Synthesis and gas transport properties of new high glass transition temperature ring-opened polynorbornenes

Article abstract of DOI:10.1021/ma011959p

The synthesis and polymerization of the new monomers N-(1-adamantyl)-exo-norbornene-5,6-dicarboximide (AdNDI), N-cyclohexyl-exo-norbornene-5,6-dicarboximide (ChNDI), and N-phenyl-exo-norbornene-5,6-dicarboximide (PnNDI) are reported. Copolymers of N-(1-adamantyl)-exo-norbornene-5,6-dicarboximide/norbornene, with molar compositions 50/50, 70/30, and 30/70, were also obtained. The transport of hydrogen, oxygen, nitrogen, carbon monoxide, carbon dioxide, methane, ethylene, and ethane across membranes prepared from these homopolymers and copolymers was determined at 30°C using permeation techniques. Diffusion coefficients correlate rather well with the diameter of diffusant molecules except in the case of carbon dioxide. The values of the permselectivity coefficient for different gases depend on the type of membranes. For example, the permselectivity of oxygen with respect to nitrogen, α(O₂/N₂), is ca. 5.50 for membranes prepared from N-(1-adamantyl)-exo-norbornene-5,6-dicarboximide/norbornene (50/50) copolymers, which compares favorably with the values reported for this parameter in membranes with imide groups located in the backbone. The values of α(H₂/C₂H₆) in membranes of poly(N-(1-adamantyl-exo-norbornene-5,6-dicarboximide), poly(N-cyclohexyl-exo-norbornene-5,6-dicarboximide), and poly(N-phenyl-exo-norbornene-5,6-dicarboximide) are 183, 124, and 122, respectively, whereas the values of α(C₂H₄/C₂H₆) amount to 6.9, 7.6, and 6.4, respectively. For most membranes used in this study, diffusion rather than solubility is responsible for the discrimination of gas transport. However, solubility is mainly responsible for the high permselectivity of ethylene with respect to ethane displayed by the membranes.

Full text of DOI:10.1021/ma011959p

Macromolecules 2002, 35, 4677-4684
4677
Synthesis and Gas Transport Properties of New High Glass Transition
Temperature Ring-Opened Polynorbornenes
Ar m a n d o P in ed a Con tr er a s a n d Mik h a il A. Tlen k op a tch ev
Instituto de Investigaciones en Materiales, Universidad Nacional Auto´noma de Me´xico,
Apartado Postal 70-360, CU, Coyoaca´n, Me´xico DF 04510, Me´xico
Ma r ia d el Ma r Lo´p ez-Gon za´lez a n d Eva r isto Ria n d e*
Instituto de Ciencia y Tecnologı´a de Polı´meros (CSIC), 28006 Madrid, Spain
Received November 12, 2001; Revised Manuscript Received March 12, 2002
ABSTRACT: The synthesis and polymerization of the new monomers N-(1-adamantyl)-exo-norbornene-
5,6-dicarboximide (AdNDI), N-cyclohexyl-exo-norbornene-5,6-dicarboximide (ChNDI), and N-phenyl-exo-
norbornene-5,6-dicarboximide (PhNDI) are reported. Copolymers of N-(1-adamantyl)-exo-norbornene-5,6-
dicarboximide/norbornene, with molar compositions 50/50, 70/30, and 30/70, were also obtained. The
transport of hydrogen, oxygen, nitrogen, carbon monoxide, carbon dioxide, methane, ethylene, and ethane
across membranes prepared from these homopolymers and copolymers was determined at 30 °C using
permeation techniques. Diffusion coefficients correlate rather well with the diameter of diffusant molecules
except in the case of carbon dioxide. The values of the permselectivity coefficient for different gases depend
on the type of membranes. For example, the permselectivity of oxygen with respect to nitrogen, R(O2/N2),
is ca. 5.50 for membranes prepared from N-(1-adamantyl)-exo-norbornene-5,6-dicarboximide/norbornene
(50/50) copolymers, which compares favorably with the values reported for this parameter in membranes
with imide groups located in the backbone. The values of R(H2/C2H6) in membranes of poly(N-(1-adamantyl-
exo-norbornene-5,6-dicarboximide), poly(N-cyclohexyl-exo-norbornene-5,6-dicarboximide), and poly(N-
phenyl-exo-norbornene-5,6-dicarboximide) are 183, 124, and 122, respectively, whereas the values of
R(C2H4/C2H6) amount to 6.9, 7.6, and 6.4, respectively. For most membranes used in this study, diffusion
rather than solubility is responsible for the discrimination of gas transport. However, solubility is mainly
responsible for the high permselectivity of ethylene with respect to ethane displayed by the membranes.
In tr od u ction
The separation of oxygen and nitrogen from air for
industrial combustion and to prevent oxidation, respec-
tively, is carried out using membranes’ technology.7
Also, this technology is used for the removal of hydrogen
from mixtures with nitrogen or hydrocarbons in petro-
chemical processes.7 Therefore, it is obvious the impor-
tance of correlating gas transport with the chemical
structure of the barriers utilized for gases separation.
