Optically transparent aromatic poly(ester imide)s with low coefficients of thermal expansion (1). Self-orientation behavior during solution casting process and substituent effect

Article abstract of DOI:10.1016/j.polymer.2015.07.026

A series of ester-linked tetracarboxylic dianhydrides were synthesized from trimellitic anhydride chloride and hydroquinone (HQ) analogs including various alkyl substituents. The properties of aromatic poly(ester imide)s (PEsI) prepared from the tetracarboxylic dianhydrides and 2,2′-bis(trifluoromethyl)benzidine (TFMB) were investigated for applications as novel plastic substrate materials in image display devices. The chemically imidized PEsI from trimethyl-substituted HQ was soluble even in less hygroscopic solvents such as cyclopentanone (CPN). Simple casting from the CPN solution provided a flexible and less colored film with a considerably low coefficient of thermal expansion (CTE = 11.5 ppm K^-1), a high T_g of 276 °C, high thermal stability (5% weight loss temperature in N₂: 453 °C), and very suppressed water uptake (W_A = 0.13%). The low CTE results from highly aligned main chains along the X-Y direction, which occurred during solvent evaporation after the PEsI solution was coated on a substrate. A mechanism is also proposed for such self-orientation behavior in this work. On the other hand, the thermally imidized counterpart was worthless in terms of low CTE and high transparency. Some conventional rigid tetracarboxylic dianhydrides were copolymerized with a minor content to di-tert-butyl- and diamyl-substituted PEsI systems. This approach was very effective not only for further improving some thermal properties (low CTE and high T_g) but also for suppressing film yellowness while keeping excellent solution-processability. Thus, some of the PEsIs developed in this work can be promising candidates as novel plastic substrate materials.

Full text of DOI:10.1016/j.polymer.2015.07.026

Polymer 74 (2015) 1e15
Contents lists available at ScienceDirect
Polymer
journal homepage: www.elsevier.com/locate/polymer
Optically transparent aromatic poly(ester imide)s with low
coefficients of thermal expansion (1). Self-orientation behavior during
solution casting process and substituent effect
Masatoshi Hasegawa*, Tomohiro Ishigami, Junichi Ishii
Department of Chemistry, Faculty of Science, Toho University, 2-2-1 Miyama, Funabashi, Chiba 274-8510, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of ester-linked tetracarboxylic dianhydrides were synthesized from trimellitic anhydride chlo-
ride and hydroquinone (HQ) analogs including various alkyl substituents. The properties of aromatic
poly(ester imide)s (PEsI) prepared from the tetracarboxylic dianhydrides and 2,20-bis(trifluoromethyl)
benzidine (TFMB) were investigated for applications as novel plastic substrate materials in image display
devices. The chemically imidized PEsI from trimethyl-substituted HQ was soluble even in less hygro-
scopic solvents such as cyclopentanone (CPN). Simple casting from the CPN solution provided a flexible
and less colored film with a considerably low coefficient of thermal expansion (CTE ¼ 11.5 ppm Kꢀ1), a
high Tg of 276 ꢁC, high thermal stability (5% weight loss temperature in N2: 453 ꢁC), and very suppressed
water uptake (WA ¼ 0.13%). The low CTE results from highly aligned main chains along the XeY direction,
which occurred during solvent evaporation after the PEsI solution was coated on a substrate. A mech-
anism is also proposed for such self-orientation behavior in this work. On the other hand, the thermally
imidized counterpart was worthless in terms of low CTE and high transparency. Some conventional rigid
tetracarboxylic dianhydrides were copolymerized with a minor content to di-tert-butyl- and diamyl-
substituted PEsI systems. This approach was very effective not only for further improving some ther-
mal properties (low CTE and high Tg) but also for suppressing film yellowness while keeping excellent
solution-processability. Thus, some of the PEsIs developed in this work can be promising candidates as
novel plastic substrate materials.
Received 4 June 2015
Received in revised form
14 July 2015
Accepted 18 July 2015
Available online 21 July 2015
Keywords:
Poly(ester imide)
Transparency
Coefficient of thermal expansion (CTE)
Solution processability
Self-orientation
Heat-resistant plastic substrates
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
cycles in the device fabrication processes (thermal dimensional
stability), film flexibility, and film-forming process compatibility
Optically transparent heat-resistant polymers are indispensable
materials for applications to various optoelectronic components,
e.g., plastic substrates in image display devices, liquid crystal
alignment layers, color filters, optical compensation films, optical
fibers, light-guiding plates, and optical lenses. One of the recent
important subjects is to replace current fragile inorganic glass
(solution-processability). In general, plastic substrates are superior
to current inorganic glass substrates in flexibility and thin-film
formability but inferior in heat resistance and thermal dimen-
sional stability.
Poly(ether sulfone) (PES) is known to possess the highest Tg
(225 ꢁC) among commercially available super engineering plastics.
However, PES is not adapted to the present purpose in terms of
short-term heat resistance (Tg) and thermal dimensional stability.
Plastic substrates with poor thermal dimensional stability undergo
significant thermal expansion/contraction during the multiple
heating/cooling cycles in the formation processes of indium tin
oxide (ITO) electrodes and thin-film transistors, which can be
responsible for a serious problem such as an ITO layer breakdown.
The most reliable high-temperature polymeric materials are
polyimides (PI) [1e4]. Some of aromatic PI systems also possess
excellent thermal dimensional stability based on the combination
substrates (300e700
mm thick) in image display devices by plastic
substrates (<50 m thick), thereby the display panels can become
m
drastically light and flexible. However, no reliable plastic substrate
materials are available yet because of difficultness of simulta-
neously achieving high levels of required properties, i.e., optical
transparency, heat resistance, dimensional stability against thermal
* Corresponding author.
E-mail address: mhasegaw@chem.sci.toho-u.ac.jp (M. Hasegawa).
http://dx.doi.org/10.1016/j.polymer.2015.07.026
0032-3861/© 2015 Elsevier Ltd. All rights reserved.

2
M. Hasegawa et al. / Polymer 74 (2015) 1e15
of their much higher Tg than the device processing temperatures
and the low linear coefficients of thermal expansion (CTE) along the
film plane (XeY) direction in the glassy region. However, conven-
tional aromatic PIs show intensive coloration arising from charge-
transfer (CT) interactions [5], which often disturbs their optical
applications. The coloration/decoloration mechanism of aromatic
PI films has been so far extensively studied from fundamental and
industrial interests [5]. One of the effective approaches for the
decoloration is to inhibit CT interactions by choosing non-aromatic
(cycloaliphatic) monomers either in diamines or tetracarboxylic
dianhydrides or both [6e16]. However, the use of cycloaliphatic
monomers causes some serious problems; semi- or wholly cyclo-
aliphatic PI films often possess poor thermal dimensional stability
based on their high CTE values (>60 ppm Kꢀ1) in the glassy region,
chloride (TMAC) in the presence of pyridine as an HCl acceptor in
anhydrous tetrahydrofuran (THF) as described in our previous work
[19], recrystallized from suitable solvents, and dried at 160 ꢁC in
vacuum for 12 h. The products were characterized by FT-IR and 1H
NMR spectroscopy, elemental analysis, and differential scanning
calorimetry (DSC). As a typical example, the analytical results for
the product from TMAC and 2,5-di-tert-butylhydroquinone
(25DBHQ, Fig.1) are as follows; melting point: 325 ꢁC (endothermic
peak), FT-IR (Jasco, FT/IR 4100, KBr, cmꢀ1): 3059 (CaromeH), 2959
(CalipheH), 1871/1780 (acid anhydride C]O), 1748 (ester C]O),
1487 (1,4-phenylene unit), 1H NMR [Jeol, JNM-ECP400, 400 MHz,
dimethyl sulfoxide (DMSO)-d6,
d
, ppm]: 8.70 [d, 2H, J ¼ 8.0 Hz, 6,60-
protons of phthalic anhydride (PA)], 8.61 (s, 2H, 3,30-protons of PA),
8.33 (d, 2H, J ¼ 8.0 Hz, 5,50-protons of PA), 7.41 (s, 2H, central di-
substituted phenylene), 1.30 (s, 18H, tert-butyl group). Anal. Calcd
(%) for C32H26O10 (570.54): C, 67.36; H, 4.59. Found: C, 67.19; H, 4.59.
An analog from trimethyl-substituted HQ (TMHQ, Fig. 1) was also
synthesized and characterized in a similar manner; FT-IR (KBr,
cmꢀ1): 3109 (CaromeH), 2933 (CalipheH), (1857/1782 (acid anhy-
dride C]O), 1735 (ester C]O), 1480 (1,4-phenylene unit), 1H NMR
0
even if their Tg s were very high (>300 ꢁC). Such high CTE arises
from practically three-dimensionally random chain orientation.
Most of cycloaliphatic monomers consist of non-linear/non-planar
steric structures [8,9]. As a result, the overall main PI chain linearity
is completely destroyed. Such distorted backbone structures do not
give rise to high extents of chain alignment along the XeY direction
(called “in-plane orientation”) upon thermal imidization process
(DMSO-d6, d
, ppm): 8.76e8.65 (m, 4H, 3,30,6,60-protons of PA),
[13,14,17].
Among
cycloaliphatic
monomers,
1,2,3,4-
8.33e8.30 (m, 2H, 5,50-protons of PA), 7.23 (s, 1H, central tri-
substituted phenylene), 2.16e2.09 (m, 9H, CH3), Anal. Calcd (%)
for C27H16O10 (500.41): C, 64.80; H, 3.22. Found: C, 64.64; H, 3.27.
Other relevant monomers were also synthesized in a similar
manner. In this work, the ester-containing tetracarboxylic dianhy-
drides were designated as TA-X (X ¼ hydroquinone analogs), e.g.,
“TA-TMHQ” in case of X ¼ trimethylhydroquinone (TMHQ).
cyclobutanetetracarboxylic dianhydride (CBDA) and trans-1,4-
cyclohexanediamine (t-CHDA) are rare cases with stiff/linear
structures. However, once these monomers were used, the result-
ing PIs lose solution-processability. On the other hand, a wholly
aromatic PI system derived from 4,40-(hexafluoroisopropylidene)
diphthalic anhydride (6FDA) and 2,20-bis(trifluoromethyl)benzi-
dine (TFMB) is known as a very limited case with high transparency
and good solubility, although this PI film owns no low CTE char-
acteristics, [18,19] on the basis of a non-linear/non-coplanar steric
structure of the 6FDA-based diimide unit [20]. Thus, the current
situation suggests how it is difficult to develop reliable plastic
substrate materials with the multiple target properties.
We have previously challenged to develop solution-processable
low-CTE transparent PI systems without using any cycloaliphatic
monomers and reported that the combination of an amide-linked
tetracarboxylic dianhydride including trifluoromethyl (CF3)
groups and TFMB led to excellent combined properties applicable
to the plastic substrate materials [19]. The results motivated us to
accomplish the target properties by using novel lower-cost aro-
matic tetracarboxylic dianhydrides without CF3 groups, which can
be prepared via a much simpler process as described later. The
present work highlights aromatic poly(ester imide)s (PEsIs) as the
key materials. We have so far studied low-CTE PEsIs with sup-
pressed water absorption for applications as higher-performance
dielectric substrates in flexible printed circuit boards (FPC)
[21e23]. In this case, the PI film transparency was not a great
matter of concern for FPC applications. However, we happened to
be aware that the coloration of the PEsI films was appreciably
weaker than that of conventional aromatic PIs and amide-
containing PIs [19,21]. This observation gave us an idea; an elabo-
rate chemical modification of the PEsIs may produce promising less
colored high-temperature plastic substrate materials. The present
work proposes a series of PEsIs modified by introducing various
alkyl groups and discusses a substituent effect on the target
properties.
2.1.2. Polyaddition, imidization, and film preparation
The sources, purification methods, and melting points of
monomers and raw materials are summarized in Supplementary
data 1. PEsI precursors [poly(amic acid)s (PAA)] were prepared by
equimolar polyaddition of tetracarboxylic dianhydrides and di-
amines in dry N,N-dimethylacetamide (DMAc) as described in our
previous papers [13e16,19]. The reaction scheme is shown in Fig. 2.
The formation of PAAs was confirmed from the transmission-mode
FT-IR spectra recorded on an FT/IR 4100 infrared spectrometer
(Jasco) using separately prepared thin cast films (4e6 mm thick)
with a non-uniform thickness to erase interference fringes. A
typical FT-IR spectrum is depicted in Fig. 3(a). Some specific bands
(cmꢀ1) are observed: 3259 (amide, NeH), 3057 (CaromeH), 2969/
2878 (CalipheH), 2630 (hydrogen-bonded COOH, OeH), 1735 (ester,
C]O), 1687/1534 (amide, C]O), 1490 (1,4-phenylene unit), and
1323 (CF3, CeF).
The PAAs were converted to PEsIs by different methods (thermal
and chemical imidization) (Fig. 2). When the imidized forms were
highly soluble, PAAs were chemically imidized as previously
described [14,16,19]. The imidized samples isolated as fibrous white
precipitates were re-dissolved in a fresh anhydrous solvent [e.g.,
cyclopentanone (CPN), triglyme (Tri-GL), or DMAc] at a solid con-
tent of 5e15 wt%. The homogeneous PEsI solutions were coated on
a glass substrate and dried typically at 60 ꢁC for 2 h in an air con-
vection oven and successively heated typically at 250 ꢁC for 1 h in
vacuum on the substrate. After peeling them off from the substrate,
the PEsI films (typically 20 mm thick) were annealed typically at
255 ꢁC/1 h in vacuum to remove residual stress. The thermal con-
ditions were properly adjusted for obtaining better quality of films.
The samples prepared via chemical imidization are denoted as “(C)”
from now on.
2. Experimental
2.1. Materials
PEsI films were also prepared upon thermal imidization; PAA
solution was bar-coated on a glass substrate, dried at 60 ꢁC for 2 h in
an air-convection oven, and heated typically at 200 ꢁC/
0.5 h þ 350 ꢁC/1 h in vacuum on the substrate, and successively
annealed in vacuum at 10e20 ꢁC lower temperatures than the Tg
2.1.1. Monomer synthesis
Ester-linked tetracarboxylic dianhydrides were synthesized
from various hydroquinone analogs (HQs) and trimellitic anhydride

