Citation
Evolution of material properties during the solvent-assisted recycling of covalent adaptable network polymers

Material Information

Title:
Evolution of material properties during the solvent-assisted recycling of covalent adaptable network polymers
Creator:
Soule, Drake Ashton
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Master's ( Master of science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Mechanical Engineering, CU Denver
Degree Disciplines:
Mechanical engineering
Committee Chair:
Welch, Sam
Committee Members:
Carpenter, Dana
Yakacki, Christopher
Yu, Kai

Notes

Abstract:
Research in methods to recycle polymers has recently gained momentum, driven by economic concerns and pressing demand. This study explored an epoxy polymer with covalent adaptable network and its recyclability by decomposing it at high temperatures in an ethylene glycol solvent using a closed environment. This method creates a decomposed polymer solution and once decomposed, this epoxy can again be repolymerized by heating in an open environment and allowing the ethylene glycol to volatilize, leaving only the repolymerized epoxy. By using mass measurements of the decomposed epoxy solutions throughout the repolymerization process, the evaporation rates of ethylene glycol solvent were achieved over a range of temperatures and heating times. The evolution of solution viscosity was also characterized using rheometer. A dynamic mechanical analysis was performed on thin films of repolymerized epoxy to determine the storage modulus, glass transition temperature, and rubbery modulus of the material at a range of temperatures and heating times. Bulk epoxy samples were produced, where the material was allowed to repolymerize to make samples significantly thicker than the thin samples (thicknesses of approximately 8 millimeters for the bulk sample and 0.7 millimeters for the thin sample). This bulk sample was cut into slices along its thickness and the slices were subjected to the same dynamic mechanical analysis procedure as the thin samples to identify how the storage modulus, glass transition temperature, and glassy modulus behaved at the different layers of the bulk sample. A multiscale constitutive model was established to study the repolymerization process.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Drake Ashton Soule. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item has the following downloads:


Full Text
EVOLUTION OF MATERIAL PROPERTIES DURING THE SOLVENT-ASSISTED
RECYCLING
OF COVALENT ADAPTABLE NETWORK POLYMERS
by
DRAKE ASHTON SOULE B.S., University of Colorado Denver, 2015
A thesis submitted to the
Faculty of the Mechanical Engineering Department of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Mechanical Engineering Program
2019


©2019
DRAKE ASHTON SOULE
ALL RIGHTS RESERVED


This thesis for the Master of Science Mechanical Engineering degree by
Drake Ashton Soule has been approved for the Mechanical Engineering Program by
Sam Welch, Chair Dana Carpenter Christopher Yakacki Kai Yu, Advisor
Date: May 18, 2019


Soule, Drake Ashton (M.S. Mechanical Engineering Program)
Evolution of Material Properties During the Sol vent-Assisted Recycling of Covalent Adaptable Network Polymers
Thesis directed by Assistant Professor Kai Yu
ABSTRACT
Research in methods to recycle polymers has recently gained momentum, driven by economic concerns and pressing demand. This study explored an epoxy polymer with covalent adaptable network and its recyclability by decomposing it at high temperatures in an ethylene glycol solvent using a closed environment. This method creates a decomposed polymer solution and once decomposed, this epoxy can again be repolymerized by heating in an open environment and allowing the ethylene glycol to volatilize, leaving only the repolymerized epoxy. By using mass measurements of the decomposed epoxy solutions throughout the repolymerization process, the evaporation rates of ethylene glycol solvent were achieved over a range of temperatures and heating times. The evolution of solution viscosity was also characterized using rheometer. A dynamic mechanical analysis was performed on thin films of repolymerized epoxy to determine the storage modulus, glass transition temperature, and rubbery modulus of the material at a range of temperatures and heating times. Bulk epoxy samples were produced, where the material was allowed to repolymerize to make samples significantly thicker than the thin samples (thicknesses of approximately 8 millimeters for the bulk sample and 0.7 millimeters for the thin sample). This bulk sample was cut into slices along its thickness and the slices were subjected to the same dynamic mechanical analysis procedure as the thin samples to identify how the storage modulus, glass transition temperature, and glassy modulus behaved at the different
IV


layers of the bulk sample. A multiscale constitutive model was established to study the repolymerization process.
The form and content of this abstract are approved. I recommend its publication.
Approved: Kai Yu
v


TABLE OF CONTENTS
CHAPTER I INTRODUCTION......................................................1
CHAPTER II MATERIALS AND EXPERIMENTS...........................................4
2.1 Material Synthesis....................................................4
2.2 Decomposing and Repolymerizing the Networks..............................5
2.4 Solvent Evaporation Rate Measurement..................................6
2.3 Viscosity Measurement.................................................7
2.6 Dynamic Mechanical Analysis on Bulk Sample...............................7
CHAPTER III A THERMO VISCOELASTIC CONSTITUTIVE MODEL...........................8
3.1 Degree of Conversion Ratio............................................9
3.2 The Evolution of Equilibrium Modulus.....................................9
3.3 Evolution of Relaxation Time.........................................10
3.4 The Evolution of Network Tg..........................................10
CHAPTER IV RESULTS AND DISCUSSIONS.........................................11
4.1 Material Parameters Identification...................................11
4.5 Results of Solvent Evaporation Rate..................................16
4.4 Prediction of the Evolution of Viscosity During the CAN Repolymerization.17
4.3 Prediction of the Glass Transition Behaviors During the CAN Repolymerization ....18
4.7 Results of Dynamic Mechanical Analysis on Bulk Sample....................27
vi