Chain fluctuations that give rise to the formation of
channels through which the molecules of diffusant can
migrate to a neighboring cavity are severely hindered
in the glassy state due to the fact that only local motions
are allowed. In the glassy state both diffusion and
solubility control gases separation of similar size, but
when the sizes of the diffusants largely differ, diffusion
may be the controlling step. The information at hand
suggests that glassy polymers having bulky side groups
in their structure contain large cavities that facilitate
gas permeability without negatively affecting the perm-
selectivity. These polymers have in general high glass
transition temperatures, and therefore aging processes
that might affect gas transport are not important at the
temperatures of use. Polycarbonates, polysulfones, poly-
imides, and poly(ether imide)s are polymers commonly
used in gas separation.7-11
Gas transport in membranes depends on the physical
state (glassy, rubbery, or crystalline) of the polymer
integrating the barrier membranes. Crystalline regions
in semicrystalline polymers are opaque to gas transport
whereas micro-Brownian motions in rubbery polymers
facilitate nonpermanent holes formation through which
the molecules of diffusant can easily jump. The solubility
coefficient of gases, S, in amorphous rubbery mem-
branes is an increasing function of pressure, and the
pressure dependence of S can be obtained from the
variation of the free energy involved in the mixture of
the polymer with the gas in the liquid state.1,2 Solubility
rather than diffusion controls the permselectivity of
rubbery membranes.
In the glassy state, the solubility coefficient is a
decreasing function of pressure. This behavior is usually
interpreted in terms of the dual mode model3 that
considers glassy membranes as a continuum phase in
which microvoids or holes that account for the excess
volume are dispersed. According to the model, solubility
occurring in the continuous phase obeys Henry’s law
whereas the microvoids act as Langmuir sites where the
gas is adsorbed. Gas transport takes place through
absorption and adsorption sites, but a failure of the
model is that conjugated transport between Henry and
Langmuir sites is not considered.4 Despite these short-
comings, the dual model is commonly used to interpret
gas transport in glassy membranes. Recent permeation
experiments show that gas transport through linear
low-density polyethylene (LLDPE) can also be inter-
preted in terms of the dual mode model.5,6
It is important to devise methods that permit the
prediction of the performance of membranes for gas
separation as a function of the chemical structure.
Molecular dynamics may be an important tool for this
task, though its use is restricted to rubbery membranes.
Actually, molecules of diffusant spend a large time
moving in cavities in glassy membranes so that the
10.1021/ma011959p CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/09/2002

4678 Contreras et al.
Macromolecules, Vol. 35, No. 12, 2002
computing time necessary to reach the diffusive regime
may be prohibitively large. To circumvent this problem,
the transition-state approach (TSA) was formulated.12
The theory assumes that the molecules of gas migrate
through polymer structures in a sequence of hops
between local minima of the potential. Despite the
promising results, some refinements are still necessary
in order that the theory can be used for realistic
simulation of gas transport in terms of the chemical
structure of membranes.13-16
Another approach that can be useful for the design
of optima membranes for gas separation is to investigate
in a systematic way the effect of chemical structure on
gas transport. Norbornene monomers are attractive due
to their facile functionalization and high reactivity using
ring-opening metathesis polymerization (ROMP).17-19
Polynorbornenes of high molecular weight and good
mechanical properties were obtained by ROMP of nor-
bornene monomers20 using appropriate catalysts. The
capability of polynorbornenes for packaging and gas
separation has been investigated.21-25 Gas transport
studies in ring-preserved polynorbornenes prepared via
addition polymerization route have also been re-
ported.26,27
The polymers described in this work contain imide
groups in the side chains and it is a purpose of this work
to investigate whether the presence of these groups
enhances the permselectivity characteristics of polynor-
bornenes. Thus, polymers obtained from the homopo-
lymerization of N-(1-adamantyl)-exo-norbornene-5,6-di-
carboximide (PAdNDI, M1), N-cyclohexyl-exo-norbornene-
5,6-dicarboximide (PChNDI, M2), and N-phenyl-exo-
norbornene-5,6-dicarboximide (PPhNDI, M3) and from
copolymers of N-(1-adamantyl)-exo-norbornene-5,6-di-
carboximide and norbornene (NB) were used for the
preparation of membranes. Schematic representations
of the repeating units of the polymers are shown in
Figure 1. It is worthy to note that the chains contain
double bonds in the backbone that confer rigidity to
them.
In this work the permeation coefficients of hydrogen,
oxygen, nitrogen, carbon monoxide, carbon dioxide,
methane, ethane, and ethylene in M1, M2, and M3 as
well in the copolymers N-(1-adamantyl)-exo-norbornene-
5,6-dicarboximide/norbornene with molar ratios 50/50
(M4), 70/30 (M5), and 30/70 (M6) were measured with
the aim of obtaining information on the permselectivity
performance of these materials to gases.
F igu r e 1. Schematic representation of the repeating units of
poly(N-(1-adamantyl)-exo-norbornene-5,6-dicarboximide) (M1),
poly(N-cyclohexyl-exo-norbornene-5,6-dicarboximide) (M2), poly-
(N-phenyl-exo-norbornene-5,6-dicarboximide) (M3), and co-
polymers of N-(1-adamantyl)-exo-norbornene-5,6-dicarboxim-
ide with norbornene.
the temperature range 30-300 °C. FTIR spectra were obtained
on a Nicolet 510 p spectrometer.
Copolymer compositions were determined by 1H NMR
integration of the olefinic peak of N-(1-adamantyl)-exo-nor-
bornene-5,6-dicarboximide (AdNDI) (5.65 ppm) units relative
to the olefinic protons of norbornene units (5.3 ppm). Molecular
weights and molecular weight distributions with reference to
polystyrene standards were determined with a Varian 9012
GPC instrument at 30 °C in chloroform (universal column and
a flow rate of 1 mL min-1).