M. Hasegawa et al. / Polymer 74 (2015) 1e15
3
Ester-containing tetracarboxylic dianhydrides
HQs (X)
TMAC
TA-X
Tetracarboxylic dianhydrides (Comonomers)
Diamine
Fig. 1. Structures of ester-containing tetracarboxylic dianhydrides and comonomers.
without the substrate to eliminate residual stress. The imidization
infrared bands, the appearance of some imide specific bands (cmꢀ1
)
are observed: 3074 (CaromeH), 2969/2879 (CalipheH), 1786 (imide-I,
C]O), 1734 (imide-II þ ester, C]O), 1492 (1,4-phenylene unit),
1372 (imide-III, NeCarom), 1310 (CF3, CeF), and 724 (imide-IV, ring
deformation). All other thermally imidized films also showed
similar spectral features, suggesting complete imidization.
In this work, the chemical compositions of PAA and PEsI systems
are represented by means of monomer abbreviations [tetra-
carboxylic dianhydrides (A) and diamines (B)] as A/B for homo-
polymers and A1;A2/B for copolymers.
and annealing conditions were properly adjusted to form better
quality of PEsI films. In this work, the thermally imidized film
samples are represented as “(T)” from now on.
For soluble PEsI systems, the completion of imidization was
confirmed from perfect disappearance of the proton signals for the
NHCO
[d ~ 10.0 ppm in DMSO-d6] and the COOH groups
(d
¼ 12e13 ppm) in the 1H NMR spectra. Full imidization was also
confirmed from the FT-IR spectra of the separately prepared thin
films. Fig. 3(b) displays a typical FT-IR spectrum of a thin film
prepared by solution casting of a chemically imidized powder
sample. In addition of complete disappearance of the PAA-inherent
Fig. 2. Reaction schemes of polyaddition and imidization of poly(amic acid)s, and expected end-capping of the terminal amino groups during chemical imidization.

4
M. Hasegawa et al. / Polymer 74 (2015) 1e15
were collected from the second heating run for removing an in-
fluence of adsorbed water.
2.2.3. Glass transition temperature
The storage modulus (E0) and the loss energy (E00) of PEsI films
were measured by the dyꢀn1amic mechanical analysis (DMA) at a
heating rate of 5 ꢁC min
on the same TMA instrument. The
measurements were conducted at a sinusoidal load frequency of
0.1 Hz with an amplitude of 15 gf in a nitrogen atmosphere. The Tg
of PEsI films was determined from a peak temperature of the E00
curve (method-1). The Tg was also determined from an inflection
point of the TMA curve, where the specimens started to abruptly
elongate under a fixed load, as an intersection of two tangential
lines (method-2).
2.2.4. Optical properties
The light transmission spectra of PEsI films (typically 20
thick) were recorded on an ultravioletevisible spectrophotometer
(Jasco, V-530) in the wavelength ( ) range of 200e800 nm. The light
transmittance at 400 nm (T400) and the cut-off wavelength (lcut
mm
l
)
where the transmittance becomes practically null were determined
from the spectra. The yellowness indices (YI, ASTM E 313) for PEsI
films were also determined from the spectra under a standard
illuminant of D65 and a standard observer function of 2ꢁ on a color
calculation software (Jasco) on the basis of the relation:
YI ¼ 100ð1:2985x ꢀ 1:1335zÞ=y
where x, y, and z are the CIE tristimulus values. YI takes zero for an
ideal white/transparent sample. The total light transmittance (Ttot
,
JIS K 7361-1) and the diffuse transmittance (Tdiff, JIS K 7136) of PEsI
films were measured on a double-beam haze meter equipped with
an integrating sphere (Nippon Denshoku Industries, NDH 4000).
The haze (turbidity) of PEsI films was calculated from the relation:
ꢀ
.
ꢁ
Haze ¼ Tdiff Ttot ꢂ 100
The in-plane (nin or nxy) and out-of-plane (nout or nz) refractive
indices of PEsI films were measured with a sodium lamp at
589.3 nm (D-line) on an Abbe refractometer (Atago, 4T, nD range:
1.47e1.87) equipped with a polarizer by using a contact liquid
(sulfur-saturated methylene iodide, nD ¼ 1.78e1.80) and a test
piece (nD ¼ 1.92). The birefringence of PEsI films, which stands for a
relative extent of chain alignment along the XeY direction, was
calculated from the relation:
Fig. 3. FT-IR spectra of thin films for TA-25DAHQ/TFMB system: (a) PAA and (b) PI.
2.2. Measurements
2.2.1. Inherent viscosity and molecular weight
The reduced viscosities (hred) of PAAs, which can be practically
regarded as hinh, were measured in DMAc at a solid content of
0.5 wt% at 30 ꢁC on an Ostwald viscometer. The number- (Mn) and
weight-average molecular weights (Mw) for PEsIs were determined
by the gel permeation chromatography (GPC) using THF as an
eluent at room temperature on a Jasco, LC-2000 Plus HPLC system
with a GPC column (Shodex, KF-806L) at a flow rate of 1 mL minꢀ1
by ultravioletevisible detection at 300 nm (Jasco, UV-2075). The
calibration was carried out using standard polystyrenes (Shodex,
SM-105).
Dnth ¼ nin ꢀ nout
2.2.5. Mechanical properties
The tensile modulus (E), tensile strength (sb), and elongation at
break (εb) of PEsI specimens (film dimension: 30 mm long, 3 mm
wide, typically 20
mm thick, specimen numbers > 15) were
2.2.2. Linear coefficient of thermal expansion
The CTE along the film plane (XeY) direction for PEsI specimens
(15 mm long, 5 mm wide, and typically 20 mm thick) in the glassy
measured on a mechanical testing machine (A & D, Tensilon UTM-
II) at a cross head speed of 8 mm minꢀ1 at room temperature. The
specimens were cut off from high-quality film samples
(10 cm ꢂ 10 cm) without any defects such as fine bubbles. The data
analysis was carried out on a data processing program (Softbrain,
UtpsAcS Ver. 4.09).
region were measured by the thermomechanical analysis (TMA) as
an averaꢀg1e in the range of 100e200 ꢁC at a heating rate of
5
ꢁC min on a thermomechanical analyzer (Netzsch, TMA 4000)
with a fixed load (0.5 g per a unit film thickness in
m
m, typically,10 g
The measurements for the other properties [thermal stability
(5% weight loss temperature, T5d), extent of water absorption (WA),
solubility, and wide-angle X-ray diffraction) were also conducted as
previously described [16,19].
load for 20 mm-thick films) in a dry nitrogen atmosphere. In this
case, after the preliminary first heating run up to 120 ꢁC and suc-
cessive cooling to room temperature in the TMA chamber, the data