CHAPTER V CONCLUSION
40
BIBLIOGRAPHY
41
Vll


LIST OF FIGURES
Figure 1: The repolymerization of thermosets................................................2
Figure 2: Detailed structure information....................................................5
Figure 3: Decomposition and repolymerization of epoxy with fatty acid linker................6
Figure 4: Bulk sample layers and direction of spatial network structure.....................7
Figure 5: The one-dimensional rheological model for the viscoelastic mechanical properties..8
Figure 6: Normalized mass increment of non-catalyst epoxy samples with different diffusion temperatures...............................................................................12
Figure 7: Stress relaxation behavior.......................................................13
Figure 8: Time temperature super position of the epoxy.....................................14
Figure 9: Master curve fitting with 11 viscoelastic branches added into the model...........15
Figure 10: Concentration of ethylene glycol solvent over time in the decomposed polymer solution...................................................................................17
Figure 11: Predictions of the viscosity during the CAN repolymerization before gelation....18
Figure 12: Prediction of the storage modulus and tand curve................................19
Figure 13: Predictions on the tand and equilibrium modulus curve for different repolymerization times at 160 °C............................................................................21
Figure 14: The evolution of storage modulus and tan delta over time at 180 °C..............24
Figure 15: The evolution of storage modulus and tan delta over time at 200 °C..............26
Figure 16: Glass transition temperature and rubbery modulus for 160 °C, 180 °C, and 200 °C cures of thin film.........................................................................27
Figure 17: Storage modulus and tan delta for Bulk Sample A, 15 hour cure at 200 °C.........29
Figure 18: Rubbery modulus and glass transition temperature for Bulk Sample A..............30
Figure 19: Storage modulus and tan delta for Bulk Sample B, 18 hour cure at 200 °C.........31
viii


Figure 20: Storage modulus and tan delta for Bulk Sample B, 21 hour cure at 200 °C.......33
Figure 21: Rubbery modulus and glass transition temperature for Bulk Sample B............34
Figure 22: Storage modulus and tan delta for Bulk Sample C, 25 hour cure at 180 °C.......36
Figure 23: Storage modulus and tan delta for Bulk Sample C, 30 hour cure at 180 °C.......38
Figure 24: Rubbery modulus and glass transition temperature for Bulk Sample C............39
IX


LIST OF TABLES
Table 1: Material Parameters..................................................................15
Table 2: Material properties for 17-21 hour cures............................................21
Table 3: Material properties for 9-13 hour cures.............................................22
Table 4: Material properties for 3-7 hour cures..............................................24
Table 5: Material properties for Sample A, layers 1 and 2, 15 hour cure......................28
Table 6: Material properties for Sample B, layers 1-4, 18 hour cure...........30
Table 7: Material properties for Sample B, layers 1-4, 21 hour cure...........32
Table 8: Material properties for Sample C, layers 1-4, 25 hour cure...........35
Table 9: Material properties for Sample C, layers 1-4, 30 hour cure...........37
x


CHAPTER I INTRODUCTION
Conventional thermoset polymers cannot be reshaped or recycled once hardened because of their cross-linked networks, which cause a large amount of plastic waste that ends up in landfill [1-6], Recently, with the advent of covalent adaptable networks (CANs), the recycling of covalently crosslinked thermoset materials has become possible [7-11], Their reversible bonds allow the polymer network to be continuously rearranged through the bond exchange reaction (BER), thus giving thermoset polymers the properties such as surface welding, self-healing and malleability [12, 13],
CANs can be completely decomposed in suitable solvents through BERs between polymer network and solvent molecules. Excitingly, repolymerization can occur by heating the decomposed polymer solution in an open environment. The repolymerization process is shown in Figure 1. When heated in an open environment, BERs can occur between the reversible groups at each end of each segment in the polymer solution, thereby connecting the different polymer chains. As the solvent molecules evaporate, the polymer network is gradually formed. The repolymerized material has almost the same network structure and thermomechanical properties as the fresh sample; for example, the elastic modulus and strength of the recycled materials are about 98.3% and 93.3%, respectively, of the fresh ones [14, 15],
1


Figure 1: The repolymerization of thermosets can be realized via the transesterification type BERs.
The repolymerization of thermoset polymers offers a green and sustainable recycle method for thermoset materials and composites which simply involves heating the polymer solution in an open environment. However, since its existing research is limited to conceptual arguments, no systematic experimental or theoretical work has been proposed to reveal the relationship between polymer repolymerization and network dynamics or to examine the spatial network structure and mechanical properties of the repolymerizing network which limits its engineering applications.
Repolymerization is a complex physical process: The polymer solution undergoes a liquid-to-solid phase transition; the polymer chain length and crosslinking density evolve with the conversion ratio; repolymerization of decomposed polymer solution starts from the surface and involves non-uniform solvent concentration and structure developed in the thickness direction. These changes at the microscopic scale during repolymerization affect their macroscopic properties of the materials, such as Young’s Modulus, glass transition temperature and mechanical properties. Therefore, it is critical to study the evolution of material properties to achieve better performance control. The degree of conversion (DoC) is commonly used to
2


characterize and simulate the evolution of polymer systems during curing of thermoset epoxy resins [16-18], As the repolymerization proceeds, the monomers gradually grow into polymer chains. In this process, the number distribution of polymer chains with different chain lengths is described by DoC, and the material properties can be calculated by a superposition method, and then the repolymerization process is associated with the evolution of material properties [19-21], The effect of temperature on the time scale of mechanical behavior is described by the time-temperature superposition principle to study the viscoelastic behavior of thermosetting polymer during repolymerization processes [22], In addition, the glass transition temperature (Tg) of the curing system increases as the repolymerization proceeds. Non-linear viscoelastic behavior may be exhibited when the cured material enters the glassy state [19, 22], so there is a need for a model that can describe the stress-strain behavior of a cured glassy polymer upon large deformations.
In this thesis, we propose a multi-scale modeling framework combining the macro-scale mechanical properties with micro-scale spatial network structure to describe the evolution of material property during the repolymerization process. However, there are some unique polymerization phenomena for the repolymerization of CANs. First, the reversible groups at the end of the segment in the polymer solution are combined through BERs to form a polymer chain and its time scale depends on the kinetics of chemical reaction. Second, the repolymerization of decomposed polymer solution always starts from the surface as the solvent evaporates. A higher temperature usually leads to the quick formation of a stiff film on the surface of the solution, which will suppress the evaporation of the ethylene glycol (EG), and consequently inhibit the repolymerization of epoxy beneath the surface. But, reducing the temperature will lead to a lower repolymerization rate and a longer repolymerization time, which means that the optimum
3


