Você está na página 1de 28

Submitted to Macromolecules

ROTATIONAL ISOMERIC STATES MODEL OF ERYTHRO DI-


Fo
ISOTATIC POLY(NORBORNENE)
rR
Journal: Macromolecules

Manuscript ID: ma-2009-01319b

Manuscript Type: Article


ev
Date Submitted by the
19-Jun-2009
Author:

Complete List of Authors: Chung, Won Jae; Georgia Institute of Technology, School of
ie
Chemical & Biomolecular Engineering
Henderson, Clifford; Georgia Institute of Technology, School of
Chemical & Biomolecular Engineering
w.

Ludovice, Peter; Georgia Institute of Technology, School of


Chemical & Biomolecular Engineering
Co
nf
id
en
tia
l-
AC
S

ACS Paragon Plus Environment


Page 1 of 27 Submitted to Macromolecules

1
2
3
4 ROTATIONAL ISOMERIC STATES MODEL OF
5
6 ERYTHRO DI-ISOTATIC POLY(NORBORNENE)
7
Fo
8
9 Won J. Chung, Clifford L.Henderson and Peter J. Ludovice*
10
11
rR
12 School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, GA
13
14
15 30332-0100, USA
16
ev
17
18 *pete.ludovice@che.gatech.edu
19
ie
20
21 RECEIVED DATE (to be automatically inserted after your manuscript is accepted if required
22
w.

23
24 according to the journal that you are submitting your paper to)
25
26
Co

27 ABSTRACT
28
29
A new rotational isomeric states (RIS) model was developed that accurately predicts the unique
30
nf

31
32 conformation of erythro di-isotactic poly(norbornene), a polymer with various important applications in
33
microelectronics. This model reflects the helix-kink morphology previously observed for this particular
id

34
35
36
polymer synthesized via a vinyl-like mechanism using a Pd catalyst. The model is based on
en

37
38
39 conformations observed in Monte Carlo simulations of oligomers containing 11 to 19 repeat units.
40
tia

41 These simulations indicated the origin of the kinks in the helix-kink morphology is a reversal of the
42
43
44 helix symmetry, and this was incorporated into the RIS model. These kinks are kinetically trapped and
l-

45
46 can move along the polymer chain and are created and destroyed at the chain ends. This model predicts
47
48
AC

a rigid rod conformation that eventually transitions to a random coil at a degree of polymerization of
49
50
51 approximately 500, as these kinks disrupt the helical order. The resulting RIS model also reproduced the
52
S

53 extended chain behavior predicted by viscometry experiments.


54
55
56
57
58
59 KEYWORDS: RIS, poly(norbornene), helix, kink, simulation
60
ACS Paragon Plus Environment
1
Submitted to Macromolecules Page 2 of 27

INTRODUCTION
1
2
3 Poly(norbornene) (PNB) produced via a vinyl-like polymerization is a new high performance
4
5 polymer that is useful as a new photoresist material in the microelectronics industry. This particular
6
7
PNB polymer has unique properties such as a low dielectric constant (2.2 < ε < 2.4), low absorption
Fo
8
9
10 coefficient at ultraviolet wavelengths, and low optical birefringence that make it suitable for use as an
11
rR
12 interlayer dielectric, a deep UV photoresist, and a waveguide material. [1] Recent advances in
13
14
15 polymerization catalyst technology have allowed norbornene monomer to undergo a metal catalyzed
16
ev
17 vinyl addition polymerization that retains the bicyclo-heptane group in the polymer backbone.[1-4]
18
19
This retention of the bicyclo-heptane group is in contrast to the Ring Opening Metathesis
ie
20
21
22 Polymerization (ROMP) mechanism for norbornene which retains only a single cyclopentyl-ring in the
w.

23
24 polymer backbone is shown in Figure 1. [5,6] As seen in Figure 1, there are a variety of stereochemical
25
26
configurations that can result from this polymerization depending on the exact nature of the catalyst
Co

27
28
29 used and resulting monomer enchainment mechanism. First, addition across the norbornene double
30
nf

31 bond could result in the monomer being enchained in either an endo-endo, endo-exo, or an exo-exo
32
33
form. Previous NMR studies of metal catalyzed vinyl addition polymerization, and in particular the Pd
id

34
35
36 catalyzed materials discussed in detail in this work, suggest that the exo-exo form is the only
en

37
38 enchainment product of the reaction.[7] Second, the monomer stereochemical orientations with respect
39
40
to one another can result in either an erythro di-isotactic form, an erythro di-syndiotactic form, or some
tia

41
42
43 more random intermediate form (see Figure 2).
44
l-

45
The structural complexity of the bicyclo-heptane group in the polymer backbone makes it difficult
46
47
48 to experimentally characterize the stereochemical isomerism. [8-10]. Howevever, previous simulations
AC

49
50 and their comparison to viscometry experiments indicate that the Pd-catalyzed polymer is likely the
51
52
erythro di-isotactic configuration.[3] The simulation of only this configuration produces a polymer with
S

53
54
55 a highly extended conformation where the polymer dimensions scale as nearly the square of the
56
57 molecular weight. Of the various catalysts studied, only the Pd catalyst produces a similar scaling.[3] In
58
59
60 the same simulation study, the scaling of intrinsic viscosity with molecular weight of the non-
ACS Paragon Plus Environment
2
Page 3 of 27 Submitted to Macromolecules

stereoregular polymer was slighter over 1/2 indicating a random coil conformation. This same scaling
1
2
3 was observed experimentally for the “Naked” Ni-catalyzed polymer indicating that this catalyst
4
5 produced an atactic polymers. A separate study shows random coil behavior in similarly Ni catalyzed
6
7
PNB.[11]
Fo
8
9
10 More detailed modeling of the erythro di-isotactic configuration produces a helix kink
11
rR
12 conformation, where the helix is occasionally interrupted by a kink.[4] The kink appears to be a
13
14
15 reversal of the helix symmetry from left to right handed. This helix-kink morphology results in a set of
16
ev
17 unusual physical properties for the polymer that can be important to consider in its various applications.
18
19 For example, the helix-kink morphology is believed to be a critical underlying cause for the unusual
ie
20
21
22 dissolution behavior exhibited by functionalized polynorbornenes that have been investigated as matrix
w.

23
24 materials for UV sensitive photoresists that are used in the photolithography processes that are critical
25
26 for microelectronics manufacturing. [12] This work has focused on the use of molecular modeling and
Co

27
28
29 and its comparison to experiment to determine the conformational behavior of erythro di-isotactic PNB
30
and develop a rotational isomeric states (RIS) model. This model will allow better prediction of this
nf

31
32
33
polymers unique conformational behavior.
id

34
35
36
en

37
38 n
39
40
tia

41
42 n
43
44
l-

45
46
47
48
AC

49
n
50
51
52 n
S

53
54
55 Figure 1. Poly(norbornene) produced via the vinyl-like polymerization (top) and ring opening
56
57
58 metathesis polymerization method (bottom).
59
60
ACS Paragon Plus Environment
3
Submitted to Macromolecules Page 4 of 27

7
1
2 5 3
4
3 6 2
1
4
5
6
7
Fo
8
9 Figure 2. Structure of PNB. The erythro di-isotactic (left) and erythro di-syndiotactic (right) trimers are
10
11
shown to illustrate possible stereochemical variation.
rR
12
13
14
15 This aforementioned helix-kink morphology present in erythro di-isotactic PNB induces a level
16
ev
17 of structural order in PNB that is intermediate between a crystal and a truly amorphous glass.[4,13] The
18
19
PNB helical conformation is similar to substituted poly(acetylenes) and this is presumably due to the
ie
20
21
22 alternating double bonds these polymer share.[4] This intermediate order appears as a split in the
w.