M. Hasegawa et al. / Polymer 74 (2015) 1e15
5
3. Results and discussion
dianhydride (PMDA)] are used for lowering the CTE, it is usually
inevitable to deteriorate other desired properties (e.g., film trans-
parency and solution-processability). As illustrated in
Supplementary data 2, previous studies revealed that structural
modifications by introducing CF3 groups into conventional rigid
tetracarboxylic dianhydrides [PMDA and 3,30,4,40-biphenylte-
tracarboxylic dianhydride (s-BPDA)] were effective for improving
some focused properties (typically, dielectric constant) while
keeping low CTE characteristics, although the PI film transparency
was not always the target in these attempts [25,26]. 3,300,4,400-
p-Terphenyltetracarboxylic dianhydride (TPDA) [27e29] and
2,3,6,7-naphthalenetetracarboxylic dianhydride (NTDA) [17] are
also known as conventional rigid aromatic tetracarboxylic dia-
nhydrides profitable for low CTE generation. However, there are so
far no reports where these monomers were modified with some
substituents probably owing to a synthetic problem. On the other
hand, TA-HQ can be modified much more flexibly because various
substituted HQs are commercially available as the raw materials.
Thus, it is possible to widely vary the intermolecular forces upon
this approach while retaining the rigid structure inherent toTA-HQ.
3.1. Molecular design
The PI properties are primarily governed by the structures of
monomers chosen. Taking into account our goal (to simultaneously
achieve low CTE, high transparency, excellent solution-
processability, and sufficient film flexibility), the suitable diamine
monomers are practically limited to TFMB. A rod-like structure as
seen typically in TFMB is indispensable for low CTE generation. The
electron-withdrawing CF3 groups of TFMB contribute to a blue-shift
of the intramolecular CT absorption band, as a result, a decrease in
the film coloration [5]. The bulky CF3 groups with extremely low
polarizability are also valid for disturbing dense interchain stacking
and crystallization, as a result, they are often effective for solubility
improvement. The position of the CF3 substituents also affects the
solubility; it is most likely that TFMB consists of a distorted (non-
coplanar) biphenylene structure owing to steric hindrance between
the 2,20-substituted CF3 groups and the 6,60-hydrogen atoms [24].
The expected loose chain stacking is also advantageous for reducing
intermolecular CT interactions as a secondary factor of film color-
ation [5]. The replacement of the CF3 groups in TFMB by CH3 groups
(use of m-tolidine) inevitably causes intensive coloration because of
a decrease in the electron-withdrawing ability. An isomer of TFMB,
3,30-CF3-substituted counterpart is probably inferior to TFMB in the
polymerization reactivity with tetracarboxylic dianhydrides by a
steric hindrance effect of the 3,30-CF3 groups adjacent to the
functional amino groups.
The success of this work depends on the molecular design of
tetracarboxylic dianhydride monomers. Fig. 4 illustrates an impact
of para/meta isomerism at the diimide units on CTE. Here, the
diamine component is fixed to TFMB or p-PDA with linear struc-
tures. In fact, the rod-like or all para-linked PI systems
[Fig. 4(a)e(c)] have much lower CTE values than the isomeric sys-
tems with distorted backbone structures [(d)e(f)]. The results
imply how the “overall” main chain rigidity/linearity is important
for deriving low CTE characteristics. This means that not only the
diamines but also the tetracarboxylic dianhydrides should possess
rigid/linear structures. However, as far as conventional rigid aro-
matic tetracarboxylic dianhydrides [typically, pyromellitic
3.2. Properties of homo PEsIs
3.2.1. PEsIs from non- and mono-substituted TA-HQs
We have previously reported the thermal and mechanical
properties of thermally imidized PEsI films prepared from TFMB
with non-substituted and methyl/methoxy-substituted TA-HQs
without interests on the film transparency [22]. Then we began on
evaluating the film transparency of these systems as the first step of
the present work. The properties are listed in Table 1. Regardless of
the substituents, the TA-HQ analogs showed sufficiently high
reactivity with TFMB to form high molecular weights of PAAs with
hinh > 2 dL gꢀ1, which is much higher than an empirical criterion for
sufficient film-forming ability (hinh ~ 1 dL gꢀ1). The high hinh values
were also maintained after chemical imidization as in the other
substituted TA-HQs/TFMB systems, corresponding to actual high
molecular weights [e.g., hinh ¼ 5.20 dL gꢀ1, Mn ¼ 9.56 ꢂ 104,
Mw ¼ 2.19 ꢂ 105 for the chemically imidized TA-25DAHQ/TFMB
powder sample]. The chemical imidization process compatibility
could be the key to our goal because PI films prepared via this
(a)
(d)
CTE = –5 ppm K–1
CTE = 36 ppm K–1
(e)
(b)
CTE = 12 ppm K–1
CTE = 58 ppm K–1
(f)
(c)
CTE = 3 ppm K–1
CTE = 51 ppm K–1
Fig. 4. Impact of overall chain linearity on CTE: (aec) PIs derived from conventional rigid tetracarboxylic dianhydrides with rod-like diamines and (def) the counterparts from the
distorted tetracarboxylic dianhydride isomers. The broken arrows stand for main chain linearity.

6
M. Hasegawa et al. / Polymer 74 (2015) 1e15
Table 1
Film properties of homo PEsIs derived from TFMB with non-/mono-substituted TA-HQs.
Tetracarboxylic dianhydride
hinh (dL gꢀ1
)
Curec
Solventd (PI content)
Appearance of PI films
T400 (%)
lcut (nm)
CTE (ppm Kꢀ1
)
TA-HQ
TA-MHQ
2.93a
3.79a
4.64a
1.23b
3.87a
5.96a
2.63b
T
T
C
e
Slightly turbid
54.0
354
365
23.2
20.3
35.7
e
NMP (7 wt%)
Turbid
13.4
1.3
TA-MeOHQ
T
C
e
Turbid
Turbid
34.0
15.5
DMAc (8.5 wt%)
394
a
Data for PAA solutions just after polymerization.
Data for chemically imidized powder samples.
Cure (imidization) processes. T: thermal imidization, C: chemical imidization in homogeneous solution.
Solvents for re-dissolution of chemically imidized powder samples.
b
c
d
process tend to display appreciably better properties (higher
transparency and lower CTE) than the counterparts via the con-
ventional two-step process (thermal imidization of PAA films) as
discussed later [14,16,19]. The non-substituted TA-HQ/TFMB system
was not meet for the chemical imidization process because of
gelation based on poor solubility of the imidized form, which dis-
turbs full imidization. On the other hand, a positive effect of the
substituents (eCH3 and eOCH3) was observed; the solution ho-
mogeneity was kept during chemical imidization. In fact, the
chemically imidized powder sample (C) for TA-MHQ/TFMB was
soluble only in hot amide solvents [N,N-dimethylformamide (DMF),
DMAc, and N-methyl-2-pyrrolidone (NMP)] at a solid content of
1 wt%. The methoxy-substituted counterpart owned improved
valuable to our aim. On the other hand, the use of the dimethyl-
substituted isomer at different positions (TA-26DMHQ) enabled
us to perform chemical imidization in a homogeneous state. In this
system, the film transparency (T400) was significantly improved just
by altering the imidization route from thermal to chemical process
as shown in Table 2. The chemical imidization process was also
more profitable for decreasing CTE than thermal imidization as
observed in the TA-MeOHQ/TFMB system.
Trimethyl-substituted TA-TMHQ, which corresponds to a com-
bined structure of 2,3- and 2,6-dimethyl-substituted monomers,
was used as the next approach. The thermally imidized PEsI film
possessed no remarkable properties; poor transparency and a
common level of CTE (>50 ppm Kꢀ1). On the other hand, in spite of
the same chemical composition, the DMAc-cast PEsI film (C) ach-
ieved excellent combined properties: drastically enhanced trans-
parency (T400 ¼ 61.7%), a very low CTE (13.4 ppm Kꢀ1), a high tensile
modulus (5.7 GPa), sufficient film flexibility (εmb ax ¼ 14%), and high
thermal stability (T5d > 450 ꢁC in N2). This PEsI film (C) possessed a
single thermal transition (Tg) irrespectively of the analytical
methods: at 261 ꢁC by DMA (275 ꢁC by TMA). The low CTE and the
high tensile modulus are closely related to a high extent of in-plane
chain orientation in accordance with a high level of the Z-direction
solubility; it was soluble even in hot non-amide solvents [g-
butyrolactone (GBL), DMSO, m-cresol, and CPN] in addition to
common amide solvents. The chemical imidization process
compatibility also brought about another merit: an appreciably
decreased CTE (15.5 ppm Kꢀ1) for TA-MeOHQ/TFMB. However, the
use of these mono-substituted TA-HQs also caused a significant
decrease in the film transparency (T400) as listed in Table 1. In
particular, the methoxy-substituted system is beyond the present
candidates owing to its film coloration, which is probably related to
initial coloration of TA-MeOHQ monomer. The incorporation of the
methyl group toTA-HQ/TFMB also seems to somewhat decrease the
birefringence (Dnth ~ 0.1) as demonstrated from good correlations
(Dnth vs. CTE and modulus) in Fig. 5. Table 2 also reveals that the use
film transparency. However, the magnitude of a
red-shift by introducing the singular methyl group to TA-HQ/TFMB
l
cut-based spectral
of CPN (b.p. 131 ꢁC) as a casting solvent instead of DMAc (b.p.
165 ꢁC) was appreciably effective to improve the film transparency.
In our experience, less volatile amide solvents such as NMP (b.p.
202 ꢁC) are unfavorable from the viewpoint of the film trans-
parency. Such extra coloration as observed in NMP-cast PEsI films
may originate from a trace amount of a thermally decomposed
NMP residue. The CPN-cast PEsI film (C) for TA-TMHQ/TFMB also
displayed a further decreased CTE as listed in Table 2. The pre-
dominance of the chemical imidization process is undoubted from
Fig. 6 [chemical imidization (Y-axis) vs. thermal imidization (X-axis)
for the present systems]; the comparison for CTE revealed that the
data points are basically located below the Y ¼ X line where the CTE
values do not depend on the imidization methods at all, and those
for T400 are all positioned from the Y ¼ X line upward. The sup-
pressed coloration of the PEsI films (C) is probably attributed to not
only the film formation process at much lower temperatures than
for thermal imidization but also an end-capping reaction of the
thermally unstable terminal amino groups by Ac2O during chemical
imidization as depicted in Fig. 2. The control of film turbidity is also
an important subject. As mentioned above, the PEsI systems
derived from TFMB with non-, mono-, and di-substituted TAHQs
led to rather hazy films irrespectively of the imidization pathways.
On the other hand, the combination of the use of trimethyl-
substituted TA-TMHQ and chemical imidization allowed the for-
mation of a completely clear PEsI film with a very low haze value of
1.48% in contrast to the thermally imidized counterpart. Both of the
(
Dlcut ¼ 11 nm) was much lower than that by the incorporation of
the methoxy group (Dlcut ¼ 40 nm). The results motivated us to
investigate how plural methyl substituents influence the film
transparency, solubility, and the other target properties.
3.2.2. PEsIs from multi-methyl-substituted TA-HQs
As the next step, we investigated how the position/number of
the methyl groups influences the target properties with expecta-
tion of property improvement. The properties are summarized in
Table 2. Unexpectedly, the introduction of an additional methyl
group to TA-MHQ, i.e., the 2,3-dimethyl substitution destroyed the
chemical imidization process compatibility because of gelation.
Thermal imidization of the corresponding PAA film led to a highly
hazy and colored PEsI film, although it maintained a still relatively
low CTE (~40 ppm Kꢀ1). Similarly to non-substituted TA-HQ/TFMB,
the TA-23DMHQ/TFMB PEsI film (T) displayed double thermal
transition behavior with the E00 peaks at 340 ꢁC (Tg) and 212 ꢁC
(probably sub-Tg) in the DMA curve. In some PEsI systems, sub-Tg
appears distinctly around 210e230 ꢁC in the DMA curves [30].
Compared to DMA, TMA is often insensitive to sub-Tg, so that only Tg
can be detected as an inflection point corresponding to a softening
temperature in the TMA curves. Actually, a single transition (only
Tg) was observed at 358 ꢁC for this PEsI film (T) by TMA. The film
properties of TA-23DMHQ/TFMB indicate that this system is not

