
itation during the imidization step.
The glass transition temperature, Tg, was strongly affected by its
chemical structure, as seen in Table 1 and Fig. 3. For three different
dianhydrides, the same trend of Tgs was found: HAB > APAF >
TBAHPB > 6FBAHPP. The rigid biphenyl structure of diamine HAB
yielded HPIs with the highest Tgs, whilst the large and very flexible
ether-containing 6FBAHPP diamine produced HPIs with the lowest Tg.
On the other hand, polymers containing BPDA dianhydride showed
higher Tg than those derived from 6FDA and BPADA dianhydrides. The
latter exhibited the lowest values in every case.
This trend can be clearly seen in Fig. 3, on DSC thermograms for
several of these polyimides. Note that, in most cases, Tg is followed
by an exothermic peak around 375e475 ꢀC, attributed to the
intramolecular cyclization to the carboxybenzoxazole intermediate
followed by decarboxylation to the final PBO [14]. Similar behavior
has been observed before for some ortho-substituted aramids
[20,21]. Thus, aramids containing a cyano, nitro or halogen group
ortho to the amide nitrogen, undergo thermal rearrangement at
high temperatures to benzoxazole polymers. This intramolecular
cyclization reaction is also characterized by an exothermic transi-
tion. Pearce et al. found that the temperature range for these
exothermic peaks was influenced to a certain extent by the polymer
structure [21]. Note from Fig. 3 that chemical structure also plays
a role in the exothermic transitions in HPIs.
Generally, the broad exothermic peaks slightly shift towards
high temperatures as Tg increases, the shift being more pronounced
for the most rigid polyimides, with the highest Tg values. The
observed enthalpy changes of the cyclodecarboxylation reaction
generally ranged between 3.0 and 13.3 J/g, although in some cases
(APAFeBPADA and APAFeBPDA polyimides), this enthalpy change
could not be clearly analyzed (see Fig. 3 and Table 2). HABe6FDA
polyimide, the most rigid polymer in this series (Tg ¼ 331 ꢀC),
exhibited the largest heat flow per mass for the rearrangement
reaction (13.3 J/g). On the contrary, 6FBAHPPeBPADA polyimide,
with the lowest Tg value (234 ꢀC), presented the smallest enthalpy
change (3.0 J/g). Hence, the most flexible BPADA containing
ship between chain mobility and Tg of HPI and/or TTR
.
In order to determine TTR by TGA, we defined three tempera-
tures with significant changes in the first slope in the TGA curve.
First, TTR1 was the starting temperature of the weight loss defining
the temperature at which polymer chains started rearranging.
Second, TTR2 was the temperature at the maximum point of weight
loss or maximum amount of CO2 evolution. Third, TTR3 was the
temperature at the end of the weight loss and showed the end of
the rearrangement process. The identification of these tempera-
tures was carried out in the first derivative thermograms (DTG).
Thus TTR1 was analyzed from the temperature at the point where an
increment in the DTG curve was detected, and TTR2 and TTR3 from
the temperatures at the peak and at the end of the weight loss
shown in the DTG curve. In some cases, simultaneous mass spec-
troscopy analysis (TGA-MS) was also carried out to confirm the CO2
evolution starting point. A typical TTR1 w TTR3 analysis procedure is
shown in Fig. 2. A summary of TTRs for all polymer samples is
compiled in Table 1.
3.4. Relationship between Tg and TTR
We have analyzed thermal conversion temperatures
(TTR1 w TTR3) for all polymers as a function of Tg to foresee the effect
of polymer structure and chain flexibility in the conversion process
and for a better understanding about the relationship between Tg
and TTRs.
3.4.1. Effect of chain flexibility
Four ortho-hydroxy diamines with different degrees of flexibility,
that is, HAB, APAF, TBAHPB and 6FBAHPP, were initially considered
Fig. 2. Typical TTR1 w TTR3 analysis procedure in TGA and DTG curves for HABe6FDA
polyimide with mass curve of CO2 on the bottom.
Fig. 3. DSC curves of HPI films, at a heating rate of 20 ꢀC/min in N2, determined by the
second heating scan. Glass transition temperature (Tg) values are shown for nine HPIs.