Membranes were prepared by casting from chloroform
solutions. Gas permeation through the membranes was mea-
sured at 30 °C using a thermostated experimental device
described in detail elsewhere.13 The diameter of permeation
area of the membranes was 1.7 cm. Thicknesses of 110-180
µm, depending on the membrane, were used. The permeation
apparatus is made up of a high-pressure or upstream chamber
separated from a low-pressure or downstream chamber by the
membrane. A pressure sensor MKS of range 10-4-1 mmHg
measures the pressure in the downstream chamber, and a
pressure transducer Gometrics of 0-10 bar range is used to
measure the pressure in the upstream chamber. Vacuum is
made in the two chambers, and the gas contained in a reservoir
at a pressure close to that to be used in the upstream chamber
is allowed to flow suddenly to the downstream chamber. Then
the flow of gas from the upstream to the downstream chamber
is monitored by the variation of pressure of gas in the latter
chamber as a function of time. Before performing each experi-
ment, the inlet of air into the evacuated downstream chamber
was measured as a function of time and further subtracted
from the curves representing the pressure of permeant against
time in the downstream chamber.
Exp er im en ta l P a r t
Ma ter ia ls. exo-Norbornene-5,6-dicarboxylic anhydride (exo-
NDA, 1) was prepared via Diels-Alder condensation of cyclo-
pentadiene and maleic anhydride and by the thermal isomer-
ization of the corresponding endo isomer according to the
literature.28 1-Adamantanamine, cyclohexylamine, aniline, and
other chemicals were purchased from Aldrich Chemical Co.
1,2-Dichloroethane, dichloromethane, chlorobenzene, and tolu-
ene were dried over anhydrous calcium chloride and distilled
under nitrogen over CaH2. RuCl3‚xH2O was used as received.
RuCl2(PPh3)3 and (PPh3)2Cl2RudCdCH(t-Bu) were prepared
according to the literature.29,30
Mea su r em en ts. 1H NMR and 13C NMR spectra were
recorded on a Varian spectrometer at 300 and 75.5 MHz
frequencies, respectively, in CDCl3 with tetramethylsilane
(TMS) as internal standard.
The glass transition temperatures were measured under
nitrogen with a Du Pont 2100 instrument, at a heating rate
of 10 °C/min. The samples were encapsulated in standard
aluminum DSC pans in duplicate. Each pan was run twice on
Mon om er s Syn t h esis. N-(1-Adamantyl)-exo-norbornene-
5,6-dicarboximide (AdNDI, 3a ). exo-NDA 1 (12 g, 70 mmol)

Macromolecules, Vol. 35, No. 12, 2002
Ring-Opened Polynorbornenes 4679
asym str), 2877 (C-H sym str), 1770 (CdO), 1594 (CdC str),
1454 (C-H def), 1382 (dCH- def), 1329 (C-H def), 1289 (C-H
def), 1188 (C-N str), 975 (C-C skel), 799 cm-1 (CdC-H def).
Meta th esis P olym er iza tion s of Mon om er s. Polymeriza-
tions were carried out in glass vials under a dry nitrogen
atmosphere at 60 °C. Polymerizations were terminated by
adding benzaldehyde under a nitrogen atmosphere. After
cooling, the solutions were poured into an excess of methanol.
The polymers were purified by solubilization in chloroform
containing a few drops of 1 N HCl and precipitation into
methanol, and they were further dried in a vacuum oven at
40 °C to constant weight.
P olym er s Syn th esis. Poly(N-1-adamantyl-exo-norbornene-
5,6-dicarboximide) (PAdNDI, M1). 1 g (3.36 mmol) of 3a and
0.033 g (0.33 mmol) of (PPh3)2Cl2RudCdCH(t-Bu) were stirred
in 3.4 mL of chlorobenzene at 60 °C for 8 h. The polymer
obtained (Figure 4) was soluble in chloroform, toluene, and
dichloromethane. 1H NMR (300 MHz, CDCl3): δ (ppm) ) 5.67
(2H, s, trans), 2.76 (2H, s), 2.61 (2H, s), 2.35 (6H, s), 2.33 (3H,
s), 1.68-1.64 (8H, d). 13C NMR (75 MHz, CDCl3): δ (ppm) )
179.79, 133.90 (cis), 132.05 (trans), 60.63, 50.73, 46.33, 41.70,
39.25, 36.15, 29.71. FT-IR: 3038, 2924, 2855, 1768, 1700, 1665,
F igu r e 2. Synthesis of monomers N-(1-adamantyl)-exo-nor-
bornene-5,6-dicarboximide (3a ), N-cyclohexyl-exo-norbornene-
5,6-dicarboximide (3b ), and N-phenyl-exo-norbornene-5,6-
dicarboximide (3c).
1455, 1372, 1340, 1304, 1194, 971, 762, 748 cm-1
.
Poly(N-cyclohexyl-exo-norbornene-5,6-dicarboximide)(PChNDI,
M2). The polymerization procedure described above was used.
1H NMR (300 MHz, CDCl3) (Figure 5): δ (ppm) ) 5.73 (2H,
s), 3. 98 (1H, m), 2.92 (2H, d), 2.66 (2H, d), 2.14-2.11 (2H, d),
was dissolved in 120 mL of toluene. 11 g (72 mmol) of
adamantanamine in 120 mL of toluene was added dropwise
to the stirred solution of exo-NDA. The reaction was main-
tained at 90 °C for 1 h and then cooled to room temperature.
A precipitate was filtered and dried to give 22 g (70 mmol) of
amic acid. Obtained amic acid 2 (22 g, 70 mmol), anhydrous
sodium acetate (3.3 g, 40 mmol), and acetic anhydride (65 g,
640 mmol) were heated at reflux for 2 h and then cooled. The
solid crystallized on cooling was filtered, washed several times
with water, and dried in a vacuum oven at 50 °C overnight.
Pure monomer 3a (Figure 2) was obtained after twice recrys-
1.80-1.58 (6H, m), 1.27 (4H, m).