repolymerization temperature of the epoxy needs to be found. And the most challenging task is understanding how to describe the length evolution of chain segments during the repolymerization. In this thesis, a micro-scale model is first established based on the kinetics of BERs to describe the develop of length of chain segments during the repolymerization, and consequently determine the evolution of DoC. The DoC is then used as an internal variable to characterize the reaction system. The time-temperature-DoC superposition is used to capture the thermomechanical behaviors of the cured polymers. The model shows good prediction on the conversion rate of functional groups, mechanical properties, and repolymerization rate of polymer network. It also helps to reveal the influence of different material and processing variables, such as time and temperature.
CHAPTER II MATERIALS AND EXPERIMENTS
2.1 Material Synthesis
The material used in this study was the epoxy thermosetting polymer developed by Leibler and coworkers [16], The epoxy samples were synthesized from a catalyst (metal catalyst Zn(Ac)2 (Sigma Aldrich, St. Louis, MO)), monomers diglycidyl ether of bisphenol A (DGEBA, Sigma Aldrich), and crosslinkers (fatty acids Pripol 1040 (Croda, Houston, TX)). Detailed chemical structures of each reagent are shown in Figure 2. For polymer synthesis, the catalyst was first mixed with a fatty acid (10 mol% to COOH groups) in a beaker. The temperature was gradually increased from 100 °C to 150 °C while maintaining the mixture under vacuum until no gas evaporation was observed and the catalyst particles were completely decomposed (3 hours). DGEBA was then added to the previous fatty acid mixture containing solubilized catalyst (stoichiometry between COOH and epoxy is 1:1) in a round-bottom flask and manually stirred at
4


130 °C until the mixture became homogeneous. Finally, the mixture was poured into a mold and placed in an oven for 6 hours at 130 °C.
ch3 ch3
Zn2+xH20
rv^o-
o
o^ 0
DGEBA
Zn(Ac)2
(CHjVCOOH
hoc OOH
.(CHj^-COOH
,(CH2)7-COOH
y c8h17
C6H13
Fatty acids(23 wt% dimers)
T T c8H-i7
CfjH13 C6H13
Fatty acids(77 wt% dimers)
Figure 2: Detailed structure information about the chemical reagents used in this study.
2.2 Decomposing and Repolymerizing the Networks
For the decomposition, epoxy samples were immersed in ethylene glycol solvent (volume
the network and participate in the transesterification type BERs with polymer chains, which effectively break the long chains into short segments at the position of ester groups on the backbone. The container was sealed to avoid solvent evaporation. After being heated in 180 °C for 6 hours, the polymer was fully decomposed. For the repolymerization, the decomposed polymer solution was poured into a mold and heated at a given temperature (160 °C, 180 °C, and 200 °C) in an open environment. The BER proceeded stepwise between short segments by combining their functional groups to form a repolymerized polymer network. The schematic view of the decomposition and repolymerization of epoxy with fatty acid linker was shown in Figure 3.
ratio between polymer and solvent was 1:2). At high temperature, solvent molecules diffuse into
5


Figure 3: Decomposition and repolymerization of epoxy with fatty acid linker.
2.4 Solvent Evaporation Rate Measurement
Masses for epoxy samples and ethylene glycol were measured before the samples were decomposed in the ethylene glycol solution to create a decomposed polymer solution using the same method discussed in section 2.2. Once the epoxy mixture was completely decomposed it was poured into a mold with a known mass. The mass of the mold containing the decomposed polymer solution was measured and the known mass of the empty mold was subtracted from the total mass, leaving the mass of the decomposed polymer solution. Assuming a uniform ratio of epoxy to ethylene glycol solution, the mass of the decomposed polymer solution was used to calculate the masses of the epoxy sample and the ethylene glycol solution. Once these masses were calculated, the mold containing the decomposed polymer solution was again heated at a given temperature (160 °C, 180 °C, and 200 °C) in an open environment. The mold allowed the ethylene glycol to evaporate from the decomposed polymer solution and the mass of the mold was measured and recorded after each hour it was exposed to the heated environment. The difference in mass after each hour was due to the evaporating ethylene glycol and after each interval the new percent by mass of ethylene glycol in the decomposed polymer solution was calculated as the solution repolymerized.
6


2.3 Viscosity Measurement
In the repolymerization, the resin viscosity was measured before gelation to identify the evolution of chain segment length. After being heated for different lengths of time, the viscosity was measured by using a rheometer (TA Instruments, AR-G2, New Castle, DE, USA) with parallel plate geometry (20-mm diameter, 1,000-pm gap). During the measurements, temperature was maintained constant at room temperature while the shear rate was increased from 10'3 to 200 s'1.
2.6 Dynamic Mechanical Analysis on Bulk Sample
A silicone mold was used to cure the polymer solution into a bulk sample which was placed in an oven to cure at temperatures of 180 and 200 °C at different heating times. Once cured, the sample was sliced in layers perpendicular to the bulk sample’s thickness, as shown in Figure 4. The same Dynamic Mechanical Analysis (DMA, Model Q800, TA Instruments, New Castle, DE, USA) process described in section 2.4 was then conducted on these layers to determine the spatial network structure and mechanical properties of bulk thermoset during the repolymerization.
Layer 1 Layer 2 Layer 3 Layer 4
Figure 4: Balk sample layers and direction of spatial network structure.
Increasing
network
crosslinking
density.
7


CHAPTER III A THERMO VISCOELASTIC CONSTITUTIVE MODEL
A multi-branch constitutive model is applied to study the viscoelastic behaviors of repolymerized epoxy CANs. Figure 5 shows the one-dimensional rheological analogy of this model consisting of two parts: the equilibrium branch and the N non-equilibrium branches that describe the relaxation of viscoelastic polymer chains. As the number of crosslinks increases, the number of balanced branches increases, which can be described by a phase evolution model [23], The newly added cross-linking will prevent the old cross-linking from relaxing at the rate of the old relaxation rate [21]; therefore, according to the time-temperature-DoC stack, the relaxation time will shift to the new relaxation time. Following Figure 5, The total stress oeq can be calculated by summing the stresses in the equilibrium branch and in the non-equilibrium branches.
Figure 5: The one-dimensional rheological model for the viscoelastic mechanical properties modeling. Red and blue parts of these branches respectively describe the equilibrium branches and viscoelastic branches.
The true stress o of the model is the summation of each paralleled components.
er
u.
eq
(1)
where oeq and ovj are respectively the true stresses of the equilibrium branch and viscoelastic branches. Each branch has the same stretch, which equals to the total stretch of the multi-branched model X.
8