23
24
amorphous halo that is typically observed in the wide angle x-ray diffraction (WAXD) for these
25
26
polymers.[4,14,13] The splitting of the first peak in the amorphous halo into two peaks in the Wide
Co

27
28
29 Angle X-ray Diffraction Scattering (WAXD) pattern for PNB is indicative of the formation of
30
nf

31
32
intermediate range order due to strong intramolecular interactions [4]. This extended chain behavior
33
that mimics the conformation of a rigid-rod appears to be unique to the erythro di-isotactic isomer of
id

34
35
36 PNB. Simulation of PNB without this alternating bridge-head carbon showed a much less extended
en

37
38
39 conformation that was similar to a random coil conformation.[3]
40
tia

41 In this work we develop an RIS model that includes a more accurate description of this unique
42
43 helix-kink conformation that was described in the previous work.[4] This work uses Monte Carlo
44
l-

45
46 models to characterize the conformation of the kink as a switching of the symmetry of the helices over a
47
48 few repeat units. This more accurate RIS model is validated against experimental viscometry results.
AC

49
50
Such an RIS model is useful in calculating the dimensions of chains in solution and for generating initial
51
52
conformational guesses for bulk simulations for poly(norbornene). The framework of the RIS model
S

53
54
55 also has application for other polymers that may adopt the helix-kink framework such as various
56
57
58
59
60
ACS Paragon Plus Environment
4
Page 5 of 27 Submitted to Macromolecules

poly(acetylenes) and poly(norbornene) produced by Ring Opening Metathesis Polymerization (ROMP)


1
2
3 catalysts.
4
5 SIMULATION METHOD
6
7
Force Field
Fo
8
9
10 Previous modeling work done on PNB used ab initio calculations to customize bond stretch and
11
rR
12 bond angle parameters to include the effect of the ring strain present in bicyclo-heptane groups in
13
14
15 PNB.[2] Such ring strain is not typically included in generic force fields. The bond angle parameters
16
ev
17 from this work were unrealistically high, relative to generic force fields, even accounting for the ring
18
19 strain in bicyclo-heptane. We believe that these angle bend force constants are artificially high due to
ie
20
21
22 the additional coupling of bonded force terms in the bicyclo-heptane groups in PNB. Quantum
w.

23
24 mechanics programs simply estimate the second derivative of the energy with respect to a particular
25
26 bond angle. In simple hydrocarbons this derivative is independent from other bonded force field terms.
Co

27
28
29 However, these derivatives are highly dependent on bond angle and bond length terms in a bicyclic
30
group such as those found in PNB. This second derivative estimate actually includes some of these
nf

31
32
33
additional bond angle and bond length terms and is overestimated by using the second derivative from
id

34
35
36 quantum calculations. Therefore the force field used in this work utilized the bond stretch parameters
en

37
38 from these ab initio calculations, but the bond angle terms used were taken from a generic force field.
39
40
The force field used in this work utilizes bond stretch parameters taken from the customized force
tia

41
42
43 field from ab initio calculations, the angle bending parameters from CHARMM 3.03,[15] and the non-
44
l-

45 bonded parameters from the Dreiding 2.21 force field.[16] This composite force field with the bond
46
47
48 stretch, angle bending, and non-bonded parameters is compared against the MNDO calculated energy
AC

49
50 values for the erythro di-isotactic dimer seen in Figure 3.[17] The difference between the two was then
51
52 fitted as the intrinsic torsion potential also shown in Figure 3. Here, the intrinsic torsional potential
S

53
54
55 includes both the intrinsic energy of the rotation of the sigma bonding orbitals as well as some quantum
56
57 correction of the torsional potential. MNDO semi-empirical quantum calculations were used because
58
59
60
their simplicity allowed the easy calculation of the potential energy as a function of the central torsional
ACS Paragon Plus Environment
5
Submitted to Macromolecules Page 6 of 27

angle for numerous angles. However, the MNDO results in Figure 3 also compared very well to more
1
2
3 rigorous Density Functional Theory calculations.[2] The results are also essentially identical to
4
5 previous semi-empirical calculations using the AM1 model.[2]
6
7 20
Fo
8
9
10 15
11
rR
12

Energy (kcal/mole)
13 10
14
15
5
16
ev
17
18 0
19
ie
20
21 -5
22
w.

23
24 -10
25 0 50 100 150 200 250 300 350

26 Torsion Angle (Degrees)


Co

27
28
29 Figure 3. Torsion potential of erythro di-isotactic norbornene dimer. The force field with the
30
customized bond stretch terms, CHARMM 3.03 angle bend terms, along with the non-bonded
nf

31
32
33
parameters from Dreiding 2.21 is compared against the MNDO values. The dots represent the MNDO
id

34
35
36 values, the short dashed line represents the intrinsic torsion potential, the short-long dashed line
en

37
38 represent potential without the intrinsic torsion potential, and the solid line represent the final force field
39
40
with the intrinsic torsion potential.
tia

41
42
43
44 The shift dihedral functional form was selected to describe the intrinsic torsion potential for the
l-

45
46 erythro di-isotactic PNB in Equation 1.
47
48
AC

E Φ = ∑ K Φ ,n [1 + cos(nΦ − Φ 0,n )]
6
1
49 [1]
50 n =1 2
51
52
The parameters KΦ,n, Φ, and Φ0,n in Equation 1 are the dihedral force constant, torsion angle, and phase
S

53
54
55 shift, respectively. This equation was used to fit the MNDO results seen in Figure 3 and the optimized
56
57 parameters are listed in Table 1.
58
59
60
ACS Paragon Plus Environment
6
Page 7 of 27 Submitted to Macromolecules

Table 1: Intrinsic torsion parameters for erythro di-isotatic PNB using the shift dihedral form (Φ0,n =
1
2 180° to reflect the intrinsic torsional energy maximum at 180°).
3
4 n Kφ,i φn,i
5 1 -3.6002 0.1547
6 2 0.5303 -0.6704
7 3 -2.7103 -0.0344
Fo
8
4 -0.0377 10.1365
9
10 5 -2.0476 0.2865
11 6 -0.9161 0.6188
rR
12
13
14 Monte Carlo Model
15
16
ev
Metropolis Monte Carlo (MC) [18] simulations with the pivot algorithm [19] followed by energy
17
18
19 minimization were used to simulate isolated chains of PNB. The intramolecular interactions in PNB
ie
20
21 dominate over the intermolecular interactions because of the bulky backbone structure of this polymer.
22
w.