M. Calle et al. / Polymer 53 (2012) 2783e2791
2789
TBAHPBeBPDA (Tg ¼ 310 ꢀC) polyimides, although the resulting
polybenzoxazoles degraded at almost the same temperature.
Therefore, TTRs seem to be controlled by the type of dianhydride
moiety. This behavior generally agrees well with the previous
findings by Likhatchev and co-workers [1] that described a shift to
low temperatures for TTR1 and TTR2 when increasing flexibility of
the dianhydride moiety.
To further elucidate the relationship between chemical struc-
ture of HPI, Tg and TTRs, the effect of different degrees of flexibility in
the diamine unit was analyzed. For 6FDA containing polyimides,
the shift of the starting rearrangement temperature, TTR1, range
from 300 ꢀC for 6FBAHPPe6FDA polyimide (Tg ¼ 280 ꢀC), the most
flexible polymer in the 6FDA series, to 347 ꢀC for HABe6FDA
(Tg ¼ 331 ꢀC). Moreover, TTR2 also augments as a function of Tg
and is parallel to TTR1, in the range of 411 ꢀCe452 ꢀC. This trend in
TTR3 shows a deviation for TBAHPBe6FDA polyimide, with lower
TTR3 value than expected.
Table 2
Theoretical versus found carbon dioxide weight loss (in %) by TGA, as well as heat
flow per mass (J/g) by DSC (in N2), corresponding to the rearrangement reaction.
Polymer code
Tg (ꢀC)
D
(J/g)
H
Fw
CO2 wt. loss
theoretical
CO2 wt. loss
found
(g/mol)b
HABe6FDA
HABeBPADA
331
286
310
304
245
295
280
234
322
313
252
13.3
5.3
8.6
10.8
7.7
5.3
624.44
700.69
724.80
874.82
951.07
808.63
958.66
1034.90
639.48
777.47
850.71
14.09
12.55
12.14
10.06
9.25
10.88
9.18
8.50
14.17
11.14
17.01
13.14
10.20
10.97
9.86
TBAHPBeBPDA
TBAHPBe6FDA
TBAHPBeBPADA
6FBAHPPeBPDA
6FBAHPPe6FDA
6FBAHPPeBPADA
APAFeBPDA
5.5
3.0
7.93
a
_
13.76
11.32
10.34
13.50
11.25
13.47
APAFe6FDA
APAFeBPADA
3.3
a
_
a
Could not be analyzed from DSC thermograms.
Molecular weight of the repeating unit.
b
Fig. 5 shows DTG curves of 6FDA dianhydride containing poly-
imides. As seen, HPI with the TBAHPB diamine presented the
lowest thermal stability in the series, with degradation tempera-
ture of the resulting polybenzoxazole around 490 ꢀC. This fact has
been observed before for analogous polyimides containing di-tert-
butyl side groups [20], and can be attributed to the loss of tert-butyl
moieties that starts weight loss prior to the generalized degrada-
polymers seemed to show the lowest enthalpy change as compared
to 6FDA and BPDA counterparts. These results indicate that the
amount of
DH at the rearrangement process depends on the
mobility of polymer chain (Tg) and the weight loss during the
conversion process.
These transitions are actually related to the first weight loss
detected by TGA, corresponding to the evolution of CO2 during the
rearrangement process as depicted in Fig. 2. If we compare thermal
conversion temperatures (TTR1 w TTR3) and Tg for this set of poly-
imides derived from three dianhydrides BPDA, 6FDA and BPADA,
respectively, it can be clearly noticed that TTRs are greatly affected
by Tg. Hence, as Tg increases with rigidity of dianhydride for any
diamine, TTR1, TTR2 and TTR3 also move to higher temperatures.