For viscoelastic branches, the stress can be calculated by:
Evj In(Aevj) = E
(2)
where Evj and Xvje are respectively the elastic modulus and stretch ratio of the spring, ib, and Xvjv are respectively the relaxation time and stretch ratio of the dashpot in the jth viscoelastic branch.
The evolution of network structure and thermomechanical properties are commonly linked to the degree of conversion (DoC) ratio of functional groups [24, 25], During the repolymerization, each chain connection reaction generates a solvent molecule that evaporates out of the system. The chain segment is increased correspondingly. In this thesis, we take conversion ratio degree to be linear with the normalized solvent evaporation rate, namely:
where y(t) is the solvent evaporation during the re-polymerization, and y0 is the initial solvent content. Thus, the solvent evaporation rate and degree of conversion ratio can be experimentally measured.
The dependence of the viscoelastic properties of thermosetting systems on the degree of cure has been widely investigated, both in terms of experimental results and modelling approach The rubbery modulus during the re-polymerization is scaled to the correlation length of chain segments fa by E0~T/Zq[26]. fa is further scaled to the DoC as Z0~(l — p2)~av[21, 28], with av being a correlation exponent. Therefore, the evolution of rubbery modulus can be written as:
3.1 Degree of Conversion Ratio
p(t) = 1 - y(t)/y0,
(3)
3.2 The Evolution of Equilibrium Modulus
E0(t) = EOD[l-p2(t)]3a».
(4)
9


3.3 Evolution of Relaxation Time
The relaxation time of each Maxwell element is a function of temperature. The time-temperature relaxation at temperature T in the viscoelastic branches also follow the time-temperature superposition principle (TTSP), which can be represented by [29]:
Tvj(T) = rvjoav(T) (5)
where tvjo is the relaxation time of the reference sample at the reference temperature Tg. Here we choose the polymer cured at XX hours as the reference state; av(T) is the temperature shift factor which can be described by WLF equation (6a) above the reference temperature Tg and by Arrhenius equation (6b) below the Tg [30, 3T|.
In [av(T)\ = -
AFc_
kjj
T + 273.15
1
Tg + 273.15
atT (6a)
logK(r)] = - I9} atT >T9 (6b)
L2 "t 1 ig
where A and T'c are material specific constants. Ci and C2 are fitting parameters. A is a reference temperature with av(Ts)= 1. It can be taken to be the crossing point of two curves represented on a a vs T plot.
3.4 The Evolution of Network Tg
The glass transition temperature is an important indication of the extent of curing and of the viscous property. As the curing progresses, the small monomers grow into large molecules and are connected to each other. The entire system has widely distributed polymer chains of different molecular weights. As mentioned above, the macroscopic properties measured in the experiments are the average behavior of these chains and the DoC is a suitable evaluation parameter to characterize the average properties of the overall system. The relationship between the Tg and the DoC has been widely studied. Gan et al. [19] developed a viscoelastic type model
10


for the curing polymer by using the empirical Dibenedetto equation [32, 33.]. The model considers the effect of cross-linking on the mobility of the curing system and can successfully capture the Tg evolution of various curing systems. Here, we use this viscoelastic model to consider the Tg changes during the curing:
Er
Ta Rlnig^l -pY +g2] ^
where Er is the activation energy of transition from the glassy to the rubbery state, R is the gas constant, £ is the parameter for accounting the effects of chain entanglement, gi, g2 are two material constants.
CHAPTER IV RESULTS AND DISCUSSIONS
4.1 Material Parameters Identification
The diffusivity of EG molecules in epoxy samples were determined by diffusion data of non-catalyst epoxy and the transportation of EG solvent can be described by using Fick's laws of diffusion [34]:
du d2u
aF = 0(r)^
(8)
where D(T) is the molecular diffusivity and its temperature dependency follows the Arrhenius equation:
En
D(T) = D0exp
(9)
where Do is the reference diffusivity, Eas is the activation energy, R is the gas constant R =
8.314 J moE1 K f The Eq. (8) can be solved numerically by using the finite difference method, where the time and space derivatives are respectively calculated using forward difference and central difference. The normalized mass increment of non-catalyst epoxy samples at different
11


temperature is shown in Figure 6. When the diffusion coefficient Eas = 34.2kJ / mol and the pre-
factor Do = 3.23 x 10'6m2 /s, the model predictions agree well with the experimental data.
0.04
■ experiment 140°C • experiment 160°C a experiment 180°C
0.03---------Prediction 140°C
-----Prediction 160°C
-----Prer1'" ~~
(/>
CD
0.01 -
0.00
-f---'-----T----T----,---T-----,----,----n-
0 50 100 150 200
-50
Time(min)
Figure 6: Normalized mass increment of non-catalyst epoxy samples with different diffusion temperatures. Note: experimental results are plotted in dots and solid lines plot the model predictions.
We can use the method developed by Yu et al. to determine the BER energy barrier Eab by performing a series of stress relaxation tests on the fresh epoxy sample at different temperatures. During the stress relaxation tests, the normalized relaxation modulus Er can be characterized by the following exponential function:
where the relaxation time r can be measured from the experimental relaxation curves when normalized relaxation modulus declines to 1/e (-36.8%). The relationship between relaxation time and the BER energy barrier is:
where the ko is a correlation exponent and the coefficient determine the energy barrier Fah. Figure 7a shows the stress relaxation curves of fresh epoxy CANs at different temperatures. The dots in Figure 7b show the measured relaxation times as a function of temperature. By fitting the
(10)
(11)
12


relaxation time, the energy barrier for the fresh epoxy sample is determined to be Eab=8.15KJ/mol.
â– o
O
c
o
-*—•
TO
X
03
O'
~o
CD
N
TO
E
(a)
CO
CL
50 n 40-30-20-
u o
c.
o
'ro 10-
x
to
o-
----Fit
O Exp
Ea(BER)=8.15KJ/mol
"0 - - - _
O--
120 140 160 180 200
Temperature(°C)
220
(b)
Figure 7: (a) Stress relaxation behavior of fresh epoxy CAN. The relaxation time is determined to be the timing point when the relaxation modulus is reduced 63.2% (the dash line), (b) The relationship between relaxation time and temperature.
The relaxation curves which represent the contribution of viscoelastic branches are plotted in Figure 8a. Based on the TTSP, each curve is shifted to the reference temperature Ts (17 °C) to form a master relaxation curve (Figure 8b). The corresponding shifting factors are plotted in Figure 8c as a function of temperature. At temperatures above and below Ts, the shift factors are respectively captured by the WLF and Arrhenius equations (shown in 3.3). The corresponding parameters were determined to be AFc/kb=-40070, Cl=8.658, and C2=15.76.
13


