Synthesis and characterization of processable fluorinated aromatic poly(benzamide imide)s derived from cycloalkane substituted diamines, and their application in a computationally driven synthesis methodology

Article abstract of DOI:10.1016/j.polymer.2018.12.042

A new set of four fluorinated aromatic poly(benzamide imide)s was synthesized with the objective of preparing condensation polymers with good processability while maintaining high chemical resistance, thermal stability and good mechanical properties. Their synthesis was carried out using commercially available cycloalkane substituted diamines, 3,5-dinitrbenzoyl chloride and 4,4'-(hexafluoroisopropylidene)diphthalic anhydride (6FDA) as starting materials. The research was focused on the influence of pendant groups and the testing of a computational methodology that could help to predict polymer properties based on their structure and available QSPR models. To this end, extensive tests were performed on these new poly(imide)s including, IR, NMR, X-ray diffraction, UV absorption, thermal stability, mechanical tests, gas permeation and solubility analyses. The results showed that the materials obtained present properties similar to many available commercial poly(imide)s. On the other hand, computational simulations demonstrated that the strategy followed could be useful in reducing the number of compounds to be synthesized and at the same time, maximize the success ratio of finding polymers with certain desired properties.

Full text of DOI:10.1016/j.polymer.2018.12.042

Accepted Manuscript
Synthesis and characterization of processable fluorinated aromatic poly(benzamide
imide)s derived from cycloalkane substituted diamines, and their application in a
computationally driven synthesis methodology
German Perez, Claudio A. Terraza, Deysma Coll, Pablo Ortiz, Manuel Aguilar-Vega,
Darío M. González, Luis H. Tagle, Alain Tundidor-Camba
PII:
S0032-3861(18)31153-4
DOI:
https://doi.org/10.1016/j.polymer.2018.12.042
JPOL 21129
Reference:
To appear in: Polymer
Received Date: 9 September 2018
Revised Date: 5 December 2018
Accepted Date: 25 December 2018
Please cite this article as: Perez G, Terraza CA, Coll D, Ortiz P, Aguilar-Vega M, González DaríM,
Tagle LH, Tundidor-Camba A, Synthesis and characterization of processable fluorinated aromatic
poly(benzamide imide)s derived from cycloalkane substituted diamines, and their application in
a computationally driven synthesis methodology, Polymer (2019), doi: https://doi.org/10.1016/
j.polymer.2018.12.042.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Graphical Abstract

ACCEPTED MANUSCRIPT
Synthesis and characterization of processable fluorinated aromatic poly(benzamide imide)s
derived from cycloalkane substituted diamines, and their application in a computationally
driven synthesis methodology
German Pereza, Claudio A. Terrazaa, Deysma Collb, Pablo Ortizb, Manuel Aguilar-Vegac, Darío M.
Gonzálezd, Luis H. Taglea and Alain Tundidor-Cambaa*
aResearch Laboratory for Organic Polymers (RLOP), Department of Organic Chemistry, Pontificia
Universidad Católica de Chile, P.O. Box 306, Post 22, Santiago, Chile.
bNúcleo de Química y Bioquímica, Facultad de Ciencias, Universidad Mayor, Santiago, Chile
cUnidad de Materiales. Laboratorio de Membranas. Centro de Investigación Científica de Yucatán
A.C., Chuburna de Hidalgo, Mérida, Yucatán, México.
dInstituto de Química, Pontificia Universidad Católica de Valparaíso, Campus Curauma. Chile
*Corresponding author. Tel.: +56223541199. E-mail address: atundido@uc.cl
ABSTRACT
A new set of four fluorinated aromatic poly(benzamide imide)s was synthesized with the objective
of preparing condensation polymers with good processability while maintaining high chemical
resistance, thermal stability and good mechanical properties,. Their synthesis was carried out using
commercially available cycloalkane substituted diamines, 3,5-dinitrbenzoyl chloride and 4,4'-
(hexafluoroisopropylidene)diphthalic anhydride (6FDA) as starting materials. The research was
focused on the influence of pendant groups and the testing of a computational methodology that
could help to predict polymer properties based on their structure and available QSPR models. To
this end, extensive tests were performed on these new poly(imide)s including, IR, NMR, X-ray
diffraction, UV absorption, thermal stability, mechanical tests, gas permeation and solubility
analyses. The results showed that the materials obtained present properties similar to many
available commercial poly(imide)s. On the other hand, computational simulations demonstrated that
the strategy followed could be useful in reducing the number of compounds to be synthesized and at
the same time, maximize the success ratio of finding polymers with certain desired properties.
Keywords: Aromatic polyimides, gas separation membranes, computationally driven synthesis
methodology
1. Introduction

ACCEPTED MANUSCRIPT
Fluorinated polymers are characterized by their high resistance to different chemicals such as acids,
bases and many common solvents [1-3]. They also have high thermal resistance and mechanical
properties that allow them to be used in membranes, coatings and other applications [4-6]. But
precisely because of these properties, their processability is limited, making the synthesis and
manufacturing processes more complex and expensive. Poly(imide)s (PIs) on the other hand, and
aromatic poly(imide)s in particular, are also highly resistant to chemical action, heat and
mechanical stress, but are easier to process [7-9]. The combination of these two types of polymers
into one has proved to yield materials with excellent properties which have already been applied in
many fields of the industry [10, 11]. However, there are a lot of structural and chemical
composition related factors affecting such properties [12], so that it becomes harder to consider and
to estimate their importance and impact when designing polymers for specific applications.
Recently, virtual screening techniques and computational approaches are being used for rational
design and synthesis of new polymers [13, 14], thus reducing established trial and error procedures.
Since these materials are often designed and synthesized aiming for definite applications, their
desired structural characteristics and properties (glass transition temperature, solubility, mechanical
behaviour, porosity, etc.) are consequently conditioned and restrained within certain limits. Usually,
a knowledge-based strategy is used to identify and to test some outstanding synthesis candidates.
Even though new materials are continuously found, the exploration of the universe of all possible
structures using this approach is rather small, biased and most likely far from optimal.
The present study has the dual objective of presenting a new set of fluorinated aromatic PIs
incorporating cycloalkanes as interchain spacers and at the same time, provide proof of concept of a
simple property-driven computational methodology to identify new materials. The difference
between these PIs resides only in the pendant group. Their physical, thermal and mechanical
properties were measured with different techniques. The results were compared with simulated
properties, trying to establish a correlation between them so that when proposing new compounds,
their properties can be estimated with a higher degree of accuracy. That way it would be possible to
discard compounds with lower possibilities of success before the beginning of synthesis phase.
2. Experimental
2.1.Materials
Cyclopentylamine (A5, 99%, AKScientific), cyclohexylamine (A6, 99%, Merck), cyclooctylamine
(A8, 97%, Acros Organics), 1-adamantanamine (A10, 96%, Acros Organics), 3,5-dinitrobenzoyl
chloride (DNBC, 98%, Sigma-Aldrich). Tetrahydrofuran (THF, 99.5%, Merck KGaA), toluene