13C NMR (75 MHz, CDCl3):
δ (ppm) ) 178.49, 132.05 (trans), 51.29, 50.63, 46.10, 42.00,
28.73, 25.81, 25.07. FT-IR: 3036, 2934, 2893, 1768, 1700, 1665,
1452, 1370, 1346, 1304, 1297, 1186, 987, 762, 748 cm-1
.
Poly(N-phenyl-exo-norbornene-5,6-dicarboximide) (PPhNDI,
M3). Poly(PhNDI) was prepared by a similar procedure as
above for 3a . 1H NMR (300 MHz, CDCl3): δ (ppm) ) 7.44-
7.25 (5H, m), 5.78 (2H, s), 3.13 (2H, s), 2.86 (2H, s), 2.20 (1H,
s), 1.68 (1H, s). 13C NMR (75 MHz, CDCl3): δ (ppm) ) 177.19,
132.33, 129.01, 128.22, 126.34, 50.31, 46.20, 42.22. FT-IR:
3034, 2930, 2877, 1775, 1594, 1459, 1385, 1329, 1290, 1165,
980, 790 cm-1
1
tallization from ethanol: yield ) 85%, mp ) 159-161 °C. H
NMR (300 MHz, CDCl3): δ (ppm) ) 6.25 (2H, s), 3.22 (2H, d),
2.48 (2H, d), 2.41 (1H, d), 2.11 (1H, s), 1.68-1.72 (1H, m),
1.33-1.44 (1H, d). 13C NMR (75 MHz, CDCl3): δ (ppm) )
181.52, 138.72, 65.94, 52.94, 48.91, 44.53, 42.21, 36.72, 30.76.
FT-IR: 3062 (CdC-H str), 2911 (C-H asym str), 2883 (C-H
sym str), 1762 (CdO), 1667 (CdC str), 1455 (C-H def), 1337
(dCH- def), 1373 (C-H def), 1290 (C-H def), 1200 (C-N str),
975 (C-C skel), 784 cm-1 (CdC-H def).
.
Cop olym er s Syn th esis. The monomer 3a was copolymer-
ized with NB using (PPh
3)2Cl2RudCdCH(t-Bu), utilizing a
similar procedure as above for 3a . 1H NMR (300 MHz,
CDCl3): δ (ppm) ) 5.69 (2H, m, trans), 5.51 (2H, m, cis), 5.32
(2H, s, trans), 5.18 (2H, s, cis), 2.71-2.36 (12H, m), 2.07 (3H,
s), 1.85-1.61 (10H, m), 1.33 (4H, m), 1.04 (4H, m). 13C NMR
(75 MHz, CDCl3): δ (ppm) ) 179.96, 133.90 (cis) 132.05 (trans)-
, 60.91, 50.98, 46.45, 43.11, 41.36, 39.27, 38.39, 36.18, 32.19,
29.74. FT-IR: 3027, 2934, 2860, 1766, 1665, 1452, 1370, 1346,
N-Cyclohexyl-exo-norbornene-5,6-dicarboximide (ChNDI, 3b).
exo-NDA 1 (12 g, 70 mmol) was dissolved in 120 mL of toluene.
An amount of 7 g (72 mmol) of cyclohexylamine in 120 mL of
toluene was added dropwise to the stirred solution of exo-NDA.
The reaction was maintained at room temperature 1 h. A
precipitate was filtered and dried to give 18.6 g (72 mmol) of
amic acid 2b. The amic acid obtained (18.6 g, 72 mmol),
anhydrous sodium acetate (3.3 g, 40 mmol), and acetic
anhydride (65 g, 640 mmol) were heated at reflux for 2 h and
then cooled. The solid crystallized on cooling was filtered,
washed several times with water, and dried in a vacuum oven
at 50 °C overnight. Pure monomer 3b (Figure 2) was obtained
after twice recrystallization from methanol: yield ) 87%, mp
) 130-131 °C. 1H NMR (300 MHz, CDCl3) (Figure 3): δ (ppm)
) 6.27 (2H, s), 3. 94 (H, m), 3.25 (2H,d), 2.60 (2H, d), 2.12-
2.16 (2H, m), 1.79-1.84 (1H, m), 1.46-1.60 (2H, m), 1.33-
1.44 (1H, d). 13C NMR (75 MHz, CDCl3): δ (ppm) ) 178.19,
137.83, 51.59, 47.38, 45.38, 42.52, 28.74, 25.79, 25.02. FT-IR:
3062 (CdC-H str), 2984 (C-H asym str), 2857 (C-H sym str),
1762 (CdO), 1664 (CdC str), 1465 (C-H def), 1346 (dCH-
def), 1370 (C-H def), 1255 (C-H def), 1199 (C-N str), 975
(C-C skel), 784 cm-1 (CdC-H def).
1187, 970, 786 cm-1
.
The glass transition temperatures of the M1, M2, M3, M4,
M5, and M6 polymers measured with a DSC scanning calo-
rimeter were respectively 271, 129, 233, 117, 202, and 80 °C.
Resu lts
Some membranes exhibit a rather low permeability
to certain gases that in some circumstances difficult to
know whether steady-state conditions have been reached.
To circumvent this problem, the equation
(-1)n
∞
p0ALST
Dt
L2
1
6
2
p(t) ) 0.2786
-
-
×
∑
(
π2
n2
Dn2π2t
V
n)1
exp -
(1)
)
)
(
L2
N-Phenyl-exo-norbornene-5,6-dicarboximide (PhNDI, 3c).