M. Hasegawa et al. / Polymer 74 (2015) 1e15
7
Table 2
Film properties of homo PEsIs derived from TFMB with di-/trimethyl-substituted TA-HQs.
Tetracarboxylic hinh (dL gꢀ1
dianhydride
)
Cure Solvent
(PI content)
Appearance of T400 (%)
PI films
lcut (nm)
D
nth Tg/sub-Tg CTE
E (GPa) εb ave/max sb (GPa) T5d (N2) Td5 (air)
f
(ꢁC)
(ppm Kꢀ1
)
(%)
(ꢁC)
(ꢁC)
c
c
TA-23DMHQ
TA-26DMHQ
2.39a
0.64a
0.95b
4.28a
T
T
C
T
C
C
Highly turbid
Turbid
0
0
432
—
—
340/212d 42.6
3.43
3.37
4.04
6/7
0.10
0.11
0.11
461
422
358/—e
—/232d
80.5
49.3
54.3
13.4
11.5
40/46
5/7
450
464
453
454
453
429
443
414
414
419
292/240e
CPN (11 wt%)
Slightly Turbid 39.5
359
433
360
358
0.069 —/217d
—/234e
c
TA-TMHQ
Turbid
DMAc (7.3 wt%) Clear
CPN (5.0 wt%) Clear
0
—
—/235d
328/240e
4.28a
6.57b
5.80a
8.44b
61.7
64.8
0.096 261/—d
275/—e
5.70
5.50
8/14
0.19
0.22
0.113 257/—d
276/—e
15/27
a
Data for PAA solutions just after polymerization.
b
Data for chemically imidized powder samples. For chemical imidization, the same PAA as that for thermal imidization was used unless the data for PAA was indicated in
the cell.
c
The refractive index measurements were difficult because of intensive turbidity.
Data determined by DMA (method-1).
Data determined by TMA (method-2).
Blank data mean that the transitions were not detected by the measurements.
d
e
f
turbid (T) and the clear PEsI film (C) for TA-TMHQ/TFMB are
regarded as being non-crystalline on the basis of a broad halo in the
WAXD pattern, as well as the turbid films of the other systems.
Thus, crystallization is not responsible for the film turbidity
observed here. The CPN-cast TA-TMHQ/TFMB film also possessed a
very low extent of water absorption (WA ¼ 0.13%) as another
feature.
light transparency and a high CTE value even when the film was
prepared via chemical imidization. The results might reflect that
the norbornane side group was too bulky to promote the casting-
induced in-plane orientation behavior responsible for low CTE
characteristics. If that is the case, further investigations by incor-
porating other bulkier substituents seem to be hopeless for aiming
at the present goal.
A thermally imidized PEsI film obtained from mono-tert-butyl-
substituted TA-BHQ and TFMB also gave some disappointing results
without remarkable properties except for a rather high εmb ax value.
On the other hand, a specific substituent effect on CTE was
observed when the TA-BHQ/TFMB film was prepared via chemical
3.2.3. PEsIs from TA-HQs with bulkier side groups
Some side groups bulkier than the methyl substituent were also
incorporated into TA-HQ/TFMB. Table 3 summarizes the film
properties. TA-NBHQ including an HQ-norbornane fused-ring
structure provided less valuable properties as judged from its poor
100
(a)
80
80
(a)
70
60
50
40
30
20
10
0
60
40
20
0
0
20
40
60
80
100
CTE (ppm K–1) (T)
0
0.05
0.1
0.15
0.2
7
6
5
4
3
2
100
80
60
40
20
0
(b)
(b)
0
0.05
0.1
0.15
0.2
0
20
40
60
80
100
Birefringence (Δnth)
T400 (%) (T)
Fig. 5. Relationships between birefringence (Dnth) and some properties: (a) CTE and
(b) tensile modulus. ( ) TA-HQs/TFMB homo PEsIs, ( ) TA-25DBHQ/TFMB-based co-
PEsIs, ( ) TA-25DAHQ/TFMB-based co-PEsIs, ( ) chemically imidized 6FDA/TFMB and
Fig. 6. Effects of imidization method on some properties: (a) CTE and (b) light
transmittance at 400 nm (T400). ( ) TA-HQs/TFMB homo and copolymer systems and
A
(
) PES.
(
) 6FDA/TFMB. The dotted line denotes the relation: Y ¼ X.