(a) (b)
(c)
Figure 8: (a) The relaxation moduli of viscoelastic branches; (b) Master curve after shifting each temperature to reference temperature (17 °C). (c) Shift factor in each temperature. Red circles are the experimental data and the black solid line is the predictions of Arrhenius and WLF equation.
The capability of the applied multi-branch model in capturing the complicated stress relaxation process originates from the additionally introduced nonequilibrium branches. By adding nonequilibrium branches gradually, the simulated curve is gradually approaching the master curve. As shown in Figure 9, when adding 11 nonequilibrium branches, the model prediction curve rejoins with the experimental relaxation master curve. The corresponding Ebi and Tbi values are listed in Table 1.
14


Figure 9: Master curve fitting with 11 viscoelastic branches added into the model. Red circle line is model prediction and the black solid line is the experimental master curve.
Table 1: Material Parameter.
Description Symbol Value
Known constants Gas constant R 8.31J moE1 K 1
Boltzmann constant kb 1.38 x 10~23 m2 kg s~2 K 1
Avogadro constant Na 6.02xl023
Directly measured parameters Energy barrier for diffusivity Eas 34.2kJ / mol
Energy barrier for BER Fab 8.15kJ / mol
Evaporation rate of EG at 160 5.122 x 10"5mol/m2/s
Evaporation rate of EG at 180 7.488 x 10"5mol/m2/s
Evaporation rate of EG at 200 9.750 x 10"5mol/m2/s
Calculated parameters Number of viscoelastic branches N 11
Parameter of Arrhenius equation AFc/kb -40070
Elastic moduli in each viscoelastic Evn 171.591,180,210,160,30,11,5,1
branche .1,1,0.801,1.099
Relaxation times at Ts in each Tvn 1 x 10"7,1 xlO"6,1.5 xlO"5,6x10"
viscoelastic branche 5,6xl0"4,3xl0"3,2.5x10"
Parameter of WLF equation Ci 2,2.5xlO"1,0,2.5,25,600 MPa 8.658
Parameter of WLF equation C2 15.76
15


4.5 Results of Solvent Evaporation Rate
The mass of epoxy and ethylene glycol in the decomposed polymer solution was calculated based on the ratio of epoxy and ethylene glycol masses in the mixture. For the sample cured at 160 °C, 24.46g of epoxy and 47.01g of ethylene glycol was used for a total mixture mass of 71.47g. 34.2% of the decomposed polymer solution’s mass was epoxy while the remaining 65.8% was the ethylene glycol. The mass of the decomposed polymer solution in the mold was measured to be 1.56g before it began exposure to heat. The epoxy, 0.53g of the decomposed polymer solution’s mass of 1.56g, was expected to remain constant during the evaporation process while the 1.05g of ethylene glycol decreased as it evaporated. The evaporation rate can be seen in Figure 10 which compares the concentration of ethylene glycol solvent over time between samples heated at 160 °C, 180 °C, and 200 °C.
The second sample was cured at 180 °C and had a decomposed polymer solution with a ratio of 15.Og of epoxy sample to 28.49g of ethylene glycol solution which combined for a total mass of 43.49g. The mass of the decomposed polymer solution in the mold was 2.07g. The concentration by mass of epoxy in the mold is 34.4% to make 0.71g while the remaining 65.6% of the mass was ethylene glycol for 1.35g. The evaporation rate of the 180 °C sample can be seen in Figure 10.
The decomposed polymer solution used to cure at 160 °C was also used to cure at 200 °C (the mixture was 24.46g of epoxy and 47.01g of ethylene glycol). The starting mass of the 200 °C sample was 1.53g, meaning 34.2% of the decomposed polymer solution’s mass in the mold was 0.52g of epoxy. The remaining 65.8% of the decomposed polymer solution’s mass in the mold was l.Olg of ethylene glycol. The evaporation rate of the 200 °C can be seen in Figure 10.
16


Time (h)
Figure 10: A plot of the concentration of ethylene glycol solvent over time in the decomposed polymer solution at 160 °C, 180 °C, and 200 °C.
The curve fitting tool in MATLAB was used to determine the fitted curves for the recorded data using the format:
y = y0e“ (12)
where y is the percent of ethylene glycol solvent by mass, y0 is some initial value of ethylene glycol solvent, t is the cure time, and r is the relaxation time of the system. These relaxation times were found to be 5.998 hours, 4.542 hours, and 2.361 hours for the 160 °C, 180 °C, and 200 °C cures respectively and it is at these times that the viscoelastic nature of the polymer pulls the system back into a solid state. The jump from 160 °C to 200 °C cuts this relaxation time by more than half, and is the first physical evidence showing that as curing temperatures increase so does the repolymerization rate, both driving the material back toward the mechanical properties it demonstrates before decomposition in the ethylene glycol solvent.
4.4 Prediction of the Evolution of Viscosity During the CAN Repolymerization
The evolution of viscosity during the CAN repolymerization is predicted by our model. The epoxy solution was subject to heating at different temperature (160 °C, 180 °C, 200 °C) for repolymerization. As the solvent volatilizes, the segments in the solution repolymerize by BER
17


and the solution viscosity is gradually increased which is shown in Figure 11. Changes in the viscosity can be taken to be an indicator of the repolymerization degree. It can be seen that our model fits the experimental results well.
T------'------1------'-------1-
0 3600 7200
Time(s)
Figure 11: Predictions of the viscosity during the CAN repolymerization before gelation at 160 °C 180 °C and 200 °C respectively
4.3 Prediction of the Glass Transition Behaviors During the CAN Repolymerization
The thin samples tested were created from the same decomposed polymer solutions that were used for the ethylene glycol solvent evaporation rate measurement in section 4.5. The thin samples were tested as soon as they were cured enough to remove them from the glass plate they were cured on without damaging the sample. If removed from the plate too soon, the samples would break or only portions of the sample would separate from the plate. The samples had to cure longer at lower temperatures to achieve enough stability to be removed from the curing plate. An initial DMA test was performed on a thin film sample of fresh epoxy that had not yet been decomposed and repolymerized. This test returned a tan delta (tand) of 1.176508, a glass transition temperature (Tg) of 27.52858 °C, and a rubbery modulus (Er, the value of the storage modulus when the sample reached 80 °C during the DMA test) of 4.350665 MPa.
18