23
24
Therefore, the isolated polymer is a reasonable model for the polymer conformation. Primarily, the
25
26 intramolecular torsional barriers and backbone steric hindrance determines the polymer conformation
Co

27
28 while the surrounding polymer produces only minor structural perturbation [3,4]. Similar isolated chain
29
30
simulations have been used to accurately reproduce experimental conformations in other polymer
nf

31
32
33 systems that do not have strong inter-chain interactions. [20-22].
id

34
35 The modeled erythro di-isotactic PNB chain began in its extended conformation (torsion angles at
36
en

37
38 180o) or at random conformations. Next, a movable backbone torsion angle is randomly selected and
39
40 randomly perturbed followed by energy minimization. Energy minimization began with the conjugate
tia

41
42
43
gradient [23] algorithm followed by a quasi-Newton method with a Broyden, Fletcher, Goldfarb,
44
l-

45 Shanno (BFGS) update of the Hessian [24]. Each minimization step, included in the MC moves,
46
47 proceeded until the potential energy gradient was less the 0.001 kcal/(mole Å). Single chains of 11, 15,
48
AC

49
50 and 19 repeat units of erythro di-isotatic PNB were simulated. These chains were simulated for 2000
51
52 MC moves and resulted in 3 typical conformations with an acceptance ratio of approximately 15%. MC
S

53
54 simulations were performed on four different initial conformations that included isolated PNB chains in
55
56
57 the extended conformation and three additional conformations that used backbone torsion angles in a
58
59 random uniform distribution. Five MC simulations were performed on each of the four initial
60
ACS Paragon Plus Environment
7
Submitted to Macromolecules Page 8 of 27

conformations, totaling 20 MC simulations for each chain length. Each of these were run for a total of
1
2
3 2000 MC moves which was sufficient to converge to a local energy minimum.
4
5
6
7
Rotational Isomeric States Model
Fo
8
9
10 Rotational Isomeric States (RIS) [25,26] models are first order Markov models that describe a
11
rR
12 polymer conformation in terms of a torsion angle distribution, which assumes that each torsion angle is
13
14
15 determined only by the one previous backbone torsion angle. Typically this distribution is extracted
16
ev
17 from a model of a small oligomer of the polymer that includes only two adjacent torsion angles. The
18
19 long-range effects of the bulky bicyclo-heptane ring in PNB make this assumption unreasonable; and
ie
20
21
22 larger structures that include more than just neighboring torsion angles must be used to produce these
w.

23
24 probabilities [4,9,10]. RIS states maps were calculated previously for a trimer, pentamer, and heptamer
25
26 for the erythro di-isotatic PNB that showed drastically different behavior as a function of molecular
Co

27
28
29 weight.[4]
30
The aforementioned MC model was used to sample the conformational space of erythro di-isotatic
nf

31
32
33
PNB and determine the probability distribution of backbone torsion angles using larger oligomers to
id

34
35
36 include long-range effects. This torsioinal angle distribution, that includes these long-range effects, is
en

37
38 then used in a reduced RIS model that models the distribution of a torsion angle as a function of its
39
40
adjacent torsion angle only. The major advantage of the RIS model is that it can be used to efficiently
tia

41
42
43 model dilute solution behavior or to generate the initial conformations for bulk modeling in periodic
44
l-

45 boundary conditions.[27-33] Using the RIS matrix generation scheme [25,26] the unperturbed mean
46
47
48 square of the end-to-end distance, r 2 , as well as the unperturbed mean square of the radius of
AC

0
49
50
51 gyration, s 2 , can also be quickly determined as a function molecular weight.
0
52
S

53
54
55
56 RESULTS AND DISCUSSION
57
58
59 Monte Carlo Simulation
60
ACS Paragon Plus Environment
8
Page 9 of 27 Submitted to Macromolecules

Metropolis MC simulation with the pivot algorithm was conducted for erythro di-isotactic PNB
1
2
3 with chain lengths of 11, 15, and 19 repeat units corresponding to molecular weights of 848, 1412, and
4
5 1788 g/mole, respectively at a temperature of 298 K. The energy evolution of the MC simulation on a
6
7
chain length of 11 repeat units for 2000 MC moves equilibrated after only 250 MC moves for several
Fo
8
9
10 different starting configurations where the initial chain was in its extended or random conformations and
11
rR
12 all had the same equilibrium energy of 432 kcal/mole with an acceptance ratio of approximately 15%.
13
14
15 Due to the fact that MC simulation involves abrupt transitions, the effect of the number of MC moves
16
ev
17 was examined. For the 11 repeat unit erythro di-isotactic PNB, 2000, 5000, and 10,000 total MC moves
18
19 were compared. For all three cases, the equilibrium conformation is achieved after about 250 MC
ie
20
21
22 moves with a resulting energy of 432 kcal/mole and no abrupt transitions or new minima were observed.
w.

23
24 Therefore, 2000 MC moves on the Metropolis MC simulation are sufficiently long enough to reach
25
26 equilibrium.
Co

27
28
29 Independent of the starting conformation, the final conformation always consisted of repetitive
30
torsion angles of either ~120o or ~240o, and both configurations had the same equilibrium energy of 432
nf

31
32
33
kcal/mole. With the exception of the torsion angles near the ends of the polymer chain, all initial
id

34
35
36 conformations produced helices with repeating angles of ~240°, or 120°. Both helices have
en

37
38 approximately 16 repeat units per turn and they are mirror images of each other since their backbone
39
40
torsion angles are symmetric about 1800. No kinks were observed in these helices because this 11
tia

41
42
43 repeat oligomer is simply not long enough to show any kinks in the helices.
44
l-

45
Previously, an RIS model was developed by calculating a RIS state map for the heptamer erythro
46
47
48 di-isotactic PNB [4]. The energies were calculated by rotating the two middle adjacent torsion angles in
AC

49
50 increments of 10o with the customized force field [2]. The energy was minimized at each point with
51
52
respect to the 4 external torsion angles. The RIS state map is only an approximate characterization of
S

53
54
55 the energetic states because the 4 external torsion angles are in local minima. This model produced
56
57 helical regions with occasional interruptions of the helices with kinks with the probability of a helix
58
59
60 angle followed by a kink of approximately 3.27% as determined from the Boltzmann factor using the
ACS Paragon Plus Environment
9
Submitted to Macromolecules Page 10 of 27

relative energies of the helix and kink conformations.[2] Both methods produced similar conformations
1
2
3 that include helices, but the previous work did not model the kink as a transition between left and right
4
5 handed symmetry of the helix. This previous model simply included a single torsion angle perturbation
6
7
without a symmetry change that was not energetically realistic. Helix-kink morphology was not seen in
Fo
8
9
10 the 11 repeat unit oligomer of this work because the chain was not long enough to exhibit a kink due to
11
rR
12 the low probability of the occurrence of the kink. For this reason, oligomers with 15 and 19 repeat units
13
14
15 were also simulated.
16
ev
17 The simulation of an oligomer of 15 repeat units showed 3 possible conformations. Helical
18
19 conformations consisting of either repeating torsion angles of 120o or 240o where the final accepted
ie
20
21
22 conformations are identical to the results from the 11 repeat units chains with an average equilibrium
w.