8
M. Hasegawa et al. / Polymer 74 (2015) 1e15
Table 3
Film properties of homo PEsIs derived from TFMB with TA-HQs containing bulky groups.
Tetracarboxylic hinh
dianhydride
Cure Solvent
(PI content)
Appearance T400 (%)
of PI films
lcut (nm)
D
nth Tg (ºC) CTE
E (GPa) εb ave/max sb (GPa) T5d (N2) (ꢁC) Td5 (air) (ꢁC)
(dL gꢀ1
)
(ppm Kꢀ1
)
(%)
c
TA-NBHQ
1.89a
2.63b
2.80a
3.24b
0.95a
1.18b
3.23a
5.20b
T
C
T
C
T
C
T
C
e
Turbid
DMAc (15.1 wt%) Turbid
Turbid
0
e
—
219d 67.3
230e
3.97
4.59
3.46
5.58
3.38
3.48
2.80
3.48
23/48
7/14
0.11
0.12
0.09
0.23
0.12
0.10
0.13
0.15
452
444
450
489
441
435
431
449
438
444
419
430
405
402
392
404
12.2
0
372
e
0.077 219d 48.4
238e
c
TA-BHQ
e
—
226d 56.8
232e
19/59
15/25
8/10
DMAc (10.7 wt%) Slightly
turbid
58.7
59.6
71.6
51.3
70.9
359
360
354
360
362
0.117 234d 10.1
243e
TA-25DBHQ
TA-25DAHQ
e
Clear
DMAc (15.1 wt%) Clear
Clear
DMAc (10.6 wt%) Clear
0.033 272d 57.6
284e
0.043 254d 33.5
264e
5/6
e
0.004 261d 74.6
283e
17/27
17/23
0.027 266d 42.6
270e
a
Data for PAA solutions just after polymerization.
b
Data for chemically imidized powder samples. For chemical imidization, the same PAA as that for thermal imidization was used unless the data for PAA was indicated in
the cell.
c
The refractive index measurements were difficult because of intensive turbidity.
Data determined by DMA (method-1).
Data determined by TMA (method-2).
d
e
imidization; it exhibited an outstandingly low CTE (10.1 ppm Kꢀ1
)
3.3. Influence of side group bulkiness
with a comparatively high T400 value of 58.7%. Among all the PEsI
films (C and T) examined in this work, only the TA-BHQ/TFMB film
(C) exhibited a sign of crystallization in the WAXD pattern.
The addition of one more tert-butyl group into TA-BHQ (i.e., TA-
25DBHQ) was effective for further improvement of the trans-
parency; casting from the DMAc solution led to an almost colorless
PEsI film with a high T400 (71.6%), a high Ttot (88.9%), and a very low
YI (2.65) although a certain degree of increase in CTE was
inevitable.
The TA-25DAHQ system (C) with bulkier amyl substituents
displayed the highest transparency in this work as demonstrated
from the following optical data: T400 ¼ 70.9%, Ttot ¼ 89.4%, and
YI ¼ 2.54, and Haze ¼ 1.69% for the DMAc-cast film.
Fig. 7 shows how the substituents affect some important
properties of the PEsI films (C). Here, the bars in the column charts
are roughly arranged in the order of the side group bulkiness.
Getting rid of highly turbid films, a trend was observed for the T400
-
based film transparency; the bulkier substituents seem to be
favorable for enhancing the film transparency [Fig. 7 (a)]. To cancel
an effect of intermolecular interactions on the transparency, the
ultravioletevisible absorption spectra in solutions were also
compared for the chemically imidized soluble PEsIs. The molar
extinction coefficient at 400 nm (ε400, in base Mꢀ1 cmꢀ1) in GBL at
room temperature increased in the following order: 3.84 (TA-
TMHQ/TFMB)
< 5.22 (TA-BHQ/TFMB) < 6.08 (TA-25DAHQ/
Table 4 summarizes the qualitatively evaluated solubility of the
chemically imidized powder samples. TA-25DAHQ/TFMB possessed
the highest solubility in this work; it was soluble even in non-
amide less hygroscopic solvents at room temperature similarly to
6FDA/TFMB known as a typically soluble PI. Thus, the exceedingly
bulky di-amyl groups behave like trifluoromethyl groups as a
typical solubility-promoting substituent.
TFMB) < 9.80 (TA-NBHQ/TFMB) < 13.1 (TA-MeOHQ/TFMB) < 13.7
(TA-25DBHQ/TFMB) < 15.9 (TA-26DMHQ). Actually, no clear cor-
relation between T400 (film) and ε400 (in GBL) was observed (the
figure is not shown here). This suggests that the enhanced film
transparency with increasing substituent bulkiness results from
disturbed intermolecular interactions rather than the intra-
molecular reason (weaker visible absorption of the isolated chains).
Table 4
The solubility of chemically imidized powder samples for homo TA-HQs/TFMB systems. The tests were carried out by adding the samples (10 mg) in solvents (1 mL).
Tetracarboxylic dianhydride
NMP
DMAc
DMF
DMSO
m-cresol
CPN
GBL
DOX
THF
CF
Acetone
Tri-GL
Temperature in solubility test upon heating (ꢁC)
150
150
140
150
150
130
150
100
60
50
50
150
TA-MHQ
þ
þ
þ
þ
e
e
þ
þ
þ
þ
þ
þ
þ
þ
e
þ
þ
e
þ
þ
þ
þ
e
e
e
e
e
e
TA-MeOHQ
TA-26DMHQ
TA-TMHQ
TA-NBHQ
TA-BHQ
TA-25DBHQ
TA-25DAHQ
6FDA
þ
þ
þ
e
e
a
a
þþ
þ
þþ
þ
þ
þ
þ
þ
þþ
þ
þ
e
e
a
þ
þ
þ
þ
þ
þ
þ
þ
e
e
e
e
a
a
b
b
b
b
þ
þ
þ
þ
þ
þ
e
e
e
e
b
þ
þ
þ
þ
þ
þ
þ
þ
þ
þþ
e
e
þ
b
b
b
þþ
þþ
þþ
þþ
þþ
þþ
þ
þ
þ
þþ
þþ
þþ
þþ
þ
þþ
þ
þþ
þþ
þþ
(þþ) Soluble at room temperature, (þ) soluble upon heating at established temperatures, ( ) deformed or swelled, and (ꢀ) insoluble. DOX ¼ 1,4-dioxane, CF ¼ chloroform.
a
Gelation after the hot homogeneous solution was kept at room temperature for a few days.
Gelation after the hot homogeneous solution was kept at room temperature for a few hours.
b

M. Hasegawa et al. / Polymer 74 (2015) 1e15
9
80
60
40
20
(a)
Turbid
Turbid
Turbid
0
5
(b)
4
3
2
60
(c)
50
40
30
20
10
0
Fig. 7. Influence of side group bulkiness on some properties for chemically imidized PEsIs: (a) T400 for PEsI films, (b) relative level of solubility (5-step evaluation, see Table 7), and
(c) CTE.
Prior to discussion on the solubility in Fig. 7(b), we show the
qualitatively tested solubility of the PEsI powder samples (C) in
various solvents (Table 4). In general, amide solvents have powerful
dissolution ability for common PIs, whereas less polar solvents such
as ether solvents (e.g., Tri-GL) are often far away from good sol-
vents. In this table, the solvents used are roughly arranged in the
order of the dissolution ability. The results revealed that the present
PEsIs (C) are essentially soluble in a variety of solvents. It is obvious
that the incorporation of bulkier substituents appreciably improves
the solubility as illustrated typically in TA-25DAHQ/TFMB. In this
work, the solubility was ranked on the basis of a criterion (see
Table 7). Indeed, the solubility level gradually is enhanced with
increasing side group bulkiness as illustrated in Fig. 7(b). Accord-
ingly, this effect can be also interpreted in terms of the weakened
intermolecular interactions accompanied with an increase in the
substituent bulkiness.
An influence of the substituents on CTE is depicted in Fig. 7 (c).
In contrast to the relatively clear trends of the transparency and the
solubility, no monotonous change was observed for CTE. However,
there seems to be some appropriate/favorable bulkiness (or mo-
lecular shape ?) of the substituents for reducing CTE. Consequently,
the diamyl and di-tert-butyl substituents were too bulky to pro-
mote the casting-induced in-plane orientation effect.
in contrast to the chemically imidized systems. The results are
consistent with a negative impact of substituents as observed in our
previous work [31]; the PI system derived from s-BPDA and 2,5-
dimethyl-1,4-phenylenediamine (DMPDA) with excessive alkyl
substituents showed no significant in-plane orientation behavior
upon thermal imidization in contrast to the non-substituted
counterpart (s-BPDA/p-PDA). The results of Fig. 8(b) probably
imply that the combination of backbone linearity/stiffness and
intimate interchain stacking is indispensable for reducing CTE upon
thermal imidization. It is reasonable to assume that the increase in
the substituent bulkiness disturbs dense chain stacking, conse-
quently weakens interchain interactions. On the basis of this
consideration, a gradual decrease in Tg [Fig. 8(a)] can be explained
without contradiction.
3.4. In-plane orientation behavior quite different by imidization
methods
As described above, it is undoubted that CTE is primarily
dominated by the degree of in-plane orientation in the PEsI films
[see Fig. 5(a)]. For further improvement of thermal expansion
property, it is important to examine how the in-plane orientation
proceeds concomitantly with solvent evaporation from the coated
PEsI solutions on substrates. Fig. 9 depicts a schematic illustration
for an in-plane orientational change during the film formation by
Fig. 8 accounts for the substituent bulkiness effect on CTE and Tg
for the thermally imidized PEsIs. As depicted in Fig. 8 (b), the CTE
almost monotonously increased as the substituents become bulkier
simple drying of
a coated PEsI solution (solid line). For a

10
M. Hasegawa et al. / Polymer 74 (2015) 1e15
Table 5
Film properties of TA-25DBHQ/TFMB-based copolymer systems.
Comonomer hinh (dL gꢀ1) Cure Solvent
Appearance T400 (%)
of PI films
lcut (nm)
D
nth Tg (ºC) CTE
E (GPa) εb ave/max sb (GPa) T5d (N2) (ꢁC) Td5 (air) (ꢁC) WA (%)
(mol%)
(PI content)
(ppm Kꢀ1
)
(%)
Homo PEsI
PMDA (0)
PMDA (30)
Randoma
0.95c
1.18d
1.79c
C
T
C
C
C
C
T
C
C
C
C
C
C
C
DMAc (15.1 wt%) Clear
71.6
2.2
354
386
365
365
369
369
374
372
358
383
358
357
363
361
0.043 254e 33.5
264f
3.48
3.84
5.21
4.98
5.23
4.36
3.64
4.34
5.26
5.26
3.88
4.77
4.18
4.58
5/6
0.10
0.13
0.18
0.18
0.21
0.16
0.14
0.16
0.17
0.17
0.14
0.18
0.19
0.21
435
439
461
465
472
463
444
484
483
468
460
426
454
458
402
408
421
422
428
430
412
432
427
432
419
418
418
421
—
Turbid
Clear
Clear
Clear
0.018 257e 50.4
315f
9/16
3.80d
CPN (7.0 wt%)
CPN (6.1 wt%)
CPN (8.0 wt%)
50.9
50.1
35.3
66.6
52.8
63.8
62.1
58.1
70.1
69.5
57.8
63.2
0.096 272e 13.0
298f
11/22
11/16
12/22
12/21
13/25
12/19
9/17
0.04
0.08
PMDA (30)
SCb
PMDA (40)
Random
s-BPDA (30)
Random
s-BPDA (50)
Random
2.42c
6.25d
2.14c
0.097 275e 12.1
295f
0.104 273e
292f
8.6
2.02c
4.19d
1.01c
CPN (10.8 wt%) Clear
Clear
DMAc (8.6 wt%) Clear
0.091 266e 18.0
282f
—
0.003 295e 56.0
310f
1.77d
0.061 263e 33.1
275f
TA-HQ (50)
Random
1.44c
2.53d
1.49c
3.27d
CPN (8.3 wt%)
CPN (8.2 wt%)
Turbid
Clear
0.107 237e 17.1
257f
NTDA (30)
Random
0.116 290e
302f
7.4
8/13
TA-BHQ (50) 1.11c
CPN (13.8 wt%) Clear
0.071 247e 25.2
255f
9/14
0.08
0.10
Random
1.71d
TA-BHQ (70) 2.19c
CPN (7.8 wt%)
CPN (8.1 wt%)
CPN (8.7 wt%)
Clear
Clear
Clear
0.098 245e 20.5
253f
18/37
13/24
18/35
Random
3.95d
TA-TMHQ (50) 2.79c
Random
0.085 270e 16.1
278f
TA-TMHQ (70) 2.62c
Random
0.091 264e 18.2
278f
a
Random copolymerization was carried out by adding mixed powder of tetracarboxylic dianhydrides into TFMB solution.
Sequence-controlled (SC) copolymerization was carried out by first adding TA-25DBHQ powder into TFMB solution, stirring 24 h, then adding PMDA powder.
Data for PAA solutions just after polymerization.
Data for chemically imidized powder samples. For chemical imidization, the same PAA as that for thermal imidization was used unless the data for PAA was indicated in
b
c
d
the cell.
e
Data determined by DMA (method-1).
Data determined by TMA (method-2).
f
comparison, the case of conventional two-step process (PAA casting
and successive thermal imidization, broken lines) is overlaid on this
figure. First, we mention the orientational behavior for the PI films
(T). An initial strategy for reducing CTE was to investigate the
structureeCTE relationship [32e34]. In order to fully understand
the principle of low CTE generation, initial interests have been
directed to a role of thermal imidization on the in-plane chain
alignment [31,35]. In particular, the s-BPDA/p-PDA system has
thoroughly been investigated since it is a practically useful heat-
resistant material with a very low CTE. Some experimental
Table 6
Film properties of TA-25DAHQ/TFMB-based copolymer systems prepared via chemical imidization.
Comonomer (mol%) hinh (dL gꢀ1) Solvent (PI content) T400 (%) nth Tg (ºC) CTE (ppm Kꢀ1) E (GPa) εb ave/max (%) sb (GPa) T5d (N2) (ꢁC) Td5 (air) (ꢁC) WA (%)
lcut (nm)
D
Homo PEsI
PMDA (0)
PMDA (30)
Random
PMDA (40)
Random
s-BPDA (30)
Random
s-BPDA (50)
Random
NTDA (30)
Random
TA-25DBHQ (25)
Random
TA-25DBHQ (50)
Random
3.23a
5.20b
3.65a
9.82b
2.37a
DMAc (10.6 wt%)
CPN (6.9 wt%)
70.9
362
368
363
370
372
383
359
358
355
0.027 266d 42.6
270e
3.48
4.82
4.86
4.06
4.50
3.49
3.38
3.70
3.82
17/23
10/22
11/25
13/20
14/23
15/28
12/23
10/20
10/19
0.15
0.17
0.18
0.17
0.18
0.15
0.14
0.15
0.15
449
454
452
430
465
481
401
399
441
404
415
410
423
436
405
398
398
408
46.3
57.3c
49.6
0.096 280d 14.7
289e
0.08
Tri-GL (5.3 wt%)
CPN (7.2 wt%)
0.086 269d 13.2
294e
4.82a
17.7b
2.75a
6.56b
2.19a
3.21b
1.92a
3.57b
3.30a
6.67b
3.62a
7.82b
64.5
65.0
53.1
69.1
71.1
70.2
0.096 262d 19.1
280e
0.09
0.06
0.31
0.11
0.11
0.14
DMAc (8.9 wt%)
CPN (9.1 wt%)
0.092 270d 20.4
284e
0.081 275d 15.8
290e
Tri-GL (9.0 wt%)
Tri-GL (9.2 wt%)
Tri-GL (8.0 wt%)
0.051 262d 34.8
266e
0.063 260d 27.7
269e
TA-25DBHQ (75)
Random
0.063 269d 27.0
273e
a
Data for PAA solutions just after polymerization.
Data for chemically imidized powder samples.
Data for Tri-GL-cast PEsI film.
Data determined by DMA (method-1).
Data determined by TMA (method-2).
b
c
d
e