PhNDI was prepared according to literature.31 Pure monomer
3c (Figure 2) was obtained after twice recrystallization from
resulting from the integration of Fick’s second law,32
using appropriate boundary conditions, was fitted to the
experimental results. In this equation p(t) and p0 which
denote the pressures of gas in the downstream and
upstream chambers, respectively, are given in cmHg,
A and L which represent the area and thickness of the
1
toluene: yield ) 81%, mp ) 195-196 °C. H NMR (300 MHz,
CDCl3): δ (ppm) ) 7.49-7.25 (5H, m), 6.34 (2H, t), 3.40 (2H,
t), 2.82 (2H, t), 1.64-1.47 (2H, m). 13C NMR (75 MHz,
CDCl3): δ (ppm) ) 177.19, 137.93, 129.11, 128.05, 126.13,
47.31, 45.20, 43.22. FT-IR: 3064 (CdC-H str), 2946 (C-H

4680 Contreras et al.
Macromolecules, Vol. 35, No. 12, 2002
F igu r e 3. 1H NMR spectrum of N-cyclohexyl-exo-norbornene-5,6-dicarboximide.
Errors ∆ involved in the determination of D by the time
lag method were obtained by means of the following
expression
|Lꢀ(L)| |ꢀ(θ)|L2
∆ )
+
/D
(4)
(
)
6θ2
3θ
For most gases the error involved in the determina-
tion of the diffusion coefficient was less than 10%.
However, for ethane and ethylene the error may be of
the order of 25%.
By considering that the permeability coefficient, P,
is the solubility coefficient times the diffusion coefficient
(P ) DS), its value can be obtained from eq 2 by means
of the expression
F igu r e 4. Synthesis of polymers via ROMP.
dp(t)
dt
VL
p0AT
P ) 3.59
lim
(5)
tf∞
membrane in cm2 and cm, respectively, the volume of
the downstream chamber, V, in cm3, the solubility
coefficient of the gas, S, in cm3 (STP)/(cm3 cmHg), and
D, the diffusion coefficient, in cm2/s. Illustrative plots
showing the fitting of eq 1 to the experimental values
of the less permeable gas in the membrane M4, ethane,
are represented in Figure 6.
All curves exhibit a transient zone followed by a
region at long times where the pressure of the down-
stream chamber is a linear function of time. Once
steady-state conditions are reached, eq 1 can be written
as
Here P is given in barrers {1 barrer ) [10-10 cm3 (STP)
cm/(cm2 scmHg)]} provided that p0 and p(t) are given
in cmHg and the rest of magnitudes in cgs units. The
values of D obtained from eq 3 were compared with
those determined by fitting eq 1 to the experimental
results in order to know whether steady-state conditions
were reached. As an example, the values obtained for
the diffusion and permeability coefficients of ethane,
directly calculated from the curve p(t) vs t in the
membrane M4, were 0.14 barrers and 4.7 × 10-9 cm2/s,
respectively. These values compare very favorably with
those of 0.15 barrers and 4.0 × 10-9 cm2/s obtained by
fitting eq 1 to the experimental results. Even better
agreement was found for the transport coefficients of
the other gases in the different membranes.
Values of the permeability coefficient of the gases in
the membranes are given in Table 1. An inspection of
the table shows that all the membranes exhibit nearly
the same behavior as far as the permeability of gases
is concerned. Thus, hydrogen has the highest perme-
ability coefficient followed by CO2 and O2, whereas the
permeability coefficients of nitrogen, carbon monoxide,
methane, and ethylene have nearly the same value.
p0ALST
V
Dt
1
6
p(t) ) 0.2786
-
(2)
2
(
)
L
Plots of p(t) against t in the steady state are straight
lines intercepting the abscissa axis at (Dθ/L2) - 1/6 ) 0,
where θ is the time lag. Therefore, the diffusion coef-
ficient can be obtained from the expression33
L2
6θ
D )
(3)

Macromolecules, Vol. 35, No. 12, 2002
Ring-Opened Polynorbornenes 4681
F igu r e 5. 1H NMR spectrum of poly(N-cyclohexyl-exo-norbornene-5,6-dicarboximide).
Ta ble 2. Diffu sion Coefficien ts, in 10-8 cm 2/s, of Differ en t
Ga ses in th e Mem br a n es Used in Th is Stu d y
gas
H2
M1
M2
M3
M4
M5
M6
160
212
132
244
323
2060
O2
6.4
10.4
6.3
11.2
5.8
8.3
N2
CO
CO2
CH4
C2H6
C2H4
3.3
2.0
2.4
0.9
0.7
0.8
3.4
2.8
3.4
1.7
1.7
0.8
2.2
1.4
1.8
0.7
0.4
0.3
2.2
3.2
3.6
1.1
0.5
0.5
1.0
1.3
2.1
0.8
0.7
0.3
0.6
1.4
0.6
Values of the diffusion coefficient are presented in
Table 2. It can be seen that hydrogen may exhibit
diffusion coefficients 2 orders of magnitude or higher
than those of the other gases. The high permeability of
hydrogen in the membranes is mainly due to the high
value of D that overcomes the rather low solubility of
the gas in these materials. Independent of the type of
membrane, the diffusion coefficient follows the trends
D(H2) > D(O2) > D(CO2) = D(N2) = D(CO) = D(CH4) >
D(C2H4) = D(C2H6). As for the effect of the structure of
the membranes on the diffusion coefficient of hydro-
gen, the results indicate that D(H2,M6) > D(H2,M5) >
D(H2,M4) > D(H2,M2) > D(H2,M1) > D(H2,M3). This
trend may not be hold for other gases.