ACCEPTED MANUSCRIPT
(TOL, 99.5%, Merck), anhydrous N,N-dimethylacetamide (DMAc, 99.8%, Merck KGaA), ethanol
(EtOH, 99.5%, Merck KGaA), dimethyl sulfoxide-d6 (DMSO-d6, 99.8%, Merck). 4,4'-
(hexafluoroisopropylidene) diphthalic anhydride (6FDA, 99%, Merck). Hydrochloric acid (HCl,
37%, Sigma-Aldrich), triethylamine (TEA, 99%, Merck KGaA), hydrazine monohydrate (N2H4,
98%, Sigma-Aldrich), anhydrous pyridine (Py, 99.8%, Sigma-Aldrich), acetic anhydride (Ac2O,
98%, Sigma-Aldrich), Pd/C (10% Pd, Merck).
2.2.Instrumentation and measurements
1
Nuclear Magnetic Resonance (NMR): All compounds synthesized were characterized by both H
13
1
NMR and C NMR spectroscopy. H NMR was also used to determine the monomer conversion.
NMR spectra were recorded on a BRUKER AVANCE III HD-400 spectrometer at 25 °C on
DMSO-d6 solutions of approximately 50 mg/mL for polymers and 100 mg/mL for all other
compounds. All chemical shifts were reported in ppm (δ) relative to tetramethylsilane (TMS).
Fourier Transform Infrared Spectroscopy (FT-IR): FT-IR spectra were acquired with a Nicolet
8700 (ThermoScientific) FT-IR spectrometer or a BRUKER VECTOR 22 mid-IR spectrometer. In
both cases, the spectra were obtained on KBr pellets (containing approximately 5 mg of samples),
in transmittance mode over a wavenumber range of 4000 - 400 cm-1, with 32 scans using a
resolution of 4 cm-1.
X-ray diffraction (XDR): XDR analyses were carried out in a Siemens 5000 X-ray diffractometer
using Cu - K
α radiation (λ=1.54 Å) at a scanning rate of 0.04 °/min (2θ = 0° - 65°), operated at 34
kV and 25 mA.
Ultraviolet-visible (UV-vis): Spectra were recorded on an AE-UV1608 spectrophotometer at room
temperature in a wavelength interval of 250 - 900 nm.
Gel Permeation Chromatography (GPC): The GPC measurements were performed on a Shimadzu
DGU-20A3R equipment with a Shimadzu RID-10A RI detector and a combination of two columns
of Shimadzu Shim-pack GPC-800DP and GPC-80MD. The molecular weight and polydispersity
index (DPI) were calculated according to the polystyrene standards.
Thermogravimetric Analyses (TGA): These assays were performed using a TGA-8000 (Perkin
Elmer, Inc) thermogravimetric balance. Samples of 5 - 8 mg were analysed over a temperature
range of 50 to 800 °C, with a heating rate of 10 °C/min under nitrogen atmosphere.
Differential Scanning Calorimetry (DSC): DSC analyses were carried out on a Diamond DSC
(Perkin Elmer, Inc) instrument. The samples (5 - 10 mg of film) were heated from 50 to 350 °C,

ACCEPTED MANUSCRIPT
cooled to 50 °C and heated again to 350 °C at a heating rate of 10 °C/min using nitrogen as a purge
gas.
Mechanical properties: Shimadzu model AGS-X universal testing machine was used to determine
the tensile properties of the films under standard conditions as reported in the ASTM-D882-02 [15].
Solubility: For these tests, 5.0 mg of each polymer were weighed into independent glass vials and 1
mL of the solvents was added, at 25 °C. The vials were then tightly capped and put into a drum
tumbler rotating end-over-end at 5 rpm for 48 hours (as indicated in ASTM-D3132 method [16]).
After that time, for those cases where a complete dissolution was not attained, the same protocol
was repeated at a near boiling point temperature for each solvent. The solutions were analysed
visually for solubility.
Inherent viscosity (ηinh): Inherent viscosity was determined with an Ubbelhode viscometer size 50
using polymeric solutions at a concentration of 0.5 g/dL in DMAc at 30 °C.
Density: The density (ρ) of film samples was determined in a density gradient column (Techne
Corp) using a calcium nitrate (Ca(NO3)2) solution with glass standards (1.30 to 1.45 g/cm3) by the
buoyancy method. The density obtained was used to calculate the PIs fractional free volume (FFV)
using the following equation:
ꢁꢂꢃꢀꢄꢅ
FFV = ꢀ
(1)
ꢀ
ꢁꢂ
where Vsp is the specific volume (Vsp= ρ-1) and Voc is the occupied volume by the polymer chains
and it was estimated as Voc= 1.3 Vw, being Vw the Van der Waals volume of the repeating unit in the
polymer chain, which was calculated by Bondi´s group contribution method [17].
Pure gas permeability: Pure gas permeability coefficients were determined using a constant volume
variable pressure device, which was described elsewhere [18]. The effective area of the film was
1.13 cm2. Before each permeation test, the film was degassed for 24 hours. The pure gases
evaluated were He, O2, N2, CH4, and CO2, obtained from Praxair Corp. with purities > 99.99%. The
measurements were carried out at 35 °C and 2 atm upstream pressure for each pure gas.
2.3.Computational part
The structures of the repeating units were generated and optimized using DFT calculations at
B3LYP/3-21G level to provide a uniform starting point for all simulations. Synthia module from
Materials Studio 7.0 suite was used to simulate all polymers properties at 298 K in the 50 kDa to 80
kDa range.