M. Hasegawa et al. / Polymer 74 (2015) 1e15
11
Table 7
Criteria for 5-step evaluation of the target properties.
Properties
Parameters
Relative level
1
2
3
4
5
Transparency (Tr)
Low thermal expansion
property (LT)
Heat resistance (HR)
Toughness (To)
T400 (%)
<5
>70
20e30
60e50
40e60
45e35
70e75
30e20
>80
<10
CTE (ppm Kꢀ1
)
Tg (ꢁC)a
εmb ax (%)
<200
No film-forming
ability or <2
220e240
5e10
250e270
20e30
280e300
40e60
>350
>100
Solution-processability (SP) Qualitative solubility Insoluble upon heating Partially soluble Soluble in hot
Soluble in
Soluble in various
or swelled
amide solvents amide solvents at r.t. non-amide solvents at r.t.
or hot ether solvents
Samples soluble in hot CPN or hot GBL: SP-level ¼ 3.5.
a
Data determined by DMA (method-1).
evidence suggests that the in-plane orientation for s-BPDA/p-PDA
proceeds as depicted in Fig. 9(b) (1T) in the simplified heating
process [60 ꢁC/2 h þ 250 ꢁC/2 h, Fig. 9(a)] [31]; the in-plane chain
alignment is drastically enhanced in the thermal imidization pro-
cess. In contrast, PI systems with flexible/distorted backbone
structures (e.g., s-BPDA/4,40-ODA) undergo orientational relaxation
upon thermal imidization [Fig. 9 (2T)]. On the other hand, in a
system where the backbone itself remains rigid but with excessive
alkyl substituents (e.g., s-BPDA/DMPDA), the in-plane orientation
increases only slightly upon thermal imidization [Fig. 9 (3T)]. A
similar situation is also observed in the present TA-TMHQ/TFMB
system (T) as suggested from its high CTE (54.3 ppm Kꢀ1, Table 2).
However, chemical imidization of TA-TMHQ/TFMB swept away a
reputation as a “worthless material” as stated above; the curve (3C)
represents that the PEsI chains for TA-TMHQ/TFMB can highly align
along the film plane just by drying at 60 ꢁC for 2 h. Then, the half-
dried PEsI film obtained probably undergoes additional orienta-
tional enhancement by fully drying it on the substrate at 250 ꢁC
owing to an expected apparent stretching effect concomitant to a
slight thickness decrease as mentioned later. This hypothesis is
based on our previous observation; i.e., melt-induced in-plane
orientation behavior by annealing above the Tg on the substrates for
s-BPDA/TFMB [36]. Indeed, the substrates must play a great role
when annealed at 250 ꢁC in the present case; if the substrate was
removed after drying at 60 ꢁC for 2 h, the half-dried TA-TMHQ/
TFMB PEsI film most likely suffers somewhat orientational relaxa-
tion by heating at 250 ꢁC (3C′) as suggested from the fact that
CTE ¼ 13.0 ppm Kꢀ1 in case of heating on the substrate and
CTE ¼ 22.3 ppm Kꢀ1 off substrate for TA-25DBHQ(70); PMDA(30)/
TFMB copolymer.
(a)
400
350
300
250
200
Sub-Tg ?
150
90
(b)
80
70
60
50
40
30
20
10
0
Fig. 8. Influence of side group bulkiness on some properties for thermally imidized PEsIs: (a) Tg by TMA (method-2) and (b) CTE.

12
M. Hasegawa et al. / Polymer 74 (2015) 1e15
chains are highly linear/rigid. This hypothesis is based on our
previous observation; a liquid crystal-like ordered structure at the
PAA stage obviously contributes to a decrease in CTE [37]. If such a
local ordered structure (maybe undetectable on a polarizing optical
microscope) could be formed with solvent evaporation prior to
solidification in the present systems, it is likely that the in-plane
orientation is intensified cooperatively with the subsequent weak
stretching effect.
Thus, the chain orientation mechanism in the film formation
process (3C) seems to be quite different from that in the process
(1T). However, in this work we tried to find out some common
feature between them from the viewpoint of “molecular mobility”
during the film formation processes. Our previous studies revealed
that the degree of in-plane orientation is affected by some struc-
tural factors (backbone linearity/stiffness) and processing condi-
tions (heating rate, film thickness, solvents and amounts of residual
solvents in PAA films) during thermal imidization [31,36]. It is
accepted that the residual amide solvents behave as a plasticizer
[38]. Accordingly, the following factors can be regarded to be
equivalent in terms of “molecular mobility” during thermal imid-
ization: (increases in imidization temperature, heating rate, film
thickness, and residual solvent content, and a decrease in solvent
volatility). Fig. 10(a) schematically depicts the in-plane orienta-
tional behavior affected by molecular mobility during thermal
imidization, although actual situation is much more complicated
(a)
1
2
4
3
Molecular mobility during thermal imidization
Fig. 9. Changes in coating, drying, and heat treatment processes: (a) simplified tem-
perature profile and (b) presumed changes in the degree of in-plane orientation during
thermal imidization (broken lines, 1T: s-BPDA/p-PDA, 2T: s-BPDA/4,40-ODA, 3T: s-
BPDA/DMPDA) or TA-TMHQ/TFMB and solution casting of chemically imidized PEsIs
(solid lines, 3C: TA-TMHQ/TFMB on substrate, and 3C0: TA-TMHQ/TFMB off substrate
after drying at 60 ꢁC for 2 h).
Bulkiness of substituents (b)
6
5
5 < 6 < 7
7
Both of the present system (3C) and the conventional system
(1T) finally bring out low CTE. However, the shape of the curve (i.e.,
orientational behavior) for the former is quite different from that
for the latter. It is undoubted that the thermal imidization reaction
for the latter acts as a trigger for the orientational enhancement.
The trigger is attributed to an “apparent stretching” effect, which
generates by inhibiting significant film contraction on the substrate
[31]. On the other hand, there seems to be no distinguished driving
force appropriate for the actual pronounced in-plane orientation in
the sample (3C); although there must be another type of stretching
effect accompanied with thickness decrease during solvent evap-
oration as described in our previous paper [16] (see Supplementary
data 3), it is probably too weak as the driving force as inferred from
our experience that simple casting from an NMP solution of PES
Molecular mobility during PEsI solution casting
(c)
Total bulkiness of substituents
Intermolecular
force
Molecular
mobility
Fig. 10. Schematic diagrams for the degree of in-plane chain orientation as influenced
by some factors: (a) molecular mobility during thermal imidization of PAA films, (b)
molecular mobility during solution casting of chemically imidized PEsIs, and (c) total
bulkiness of substituents in the present PEsIs. 1: PMDA/p-PDA, 2: s-BPDA/p-PDA, 3: s-
BPDA/4,40-ODA, 4: s-BPDA/DMPDA or TA-TMHQ/TFMB, 5: TA-MHQ/TFMB, 6: TA-
TMHQ/TFMB, 7: TA-25DAHQ/TFMB.
caused no appreciable in-plane orientation (almost zero
Dnth).
Therefore, other effective driving forces should be considered to
account for the curious phenomenon in the sample (3C). A possible
idea assumes the presence of some local ordered structure
(Supplementary data 3), which is favorably formed when the main