F igu r e 6. Experimental (circles) and calculated (continuous
line) values of the pressure of ethane in the downstream
chamber as a function of time.
Ta ble 1. P er m ea bility Coefficien ts in ba r r er s of Differ en t
Ga ses in th e Mem br a n es Used in Th is Stu d y
gas
H2
M1
M2
M3
M4
M5
M6
12.8
16.1
11.0
10.7
18.2
12.2
O2
1.59
2.13
1.44
1.21
1.92
1.40
N2
CO
CO2
CH4
C2H6
C2H4
0.50
0.51
8.39
0.58
0.07
0.48
0.61
1.04
18.11
1.12
0.13
0.99
0.31
0.52
11.44
0.54
0.09
0.58
0.22
0.37
5.11
0.42
0.14
0.48
0.60
0.92
8.13
0.96
0.41
0.58
0.43
0.54
The apparent solubility coefficient is given by
P
D
0.60
S )
(6)
Carbon dioxide and ethylene are the gases with higher
solubility, whereas hydrogen, which has the highest
diffusion coefficient, turns out to have the lowest
solubility coefficient. The value of S for hydrogen may
be 2 orders of magnitude lower than the solubility
coefficient of carbon dioxide. For example, the apparent
solubility coefficients of H2 and CO2 in the membrane
M1 are respectively 8.0 × 10-4 cm3 (STP)/(cm3 cmHg)
and 3.5 × 10-2 in the same units. In general, S(CO2) >
S(C2H4) > S(CH4) > S(CO) > S(O2) = S(N2) > S(C2H6)
> S(H2). The relatively high value of the solubility
coefficient of ethylene (6.0 × 10-3 cm3 (STP)/(cm3 cmHg))
in comparison with that of ethane (1.0 × 10-3 cm3 (STP)/
(cm3 cmHg) is responsible for the relatively high perme-
Therefore, P(H2) . P(CO2) > P(O2) > P(N2) = P(CO) =
P(C2H4) = P(CH4) > P(C2H6).
The values of the permeability coefficient show a
significant dependence on the fine chemical structure
of the membrane. The largest and lowest permeability
coefficients of hydrogen occur in the M5 and M4
membranes, respectively. In general, the permeability
of hydrogen in the membranes follows the trends P(H2,
M5) > P(H2, M2) > P(H2,M1) = P(H2, M6) > P(H2, M3)
= P(H2,M4), where P(H2,Mi) represents the perme-
ability coefficient of hydrogen in the membrane Mi. The
permeability coefficient of the other gases may not follow
similar trends as hydrogen.

4682 Contreras et al.
Macromolecules, Vol. 35, No. 12, 2002
F igu r e 7. Natural logarithms of the diffusion coefficients of
the gases as a function of the squares of their kinetic diameters
for different membranes: (b) M1, (2) M2, (1) M3, (9) M4, (()
M5, and (×) M6.
F igu r e 8. Experimental results of the permselectivity coef-
ficient of hydrogen with respect to ethane represented as a
function of the permeability of hydrogen in the membranes:
(b) M1; (9) M2; (2) M3; (1) M4; (() M5.
substituted groups in their structure.36 It is noteworthy
that the M2 membrane is more permeable than the M1,
although the adamantyl group should hinder chain
packing more efficiently than the cyclohexyl group.
The permselectivity efficiency of a membrane to the
transport of gas A with respect to gas B is defined as
ability coefficient of the unsaturated hydrocarbon in
comparison with that of the saturated one.
Discu ssion
In the glassy state, molecules of diffusant spend a
significant interval of time erratically moving in cavities
where they are confined without contributing to the
diffusive process. Although micro-Brownian motions are
frozen in glassy molecular chains, fluctuations occur
that may give rise to the formation of channels through
which molecules slip to a neighboring cavity. Successful
slippage requires not only an appropriate velocity of the
molecules in the cavity next to the channel but also the
radius of the channel to be larger than the radii of
diffusant molecules. Therefore, it would be expected
that, aside from polar interactions, diffusant molecules
of larger radius should have lower diffusion coefficients.
Values of the diffusion coefficient are plotted as a
function of the radii of the diffusants in Figure 7. The
plots show rather good concordance between the values
of D and the kinetic diameter of the permeant molecules
in the sense that D decreases as the radius increases.
The values of the diffusion coefficient lie fairly well in
a straight line except those corresponding to carbon
dioxide that fall far below the regression line. The size
of a gas diffusant molecule is estimated from the
Lennard-J ones collision diameter (σc), determined on the
basis of the molecular interactions of a gas, and the
kinetic diameter (σk) which is close to the molecular
sieving dimension of the molecule.34,35 The former and
latter parameters are widely accepted correlation pa-
rameters for diffusivities in the rubbery state and in
the glassy state, respectively. The two diameters are
nearly similar except for CO2, for which the values of
(σk) and (σc) are 3.30 and 4.00 Å, respectively. The
correlation shown in Figure 7 is not much better if the
P(A) D(A) S(A)
R(A/ B) )
)
(7)
P(B) D(B) S(B)
As was indicated in the Introduction, membranes are
commonly used to separate hydrogen from both hydro-
carbons and nitrogen in petrochemical processes. Let
us discuss briefly the performance of these membranes
regarding the permselectivity of hydrogen with respect
to ethane, ethylene, methane, and carbon monoxide. A
clear correlation between permselectivity and perme-
ability is not clearly seen. Actually, the rule according
to which the higher is the permeability the lower is the
permselectivity does not seem to hold for R(H2/C2H6).