ACCEPTED MANUSCRIPT
2.4.Precursors synthesis and characterization
Precursors C5, C6, C8 and C10 were obtained via acyl chloride DNBC reaction with amines A5,
A6, A8 and A10, respectively.
Scheme 1. Synthesis of precursors.
According to Scheme 1, precursors were synthesized by reaction of DNBC with an excess of the
primary amine in a 1:1.2 molar ratio. A stock solution (approximately 0.03 g/mL) of the amine in
THF was prepared at room temperature with magnetic stirring in a one-neck round-bottom flask.
Weak base TEA was then added to the solution in a 1:1 molar ratio with respect to DNBC.
In a separate flask, a second solution of DNBC in THF (approximately 0.08 g/mL) was prepared
and then poured into a pressure-equalizing dropping funnel. A Liebig straight-type condenser was
attached to the top of the funnel and the assemble was then coupled to the flask. At reflux
temperature and vigorous stirring the reaction started at a 30 drop/min rate, and once the dropping
ceased, it was maintained for at least 12 hours.
After that time, the hot mixture was filtered under reduced pressure to separate the
triethylammonium chloride formed. The filtrate was concentrated in a rotatory evaporator to around
50 mL and poured into 300 mL of a cold HCl aq. solution 30% v/v. The precipitate was separated
by reduced pressure filtration and washed with distilled water. The coarse product was purified by
recrystallization in toluene. The entire process for each precursor was repeated three times, with
overall yields between 83% and 86%, throughout all precursors synthesized.

ACCEPTED MANUSCRIPT
C5: Yield: 85%. FT-IR (KBr,
(C-H stretch aliph.); 1646 (C=O stretch, amide); 1540, 1345 (N-O stretch, nitro); 728 (1,3,5-arom.
subs.). (400 MHz, DMSO-d6) 9.05 (2 H, s(3)), 8.99 (1 H, d(6)), 8.93 (1 H, s(1)), 4.27 (1 H,
dd(7)), 1.93 - 1.74 (4 H, m(8)), 1.60 (4 H, d(9)).
(C2), 137.64 (C4), 128.00 (C3), 121.05 (C1), 52.02 (C7), 32.44 (C8), 24.21 (C9).
ν
, cm-1), 3303 (N–H stretch, amide); 3107 (C–H stretch, arom.); 2957
δ
H
δ
(400 MHz, DMSO-d6) 162.00 (C5), 148.54
C
C6: Yield: 86%. FT-IR (KBr,
(C-H stretch aliph.); 1644 (C=O stretch, amide); 1539, 1345 (N-O stretch, nitro); 728 (1,3,5-arom.
subs.). (400 MHz, DMSO-d6) 9.05 (2 H, s(3)), 8.93 (2 H, s(1,6)), 3.80 (1 H, s(7)), 2.02 – 1.04
ν
, cm-1), 3296 (N–H stretch, amide); 3106 (C–H stretch, arom.); 2941
δ
H
(10 H, m(8,9,10)). δC (101 MHz, DMSO-d6) 161.43 (C5), 148.55 (C2), 137.73 (C4), 127.97 (C3),
121.04 (C1), 49.58 (C7), 32.66 (C8), 25.68 (C10), 25.33 (C9).
C8: Yield: 83%. FT-IR (KBr, ν
, cm-1), 3322 (N–H stretch, amide); 3114 (C–H stretch, arom.); 2929
(C-H stretch aliph.); 1646 (C=O stretch, amide); 1537, 1346 (N-O stretch, nitro); 729 (1,3,5-arom.
subs.). δ H (400 MHz, DMSO-d6) 9.05 (2 H, s (3)), 8.95 (1 H, s (6)), 8.93 (1 H, s (1)), 4.05 (1 H, s
(7)), 1.67 (14 H, ddd (8,9,10,11)).
δ
(101 MHz, DMSO-d6) 161.22 (C5), 148.53 (C2), 137.85
C
(C4), 127.99 (C3), 121.01 (C1), 50.58 (C7), 31.89 (C8), 27.25 (C10), 25.58 (C11), 23.98 (C9).
C10: Yield: 86%. FT-IR (KBr,
2915 (C-H stretch aliph.); 1729 (C=O stretch, amide); 1540, 1346 (N-O stretch, nitro); 723 (1,3,5-
ν
, cm-1), 3288 (N–H stretch, amide); 3094 (C–H stretch, arom.);

ACCEPTED MANUSCRIPT
arom. subs.).
δ
(400 MHz, DMSO-d6) 8.98 (2 H, s(3)), 8.93 (1 H, s(1)), 8.17 (1 H, s(6)), 2.15 –
H
1.72 (15 H, m(8,9,10)).
δ
(101 MHz, DMSO-d6) 162.30 (C5), 148.62 (C2), 139.41 (C4), 128.04
C
(C3), 120.68 (C1), 53.10 (C7), 41.32 (C8), 36.58 (C10), 29.52 (C9).
2.5.Monomers synthesis and characterization
Monomers M5, M6, M8, M10 were obtained by reduction of nitro groups in precursors C5, C6, C8
and C10, respectively.
Scheme 2. Synthesis of monomers.
Precursors (approximately 3 g) were dissolved in 100 mL of EtOH and treated with 10% Pd/C (300
mg) in a flask with a coupled reflux condenser. The mixture was stirred at reflux temperature,
adding 8 mL of N2H4 (0.5 mL/min). The resulting mixture was kept at boiling temperature and left
to stir for 12 hours. The catalyst was then removed by filtration over celite and the filtrate was
concentrated in vacuum and precipitated in water to provide the coarse product. Further purification
was conducted by sublimation. The process was also repeated three times for each monomer with
average yields of 90 to 95%.
M5: Yield: 93%. FT-IR (KBr, ν
, cm-1), 3417, 3387 (N–H stretch, primary amine); 3317 (N–H
stretch, amide); 3055 (C–H stretch, arom.); 2954 (C-H stretch aliph.); 1651 (C=O stretch, amide);
1265 (C-N stretch, amine); 763 (1,3,5-arom. subs.). δ H (400 MHz, DMSO-d6) 7.82 (1 H, d(7)), 6.21
(2 H, d(4)), 5.94 (1 H, s(1)), 4.80 (4 H, s(2)), 4.15 (1 H, dd(8)), 1.82-1.66 (4 H, m(9)), 1.50 (4 H,