23
24 energy of 608.5 kcal/mole. A helix-kink conformation similar to what was seen in previous simulations
25
26 was also observed and the torsion angle distribution for this conformer is seen in Figure 4. The average
Co

27
28
29 potential energy of the conformation of the helix with a single kink is 612.0 kcal/mole. No simulation
30
contained more than one kink in the polymer chain. This helix-kink conformation consists of repetitive
nf

31
32
33
torsion angles of ~120° followed with the kink transition angles in the order of ~100°, ~200°, and ~260°
id

34
35
36 followed by another set of repetitive torsion angle of ~240°. The conformation is changing handedness
en

37
38
39
in the helices from a left-handed helix to that of a right-handed helix with the kink being the transition
40
tia

41 angles in between the two regions of different handedness. Assuming there is a negligible difference in
42
43 the entropy between the helix and the transition between helices we can use a Boltzmann factor to
44
l-

45
46 calculate the relative population of the kink. The probability of the chain going from having no kink to
47
48 having a kink is 0.266% using Boltzmann’s factor of the conformational energy difference, but the
AC

49
50 occurrence of helices in our oligomer simulations is higher because the kinks can be formed at the chain
51
52
ends and this occurs at an artificially high rate in the simulations because they are relatively short. This
S

53
54
55 may be the only way to observe these kinks as they would be less prevalent in long chain constrained by
56
57
the surrounding bulk polymer.
58
59
60
ACS Paragon Plus Environment
10
Page 11 of 27 Submitted to Macromolecules

350
1
2
3 300

Torsion Angle (Degrees)


4
5 250
6
7 200
Fo
8
9
150
10
11
rR
12 100
13
14 50
15
16
ev
0
17 0 500 1000 1500 2000
18 Monte Carlo Move #
19
ie
20
21 Figure 4. MC simulation on erythro di-isotatic PNB with 15 repeat units. Torsion angle distribution as
22
w.

23 a function of MC move is shown with the two torsion angles to each of the ends excluded. The
24
25
26 equilibrium conformation has an energy of 612.0 kcal/mole that consists of repetitive torsion angle of
Co

27
28 120o or 240o.
29
30
nf

31
32
33 The torsion angles that make up the kink are identical in a relative sense but have different
id

34
35
36 absolute values depending on the symmetry of the helix transition. Going from a helical region
en

37
38 consisting of repeated torsion angles of 120° to a helical region made up of 240° repeated torsion
39
40
angles, the kink is in the order of 100°, 200°, and 260° torsion angles. However, going from a helical
tia

41
42
43 region consisting of repeated torsion angles of 2400 to a helical region of repeated torsion angles of 120°
44
l-

45
46 the transition exhibits the following torsion angles in sequence of 260°, 160°, and 100°. The sequential
47
48 transition angles between helices consisting of repeated helical angles of 120° or 240° as seen below.
AC

49
50
51
52 − ⎡⎣120D ⎤⎦ − 100D − 200D − 260D − ⎡⎣ 240D ⎤⎦ − 260D − 160D − 100D − ⎡⎣120D ⎤⎦ −
n1 n2 n3
S

53
54
55 kink kink
56
57 An abrupt transition is seen around MC move number 1450 and 1750 in Figure 4. However, there
58
59 are no significant energetic transitions associated with this torsion angle change as seen in the energy
60
ACS Paragon Plus Environment
11
Submitted to Macromolecules Page 12 of 27

plot for this simulation in Figure 5. A typical energetic distribution is shown for an oligomer without a
1
2
3 kink and an oligomer with a single kink. The oligomer that contains a kink is always of slightly higher
4
5 energy. This abrupt transition is associated with a translation of the kink along the axis of the helix.
6
7
This illustrates a mechanism for the creation and annihilation of kinks at the ends of the chains. The
Fo
8
9
10 energy before MC move 1450 and after MC move 1750 is essentially the same at a value of 612.0
11
rR
12 kcal/mole, which is the equilibrium value of the conformation that has the repeated torsion angle of 120°
13
14
15 interrupted with the kink transition angles followed by a helical region that consists of repeated torsion
16
ev
17 angle of 240°. The torsion angle distribution before MC move 1450 (neglecting the two torsion angles
18
19
nearest both ends of the oligomer) is given below.
ie
20
21
22 − 120 0 − 120 0 − 120 0 − 120 0 − 120 0 − 100 0 − 200 0 − 260 0 − 240 0 − 240 0 −
w.

23
24
25
kink
26
Co

700
27
28
29
30 680
nf

31
Energy (kcal/mole)

32
33 660
id

34
35
36 640
en

37
38
39
620
40
tia

41
42
43 600
0 500 1000 1500 2000
44
l-

45 Monte Carlo Move #


46
47 Figure 5. MC simulation on erythro di-isotatic PNB with 15 repeat units. Solid line represents final
48
AC

49
50 conformation consisting of repeated torsion angle of 120° and the dotted line represents the
51
52 conformation that has the repeated torsion angle of 120° interrupted with the kink transition angles then
S

53
54
55 back in to helical region that consisted of repeated torsion angle of 240°, with equilibrium energies of
56
57 608.5 and 612.0 kcal/mole, respectively.
58
59
60
ACS Paragon Plus Environment
12
Page 13 of 27 Submitted to Macromolecules

1
2
The same distribution after MC move 1750, where the kink and the associated transition angles have
3
4
5 shifted to the right of the oligomer chain as seen below.
6
7
Fo
8 − 120 0 − 120 0 − 120 0 − 120 0 − 120 0 − 120 0 − 100 0 − 200 0 − 260 0 − 240 0 −
9
10
11
kink
rR
12
13 The energy barrier for shifting of the kink, moving down or up along the chain is approximately 4.0
14
15 kcal/mole which is the maximum in the energy difference between the energy between moves 1450 and
16
ev
17
18 1750 and the average energy before and after this period from Figure 5.
19
ie
20 In order to determine if this behavior observed in the oligomer of 15 repeat units was
21
22 representative of the polymer chain, an even longer oligomer of 19 repeat units was also simulated. This
w.