Actually, TTR1 w TTR3 seem to increase with Tg and are parallel to
each other. For four different diamines, the same order of TTRs is
present: BPDA > 6FDA > BPADA. In fact, only 6FBAHPPeBPDA
polymer (Tg ¼ 295 ꢀC), in spite of its high Tg, shows a decrease in
TTR2 and TTR3 in comparison with 6FBAHPPe6FDA counterpart
(Tg ¼ 280 ꢀC). As an example, TGA thermograms of TBAHPB diamine
containing polyimides are shown in Fig. 4. Temperatures at the first
maximum weight loss or maximum rate of CO2 evolution, TTR2s, are
indicated as a guide. Note that, the first weight loss peak in the DTG
curve clearly moves to a high temperature, as the rigidity of the
dianhydride unit increases. Hence, the BPADA containing polyimide
(Tg ¼ 245 ꢀC) undergoes the cyclization reaction at lower tempar-
atures than the more rigid TBAHPBe6FDA (Tg ¼ 304 ꢀC) and
tion of the polymer chain. Unlike the rest of HPIs in the series, TTR3
,
for TBAHPBe6FDA seems to overlap with the early onset of
degradation of the polymer backbone. This fact could explain the
unexpected lower TTR3 value observed for this hydroxy-poly(ether-
imide) within this series.
Note that chemical structure and rigidity of diamine and dia-
nhydride monomers seem to play a major role in the conversion
from HPI to PBO. As a general trend, TTR1 w TTR3 shift to high
temperatures as a function of increasing monomer rigidity, and
thus, due to increasing Tg.
Hence, it was found that in every case TTR1 > Tg, meaning that
polymer chains started rearranging in the rubbery state, where
sufficient free volume and adequate segmental mobility already
existed for the rearrangement reaction to occur.
Another interesting fact is the range of detected TTR1 values
moving from 290 ꢀC for the most flexible polymer, 6FBAHPPeBPADA
(Tg ¼ 234 ꢀC), containing ether linkages in either, diamine and dia-
nhydride segments, to 347 ꢀC for the very rigid HABe6FDA
polyimide.
100
TBAHPB
6FDA
BPADA
80
BPDA
60
40
6FBAHPP
HAB
APAF
426
396
450
482
20
424
469
488
0
400
600
200
400
600
o
Temperature ( C)
o
Temperature
C
Fig. 4. TGA and DTG curves of TBAHPB diamine containing polyimides, at a heating
Fig. 5. DTG curves of 6FDA dianhydride containing polyimides, at a heating rate of
10 ꢀC/min in N2. TTR3 values are pointed out.
rate of 10 ꢀC/min in N2.

2790
M. Calle et al. / Polymer 53 (2012) 2783e2791
TTR2, the temperature at the maximum rate of CO2 evolution,
considered to be the most effective and appropriate temperature to
successfully accomplish the rearrangement process in TR-polymers,
is ranging between of 380e450 ꢀC (see Table 1). For a better
understanding between chemical structure of HPIs and TTR2, we
have also examined the gap between Tg and TTR2. TTR2eTg values
usually range from 100 to 150 ꢀC. On analyzing TTR2eTg as a func-
tion of Tg, it can be seen that, with few exceptions, the distance
between Tg and TTR2 increased with decreasing rigidity of dianhy-
dride. HPIs with the lowest Tg values within the series
(6FBAHPPeBPADA, Tg ¼ 234 ꢀC; TBAHHPBeBPADA, Tg ¼ 245 ꢀC;
APAFeBPADA, Tg ¼ 252 ꢀC) exhibited the largest TTR2eTg gaps.
Therefore, as observed for TTR1, TTR2 moves along with Tg, and
100
50
0
APAF/DAM-6FDA (2:8)
APAF-6FDA
APAF/DAM-6FDA (5:5)
430
432
328
454
T
TR2eTg increases for the most flexible polyimides in the series.
Temperature range of the rearrangement process, TTR, deter-
mined as TTR3eTTR1 was also considered for a more in depth anal-
ysis of the conversion process. Large TTR indicates broad
distribution of rearrangement process and vice versa. Nonetheless,
a clear trend between TTR and Tg does not seem to be present. Peak
335
373
D
300
450
600
750
o
Temperature ( C)
D
Fig. 6. TGA and DTG curves for APAF/DAMe6FDA series of copolyimides.
D
sharpening or broadening during CO2 evolution does not appear to
be directly governed by variations in the Tg, taking place indistinctly
along the polymer series.