The multi-branched model was used to predict the storage modulus and tanS values during the glass transition process by DMA testing on the samples. For the ID multi-branched model, the temperature dependent storage modulus Es, loss modulus Ei, and tand are respectively expressed as:
Es
Ei
Y Ebi(Q2Tli y EVjai2Tlj
2-t 1 + o)2tL 2—i 1 + m2r^-
i=1 Di j=1 Vj
m n
ZEfoitOT \ ' Ev j CO Tv j
1 + CO2 tL 2—i 1 + (JL)2T^j
i = 1 DL j = 1 VJ
(13a)
(13b)
(13c)
where co is the testing frequency (1Hz in the experiments). From Figure 12, it can be seen that the predictions of storage modulus and tand agree with experimental data ranging from -20 to 80 °C.
Figure 12: Prediction of the storage modulus and tanS curve. Red circles indicate the predicted behavior while the black solid lines are experimental data.
The equilibrium modulus and Tg of CANs for different repolymerization times at 160 °C were predicted by our model and presented in Figures 13a and 13b. The corresponding storage modulus and tand curve were predicted by the multi-branched model and presented in Figures 13c through 13g. The mechanical properties of the 160 °C cure can be seen in Table 2.
19


Storage Modulus(MPa)
°o °- So O
e;|9Q uei

Storage Modulus(MPa)
Oo on ou
enea uei
to
o
Storage Modulus(MPa)
o, o, o, o,
Tg
Storage Modulus(MPa)
o. Om Ou
Eeq(MPa)
310


(g)
Figure 13: Predictions on the (a) tcind and (b) equilibrium modulus curve for different repolymerization times at 160 °C; (c) - (g) Prediction on the storage modulus and tanS curve of CANs at 17, 18, 19, 20, 21 hours ’repolymerization respectively. Red circles are predicted behavior, black solid lines are experimental data.
Table 2: Material properties for 17-21 hour cures
17 Hour 18 Hour 19 Hour 20 Hour 21 Hour
Tan Delta 1.103934 0.9125098 0.9078816 0.8486876 0.8736528
Tg(°C) 15.98789 11.47278 10.60691 11.36923 14.86556
Embbery (MPa) 1.1855 0.7013098 0.8959832 0.9048311 1.065385
It can be seen that Tg and Er trend upward to both increase as the sample cures longer (with the exception of the 17 hour data which stands out as an anomaly for this sample), indicating that as the solvent volatizes the network density increases. The glass transition temperature at the 21 hour mark is at 54% of the fresh sample, meaning the crosslinking in this sample has not yet reached the density of the fresh sample. The rubbery modulus of the 21 hour cure stands at 24% of the strength of the fresh sample.
After 9 hours of curing at 180 °C the next epoxy sample was ready to be tested. This time the sample would be tested after 9, 10, 11, 12, and 13 hours of curing and the material properties recorded during the dynamic mechanical analysis can be seen in Table 3. The hour by hour
21


comparison of the tan delta and storage modulus recorded by the DMA machine can be seen in Figures 14a through 14e. A comparison of all recorded tan delta and all recorded storage modulus for the 180 °C cure can be seen in Figures 14f and 14g respectively.
Table 3: Material properties for 9-13 hour cures
9 Hour 10 Hour 11 Hour 12 Hour 13 Hour
Tan Delta 1.125106 1.058598 0.98586 1.002632 0.9333222
Tg (°C) 6.234282 13.68666 18.80338 18.61292 21.67086
Erubbery (MPa) 1.3743 3.5375 3.5799 3.8911 3.5925
Note that at 180 °C the mechanical properties of the repolymerized sample are much stronger than those of the 160 °C. This shows that the polymerization rate significantly increases as the temperature of the process increases. The glass transition temperature and rubbery modulus more than double from the 9 hour cure to the 10 hour cure. Recall that the earliest this sample could safely be removed from the curing plate was at the 9 hour mark, when the sample was still relatively weak. The Tg and Er values for this sample get much closer to the strength for the fresh sample at 79% and 83% respectively. This indicates the network properties are increased with higher temperatures and faster curing times.
22


Storage Modulus (MPa) Storage Modulus (MPa) Storage Modulus (MPa)
(C
-t-<
CD
Temperature (°C)
(e)
(f)
23
Tan Delta Tan Delta


(g)
Figure 14: The evolution of storage modulus and tan delta at each hour interval (a) - (e) and all tan delta (f) and all storage modulus (g) produced by dynamic mechanical analysis for the 180 °C cure.
The final thin film sample cured at 200 °C and was ready to be tested after 3 hours of curing. The additional dynamic mechanical analysis testing also took place after 4, 5, 6, and 7 hours of curing. The material properties recorded during these tests can be seen in Table 4. The evolution of storage modulus and tan delta on an hour by hour basis can be seen in Figures 15a though 15e. The comparison of all tan delta and all storage modulus recorded for each cure time can be seen in Figures 15f and 15g respectively.
Table 4: Material properties for 3-7 hour cures
3 Hour 4 Hour 5 Hour 6 Hour 7 Hour
Tan Delta 1.09131 0.996882 0.9243482 1.042304 0.9908801
Tg (°C) 14.33166 19.15065 30.60437 25.64553 33.00221
Embbery (MPa) 1.1537 1.748 2.3108 2.924 1.8215
The Tg for the sample cured at 200 °C ended up exceeding that of the fresh sample by
24