M. Hasegawa et al. / Polymer 74 (2015) 1e15
13
because of significantly varying molecular mobility with the
progress of imidization. Rod-like PI systems (e.g. PMDA/p-PDA),
which is unfavorable in terms of molecular mobility, should be
prepared under severer conditions (for example, at higher imid-
ization temperatures or with higher heating rates) for maximizing
the in-plane orientation. On the contrary, flexible PI systems (e.g., s-
BPDA/4,40-ODA) prefer milder imidization conditions to suppress
orientational relaxation. On the other hand, the in-plane orienta-
tion can be maximized under a thermal imidization condition with
a medium intensity of molecular mobility for semi-rigid PI systems
(e.g., s-BPDA/p-PDA). Thus, such optimum intensity of the molec-
ular mobility for maximizing the in-place orientation shifts with
the backbone linearity/stiffness.
On the other hand, we also found out some factors advanta-
geous for enhancing the in-plane orientation by simple casting of PI
solutions, i.e., the increases in PI molecular weights, solvent vola-
tility, and the decreases in drying temperature and heating rate
[16]. The results suggest that the in-plane orientation tends to in-
crease with decreasing molecular mobility during solution casting
(under milder conditions). Therefore, the in-plane orientation can
be roughly represented as a monotonously decreasing curve with
increasing molecular mobility [Fig.10(b)], although the actual curve
may possess a “peak” at a low molecular mobility. This figure also
reflects the results of the present PEsI systems with three different
substituents (mono-methyl, tri-methyl, and di-amyl groups).
Fig. 10(c) suggests the presence of an optimum substituent size for
maximizing the casting-induced in-plane orientation. Assuming
that an increase in the substituent bulkiness corresponds to a
decrease in intermolecular forces, in other words, an increase in
molecular mobility during the casting process, there seems to be a
roughly common feature in the in-plane orientation behavior even
when the films were prepared via different routes [see the curve in
Fig. 10(c) and curve 2 in Fig. 10(a)]. Thus, seeking out an optimum
condition from the viewpoint of molecular mobility must be the
strategy common to the different imidization methods (C and T).
improved by using PMDA comonomer (see Supplementary data 4)
against our initial concern coming from the fact that homo PMDA/
TFMB is essentially insoluble. As expected, an increase in the PMDA
content to 40 mol% further decreased the CTE, but it also caused
inevitably a significant deterioration in the transparency.
The control of chain sequence occasionally brings about a pos-
itive effect on some properties. For example, sequence-controlled
copolymerization consisting of PMDA, 4,40-ODA, and p-PDA pro-
vides lower CTE and higher εb than the corresponding random
copolymers [40]. Besides, a siloxane-containing block-like copol-
ymer produces a much lower modulus than the corresponding
random copolymer [41]. Such technique was also applied to the
present copolymer system. Unfortunately, the effect of sequence
control was not so pronounced at a PMDA content of 30 mol%
(Table 5).
s-BPDA was also used as another comonomer. The copolymer
film (C) at s-BPDA ¼ 30 mol% afforded well-balanced properties; a
low CTE (18.0 ppm Kꢀ1) close to that of copper foil and relatively
high transparency (T400 ¼ 66.6%), although the modification effect
got inconspicuous at s-BPDA ¼ 50 mol%. A similar positive copo-
lymerization effect was also observed by using TA-HQ comonomer.
On the other hand, it should be noted that the modification using
NTDA (30 mol%) is capable of drastically decreasing the CTE
(7.4 ppm/K) and significantly increasing the Tg (290 ꢁC) while
maintaining excellent solubility (see Supplementary data 4) and
good transparency, which is obviously higher than for the corre-
sponding PMDA-modified copolymer.
The copolymer systems modified with conventional rigid
monomers (PMDA, s-BPDA, and NTDA) showed a CTEeDnth curve
different from that of the homo substituted TA-HQs/TFMB as
depicted in Fig. 5 (a). The downward-shifted curve for the co-
polymers may imply that these conventional rigid monomers have
distinctly greater ability for enhancing the in-lane orientation than
the substituted TA-HQs.
As mentioned above, two homo PEsI systems (C), i.e., TA-BHQ/
TFMB and TA-TMHQ/TFMB possess considerably low CTE values
(~10 ppm Kꢀ1). According to the results, we also used TA-BHQ and
TA-TMHQ as comonomers for improving the insufficient thermal
expansion property of TA-25DBHQ/TFMB; they were copoly-
merized with a high content of 50 or 70 mol%. This approach was
also effective for affording well-balanced properties with respect to
low CTE and high transparency. In particular, the modification us-
ing TA-BHQ is superior in the transparency to that with TA-TMHQ;
the former is capable of lowering the CTE without practical dete-
rioration of the transparency as shown in Table 5.
3.5. Property improvement by copolymerization
3.5.1. TA-25DBHQ/TFMB-based copolymers
As described above, the pronounced feature of the TA-25DBHQ/
TFMB system (C) is outstanding transparency although there is
room for further improvement of the thermal expansion property
(CTE ¼ 33.5 ppm Kꢀ1). Copolymerization using some rigid tetra-
carboxylic dianhydrides with a minor fraction may contribute to a
decrease in the CTE while avoiding significant deterioration of the
transparency. Table 5 summarizes the film properties of TA-
25DBHQ/TFMB-based copolymer systems.
Our first attempt was to copolymerize using PMDA known as
the most familiar rigid tetracarboxylic dianhydride, which is based
on the fact that the homo PMDA/TFMB system brings about a
considerably low CTE upon thermal imidization [17,39]. Against our
expectation, no significant effect on CTE was observed by random
copolymerization (PMDA ¼ 30 mol%) when the film was prepared
via thermal imidization. This probably reflects too high molecular
mobility during thermal imidization in this copolymer, which ari-
ses from weakened intermolecular force by the bulky substituents
(di-tert-butyl groups), thereby orientational relaxation becomes
preferential. The PMDA-modified copolymer film (T) was also
poorly transparent. However, in contrast to such worthless results,
the chemically imidized counterpart exhibited a significantly
decreased CTE (13.0 ppm Kꢀ1) while maintaining relatively high
transparency. The suppressed water uptake is also another feature
of this system, which probably results from the presence of hy-
drophobic substituents (CF3 and the di-tert-butyl groups). It should
be noted that the solubility of the powder sample (C) was obviously
Besides, among the TA-25DBHQ/TFMB-based copolymer sys-
tems (C), the solubility was most effectively enhanced when s-
BPDA was used as a property modifier (see Supplementary data 4).
3.5.2. TA-25DAHQ/TFMB-based copolymers
The predominant feature of TA-25DAHQ/TFMB (C) is its distin-
guished solubility in addition to its outstanding transparency.
Because of the excellent solubility, solution casting can be made
even by means of Tri-GL with a much lower dissolution power than
amide solvents. Empirically speaking, the film coloration also de-
pends on casting solvents, and it tends to increase in the following
order: Tri-GL < CPN < DMAc << NMP. Therefore, the Tri-GL-process
compatibility has great superiority in enhancing the transparency.
The properties of TA-25DAHQ/TFMB-based copolymers are
summarized in Table 6. For the TA-25DAHQ/TFMB copolymer
modified with 30 mol% PMDA, solution casting using CPN caused a
significantly decreased CTE although the transparency
(T400 ¼ 46.3%) somewhat deteriorated. On the other hand, the use
of Tri-GL instead of CPN improved indeed the transparency
(T400 ¼ 57.3%). Even when the PMDA content was increased to

14
M. Hasegawa et al. / Polymer 74 (2015) 1e15
Modifications using s-BPDA and NTDA comonomers produced
15
10
5
well-balanced properties (low CTE, maintained transparency, very
low amounts of water absorption, and sufficient εb, Table 6) without
sacrificing the solubility (see Supplementary data 5).
The copolymerization approach also produced another positive
effect on film yellowness control. A linear relationship is roughly
observed between T400 and YI as depicted in Fig. 11. It should be
noted that some of the data points for the TA-25DAHQ/TFMB-based
copolymers with minor fraction of PMDA or s-BPDA (surrounded by
broken circle) deviate distinctly downward from the YIeT400 cor-
relation (broken line). This implies that the use of these rigid co-
monomers with minor fractions was very effective to restrain the
film yellowing although the discernible decreases in T400 were
actually inevitable as mentioned above. A similar effect was also
observed for TA-25DBHQ/TFMB-based copolymers. The unexpect-
edly suppressed film yellowness for these copolymers is closely
related to the absence of a weak absorption band (tail) in the
wavelength range (420e480 nm), which corresponds to the com-
plementary colors of yellow [19].
5
4
1
3
0
40
50
60
70
80
T400 (%)
Fig. 11. Correlation between T400 and YI: (1) TA-25DAHQ(70);PMDA(30)/TFMB, (2) TA-
25DAHQ(60);PMDA(40)/TFMB, (3) TA-25DBHQ(70);PMDA(30)/TFMB (random), (4) TA-
25DBHQ(70); s-BPDA(30)/TFMB, and (5) TA-25DBHQ(50); s-BPDA(50)/TFMB.
100
Target Area
90
80
70
60
50
40
30
20
10
0
3.6. Position of our PEsIs as low-CTE transparent materials and
performance balance
1
2
Fig. 12 displays a CTEetransparencyesolubility diagram, which
stands for the status of our PEsIs with respect to the practical use.
The target area (CTE ꢃ 15 ppm Kꢀ1 and T400 ꢄ 80%) is also indicated
in this figure. PES and all other conventional colorless engineering
plastics are removed from the present candidates because of their
too high CTE. Most of conventional or previously reported colorless
PIs (e.g., 6FDA/TFMB and cycloaliphatic PIs) possess no satisfacto-
rily low CTE [13e15,18,19]. A general trend is observed in this
figure; conventional PIs abruptly lose the transparency and the
solubility below CTE ~20 ppm Kꢀ1 as typically in the s-BPDA/p-PDA
system. The s-BPDA/t-CHDA system is a limited case possessing
high transparency and low CTE [29]. However, its essential insol-
ubility prevents us from applying the chemical imidization process.
On the other hand, it should be noted that our PEsIs (B, Fig. 12) was
capable of escaping from such general tendency and simulta-
neously accomplished low CTE, high transparency, and excellent
solution-processability. For example, the data points for TA-TMHQ/
TFMB (#1) and TA-25DBHQ(70); NTDA(30)/TFMB (#2) are located
Soluble &
Insoluble
Area
Insoluble
Area
-10
-10
0
10
20
30
40
50
60
70
CTE (ppm K−1)
Fig. 12. CTEetransparencyesolubility diagram: (B) some selected systems of this
work, 1: TA-TMHQ/TFMB and 2: TA-25DBHQ(70); NTDA(30)/TFMB], ( , ) soluble PIs,
) insoluble PIs, ( ) PES. Open and closed symbols denote soluble and
A
insoluble materials, respectively.
◊
▫
(-,
,
40 mol%, the copolymer system could be processed using Tri-GL
owing to the originally outstanding solubility of the pristine TA-
25DAHQ/TFMB system.
Tr
Tr
SP
SP
Tr
LT
LT
To
To
HR
HR
(a)
(b)
Tr
Tr
LT SP
LT SP
LT
SP
To
To
To
HR
HR
HR
(c)
(d)
(e)
Fig. 13. Performance balance represented by 5-level evaluation for the systems derived from TFMB with different tetracarboxylic dianhydrides: (a) TA-TMHQ, (b) TA-25DBHQ(70);
NTDA(30), (c) CBDA, (d) 6FDA, and (e) PMDA. The criteria and abbreviations of the properties are shown in Table 7.