As shown in Figure 8, the permselectivity coefficient of
hydrogen with respect to ethane increases as the
permeability coefficient of hydrogen in the membranes
increases, reaching a maximum for the membrane M1,
and then decreases. As was indicated above, the dif-
fusive process is responsible for the elevated permse-
lectivity of the membranes toward hydrogen. Actually,
the high diffusion coefficient of hydrogen overcomes the
poor solubility of this gas in the membranes.
Values of R(H2/N2), R(H2/CO), R(H2/CH4), R(H2/C2H4),
and R(H2/O2) are plotted as a function of the perme-
ability coefficient of hydrogen in Figure 9. The values
of R(H2/C2H6), R(H2/CH4), and R(H2/N2) for the M1, M2,
and M3 membranes are much higher than those re-
ported for poly(vinyltrimethylsilane), poly(trimethylsi-
lylnorbornene), and fluorine-containing ring-opened
polynorbornenes.22,23 It is worth noting that the mem-
brane M4 is highly selective to hydrogen with respect
to nitrogen. The values of R(H2/CO), R(H2/CH4), and
R(H2/C2H4) are quite similar, and they do not show a
clear dependence on the type of membrane.
diameter of carbon dioxide is taken to be as (σkσc)1/2
)
3.63 Å. Therefore, repulsive interactions between the
polar CO2 and the polar matrices presumably delay the
diffusion of the gas through the membranes.
Double bonds in the molecular chains confer rigidity
to them alleviated by motions about two single bonds
located in the structural unit. Conformational transi-
tions on the cyclopentyl moieties that hinder chain
packing cannot be ruled out. However, the polymers
used in this study exhibit rather low permeability in
comparison with other glassy polymers containing bulky
The values of R(H2/CO2) lie in the interval 0.9-2.1
although the results obtained for D(H2)/D(CO2) are 67,
62, 73, 68, and 154, respectively, in the membranes M1,
M2, M3, M4, and M5. However, the solubility coefficient
of CO2 is significantly higher than that of H2, the values
of S(CO2)/S(H2) being 44, 70, 76, 32, and 70, respectively,

Macromolecules, Vol. 35, No. 12, 2002
Ring-Opened Polynorbornenes 4683
Con clu sion s
Membranes prepared from poly(N-1-adamantyl-exo-
norbornene-5,6-dicarboximide) (M1), poly(N-cyclohexyl-
exo-norbornene-5,6-dicarboximide) (M2), poly(N-phenyl-
exo-norbornene-5,6-dicarboximide) (M3), and N-(1-
adamantyl)-exo-norbornene-5,6-dicarboximide/norbor-
nene copolymers, with molar ratios 50/50 (M4), 70/30
(M5), and 30/70 (M6), exhibit permselectivity charac-
teristics strongly dependent on the nature of the gases
to be separated. The membranes M1, M2, and M3 dis-
play optimal characteristics for the separation of hy-
drogen from ethane. For the separation of hydrogen
from methane all the membranes, except the M2, may
be suitable. All the membranes exhibit rather high
permselectivity for the separation of hydrogen from
nitrogen, carbon monoxide, and ethylene. Only the
membrane M3 and that prepared from M4 present
values of R(O2/N2) that could make them suitable for
the separation of oxygen from nitrogen. The other mem-
branes show a rather bad performance for the separa-
tion of these gases. In general, the membranes exhibit
permselectivity characteristics higher than those re-
ported for other polynorbornenes. Although the results
reported here are promising, more work is necessary to
reach conclusions with respect to the capability of these
materials to be used for separations in industrial
processes. Attempts are being made to rationalize the
solubility and diffusion coefficients reported in this work
by simulation of these coefficients in terms of the
chemical structure of the membranes.
F igu r e 9. Variation of the (b) R(H2/N2), (9) R(H2/CO), (2)
R(H2/CH4), (1) R(H2/C2H4), and (() R(H2/O2) with the perme-
ability of hydrogen in different membranes.
Ack n ow led gm en t. We thank the CONACyT for
generous support of this research with Contract NC-
204. The authors thank Miguel Angel Canseco, J uan
Manuel Garcia Leon, and Ruben Gavino for their
assistance in thermal, GPC, and NMR analysis. Support
of this work by the CAM (Spain) through Grant Bio-
009-2000 is gratefully acknowledged.
F igu r e 10. Permselectivity of oxygen with respect to nitro-
gen as a function of the permeability coefficient of oxygen
in the membranes: (b) M1; (9) M2; (2) M3; (1) M4; (() M5;
(#) M6.
Refer en ces a n d Notes
(1) Flory, P. J . Principles of Polymer Chemistry; Cornell Univer-
sity Press: Ithaca, NY, 1953.
(2) Petropoulos, J . H. Pure Appl. Chem. 1993, 65, 219.
(3) Vieth, W. R.; Sladek, K. J . J . Colloid Sci. 1965, 20, 1014.
(4) Wen, W.-Y. Chem. Soc. Rev. 1993, 117.
in the membranes M1, M2, M3, M4, and M5. These facts
explain the rather low permselectivity of hydrogen with
respect to CO2.
Values of the permselectivity of oxygen with respect
to nitrogen are plotted as a function of the permeability
coefficient of oxygen in Figure 10. The two membranes
less permeable to both oxygen and nitrogen, M3 and M4,
have the higher values of R(O2/N2), specifically 4.7 and
5.5, respectively. Since the apparent solubility coef-
ficients of oxygen and nitrogen are similar in the M4
membrane, about 10-3 cm3 (STP)/(cm3 cmHg), the
diffusive process is responsible for the high value of
R(O2/N2) exhibited by this membrane. Actually, the ratio
D(O2)/D(N2) for the M4 membrane amounts to 5.1. In
the case of the M3 membrane the values of D(O2)/D(N2)
and S(O2)/S(N2) are 2.9 and 1.6, respectively. The values
of the permselectivity of the other membranes are of
the order of 3, less than the lower bound (4-5) admitted
for commercial separation of oxygen from nitrogen.