20% which continues to show the progressive nature of this mechanical property with respect to increased curing temperatures and cure rates. Peculiarly, the rubbery modulus didn’t continue to grow as it previously had with the other cure times. The strongest Er returned from this sample, 2.924 MPa after 6 hours, only reached 67% of the rubbery modulus for the fresh sample. In the previous samples the polymerization rate increased as the curing temperature increased, leading to the rubbery modulus of each sample to increase and approach the Er of the fresh sample. Perhaps there is a threshold for this relationship to be studied in more depth.
(a)
(b)
(c)
(d)
25


(e)
(f)
(g)
Figure 15: The evolution of storage modulus and tan delta at each hour interval (a) - (e) and all tan delta (f) and all storage modulus (g) produced by dynamic mechanical analysis for the 200 °C cure.
Figure 16 shows the relationship between Tg and Er by cure time for each of the thin film samples. Notice that the glass transition temperature tends to be higher for samples cured at higher temperature and these samples also take less time to cure. The average glass transition temperature for the 160, 180, and 200 °C cures is 12.86047, 15.80162, and 24.5469 °C respectively. This continues to show that the repolymerization rate increases with higher curing temperatures and during this process the mechanical property approaches that of the fresh sample. During the 200 °C cure the glass transition temperature approached and even surpassed
26


the 27.52858 °C of the fresh sample and if another cure at 220 °C were to take place, the average Tg of the sample may be at or higher than the fresh sample. The average rubbery modulus for the 160, 180, and 200 °C cures is 0.94997, 3.18664, and 1.99779 MPa respectively. From the 160 °C to the 180 °C samples the rubbery modulus approached the fresh sample’s Er of 4.35067 MPa before regressing at the final cure of 200 °C. Generally, the trend seen in this figure shows that Tg and Er both simultaneously increase or decrease, with the exception of a few instances where one will increase while the other will decrease. These instances are isolated and may be considered anomalies along with the regression in the average rubbery modulus from the 180 °C to 200 °C cure.
Cure Time (h)
Figure 16: Glass transition temperature and rubbery modulus for 160 °C, 180 °C, and 200 °C cures of thin film.
4.7 Results of Dynamic Mechanical Analysis on Bulk Sample
The dynamic mechanical analysis testing was done on three bulk samples, A, B, and C. Samples A and B were cured first at 200 °C before being tested, followed by sample C, which cured at 180 °C before testing. The goal of the bulk sample study is to show change in crosslinking density from the top layer (layer 1), where the crosslinking density is highest, to the bottom of the bulk sample (layer 4), where the crosslinking is least dense. This is accomplished by recording the
27


glass transition temperature and rubbery modulus of the epoxy at each layer.
Sample A was created by decomposing 24.52g of epoxy in ethylene glycol and reducing the decomposed polymer solution until the epoxy constituted 59% of the mixture’s mass (the mass of ethylene glycol in the solution was 17.13g) before the mixture was poured into the mold and cured for 15 hours at 200 °C. After 15 hours a piece of the bulk epoxy was removed from the mold but only two layers could be produced from the bulk sample because it was still too soft. Dynamic mechanical analysis testing was performed on the samples and the material properties from the two layers can be seen in Table 5. The tan delta and storage modulus for layer 1 and layer 2 can be seen in Figures 17a and 17b respectively. Both layers have their tan delta and storage modulus compared to each other’s in Figures 17c and 17d respectively.
Table 5: Material properties for Sample A, layers 1 and 2, 15 hour cure.
Layer 1 Layer 2
Tan Delta 1.03118 1.183961
Tg (°C) 25.10708 9.748312
Erubbery (MPa) 0.6183148 0.02265562
28


(a)
(b)
(c) (d)
Figure 17: Storage modulus and tan delta for layer 1 (a) and layer 2 (b). The tan delta compared for both layers (c) and the storage modulus comparedfor both layers (d). Produced by dynamic mechanical analysis for Sample A, layers 1 and 2, 15 hour cure at 200 °C.
While Sample A didn’t produce the desired number of layers, the goal of studying it was
still reached. It can be observed that the Tg of layer 1 is more than twice that of layer 2, while the
Er for layer 1 is 27 times greater than layer 2. This is the first physical evidence to demonstrate
increased crosslinking density in the system at the top layer. Figure 18 shows the rubbery modulus
and glass transition temperature by layer.
29


0.7
Layer
(a) (b)
Figure 18: (a) Rubbery modulus for layers 1 and 2 of Sample A. (b) Glass transition temperature for layers 1 and 2.
Sample B was created with 18.63g of epoxy which was decomposed in ethylene glycol solvent and then reduced until the decomposed polymer solution consisted of 59% decomposed epoxy by mass (the mass of the ethylene glycol in the solution was 12.88g). This time the sample was cured for 18 hours, again at 200 °C, before a portion of the bulk sample was removed and cut into 4 layers. The results of the dynamic mechanical analysis for Sample B after 18 hours of curing can be seen in Table 6. The tan delta and storage modulus behavior for each layer can be seen in Figures 19a through 19d. Figures 19e and 19f show comparisons of tan delta for all layers and storage modulus for all layers respectively.
Table 6: Material properties for Sample B, layers 1-4, 18 hour cure.
Layer 1 Layer 2 Layer 3 Layer 4
Tan Delta 1.185251 1.323853 1.007943 1.363466
Tg (°C) 34.22455 19.26777 21.34405 18.00127
Embbery (MPa) 0.3022709 0.06155862 0.07846831 0.01455358
30


Storage Modulus (MPa) Storage Modulus (MPa)
Temperature (°C)
(a)
(b)
(c) (d)
Temperature (°C)
(e)
(f)
Figure 19: The evolution of storage modulus and tan delta from layers 1 (a) though 4 (e) and all tan delta (e) and all storage modulus (f). Produced by dynamic mechanical analysis for Sample B, layers 1-4, 18 hour cure at 200 °C.
31


Sample B produced 4 more layers after curing an additional 3 hours for a total cure time of 21 hours. The DMA measurements from these layers can be seen in Table 7. Figures 20a though 20d indicate the evolving behavior for storage modulus and tan delta for each layer. Tan delta for all layers and the storage modulus for all layers can be seen in Figures 20e and 20f respectively.
Table 7: Material properties for Sample B, layers 1-4, 21 hour cure
Layer 1 Layer 2 Layer 3 Layer 4
Tan Delta 1.123647 1.242308 1.163087 1.197856
Tg (°C) 36.27581 20.208 20.80494 18.40949
Embbery (MPa) 2.023795 0.1287223 0.1154176 0.05926586
(a)
(b)
32
Tan Delta