Tg. In fact, Tg dropped from 313 ꢀC for APAFe6FDA homopolymer to
306 ꢀC for APAF/ODAe6FDA (5:5) and remained almost constant
when further augmenting ODA diamine content up to 80%
(Tg ¼ 304 ꢀC), which was the same as ODAe6FDA homopolymer
[22]. From these observations, it is difficult to recognize a trend, as
the chain rigidity might not be the only factor affecting Tg. As was
expected, these minor variations in Tg for this series of ODA copo-
lyimides, did not have significant repercussions on conversion
temperatures to PBO, and thus, TTR1 w TTR3 values of this copoly-
mers scarcely changed with Tg. TGA and DTG curves for APAF/
ODAe6FDA series clearly show this behavior (Fig. 7).
Notice that TTR1 w TTR3 of HPIs, regardless of the presence of
rigid DAM diamine are influenced by the changes in flexibility, Tg, In
addition, it is obvious that the addition of comonomer such as ODA
in HPIs does not significantly alter the Tg of HPIs and thus TTR
remains relatively unchanged for these polymers.
As pointed out above, rearrangement reaction of HPIs into PBOs
involves the evolution of two mol of carbon dioxide per repeat unit.
Theoretical weight loss corresponding to this evolution differs for
diverse polymer structures, decreasing as the molecular weight of
the repeating unit increases. Hence, for the four different diamines,
HPIs containing BPADA dianhydride, with the highest molecular
weight per repeating unit, showed the lowest CO2 weight losses.
Similarly, 6FBAHPP diamine containing HPI exhibited the smallest
values for every dianydride. Table 2 collects the actual and theo-
retical CO2 weight losses for this set of HPIs. With a few exceptions,
actual weight losses, determined from the first step in the TGA
curves, usually agree well with the calculated values. Observed
weight loss in HPIs containing TBAHPB diamine exceeded theo-
retical ones, probably due to some inherent instability for this
particular di-tert-butylated bis-o-aminophenol, as discussed
before.
To attain a complete and general picture on how Tg of HPI and
conversion to PBO temperature are related to each other,
TTR1 w TTR3 versus Tg for all the o-hydroxypolyimides described in
this work, have been plotted in Fig. 8. In this figure, TTR1 w TTR3 and
Tg generally show a linear relationship, signifying that as Tg
increases, TTR1 w TTR3 increase linearly.
3.4.2. Effect of diamines without hydroxyl group in copolyimides
From the findings discussed up to now, the starting conversion
temperature of the imide-to-benzoxazole, TTR1, as well as the
maximum rate of reaction (TTR2) and the final reaction temperature
(TTR3), seem to be controlled by the type of dianhydride and
diamine monomers. In order to better understand the relationship
between chemical structure, Tg and TTR, we intentionally attempted
to change flexibility of HPI precursor while retaining chemical
structure of the monomers involved in the rearrangement process
by copolymerization with two different diamines without hydroxyl
groups, namely, the highly stiff tri-methylated DAM diamine, and
the quite flexible ether-containing oxydianiline (ODA). Thus, by
incorporating DAM into the APAFe6FDA structure, the rigidity of
the polymer backbone increased. Two copolyimides with APAF and
DAM diamines with molar ratios of 5:5 and 2:8 were prepared.
Thus, Tgs for this series of copolymers increased gradually with
increasing mol fraction of DAM (see Table 1). Fig. 6 exhibits TGA and
DTG curves of this APAF/DAMe6FDA series. As seen, the first peak
due to the CO2 evolution during the conversion process, shifts to
high temperatures as the mol fraction of the more rigid comonomer
DAM increases. For the most rigid composition with the highest Tg
value, (APAF/DAMe6FDA (2:8), Tg ¼ 358 ꢀC), TTR3 becomes imper-
ceptible, reaching the onset of generalized decomposition of the
polymer backbone, around 500 ꢀC.
O
O
O
CF
CF
CF
CF
O
N
O
CF
CF
N
O
N
N
Y
X
O
O
O
HO
OH
100
50
0
APAF/ODA-6FDA (2:8)
APAF/ODA-6FDA (5:5)
APAF-6FDA
400
600
800
o
Temperature
C
By incorporating a more flexible diamine such as ODA in HPI, we
expected to improve polymer chain mobility, and thus to decrease
Fig. 7. TGA and DTG curves for APAF/ODAe6FDA series of copolyimides.

M. Calle et al. / Polymer 53 (2012) 2783e2791
2791
temperature to polybenzoxazole, which should be considered
when studying TR polymers.