(5) Compan˜, V.; Andrio, A.; Lo´pez, M. L.; Alvarez, C.; Riande,
E. Macromolecules 1997, 30, 3317.
(6) Compan˜, V.; Lo´pez-Lido´n, M.; Andrio, A.; Riande, E. Macro-
molecules 1998, 31, 6984.
(7) Kesting, R. E.; Fritzsche, A. K. Polymeric Gas Separation
Membranes; Wiley-Interscience: New York, 1993.
(8) Ohya, H.; Kudryavtsev, V. V.; Semenova, S. I. Poliimide
Membranes; Gordon and Breach Publishers: Tokyo, 1996.
(9) Koros, W. J .; Fleming, G. K. J . Membr. Sci. 1993, 83, 1.
(10) Stern S. A. J . Membr. Sci. 1994, 94, 1.
(11) Al-Masri, M.; Kricheldorf, H. R.; Fritsch, D. Macromolecules
1999, 32, 7853.
(12) Gusev, A. A.; Mu¨ller-Plathe, F.; van Gunsteren, W. F.; Suter,
U. V. Adv. Polym. Sci. 1994, 116, 207.
(13) Laguna, M.-F.; Saiz, E.; Guzma´n, J .; Riande, E. Macromol-
ecules 1998, 31, 7488.
(14) Laguna, M.-F.; Saiz, E.; Guzma´n, J .; Riande, E. J . Chem.
Phys. 1999, 110, 3200.
(15) Lo´pez, M. M.; Saiz, E.; Guzma´n, J .; Riande, E. J . Chem. Phys.
2001, 115, 6728.
Finally, it is important to know the permselectivity
coefficients of CO2 with respect to oxygen in films used
in the package industry. The pertinent results show that
the values of R(CO2/O2) are 8.5 and 7.9 for the mem-
branes M2 and M3, respectively. The values of R(CO2/
O2) for the M1, M4, and M-5 membranes are 5.3, 4.2,
and 4.2, respectively.
(16) Lo´pez, M. M.; Saiz, E.; Guzma´n, J .; Riande, E. Macromol-
ecules 2001, 34, 4999.
(17) Tlenkopatchev, M. A.; Miranda, E.; Canseco, M. A.; Gabin˜o,
R.; Ogawa, T. Polym. Bull. 1995, 34, 385.
(18) Tlenkopatchev, M. A.; Fomine, S.; Fomina, L.; Gavin˜o, R.;
Ogawa, T. Polym. J . 1997, 29, 622.
(19) Maya, V. G.; Contreras, A. P.; Canseco, M.-A.; Tlenkopatchev,
M. A. React. Funct. Polym. 2001, 49, 145.

4684 Contreras et al.
Macromolecules, Vol. 35, No. 12, 2002
(20) Ivin, K. J .; Mol, J . C. In Olefin Metathesis and Metathesis
Polymerization; Academic Press: San Diego, CA, 1997;
Chapter 13.
(21) Paul, D. R.; Yampolskii, Yu. P. Polymeric Gas Separation
Membranes; CRC Press: Boca Raton, FL, 1994.
(22) Bondar, V. I.; Kukharskii, Yu. M.; Yampolskii, Yu. P.;
Finkelshtein, E. Sh.; Makovetskii, K. L. J . Polym. Sci., Part
B: Polym. Phys. 1993, 31, 1273.
(23) Yampolskii, Yu. P.; Bespalova, N. B.; Finkelshtein, E. Sh.;
Bondar, V. I.; Popov, A. V. Macromolecules 1994, 27, 2872.
(24) Yampolskii, Yu. P.; Finkelshtein, E. Sh.; Makovetskii, K. L.;
Bondar, V. I.; Shantarovich, V. P. J . Appl. Polym. Sci. 1996,
62, 349.
(25) Dorkenoo, K. D.; Pfromm, P. H.; Rezac, M. E. J . Polym. Sci.,
Part B: Polym. Phys. 1998, 36, 797.
(27) Contreras, A. P.; Tlenkopatchev, M. A.; Ogawa, T.; Naka-
gawa, T. Polym. J . 2002, 34, 49.
(28) Kastner, K. F.; Calderon, N. J . J . Mol. Catal. 1982, 15, 47.
(29) Smith, T. A.; Aplin, R. P.; Maitlis, P. M. J . Organomet. Chem.
1985, 291, C13.
(30) Wakatsuki, Y.; Yamazaki, H.; Kumegawa, N.; Satoh, T.;
Satoh, J . Y. J . Am. Chem. Soc. 1991, 113, 9604.
(31) Asrar, J . Macromolecules 1992, 25, 5150.
(32) Crank, J . The Mathematics of Diffusion; Oxford University
Press: New York, 1975.
(33) Barrer, R. M. Trans. Faraday Soc. 1939, 35, 628.
(34) Breck, D. W. Zeolite Molecular Sieves; Wiley: New York,
1974; p 636.
(35) Costello, L. M.; Koros, W. J . J . Polym. Sci., Part B: Polym.
Phys. 1994, 32, 701.
(36) Park, J . Y.; Paul, D. R. J . Membr. Sci. 1997, 125, 23.
(26) Zhao, Ch-T.; Do Rosario Ribeiro, M.; de Pinho, M. N.;
Subrahmanyan, V. S.; Gil, C. L.; de Lima, A. P. Polymer 2001,
42, 2455.
MA011959P