23
24
25 oligomer showed essentially the same behavior observed in the oligomer of 15 repeat units. The exact
26
Co

27 same 3 possible conformations were observed, which includes helical conformations consisting of either
28
29
repeating torsion angles of ~1200 or ~2400 with an equilibrium energy of 785.2 kcal/mole and
30
nf

31
32 conformations that have a helical segment that is interrupted with the kink transition angles followed by
33
a helical segment with an equilibrium energy of 789.0 kcal/mole, as shown in Figure 6.
id

34
35
36 880
en

37
38
39
860
40
tia

41
Energy (kcal/mole)

42
43 840

44
l-

45
46 820
47
48
AC

49 800
50
51
52 780
S

53 0 500 1000 1500 2000


54
Monte Carlo Move #
55
56
57 Figure 6. Potential energy from the MC simulation of erythro di-isotatic PNB with 19 repeat units.
58
59
Solid line represents final conformation consisting of repeated torsion angle of 240o, the dotted line
60
ACS Paragon Plus Environment
13
Submitted to Macromolecules Page 14 of 27

represents the conformation with 120o repeated torsion angles, and the short-long dashed line represents
1
2
3 the conformation that has the repeated torsion angle of 120o interrupted with the kink transition angles
4
5 then back in to helical region consisting of repeated torsion angle of 240o, with equilibrium energies of
6
7
785.2, 785.2, and 789.0 kcal/mole, respectively.
Fo
8
9
10
11 Independent of the starting conformation, the final conformations that consisted of repetitive
rR
12
13 torsion angles of either ~120° or ~240°, which both had the same potential energy of 785.20 kcal/mole
14
15
16 after energy minimization. The torsion angle distributions are shown in Figure 7a and7b. Deviations
ev
17
18 from 120° or 240° in Figures 7a and 7b adopt values similar to the transition angles above. The angle in
19
ie
20
21
Figure 7a that transitions from ~120° to ~100° and then later to ~200° is the third angle from the end of
22
w.

23 the oligomeric chain (the two end angle are are not included in these figures). This illustrates the
24
25 formation of a kink from the end of the PNB chain. As this angle is third from the end transitions from
26
Co

27
28 ~100° to ~200° the adjacent angle that is fourth from the end transitions from ~120° to ~100° as seen in
29
30 Figure 7a. A similar phenomenon is observed in Figure 7b as the torsion angle located 3rd from the end
nf

31
32
33 transitions from ~240° to ~260°. These transitions are the formation of a kink at the chain end. Since
id

34
35 these kinks are a change in the symmetry of the helices their net formation or destruction can only occur
36
en

37
38
at the chain end.
39
40
tia

41
42
43
44
l-

45
46
47
48
AC

49
50
51
52
S

53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
14
Page 15 of 27 Submitted to Macromolecules

1 350 350
2
3 300
4 300
Torsion Angles (Degrees)

Torsion Angle (Degrees)


5
6 250
7 250
Fo
8
200
9
10 200
11 150
rR
12
13 150
14 100

15
16
ev
50 100
17 0 500 1000 1500 2000 0 500 1000 1500 2000
18 Monte Carlo Move # Monte Carlo Move #
19
ie
20
21
22 (a) (b)
w.

23
24
25 Figure 7. MC simulation on erythro di-isotatic PNB with 19 repeat units. Torsion angle distribution as
26
Co

27 a function of MC move is shown with the two torsion angles to each of the ends excluded for (a) helix
28
29
30 with 1200 torsion angles with an equilibrium energy of 785.2 kcal/mole. (b) helix with 240° torsion
nf

31
32 angles with an equilibrium energy of 785.2 kcal/mole.
33
id

34
35
36
en

37
38 The 3rd type of conformation included the helix-kink morphology where the helical region is
39
40 interrupted by kink transition torsion angles where the handedness of the helicex was changing. The
tia

41
42
43
equilibrium conformation has an energy of 789.0 kcal/mole that consists of repetitive torsion angle of
44
l-

45 120° followed with the kink transition angles in the order of 100°, 200°, and 260° followed by another
46
47
set of repetitive torsion angle of 240° is shown in Figure 8. Assuming a Boltzman distribution for this
48
AC

49
50 observed energy difference between the kinked and un-kinked helix suggests that the probability of the
51
52 chain possesing a kink is approximately 0.17%. This low probability indicates that observing a kink in
S

53
54
55 the simulation of our oligomers should be highly unlikely. However, as with the smaller oligomers, this
56
57 energy difference does not reflect the fact that kinks are formed at the chain ends and therefore appear
58
59 with a greater frequency in the simulations here.
60
ACS Paragon Plus Environment
15
Submitted to Macromolecules Page 16 of 27

This kinked helix structure, comprised of two angles of 240° with the remaining non-transition
1
2
3 angles at 120° forms after approximately 1400 MC moves. Prior to this no angles of 240° existed, and
4
5
6
these are formed as a kink which starts near the end of the chain moves two positions into the 120°
7
Fo
8 angles eventually forming two 240° angles in the wake of this shift. Note the angle distribution at the
9
10
end of the chain does not match the angle distribution above exactly because of angle variation at the
11
rR
12
13 chain end. However, after the transition only the angles in this distribution are observed (120°, 240°,
14
15
100°, 200° and 260°).
16
ev
17
18 350
19
ie
20 300
21
Torsion Angle (Degrees)

22
w.
250
23
24
25 200
26
Co

27 150
28
29 100
30
nf

31
50
32
33
0
id

34
0 500 1000 1500 2000
35
36 Monte Carlo Move #
en

37
38 Figure 8. MC simulation on erythro di-isotatic PNB with 19 repeat units. Torsion angle distribution as
39
40
a function of MC move is shown with the two torsion angles to each of the ends excluded. The
tia

41
42
43 equilibrium conformation has an energy of 789.0 kcal/mole that consists of repetitive torsion angle of
44
l-

45 120° followed with the kink transition angles of 100°, 200°, and 260° followed by another set of
46
47
48 repetitive torsion angle of 240°.
AC

49
50
51
52
S

53
54
From the MC simulation above the mean squared unperturbed radius of gyration s 2 averaged
0
55
56
57 over simulated chains varies as a power law function with molecular weight. The value of s 2 scales
0
58
59
60
with molecular weight to the 1.85 power from a non-linear fit of MC simulations of the three different
ACS Paragon Plus Environment
16
Page 17 of 27 Submitted to Macromolecules

molecular weights (R2=0976). An identical trend vs. molecular weight is obtained for the mean squared
1
2
3 unperturbed end-to-end distance r 2 where the scaling exponent is 1.89 with molecular weight. Both
0
4
5
6 scalings are indicative of a highly extended chain similar to the rigid-rod behavior for the erythro di-
7
Fo
8 isotactic PNB. This exponent should vary from ½ for a random coil in the θ condition to a value of 2
9
10
11 for a rigid rod. These scaling values compare well to their experimental scaling of approximately 2
rR
12
13 determined from intrinsic viscosity.[3,4].
14
15
16
Rotational Isomeric States Model
ev
17
18 RIS models are used to describe the microstructure of isolated polymer chains in detail as a
19
ie
20 function of their torsional angle distributions.[34] Additionally, RIS models are used to generate initial
21
22
w.