ACCEPTED MANUSCRIPT
m(10)).
δ
(400 MHz, DMSO-d6) 168.15 (C6), 149.32 (C3), 137.26 (C5), 102.72 (C4), 102.37
C
(C1), 51.10 (C8), 32.58 (C9), 24.04 (C10).
2
H2N
3
4
O
5 6
1
11
HN
8
7
H2N
9
10
M6: Yield: 95%. FT-IR (KBr,
stretch, amide); 3047 (C–H stretch, arom.); 2924 (C-H stretch aliph.); 1643 (C=O stretch, amide);
1265 (C-N stretch, amine); 756 (1,3,5-arom. subs.). (400 MHz, DMSO-d6) 7.70 (1 H, d(7)),
ν
, cm-1), 3419, 3387 (N–H stretch, primary amine); 3325 (N–H
δ
H
6.21 (2 H, s(4)), 5.93 (1 H, s(1)), 4.80 (4 H, s(2)), 3.67 (1 H, s(8)), 1.71-1.27 (10 H, m(9,10,11)). δ C
(50 MHz, DMSO-d6) 167.01 (C6), 148.81 (C3), 136.79 (C5), 102.19(C4), 101.86 (C1), 47.91 (C8),
32.36 (C9), 25.24 (C11), 24.92 (C10).
M8: Yield: 96%. FT-IR (KBr, ν
, cm-1), 3425, 3387 (N–H stretch, primary amine); 3332 (N–H
stretch, amide); 3047 (C–H stretch, arom.); 2924 (C-H stretch aliph.); 1635 (C=O stretch, amide);
1257 (C-N stretch, amine); 756 (1,3,5-arom. subs.). δ H (400 MHz, DMSO-d6) 7.72 (1 H, d(7)), 6.20
(2 H, s(4)), 5.93 (1 H, s(1)), 4.79 (4 H, s(2)), 3.93 (1 H, s(8)), 1.59 (14 H, m(9,10,11,12)).
δ
(50
C
MHz, DMSO-d6) 166.78 (C6), 148.79 (C3), 136.92 (C5), 102.20 (C4), 101.84 (C1), 48.81 (C8),
31.59 (C9), 26.83 (C11), 25.07 (C12), 23.52 (C10).
M10: Yield: 95%. FT-IR (KBr, ν
, cm-1), 3425, 3387 (N–H stretch, primary amine); 3325 (N–H
stretch, amide); 3070 (C–H stretch, arom.); 2924 (C-H stretch aliph.); 1635 (C=O stretch, amide);
1249 (C-N stretch, amine); 725 (1,3,5-arom. subs.). δ H (400 MHz, DMSO-d6) 7.00 (1 H, s(7)), 6.12

ACCEPTED MANUSCRIPT
(2 H, s(4)), 5.91 (1 H, s(1)), 4.79 (4 H, s(2)), 2.02 (9 H, s(9,10)), 1.64 (6 H, s(11)). δ C (101 MHz,
DMSO-d6) 168.20 (C6), 149.30 (C3), 138.39 (C5), 102.63 (C4), 102.13 (C1), 51.42 (C8), 41.42
(C9), 36.61 (C11), 29.38 (C10).
2.6. Polymer synthesis and spectroscopic characterization
The four condensation polymers: P5, P6, P8 and P10 were obtained by imidization of monomers
M5, M6, M8 and M10, respectively, with 6FDA monomer, in agreement with the procedure shown
in Scheme 3.
Scheme 3. Synthesis of fluorinated aromatic poly(benzamide imide)s.
In a three-necked round-bottomed flask equipped with mechanical stirrer and nitrogen atmosphere,
a 1:1 molar ratio mixture of the dianhydride 6FDA and the corresponding diamine, was added and
stirred using anhydrous DMAc as solvent. The DMAc volume used was split in two portions of 1
mL per mmol of each monomer. The diamine was added and dissolved first, and the flask was then
placed in an ice-water bath while adding the dianhydride and the rest of the solvent. The mixture
was stirred for 1 – 2 hours under these conditions (until complete dissolution of both monomers)
after this time, the bath was withdrawn, and the stirring continued for another 6 hours at room
temperature. Next, Ac2O and Py, in that order, were added, each in a proportion of 5 mmol per
mmol of the diamine, and the resulting mixture was stirred for two hours at room temperature and
yet another hour at 60 °C. Finally, the mixture was cooled, poured and stirred in 300 mL of water.

ACCEPTED MANUSCRIPT
The polymer, a white slightly yellowish solid, was filtered, washed thoroughly with methanol and
dried at 100 °C for 12 hours.
P5: FT-IR (KBr, ν
, cm-1), 3388 (N–H stretch, amide); 3083 (C–H stretch, arom.); 2953, 2869 (C-H
stretch aliph.); 1784 (C=O stretch asym., imide); 1719 (C=O stretch sym., imide); 1649 (C=O
stretch, amide); 1351 (C-N stretch); 717 (imide ring deform.). δ H (400 MHz, DMSO-d6) 8.42 (1 H,
d(16)), 8.21 (2 H, d(6)), 8.04 (2 H, s(5)), 7.99 (1 H, d(4)), 7.78 (2 H, s(13)), 7.74 (1 H, s(12)), 4.24
(1 H, d(17)), 1.99 – 1.40 (8 H, m(18,19)). Elem. Anal. Calc. for [C31H19F6N3O5]n (627.50)n, C,
59.34; H, 3.05; O, 12.75. Elem. Anal. Found: C, 59.15 %; H, 3.10 %; O, 12.90 %.
P6: FT-IR (KBr, ν
, cm-1), 3388 (N–H stretch, amide); 3080 (C–H stretch, arom.); 2930, 2855 (C-H
stretch aliph.); 1785 (C=O stretch asym., imide); 1720 (C=O stretch sym., imide); 1651 (C=O
stretch, amide); 1351 (C-N stretch); 717 (imide ring deform.). δ H (400 MHz, DMSO-d6) 8.35 (1 H,
d(16)), 8.22 (2 H, d(6)), 8.05 (2 H, s(5)), 7.99 (2 H, d(4)), 7.78 (2 H, s(13)), 7.74 (1 H, s(12)), 3.78
(1 H, s(17)), 1.52 (10 H, m(18,19,20)). Elem. Anal. Calc. for [C32H21F6N3O5]n (641.53)n, C, 59.91;
H, 3.30; O, 12.47. Elem. Anal. Found: C, 59.50 %; H, 3.39 %; O, 12.18 %.