M. Hasegawa et al. / Polymer 74 (2015) 1e15
15
near the target area as indicated in Fig. 12.
References
We also compared performance balance for our PEsIs with that
for other conventional TFMB-derived systems. The properties were
evaluated with five levels on the basis of criteria listed in Table 7. As
depicted in Fig. 13, the spider chart for CBDA/TFMB is expanded
biasedly on the right side owing to the absence of solution-
processability and film toughness. The 6FDA/TFMB system almost
fulfills the present target properties except for the low CTE char-
acteristics. PMDA/TFMB also gives a significantly deformed spider
chart because it does not satisfy the required properties except for
the low-CTE and high-Tg characteristics. The results manifest how it
is difficult to actualize good performance balance for the required
properties. On the other hand, it is obvious that TA-TMHQ/TFMB, as
well as TA-25DBHQ(70); NTDA(30)/TFMB copolymer, has an
outstandingly well-balanced performance.
[1] M.I. Bessonov, M.M. Koton, V.V. Kudryavtsev, L.A. Laius (Eds.), Polyimides:
Thermally Stable Polymers, Plenum, New York, 1987.
[2] M.I. Bessonov, V.A. Zubkov (Eds.), Polyamic Acid and Polyimides: Synthesis,
Transformation and Structure, CRC Press, Boca Raton, FL, 1993.
[3] S. Ando, M. Ueda, M. Kakimoto, M. Kochi, T. Takeichi, M. Hasegawa, R. Yokota
(Eds.), The Latest Polyimides: Fundamentals and Applications, second ed., NTS,
Tokyo, 2010.
[4] C.E. Sroog, Prog. Polym. Sci. 16 (1991) 561e694.
[5] M. Hasegawa, K. Horie, Prog. Polym. Sci. 26 (2001) 259e335.
[6] H. Suzuki, T. Abe, K. Takaishi, M. Narita, F. Hamada, J. Polym. Sci. Part A 38
(2000) 108e116.
[7] W. Volksen, H.J. Cha, M.I. Sanchez, D.Y. Yoon, React. Funct. Polym. 30 (1996)
61e69.
[8] T. Matsumoto, T. Kurosaki, Macromolecules 30 (1997) 993e1000.
[9] T. Matsumoto, Macromolecules 32 (1999) 4933e4939.
[10] H. Seino, T. Sasaki, A. Mochizuki, M. Ueda, High. Perform. Polym. 11 (1999)
255e262.
[11] J. Li, J. Kato, K. Kudo, S. Shiraishi, Macromol. Chem. Phys. 201 (2000)
2289e2297.
4. Conclusions
[12] A.S. Mathews, I. Kim, C.S. Ha, Macromol. Res. 15 (2007) 114e128.
[13] M. Hasegawa, K. Kasamatsu, K. Koseki, Eur. Polym. J. 48 (2012) 483e498.
[14] M. Hasegawa, D. Hirano, M. Fujii, M. Haga, E. Takezawa, S. Yamaguchi,
A. Ishikawa, T. Kagayama, J. Polym. Sci. Part A 51 (2013) 575e592.
[15] M. Hasegawa, M. Horiuchi, K. Kumakura, J. Koyama, Polym. Int. 63 (2014)
486e500.
[16] M. Hasegawa, M. Fujii, J. Ishii, S. Yamaguchi, E. Takezawa, T. Kagayama,
A. Ishikawa, Polymer 55 (2014) 4693e4708.
[17] M. Hasegawa, S. Horii, Polym. J. 39 (2007) 610e621.
[18] T. Matsuura, Y. Hasuda, S. Nishi, N. Yamada, Macromolecules 24 (1991)
5001e5005.
A series of PEsIs (TA-HQs/TFMB) with various alkyl substituents
were prepared for applications as novel plastic substrate materials
in image display devices. A significant impact of the substituent
bulkiness on the CTE and the other important properties was
observed in this work. We found out some promising candidate
systems on systematic investigations. For example, the chemically
imidized TA-TMHQ/TFMB was soluble even in non-amide solvents
such as CPN. Simple casting process from the CPN solution afforded
a flexible and less colored film with a considerably low CTE
(11.5 ppm Kꢀ1), a high Tg of 276 ꢁC (by TMA), high thermal stability
(5% weight loss temperature in N2: 453 ꢁC), and very suppressed
water uptake (WA ¼ 0.13%). The low CTE results from a high extent
of in-plane chain orientation, which took place during solvent
evaporation after the PEsI solution was coated on a substrate. On
the other hand, thermally imidized counterpart was worthless in
terms of low CTE and high transparency.
The copolymerization approach using conventional rigid tetra-
carboxylic dianhydrides as comonomers was very effective for
further improving the thermal properties of PEsI systems (C)
including bulky substituents. For example, copolymerization of
NTDA (30 mol%) to TA-25DBHQ/TFMB caused not only a drastic
decrease in the CTE (7.4 ppm Kꢀ1) but also an enhancement of the
Tg (290 ꢁC) while keeping good transparency and excellent
solution-processability. This approach was also very effective to
restrain the film yellowness. Thus, some of the PEsIs developed in
this work can be applied as novel plastic substrate materials.
[19] M. Hasegawa, T. Ishigami, J. Ishii, K. Sugiura, M. Fujii, Eur. Polym. J. 49 (2013)
3657e3672.
[20] S. Ando, T. Matsuura, S. Sasaki, Polym. Prepr. Jpn. 41 (1992) 957.
[21] M. Hasegawa, K. Koseki, High. Perform. Polym. 18 (2006) 697e717.
[22] M. Hasegawa, Y. Tsujimura, K. Koseki, T. Miyazaki, Polym. J. 40 (2008) 56e67.
[23] M. Hasegawa, Y. Sakamoto, Y. Tanaka, Y. Kobayashi, Eur. Polym. J. 46 (2010)
1510e1524.
[24] J.C. Coburn, P.D. Soper, B.C. Auman, Macromolecules 28 (1995) 3253e3260.
[25] T. Matsuura, M. Ishizawa, Y. Hasuda, S. Nishi, Macromolecules 25 (1992)
3540e3545.
[26] S.H. Lin, F. Li, S.Z.D. Cheng, F.W. Harris, Macromolecules 31 (1998)
2080e2086.
[27] T. Kikuchi, K. Uejima, H. Sato, Polym. Prepr. Jpn. 40 (1991) 3748e3750.
[28] S. Uchimura, T. Kikuchi, T. Saito, H. Sato, M. Kojima, D. Makino, Japanese
Patent, JP-2549793B.
[29] M. Hasegawa, M. Koyanaka, High. Perform. Polym. 15 (2003) 47e64.
[30] M. Hasegawa, T. Honda, S. Kurokawa, K. Shindo, J. Ishii, Polym. Prepr. Jpn. 61
(2012) 4120.
[31] M. Hasegawa, T. Matano, Y. Shindo, T. Sugimura, Macromolecules 29 (1996)
7897e7909.
[32] S. Numata, S. Oohara, K. Fujisaki, J. Imaizumi, N. Kinjyo, J. Appl. Polym. Sci. 31
(1986) 101e110.
[33] S. Numata, K. Fujisaki, N. Kinjyo, Polymer 28 (1987) 2282e2288.
[34] S. Numata, N. Kinjo, D. Makino, Polym. Eng. Sci. 28 (1988) 906e911.
[35] J.C. Coburn, M.T. Pottiger, Thermal curing in polyimide films and coatings, in:
M.K. Ghosh, K.L. Mittal (Eds.), Polyimides: Fundamentals and Applications,
Marcel Dekker, New York, 1996, pp. 207e247.
Acknowledgments
[36] J. Ishii, N. Shimizu, N. Ishihara, Y. Ikeda, N. Sensui, T. Matano, M. Hasegawa,
Eur. Polym. J. 46 (2010) 69e80.
[37] S. Ebisawa, J. Ishii, M. Sato, L. Vladimirov, M. Hasegawa, Eur. Polym. J. 46
(2010) 283e297.
[38] M.J. Brekner, C. Feger, J. Polym. Sci. Part A 25 (1987) 2479e2491.
[39] J. Ishii, A. Takata, Y. Oami, R. Yokota, L. Vladimirov, M. Hasegawa, Eur. Polym. J.
46 (2010) 681e693.
We are indebted to Mr. S. Tsukuda for his partial experimental
support.
Appendix A. Supplementary data
[40] H. Nagano, H. Nojiri, H. Furutani, Abstract of polymers for microelectronics,
Sci. Technol. (PME089), 1989, 102ꢀ103.
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.polymer.2015.07.026.
[41] M. Hasegawa, K. Morita, J. Ishii, J. Photopolym. Sci. Technol. 23 (2010)
495e499.