500
400
300
Acknowledgment
T
TR3
Financial support by the Korea Carbon Capture and Sequestration
R&D Center under the Korea CCS2020 Program of the Ministry of
Education and Science and Technology, Republic of Korea and The
World Class University (WCU) program of the National Research
Foundation (NRF) of the Korean Ministry of Science and Technology
(No. R31-2008-000-10092-0) is gratefully acknowledged.
T
TR2
T
TR1
References
250
300
350
[1] Likhatchev D, Gutierrez-Wing C, Kardash I, Vera-Graziano RJ. Appl Polym Sci
1996;59:725e35.
o
Tg ( C)
[2] Tullos GL, Mathias L. J Polym 1999;40:3463e8.
[3] Tullos GL, Powers JM, Jeskey SJ, Mathias LJ. Macromolecules 1999;32:
3598e612.
Fig. 8. TTR
o-hydroxycopolyimides.
s values as a function of Tg for 15 o-hydroxypolyimides and
[4] Guzman-Lucero D, Likhatchev D. Polym Bull 2002;48:261e9.
[5] Schab-Balcerzak E, Jikei M, Kakimoto M. Polym J 2003;35:208e12.
[6] Chen BK, Tsai YJ, Tsay SY. Polym Int 2006;55:93e100.
[7] Okabe T, Morikawa A. High Perform Polym 2008;20:53e66.
[8] Park HB, Jung CH, Lee YM, Hill AJ, Pas SJ, Mudie ST, et al. Science 2007;318:
254e8.
[9] Choi JI, Jung CH, Han SH, Park HB, Lee YM. J Membr Sci 2010;349:358e68.
[10] Jung CH, Lee JE, Han SH, Park HB, Lee YM. J Membr Sci 2010;350:301e9.
[11] Park HB, Han SH, Jung CH, Lee YM, Hill AJ. J Membr Sci 2010;359:11e24.
[12] Han SH, Misdan N, Kim S, Doherty CM, Hill AJ, Lee YM. Macromolecules 2010;
43:7657e67.
[13] Kim SJ, Han SH, Lee YM. J Membr Sci 2012;403e404:169e78.
[14] Calle M, Lee YM. Macromolecules 2011;43:7657e67.
[15] Chern, Y.T.; U.S. Patent 2006;2006/0241187.
[16] Park KH, Kakimoto M, Imai Y. J Polym Sci Part A Polym Chem 1998;36:
1987e94.
[17] Imai Y, Maeda Y, Takeuchi H, Park KH, Kakimoto MA, Kurosaki T. J Polym Sci
Part A Polym Chem 2002;40:2656e62.
[18] Imai Y, Shibasaki Y, Takeuchi H, Park KH, Kakimoto MA. High Perform Polym
2002;14:253e60.
[19] Calle M, Lozano AE, de Abajo J, de la Campa JG, Alvarez C. J Membr Sci 2010;
365:145e53.
[20] Kim S, Pearce EM, Kwei TK. Polym Adv Technol 1989;1:49e73.
[21] Kim S, Pearce EM. Makromol. Chem. suppl. 1989;15:187e218.
[22] Kamaruddin HD, Koros WJ. J Polym Sci Part B Polym Phys 2000;38:2254e67.
4. Conclusions
Upon analyzing DSC and TGA data for a total of 15 sets of
o-hydroxypolyimides and copolyimides derived from two exper-
imental and four commercial structurally different diamines, as
well as three different dianhydrides monomers, we have
attempted to elucidate the relationship between glass transition
(Tg) of the precursor hydroxy-polyimide and the conversion
temperature to polybenzoxazole in film samples. Thus, thermal
rearrangement temperature, defined as TTR1, conversion starting
temperature of the imide-to-benzoxazole, and TTR2 and TTR3, the
maximum rate of reaction and the end of the rearrangement
process temperatures, respectively, showed dependence with Tg
of HPIs. In fact, TTR1 w TTR3 seemed to be influenced by the type of
dianhydride and diamine monomers. Heat flow during the
thermal rearrangement detected by DSC showed relationship
with Tg, but was not as clear as TTRs detected by TGA. It is found
that chain rigidity of HPI influences the thermal conversion