ACCEPTED MANUSCRIPT
P8: FT-IR (KBr,
ν
, cm-1), 3388 (N–H stretch, amide); 3080 (C–H stretch, arom.); 2930, 2855 (C-H
stretch aliph.); 1784 (C=O stretch asym., imide); 1720 (C=O stretch sym., imide); 1650 (C=O
stretch, amide); 1350 (C-N stretch); 718 (imide ring deform.). δ H (400 MHz, DMSO-d6) 8.35 (1 H,
d(16)), 8.21 (2 H, d(6)), 8.04 (2 H, s(5)), 7.99 (2 H, d(4)), 7.77 (2 H, s(13)), 7.73 (1 H, s(12)), 4.02
(1 H, s(17)), 1.88 – 1.28 (15 H, m(18 – 21)).Elem. Anal. Calc. for [C34H25F6N3O5]n (669.58)n, C,
60.99; H, 3.76; O, 11.95. Elem. Anal. Found: C, 60.50 %; H, 3.53 %; O, 12.10 %.
P10: FT-IR (KBr, ν
, cm-1), 3388 (N–H stretch, amide); 3082 (C–H stretch, arom.); 2953, 2869 (C-H
stretch aliph.); 1784 (C=O stretch asym., imide); 1720 (C=O stretch sym., imide); 1650 (C=O
stretch, amide); 1351 (C-N stretch); 717 (imide ring deform.). δ H (400 MHz, DMSO-d6) 8.20 (2 H,
d(6)), 7.97 (4 H, s(5,4)), 7.76 (2 H, s(13)), 7.70 (2 H, s(12,16)), 2.05 (9 H, s(18,19)), 1.63 (6 H,
s(20)). Elem. Anal. Calc. for [C36H25F6N3O5]n (693.60)n, C, 62.34; H, 3.63; O, 11.53. Elem. Anal.
Found: C, 62.45 %; H, 3.68 %; O, 11.41 %.
3. Results and discussion
3.1. Monomers and polymers synthesis.
The entire synthesis process comprised three stages: synthesis of the N-cycloalkyl-3,5-
dinitrobenzamides; reduction of these nitro compounds to respective diamines, and synthesis of the
PIs using these diamines and 6FDA. In the first stage, commercial DNBC and a primary

ACCEPTED MANUSCRIPT
cycloalkylamine were used as starting materials. The 3,5-dinitro-N-cycloalkylbenzamide obtained
in this process (Yield: 85% - 90%), was purified by recrystallization and reduced to the
corresponding diamine using hydrogen gas, Pd/C as catalyst and ethanol as solvent (Yield: 90% -
95%). The product was also recrystalized in toluene and then sublimated. This monomer was used
in combination with commercial 6FDA, using anhydrous DMAc as solvent to obtain a poly(amic
acid) that was soluble in the reaction media. This poly(amic acid) underwent a chemical cyclization
process, and it was later precipitated in water to finally obtain the PI (Yield: above 99%).
1
13
In each stage of the process, infrared spectroscopy as well as H NMR and C NMR techniques
were used to confirm the conversion of the products and their purity as previously highlighted in the
Materials and Methods section. Using FT-IR it was possible to identify characteristic absorption
bands corresponding to specific functional groups after each stage. Overall signal ranges found for
the four nitro compounds are as follows: N-O stretching signals were identified as two strong bands,
one around 1540 - 1543 cm-1 and the other at 1344-1347 cm-1; a strong band at 1642 - 1647 cm-1
corresponding to amide C=O stretching; a strong absorption band at 3289 - 3323 cm-1
corresponding to amide N-H stretch; and C-H stretch signals for both aromatic and aliphatic
carbons at 3100 - 3109 cm-1 and 2915 – 2967cm-1 respectively. For diamines, N-O stretch band was
absent and instead FT-IR spectra showed the two bands corresponding to N-H stretching of primary
amines at 3325 – 3332 cm-1 and 3425 cm-1, a medium intensity C-N stretch band at 1085 cm-1 and a
strong band around 1650 – 1665 cm-1 corresponding to the amine N-H bending. After final stage,
FT-IR spectra showed the bands corresponding to the imide group C=O stretching close to 1780
1
and 1720 cm-1. NMR, specifically H NMR, was particularly useful to corroborate that the
compounds were obtained and to assess their purity.
3.2.Polymer solubility
Polymers solubility was tested on different solvents. They were insoluble in highly polar protic
solvents such as water and ethanol as well as in non-polar solvents such as toluene or n-hexane.
However, they were fairly soluble at room temperature in common polar and borderline polar
aprotic solvents as shown in Table 1. It is important to note that overall solubility in oxygenated
solvents was higher than in chlorinated solvents. That fact can probably be attributed to the stronger
C-F ··· O interaction between organic fluorine in the polymeric structure and the oxygen atoms in
solvent molecules, in comparison to C-F ··· Cl interactions. There is also a possibility of H bond
formation between the oxygen atom from the solvents and the hydrogen donor NH of the amide
group. Regarding polymer structures, pendant groups did not significantly affect the polymers
solubility since all of them showed similar behavior, except for P10 which was soluble in all

ACCEPTED MANUSCRIPT
solvents tested. This polymer had the bulkier cycloalkyl pendant group, which allowed an easier
access of solvent molecules to main chain polymer atoms and an increment in the number of solute
– solvent interactions. Both factors could explain P10 higher solubility.
Table 1. Here
3.3.Thermal and physical properties
Thermal analyses of these PIs indicated that all of them are thermally stable polymers with high Tg
and onset of decomposition temperatures (TDesc) above 480 °C (Table 2). The Tgs for all samples
were found in the range 290 – 304 °C and there was a trend to lower the Tg as the size and
flexibility of the pendant group increased. However, adamantane as pendant group in P10, causes a
rise in the Tg, most likely because it is a bulky rigid bi-cycle which increases the polymer chain
stiffness thus reducing segmental mobility.
Table 2. Here
Decomposition temperatures were all comprised within the 490 ± 10 °C interval. The main mass
loss observed corresponds to the cleavage of the aliphatic pendant groups as expected considering
that polymer backbones were more stable since they are mostly aromatic, with imide and amide
bonds present. Also, tensional forces acting inside the cycles, and in consequence, the stability of
pendant groups, would have had a direct effect on decomposition temperatures of the polymers. For
example, adamantane is significantly more stable than the other pendant groups and that is reflected
in a higher decomposition temperature. Cyclohexane, cyclopentane and cyclooctane, sorted in
decreasing stability order, have TDesc that are quite similar, although these three temperatures are
within the instrumental error and no difference in decomposition temperature was observed among
them. The similarity in the TDesc is ascribed to a similar thermal stability in the monocyclic pendant
structures regardless the size of the ring, at least in the poly(imide)s tested.
PIs molecular weights were within the range expected for these materials, averaging 5.7 x104 Da
and with polydispersity indexes ranging from 1.3 to 1.5. When considering the molecular weight of
their repeating units, it was found that chains were 80 to 95 units long. Density measurements
exhibited small variations from one polymer to the other although there is a clear inverse relation
between pendant group size and polymer densities.
The results are consistent with an increase in interchain distance since it leads to a larger free
volume amongst chains, as can be seen in P10.