23 conformations in periodic boundary conditions where combinations of subsequent minimizations and or


24
25 dynamics are used to relax the structure to simulate bulk behavior.[26,33] Because polymer glasses are
26
Co

27 highly constrained they relax only small amounts under typical phase space sampling techniques. For
28
29
30 this reason, RIS models are particularly important in producing accurate initial conformations because
nf

31
32 the relaxed models do not depart significantly from the vicinity of the initial conformation generated
33
id

34
[27,29-32].
35
36
Previously, preliminary RIS descriptions for the erythro di-isotactic PNB were developed based
en

37
38
39 on potential energy contour maps for much smaller oligomers than the ones simulated here. [4,8].
40
tia

41
42
These models looked at only two adjacent torsion angles. While they both characterized the elongated
43
44 nature of the polymer chain, they did not include the detailed nature of helices interrupted by kinks
l-

45
46 caused by a helix symmetry reversal. This level of order is not revealed with such small models and is
47
48
AC

49 the basis of the RIS model developed here. The statistical weight matrix, U, for the RIS model for
50
51 erythro di-isotactic PNB is given in equation 2
52
S

53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
17
Submitted to Macromolecules Page 18 of 27

1
φ 2 = 120 100 a 200 260 a 240 260 b 160 100 b
2 φ1 = 120 ⎡ 1 0.002 0 0 0 0 0 0 ⎤
= 100 a ⎢⎢ 0 ⎥⎥
3
4 φ1 0 0 1 0 0 0 0
5 φ1 = 200 ⎢ 0 0 0 1 0 0 0 0 ⎥
6 ⎢ ⎥
7 φ = 260 a ⎢ 0 0 0 0 1 0 0 0 ⎥ [2]
U= 1
Fo
8 φ1 = 240 ⎢ 0 0 0 0 1 0.002 0 0 ⎥
9 ⎢ ⎥
10 φ1 = 260 b ⎢ 0 0 0 0 0 0 1 0 ⎥
11
φ1 = 160 ⎢ 0 0 0 0 0 0 0 1 ⎥
rR
12 ⎢ ⎥
13 φ1 = 100 b ⎢⎣ 1 0 0 0 0 0 0 0 ⎥⎦
14
15
16 where the state of bond 2 is defined according to statistical weights, which depend on the state of only
ev
17
18 the previous bond 1. Some of the torsion angles repeat but the transition of one helix to another is path
19
ie
20
dependent. The superscript a and b represents the path of going from a ~120° helix to that of ~240°
21
22
w.

23 helix and ~240° helix to that of ~120° helix, respectively. Based on the aforementioned Boltzmann
24
25 probability using the energy difference from the MC simulations of the oligomer with 15 and 19 repeat
26
Co

27
28 units, the probability of a kink (i.e. going from one helix to another) is ~0.2%.
29
30 12
nf

31
32
10
33
id

34 Experimental Rigid Rod


35 8 Region
36
<r >/<s >
2

en
o

37 Random Coil
38 6
2

39
o

40 4
tia

41
42
43 2
44
l-

45
0
46 0 0.001 0.002 0.003 0.004 0.005
47
48 1/(# bonds)
AC

49
50 Figure 9. The ratio of the unperturbed mean square end-to-end distance to the unperturbed mean square
51
52
radius of gyration plotted against inverse number of bonds from the RIS model. The value of this ratio
S

53
54
55 is approximately constant at a value of 11.5 and decreases rapidly after passing a certain chain length.
56
57 In the limit, the value of this ratio is 12 for a rigid rod and a value of 6 for random coil.
58
59
60
ACS Paragon Plus Environment
18
Page 19 of 27 Submitted to Macromolecules

This RIS model was used to calculate the scaling of the mean square radius of gyration and mean
1
2
3 square end-to-end distance as a function of molecular weight by using the matrix generation scheme of
4
5 the RIS model [25,26]. The ratio of the unperturbed mean square end-to-end distance r02 to the
6
7
Fo
8 unperturbed mean square radius of gyration s 02 is plotted as a function of the inverse of the degree of
9
10
11 polymerization (number of bonds) from the RIS model is shown in Figure 9. The value of this ratio is
rR
12
13
14 approximately constant at 11.5 until a polymer containing approximately 1000 bonds is reached and the
15
16 value begins to decreases to 6. Experimentally, a transition from rigid to random coil behavior was
ev
17
18 observed at a similar chain size using the ratio of the radius of gyration to the hydrodynamic radius for a
19
ie
20
21 Pd catalyzed polymer.[9] The value of this ratio varies from 12 in the rigid rod limit to 6 in the random
22
w.

23 coil limit. These results indicate that PNB behaves like a highly extended chain at low molecular weight
24
25 which transitions to a random coil in the very high molecular weight limit as the kinks become more
26
Co

27
28 prevalent. The molecular weight of the erythro di-isotactic PNB used in the previous viscometry
29
30 experiments are in the low molecular weight range so they should exhibit the rigid rod scaling in
nf

31
32
33
viscometry [3,4]. This extended chain conformation is close to the rigid rod limit because the kink
id

34
35 causes a deviation of the chain persistence direction by only ~3° at 0K as shown in Figure 10. The
36
en

37 conformation pictured in Figure 14 is an energy minimized structure and consequently represents the
38
39
40 polymer chain in the 0K temperature limit. The polymer is certainly not this rigid, as even the helix is
tia

41
42 not perfect at typical temperatures as scene in the previous bulk simulations.[4] Since every other bond
43
44
is non-rotatable, the turns of each repeat unit in the helix are very slow and require 16 repeat units to
l-

45
46
47 make 1 turn in the helix.
48
AC

49
50
51
~300
~3
52
S

53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
19
Submitted to Macromolecules Page 20 of 27

Figure 10. The angle created by the kink that changes the helical symmetry changes the persistence
1
2
3 direction by ~3°. The helices are perfect helices at 120° and 240°, where in realistic bulk structures the
4
5 helices will not be perfect and the angle created would be slightly different.
6
7
Fo
8
9
10
11 The values of s 02 from the RIS model and the MC simulations used to develop the RIS model
rR
12
13
14 are plotted as a function of molecular weight shown in Figure 11. The scaling of s 02 with molecular
15
16
ev
17 weight is 1.96 from the RIS model and 1.85 from the MC simulations. This agreement indicates that the
18
19 RIS model has captured the appropriate scaling observed in the atomically-detailed model used in the
ie
20
21
22 MC simulations. Slightly better agreement between the RIS and MC models is obtained for r02 where
w.

23
24
25 the RIS and MC scaling are 1.92 and 1.89 respectively. However s 02 is generally more reliable as it is
26
Co

27
28 based on all the atoms in the oligomer chain rather than the end atoms.
29
30 10
6
nf

31
32
33 5
10
id

34
35
36 4
10
en

37
<s >

38
2
0

39
40 1000
tia

41
42
43 100
44
l-

45
46 10
4 5
47 1000 10 10

48 MW
AC

49
50 Figure 11. Comparison of MC results for erythro di-isotactic PNB (circles) and RIS model calculations
51
52
(triangles) for the unperturbed mean square radius of gyration.
S

53
54

The intrinsic viscosity [η ] is calculated from the universal viscosity law to relate the [η ] scaling
55
56
57
58 with molecular weight. While this law assumes a random-coil conformation, a proportionality between
59
60
ACS Paragon Plus Environment
20
Page 21 of 27 Submitted to Macromolecules

1 [h] and a function of the molecular weight (M) and the unperturbed radius of gyration s02 At the θ
.
2
3
4 condition, the following scaling law is derived from the universal viscosity law [4]
5
6 32
⎛ s 02 ⎞
7
[η ] ∝ ⎜⎜ ⎟ M12 . [3]
Fo
8 M ⎟
9 ⎝ ⎠
10
11 This proportionality is valid regardless of whether the polymer conformation is a random coil or a more
rR
12
13
14 elongated chain as observed in PNB. However, the proportionality constant in the relationship above
15
16 does vary depending on the polymer conformation, so the actual universal viscosity law cannot be used
ev
17
18
19 given that the ratio of r02 to s 02 changes with molecular weight. Using the 1.98 power scaling of
ie
20
21
22 s 02 with molecular weight and equation 6 we obtain a 1.97 power scaling of intrinsic viscosity with
w.