ACCEPTED MANUSCRIPT
The relation is not linear though since molecular weight of repeating units must also be considered.
This free volume or spacing between chains was estimated by computing fractional free volumes
(FFV) and by determining d-spacing (dsp) using wide angle X-ray diffraction (Figure 1). As shown
in Table 2, both methods indicate a similar trend although P6 has an FFV value higher than
expected, considering the overall trend.
Fig 1. Here
Nonetheless, density as well as inherent viscosity parameters, are also within the expected range for
these materials. Moreover, it was possible to make resistant yet flexible films with these PIs.
3.4.Mechanical properties
Mechanical tests were also performed on these films. They had average tensile strength and
Young’s modulus of 47 MPa and 1.1 GPa, respectively (Table 3). Results suggest that there is a
relationship between these properties and the size of polymer chains. As the chains are longer the
materials tend to deform more easily under a lower stress (decrease in Young’s modulus), more
tension is required to break the material (an increase in tensile strength), as they become more
ductile and resistant to break (increase in the elongation to break parameter). Considering that the
longer the chain, the more interactions can be established, these results were expected. However, to
make conclusive experiments more polymers are required, not just the four used in this work.
Regardless of these considerations, the values obtained for mechanical properties and the elongation
to break matched with values reported earlier for experimental aromatic PIs and other engineered
thermoplastic polymers[19, 20].
Table 3. Here
3.5.Gas permeation assays
The gas permeation capabilities to some common gases were evaluated to continue the
characterization of these materials. The results of these tests are shown in Table 4. The films formed
with these polymers had in general higher permeability to all gases as the pendant group size
increased. Two groups with a volume difference of about 50 cm3/mol can be formed based on the
values observed: P5, P6 and P8, P10. As reference, Matrimid, a commercial aromatic PI very
selective but lowly permeable to gasses was included as well as poly(p-phenylene oxide) (PPO), a
permeable commercial poly(ether) with low selectivity. Both P5 and P6 are less permeable to all the
gases analysed but the P5 selectivity is higher than all others. When compared to commercial
polymers, all the PIs studied are more permeable than Matrimid, but their selectivity is significantly

ACCEPTED MANUSCRIPT
lower in some cases. However, when compared to PPO, they outperform it in separation capabilities
although they are not as permeable. This is a commonly found trade-off between permeability and
selectivity of polymer film membranes [22].
Table 4. Here
Fig. 2 shows three Robeson upper-bound plots generated from representative polymer data [17].
Selectivity versus permeability points are indicated in each graph for the tested PI-membranes plus
Matrimid and PPO. These results illustrate how PI-membranes obtained in this work have similar
ideal-separation performance to that of commercially available polymer membranes. Permeability
to these gases increase as the pendant group size increases. Ideal selectivities show a similar trend,
although P5 appears to be the most selective of the four PIs tested. Interesting to note that even
when these ideal selectivities may not reflect real performance for separating mixtures of gases,
they do suggest that membranes made with these PIs could be a viable membrane material for gas
separations.
Fig 2. Here
3.6.Optical properties of polymer films
Strong aromatic conjugation and the presence of intermolecular charge transfer complexes often
result in fluorinated aromatic poly(imide)s with a strong yellowish coloration. As a result, these
films have low transparency in the ultraviolet-visible region. In comparison, the films obtained in
this work were only slightly colored and displayed high optical transparency. All PIs showed a low
wavelength cut-off around 350 nm. Among them, P5 exhibited a very low transmittance, under 70%
for the entire wavelength range analyzed.
However, as depicted in Figure 3, the other three PIs had significantly better optical transparency. It
appears that except for P5, as the pendant group size decreased, the material became more
transparent. Average transmittances in the 400 – 780 nm range for P5, P6, P8 and P10 were 51%,
92%, 90% and 84% respectively. The minimum wavelength at which transmittance is over 80%
was not attained for P5 but for P6, P8 and P10 transmittance values were 403, 417 and 483 nm,
respectively.
Fig 3. Here
3.7.Computational simulations

ACCEPTED MANUSCRIPT
As the final stage of this research, a set of computational simulations were performed with a group
of QSPR-based available software. The simulations started by representing the repeating unit for
each PI and optimizing their geometry. These structures were then used as input in different
programs using QSPR models to estimate a set of selected properties. Some of these results are
presented in Figure 4. These graphs show the correlation between experimental Tg, TDT50%
,
density, Young’s modulus, Tensile modulus and O2/N2 selectivity results and their computational
counterparts.
Fig 4. Here.
Since the properties of the PIs were estimated based on their repeating units and the models used
were not specifically designed for this class of poly(imide)s but rather for general purpose, it was to
be expected that the values predicted differed significantly from the experimental ones. This fact
can be observed in all graphs in Figure 4. To overcome this issue, a simple linear regression
analysis was carried out using as dependant variable the experimental results and the simulated
results as independent variable. In all cases it was possible to obtain an analytical solution as a
classical two variables linear equation Y= aX +b where Y corresponds to the experimental value; X
is the simulated data; a is the slope of the line; and b the intersect. These mathematical functions
would need further tuning before applying them in real situations since the number of experiments
performed in this work, is small and not enough as to provide a sound statistical evidence of strong
correlation.
Other properties such as the Young modulus have an inverse trend, although a relationship can still
be established. In all cases however, it is evident that in order to use these models, it is necessary to
synthesize a statistically representative set of compounds, determine the properties of interest and
develop a linear model that connects the simulated properties with the experimental ones.
4. Conclusions
By means of the proposed synthesis strategy it was possible to obtain the monomers with a yield
higher than 75% in all cases and with a high degree of purity. This set of four new diamines were
obtained by reduction of the dinitro compounds synthesised through an amidation reaction from raw
materials comprising 3,5-dinitrobenzoyl chloride, and a group of primary cycloalkyl amines. These
diamines were used in a polycondensation reaction with 6FDA to obtain a set of the corresponding
four new PIs. The resulting polymers were characterized using several spectroscopic techniques,
showing, in general, high Tg and thermal decomposition temperatures (TDT50%), in addition to good
solubility in several common solvents. All these structures are reported for the first time and are