23
24
25
molecular weight. This is consistent with the experimentally determined value of this scaling of
26
approximately 2.[3,4]
Co

27
28
29 While the RIS model does reproduce the appropriate scaling from equation 3, it does not take into
30
nf

31
32 account the dynamic variation about the torsion angle used in the RIS model. The force field employed
33
here possesses a slightly flatter energy curve in the vicinity of the RIS angles relative to many vinyl
id

34
35
36 polymers typically modeled by RIS models as seen in Figure 3. This means that there is some variation
en

37
38
39 about these angles of 120° and 240° and this warrants an investigation of the effect of this flexibility.
40
tia

41 Previous bulk simulations showed a variation of ±20° and resulted in a much less regular polymer
42
43
44 conformation than the one pictured in Figure 11.[4] To determine the effect of this torsional state
l-

45
46 flexibility, the RIS statistical weight matrix was incorporated into the amorphous builder module in the
47
48
AC

49
Cerius2 software developed by Accelrys. The combination of the van der Waals energy and the RIS
50
51 statistical weight matrix is utilized in order to prevent the chain being built from overlapping with itself.
52
S

53 Chains were generated randomly using the RIS model above with a specified tolerance on the torsional
54
55
56 angles and if the next added monomer increased the potential energy by more than 10 kcal/mole the unit
57
58 was rejected and a new torsion angle was chosen. If this procedure failed 20 times another monomeric
59
60
ACS Paragon Plus Environment
21
Submitted to Macromolecules Page 22 of 27

unit was removed from the polymer chain. The root mean square average of the radius of gyration
1
2 1

s 02
2
3 for the erythro di-isotactic PNB is graphed as a function of the number of bonds in Figure 12.
4
5
6 Note that this is converted to degree of polymerization by multiplying by two, and the degree of
7
Fo
8 polymerization is converted to molecular weight by multiplying the degree of polymerization by 94.15
9
10
11 g/mole. Similar results were also obtained for r02 vs. molecular weight. A total of 50 chains were
rR
12
13
14 simulated using the RIS torsion angle distribution, but allowing windows of either ±10° or ±200 to
15
16
ev
account for this deviation in the helical regions of the conformation. Figure 12 shows a distinct
17
18
19 reduction in the radius of gyration and the scaling with the size of the polymer. The mean squared
ie
20
21 radius of gyration (the root mean squared value is seen in Figure 12) scaled with the size of the polymer
22
w.

23
chain to the 1.94 power for the isolated chains in Figure 12 using the original RIS model. This is similar
24
25
26 to the value of 1.96 from the RIS model used in the generation of these chains. This power decreases to
Co

27
28 1.62 and 1.27 when windows of ±10° and ±20° are applied to the RIS torsion angles respectively.
29
30
Using equation 3, these correspond to scaling exponents of the intrinsic viscosity with molecular weight
nf

31
32
33 of 1.91, 1.43 and 0.9 for no flexibility, ±10° and ±20° flexibility cases respectively. Since the RIS
id

34
35
36 model without this additional flexibility is closest to the experimentally measured scaling exponent of 2,
en

37
38 we conclude that additional flexibility does not improve the accuracy of the RIS model and is not
39
40 needed in this model.
tia

41
42
43
44
l-

45
46
47
48
AC

49
50
51
52
S

53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
22
Page 23 of 27 Submitted to Macromolecules

800
1
2
700
3
4
5 600
6
7 500
Fo
8

2 1/2
9

<s >
400

0
10
11
rR
300
12
13
14 200
15
16
ev
100
17
18 0
19 0 500 1000 1500
ie
20 # bonds
21
22
w.
1

s 02
2
23 Figure 12. Simulation of of PNB determined from RIS model at various degrees of
24
25
26 polymerization. Simulations include a window of ±0°, ±10°, and ±20° variation about the local energy
Co

27
28
29
minima of the backbone torsion angles represented by ▼, ▲, and ♦, respectively. All error bars are the
30
95% confidence interval about the mean.
nf

31
32
33
id

34 CONCLUSIONS
35
36
A new RIS model was developed for the erythro di-isotactic isomer of poly(norbornene) that
en

37
38
39 reflects the helix-kink morphology previously observed for this particular polymer in these and previous
40
tia

41 simulations. Monte Carlo simulations indicate that the origin of the kink is the reversal of the helical
42
43
44 symmetry, which occurs in approximately 0.2% of the monomeric units. These kinks can form or be
l-

45
46 destroyed at the polymer ends and each one of these kinks adds approximately 4 kcals/mole to the
47
48
AC

energy of the helical structure. Similarly, the barrier to the movement of the kink down the chain also
49
50
51 appears to be approximately 4 kcals/mole, suggesting that these kinks are kinetically trapped in the
52
S

53 polymer chain and their distribution may be modified by high temperature annealing. The RIS model
54
55
and the MC simulations, on which this model is based, predict a polymer that has rigid–rod like
56
57
58 behavior that persists until a transition to random coil behavior that begins when the number of
59
60
ACS Paragon Plus Environment
23
Submitted to Macromolecules Page 24 of 27

backbone bonds exceeds 1000. The RIS model accurately reproduces the rigid-rod scaling behavior of
1
2
3 this polymer extracted from viscometry experiments for a Pd catalyzed polymer that is assumed to be
4
5 erythro di-isotactic polymer. A sensitivity analysis of including flexibility about the RIS torsional states
6
7
was conducted and this found that the rigid rod scaling observed in the viscometry experiments
Fo
8
9
10 disappeared from the model when such flexibility was incorporated. This RIS model reproduces the
11
rR
12 unique conformation of this industrially important polymer that has a molecular weight dependent
13
14
15 conformation.
16
ev
17
18
19
ie
20 ACKNOWLEDGMENT
21
22
The authors gratefully acknowledge financial support from NSF award DMR-000309236 and Promerus
w.