ACCEPTED MANUSCRIPT
part of a wider study in which the substituent effect is analysed. The polymers are thus structurally
similar to each other and to other high-performance polymers currently used for commercial
purposes. All PIs showed good cohesion and adequate mechanical strength to form films. It was
also possible to observe a relationship between the size of pendant groups and some properties of
the polymers. The effect that pendant groups have in different properties of these type of polyimides
was determined in this study, and an overall methodology is proposed to extend and adapt this
procedure for the synthesis of other types and families of polymers.
The incorporation of larger cycloalkane pendant groups into the synthesised PIs had a favourable
effect on enhancing the solubility and processability of the resultant polyimides. Their mechanical
properties and thermal resistance were in accordance to the ranges reported for these types of
materials. The resistance of these materials appears to improve as the pendant group gets larger, and
although the number of tests performed doesn’t allow to present conclusive results, it does appear to
be related to the size of polymers chains.
Thermal stability was high, and its behaviour was as expected considering the stability of pendant
groups. The lowest decomposition temperatures were observed with the more stressed structures.
It was possible to establish a mathematical relation between simulated and experimental results.
Such relations could be useful in a screening process prior to the syntheses of new materials, to
estimate the properties of synthesis candidates. This strategy required the synthesis and
characterization of a small set of polyimides. In actual studies, this set should be larger (at least 20
samples) and explore a wider variability of structural motifs, depending on the modifications and
structural features the polymers should present.
As mentioned before, the mathematical functions obtained in this work cannot be used without
further refinement since their confidence interval is too low due to the small amount of points used
for obtaining them. However, they do serve the purpose of evidentiate a mathematical relationship
between experimental and predicted properties. It also constitutes proof of concept that general-
purpose programs can be used for predicting real properties of polymers with an acceptable degree
of accuracy, once they are adjusted and fitted to real data. It is not necessary to develop new QSPR
models for each new family of polymers. The existing ones can be used, although their output needs
to be adjusted and verified with a small set of polymers with properties of interest determined
experimentally. In other words, a statistically representative sample of the desired polymers should
be synthesised and characterized, as presented in this work. To further improve the prediction
accuracy, available QSPR models specific for the type of polymers studied can be used.

ACCEPTED MANUSCRIPT
Acknowledgements
We thank research assistant Marjorie Gonzalez, and undergraduate student Sebastian Morales for
their help with the laboratory experiments. This work was funded by FONDECYT Postdoctoral
Fellowship, grant number 3160724 to German Perez, and by FONDEQUIP EQM120021 for the
NMR analysis. In memory of Dr. Jose Luis Santiago.

ACCEPTED MANUSCRIPT
References
[1]
Tundidor-Camba A, Terraza CA, Tagle LH, Coll D, Ortiz P, Pérez G, et al. Aromatic
polyimides containing cyclopropylamide fragment as pendant group. A study of the balance
between solubility and structural rigidity. Macromolecular Research. 2017;25(3):276-281.
[2]
[3]
Cardoso V, Correia D, Ribeiro C, Fernandes M, Lanceros-Méndez S. Fluorinated Polymers
as Smart Materials for Advanced Biomedical Applications. Polymers. 2018;10(2):161.
Dhara MG, Banerjee S. Fluorinated high-performance polymers: Poly(arylene ether)s and
aromatic polyimides containing trifluoromethyl groups. Progress in Polymer Science.
2010;35(8):1022-1077.
[4]
Leivo E, Wilenius T, Kinos T, Vuoristo P, Mäntylä T. Properties of thermally sprayed
fluoropolymer PVDF, ECTFE, PFA and FEP coatings. Progress in Organic Coatings.
2004;49(1):69-73.
[5]
[6]
Kharitonov A. Practical applications of the direct fluorination of polymers. Journal of
Fluorine Chemistry. 2000;103(2):123-127.
Price SC, Stuart AC, Yang L, Zhou H, You W. Fluorine substituted conjugated polymer of
medium band gap yields 7% efficiency in polymer− fullerene solar cells. Journal of the
American Chemical Society. 2011;133(12):4625-4631.
[7]
[8]
De Abajo J, De la Campa J. Processable aromatic polyimides. Progress in polyimide
chemistry I: Springer; 1999. p. 23-59.
Falkovich SG, Lyulin SV, Nazarychev VM, Larin SV, Gurtovenko AA, Lukasheva NV, et
al. Influence of the electrostatic interactions on thermophysical properties of polyimides:
Molecular dynamics simulations. Journal of Polymer Science Part B: Polymer Physics.
2014;52(9):640-646.
[9]
Sydlik SA, Chen Z, Swager TM. Triptycene polyimides: soluble polymers with high thermal
stability and low refractive indices. Macromolecules. 2011;44(4):976-980.
[10] Liaw D-J, Wang K-L, Huang Y-C, Lee K-R, Lai J-Y, Ha C-S. Advanced polyimide
materials: Syntheses, physical properties and applications. Progress in Polymer Science.
2012;37(7):907-974.
[11] Ghosh A, Sen SK, Banerjee S, Voit B. Solubility improvements in aromatic polyimides by

ACCEPTED MANUSCRIPT
macromolecular engineering. RSC Advances. 2012;2(14):5900-5926.
[12] Wood-Adams PM, Dealy JM, deGroot AW, Redwine OD. Effect of Molecular Structure on
the Linear Viscoelastic Behavior of Polyethylene. Macromolecules. 2000;33(20):7489-7499.
[13] Kanal IY, Owens SG, Bechtel JS, Hutchison GR. Efficient computational screening of
organic polymer photovoltaics. The journal of physical chemistry letters. 2013;4(10):1613-
1623.
[14] Gronau G, Krishnaji ST, Kinahan ME, Giesa T, Wong JY, Kaplan DL, et al. A review of
combined experimental and computational procedures for assessing biopolymer structure–
process–property relationships. Biomaterials. 2012;33(33):8240-8255.
[15] ASTM D882-02, Standard Test Method for Tensile Properties of Thin Plastic Sheeting.
West Conshohocken, PA: ASTM International; 2002.
[16] ASTM D3132-84(1996), Standard Test Method for Solubility Range of Resins and
Polymers (Withdrawn 2005). West Conshohocken, PA: ASTM International; 1996.
[17] Bondi A. van der Waals Volumes and Radii. The Journal of Physical Chemistry.
1964;68(3):441-451.
[18] Loría-Bastarrachea MI, Aguilar-Vega M. Membranes from rigid block hexafluoro
copolyaramides: Effect of block lengths on gas permeation and ideal separation factors.
Journal of Membrane Science. 2013;443:36-44.
[19] Dinari M, Rajabi AR. Structural, thermal and mechanical properties of polymer
nanocomposites based on organosoluble polyimide with naphthyl pendent group and layered
double hydroxide. High Performance Polymers. 2017;29(8):951-959.
[20] Aguilar-Lugo C, Santiago-García J, Loría-Bastarrachea M, Guzmán-Lucero D, Alexandrova
L, Aguilar-Vega M. Synthesis, characterization, and structure-property relationships of
aromatic polyimides containing 4, 4
Research. 2016;23(3):49.
′-diaminotriphenylmethane. Journal of Polymer
[21] Sanders DF, Smith ZP, Guo R, Robeson LM, McGrath JE, Paul DR, et al. Energy-efficient
polymeric gas separation membranes for a sustainable future: A review. Polymer.
2013;54(18):4729-4761.
[22] Robeson LM, Liu Q, Freeman BD, Paul DR. Comparison of transport properties of rubbery