23
24
25 Corporation. We also wish to thank Larry Rhodes, Larry Seger, and Robert Shick of the Promerus
26
Co

27 Corporation for helpful input and discussions.


28
29
30
REFERENCES
nf

31
32
33
(1) Goodall, B.L.; Benedikt, G.M.; McIntosh, L.H.; Barnes, D.A. and Rhodes, L.F. International
id

34
35
36 Patent WO95/14048, 1995.
en

37
38
39 (2) S. Ahmed, S.A. Bidstrup, P.A. Kohl and P.J. Ludovice. Journal of Physical Chemistry B. 102,
40
tia

41
42 9783-9790 (1998).
43
44
l-

45 (3) S. Ahmed, S.A. Bidstrup, P.A. Kohl and P.J. Ludovice. Macromolecular Symposia. 133, 1-10
46
47 (1998).
48
AC

49
50
(4) S. Ahmed, P.J. Ludovice, and P.A. Kohl. Computational and Theoretical Polymer Science. 10,
51
52
221-233 (2000).
S

53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
24
Page 25 of 27 Submitted to Macromolecules

(5) J. Kress, J.A. Osborn, K.J. Ivin, J.J. Rooney, In: M. Fontaille and A. Guyot editors. Recent
1
2
3 advances in mechanistic and synthetic aspects of polymerization. Kluwer Academic, Norwell,
4
5 1987 p. 363.
6
7
Fo
8 (6) Z. Komiya and R.R. Schrock. Macromolecules. 26, 1387 (1993).
9
10
11
(7) Goodall, B. L. in Late Transition Metal Polymerization Catalysis; Rieger, B. Ed.; Wiley-VCH:
rR
12
13
14 Weinheim, Germany, 2003; p 114.
15
16 (8) Hoskins, T.; Chung, W.J.; Agrawal, A.; Ludovice, P.J.; Henderson, C.L.; Seger, L.D.; Rhodes,
ev
17
18
19 L.F.; Shick, R.A. “Bis(trifluoromethyl)carbinol-Substituted Polynorbornenes: Dissolution
ie
20
21 Behavior,” Macromolecules, 37(12), 4512-4518 (2004).
22
w.

23
24 (9) N.R. Grove, Q. Zhao, P.A. Kohl, S.A. Bidstrup-Allen, R.A. Shick, B.L. Goodall, L.H. McIntosh
25
26
and S. Jayaraman. Advancing Microelectronics. 23, 16 (1996).
Co

27
28
29
30 (10) Haselwander, T.F.; Heitz, W.; Krügel, S.A. and Wendorff, J.H. Macromolecules 1997, 30,
nf

31
32 5345-5351.
33
id

34
35 (11) T.F. Haselwander, W. Heitz, S.A. Krügel and J.H. Wendorff. Macromol. Chem. Phys. 1996,
36
en

37
38 197, 3435.
39
40
tia

41 (12) Arndt, M.; Engehausen, R.; Kaminsky, W.; Zoumis, K. Journal of Molecular Catalysis A:
42
43 Chemical 101, 171 (1995).
44
l-

45
46
47 (13) W. Kaminsky and A. Noll. Polymer Bulletin. 31, 175 (1993).
48
AC

49
50
51
52 (14) Liu, Y.; Bo, S. “Hydrodynamic Radius Characterization of a Vinyl-Type Polynorbornene by
S

53
54 Size-Exclusion Chromatography with Static and Dynamic Laser Light Scattering Detector,”
55
56
57 Chromatographia, 59, 299-303 (2004).
58
59
60 (15) Ludovice, P. Ahmed, S.; Van Order, J.; Jenkins, J. Macromol. Symp. 146, 235-244 (1999).
ACS Paragon Plus Environment
25
Submitted to Macromolecules Page 26 of 27

(16) B.R. Brooks, R.E. Bruccoleri, B.D. Olafson, D.J. States, S. Swaminathan, and M. Karplus.
1
2
3 Journal of Computational Chemistry. 4, 187-217 (1983).
4
5
6 (17) S.L. Mayo, B.D. Olafson and W.A. Goddard, J. Phys. Chem. 32, 8897 (1990).
7
Fo
8
9 (18) M.J.S. Dewar and W. Thiel. Journal of the American Chemical Society. 99, 4899 (1977).
10
11
rR
12 (19) N. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A.H. Teller, and E. Teller. Journal of
13
14
15 Chemical Physics. 21, 1087 (1953).
16
ev
17
18 (20) N. Madras and A.D. Sokal. Journal of Statistical Physics. 50, 109 (1988).
19
ie
20
21 (21) Vacatello M, Yoon DY. Macromolecules. 25, 2502 (1992).
22
w.

23
24
(22) Baschnagel J, Qin K, Paul W, Binder K. Macromolecules. 25, 3117 (1992).
25
26
Co

27
28 (23) Freire JJ, Rey A, Bishop M, Clarke JHR. Macromolecules. 24, 6494 (1991).
29
30
(24) R. Fletcher and C.M. Reeves. Computational Journal. 7, 149 (1964).
nf

31
32
33
(25) J.E. Dennis Jr. and R.B. Schnabel. Numerical Methods for Unconstrained Optimization and
id

34
35
36
Nonlinear Equations. Prentice-Hall, Englewood Cliffs, New Jersey 1983.
en

37
38
39
40 (26) Mattice, W. L.; Suter, U. W. Conformational Theory of Large Molecules: The Rotational
tia

41
42 Isomeric State Model in Macromolecular Systems; Wiley: New York, 1994.
43
44
l-

45 (27) Flory, P. J. Statistical Mechanics of Chain Molecules; Oxford University Press: Oxford, 1989.
46
47
48
AC

49
(28) Theodorou DN, Suter UW. Macromolecules. 18, 1467 (1985).
50
51
52 (29) Ludovice PJ, Suter UW. from Computational modeling of polymers, New York: Marcel
S

53
54 Dekker, (J. Bicerano, editor)1992.
55
56
57 (30) Hutnik M, Gentile FT, Ludovice PJ, Argon AS, Suter UW. Macromolecues. 24, 5962 (1991).
58
59
60
ACS Paragon Plus Environment
26
Page 27 of 27 Submitted to Macromolecules

(31) Smith, G. D., Ludovice, P. J.; Jaffe, R. L.; Yoon, D. Y. J. Phys. Chem. 1995, 99, 164.
1
2
3
4 (32) Ludovice, P. J.; Davidson, M.; Suter, U. W. in Simulations of Macromolecular Systems, in
5
6 Supercomputer Research in Chemistry and Chemical Engineering, Jensen, K. F.; Truhlar, D. G.,
7
Fo
8 Eds.; American Chemical Society: Washington D. C., 1987.
9
10
11
(33) Ludovice, P. J.; Suter, U. W. Polym. Prepr. A. Chem. Soc. Div. Polym. Chem. 1987, 28, 295.
rR
12
13
14
15 (34) Flory, P. J. Macromolecules 1974, 7, 381.
16
ev
17
18 (35) M. Rehahn, W. L. Mattice and U. W. Suter, Advances in Polymer Science 131-132, 1 (1997).
19
ie
20
21
22
w.

23
24
25
26
Co

27
28
29
30
nf

31
32
33
id

34
35
36
en

37
38
39
40
tia

41
42
43
44
l-

45
46
47
48
AC

49
50
51
52
S

53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
27

Você também pode gostar