ACCEPTED MANUSCRIPT
and glassy polymers and the relevance to the upper bound relationship. Journal of
Membrane Science. 2015;476:421-431.

ACCEPTED MANUSCRIPT
List of figures and tables.
Fig. 1. X-Ray diffractogram of the four PIs studied. P5 (black), P6 (red), P8 (blue) and P10 (green)
follow the same notation as presented in Scheme 3
Fig. 2. Robeson plots for the PIs studied and other two commercially available polymers (Matrimid
and PPO). Threshold lines in these graphs are the Robeson upper bounds as of 2008 [21,
22]
Fig. 3. UV Spectra of the four PIs studied. P5 (black), P6 (red), P8 (blue) and P10 (green).
Continuous lines represent the transmittance of polymers. Discontinuous lines represent their
absorbance.
Fig. 4. Experimental vs computed selected properties of the polymers obtained in this work. P3 has
cyclopropane as pendant group [1]. R2 parameter was included as indication of data correlation.
Table 1. Polymers solubility in different solvents
Table 2. Physical properties of polymers
Table 3. Mechanical properties
Table 4. Permeability to common gases for polymers studied in this work and some representative
commercial polymers.

ACCEPTED MANUSCRIPT
Figure 1.
Figure 2.
Figure 3.

ACCEPTED MANUSCRIPT
Figure 4.

ACCEPTED MANUSCRIPT
Table 1.
DMSO
DMF
DMAc
NMP
DCM
CHCl3
THF
TCE
1,4-D
Polymer (189 °C)*
(153 °C)
(165 °C)
(203 °C)
(40 °C)
(61 °C)
(66 °C)
(74 °C)
(101 °C)
D3.96/P44.4 D3.8/P40.4
D3.8/P40.1 D4.1/P36
D1.8/P30.9 D1.1/P25.9 D1.8/P21
(+)
D1.7/P17.0 D0.4/P16.4
P5
P6
P8
P10
(+)
(-)
(+)
(+)
(+)
(+)
(+)
(+)
(+)
(+)
(+)
(+)
(+)
(+)
(-)
(-)
(-)
(+)
(-)
(-)
(-)
(+)
(+)
(+)
(+)
(+)
(+)1
(+)
(+)
(+)
(-)
(+)1
(+)
(+)1
(+)
(+) Soluble at RT, (+)1 Soluble when heated, (-) Insoluble even after heating
* Solvent boiling temperatures
NMP=
N-methyl-2-pyrrolidone; DMSO= dimethyl sulfoxide; DMAc= N,N-dimethylacetamide; DMF= N,N-dimethylformamide; 1,4-D= 1,4-dioxane; TCE=
trichloroethane; THF: tetrahydrofuran; CHCl3= chloroform; DCM= dichloromethane; D= Dipole moment; P= Polarity Index
Table 2.
Polymer
P5
P6
Tg [°C]
304
300
290
303
TDesc [°C]
489
ηinh [dL/g]
0.56
ρ [g/cm3]
1.396
1.386
1.373
FFV
0.134
0.152
0.141
0.154
dsp
6.14
6.14
6.33
6.39
Mn (×104) [Da] Mw(×104) [Da] DPI
5.804
5.304
6.514
5.507
8.126
6.895
8.463
8.201
1.4
1.3
1.3
1.5
490
488
498
0.36
0.63
0.40
P8
P10
1.372
Tg= Glass Transition Temperature; TDesc= Decomposition Temperature (temperature at which 10% of sample mass decomposes) ; ηinh = Inherent Viscosity; ρ = Density
at 298 K; FFV= Fractional Free Volume; dsp= d- spacing estimated by DRX; DPI= Polydispersity Index
Table 3.
Polymer Young´s Modulus Tensile Strength Elongation-to-
(GPa)
(MPa)
Break (%)
10.05 ± 4.70
6.14 ± 0.92
11.36 ± 0.58
7.07 ± 1.76
P5
P6
P8
P10
1.13 ± 0.30
1.14 ± 0.17
0.89 ± 0.19
1.33 ± 0.20
48.10 ± 5.99
37.25 ± 5.49
42.46 ± 6.37
61.43 ± 15.55
Table 4.
O2/N2
5.020
3.565
4.030
4.040
6.563
4.098
N2/CH4
1.288
1.149
1.123
1.130
1.143
0.953
CO2/ N2
20.898
16.405
17.134
19.207
31.250
14.878
CO2/ CH4
26.915
18.851
19.247
21.711
35.714
14.186
He
O2
N2
CH4
CO2
Polymer
P5
P6
P8
P10
47.177
46.450
69.230
62.825
-
4.694
4.670
6.610
6.690
2.100
16.800
0.935
1.310
1.640
1.656
0.320
4.100
0.726
1.140
1.460
1.465
0.280
4.300
19.540
21.490
28.100
31.806
10.000
61.000
Matrimid*
PPO*
-
* Data extracted from [21]

ACCEPTED MANUSCRIPT
Highlights
•
•
•
Three new aromatic polyimides containing spirocyclic moieties were obtained and evaluated
as gas separation membrane.
The spirocyclic unit is responsible for the good solubility of the polymers, minimizing the
cohesive forces between polymeric chains.
The requirements of total chain rigidity and the prohibition of rotation along the polymer
backbone, in addition to the possession of rigid components, is mandatory for attaining
polymers with high permeability and selectivity.