(e) (f)
Figure 20: The evolution of storage modulus and tan delta from layers 1 (a) though 4 (e) and all tan delta (e) and all storage modulus (f). Produced by dynamic mechanical analysis for Sample B, layers 1-4, 21 hour cure at 200 °C.
The data from the 18 hour cure for Sample B also indicates that the goal of the bulk sample study was reached. With the exception of layer 3, the Tg and Er decreased with each layer. Layer 3 produced a 2 °C increase in Tg over layer 2 as well as a 17 kPa increase in Er. Layer 4 produced lower Tg and Er values than layer 2 which suggests layer 3 was an anomaly to the study. The glass transition temperature of layer 1 is twice that of layer 4 while layer 1 also has an Er almost 21 times larger than layer 4. This shows that the rubbery modulus decreases at a significantly higher rate than the glass transition temperature over the whole depth of the bulk
33


epoxy sample. The anomaly notwithstanding, Sample B continues to demonstrate increased crosslinking density at the higher layers. It should also be noted that as Tg decreases from layer to layer, so does Er, showing that these two mechanical properties for the epoxy are tied together. Figure 21 shows the Tg and Er behavior for the 18 hour cure and compares it to the 21 hour cure.
The 21 hour cure of Sample B goes further to illustrate the relationship between the layer of the bulk epoxy sample and the material’s crosslinking density, where the higher layers in this cure displayed Tg and Er values that were greater than the lower layers. With this cure the Tg and Er both decrease from layer to layer (with the exception of a 0.6 °C increase in Tg from layers 2 to 3 which is even less of an anomaly) and Tg decreases by half from layer 1 to layer 4. Er in layer 1 is 34 times greater than the rubbery modulus in layer 4. The decrease in Tg and Er from layer 1 to layer 4 is consistent with the findings in the 18 hour cure and continues to emphasize the difference in strength between the top of the bulk sample, where it’s directly exposed to the heating environment, and the lower layers due to crosslinking density. This behavior can be seen in Figure 21.
(a) (b)
Figure 21: (a) Rubbery modulus for Sample B at 18- and 21-hour cure times for all 4 layers, (b) Glass transition temperature for Sample B at 18- and 21-hour cure times for all 4 layers.
34


Sample B was also cured at 180 °C but without success at the ratio of 59% epoxy to 41% ethylene glycol solvent by mass. Only the portion of decomposed polymer solution exposed to the air at the top of the mold would cure at this temperature, creating a layer of repolymerized epoxy at the top of the mold while the decomposed epoxy solution under the cured portion remained a viscous liquid.
To successfully repolymerize the decomposed epoxy solution in a bulk sample at 180 °C, it was decided to increase the concentration of the epoxy in the mixture, thus creating Sample C. Sample C contained 16.97g of epoxy that was decomposed in the ethylene glycol solvent and then reducing the mixture until it contained 73.75% epoxy by mass (leaving the remaining 26.25% of the mixture to be 6.04g of solvent). This sample was first cured for 25 hours before a portion of the bulk sample could be removed and cut into 4 testable layers. The results of the dynamic mechanical analysis for Sample C after 25 hours of curing can be seen in Table 8. Figures 22a through 22d show the relationship between the storage modulus and tan delta in each of the four layers. The tan delta for all layers and storage modulus for all layers can be seen in Figures 22e and 22f.
Table 8: Material properties for Sample C, layers 1-4, 25 hour cure
Layer 1 Layer 2 Layer 3 Layer 4
Tan Delta 1.234807 1.179738 1.234899 1.27985
Tg (°C) 27.98791 25.42427 14.43008 9.267545
Erubebry (MPa) 1.002956 0.3832622 0.1780851 0.06788154
35


Tan Delta _ Storage Modulus (MPa) Storage Modulus (MPa)
Temperature (°C) Temperature (°C)
(e) (f)
Figure 22: The evolution of storage modulus and tan delta from layers 1 (a) though 4 (e) and all tan delta (e) and all storage modulus (f). Produced by dynamic mechanical analysis for Sample C, layers 1-4, 25 hour cure at 180 °C.
36
Tan Delta Tan Delta


To show the development in the behavior of the tan delta and storage modulus, Sample C was cured for another 5 hours for a total of 30 hours at 180 °C. Another portion of the bulk sample was removed and yielded 4 layers for dynamic mechanical analysis testing. The measurements recorded from the DMA can be seen in Table 9 while the relationship between the storage modulus and the tan delta for each layer can be seen in Figures 23a though 23d. Comparisons of all tan delta and all storage modulus can be seen in Figures 23e and 23f.
Table 9: Material properties for Sample C, layers 1-4, 30 hour cure
Layer 1 Layer 2 Layer 3 Layer 4
Tan Delta 1.212818 1.352075 1.448953 1.331817
Tg (°C) 28.32718 13.13534 10.54569 12.02929
Embbery (MPa) 1.30335 0.2503388 0.1186616 0.2800871
(a)
(b)
37


<3
0)
Q
c
,03
(c)
(d)
(e) (f)
Figure 23: The evolution of storage modulus and tan delta from layers 1 (a) though 4 (e) and all tan delta (e) and all storage modulus (f). Produced by dynamic mechanical analysis for Sample C, layers 1-4, 30 hour cure at 180 °C.
Sample C continued the trend of Tg and Er values decreasing from the top layer to the subsequent layers. The 25 hour cure showed layer 1 with a Tg 3 times higher than layer 4 while Er for layer 1 was almost 15 times larger than layer 4. The glass transition temperature showed a wider range for Sample C than Sample B while the rubbery modulus did not have a range as wide, suggesting that the decreased curing temperature of Sample C allowed a more even cure throughout the sample. This can be contributed to the top layer not repolymerizing before solvent in the lower layers could volatize from the overall bulk sample. In the curing of Sample B the top
38


layer repolymerized quick enough that remaining solvent in the lower layers was trapped. Since both Tg and Er decreased with each layer for the 25 hour cure of Sample C, this cure also shows that crosslinking density is greater at the top layers. Figure 24 shows the Tg and Er behavior for the 25 hour cure and compares it to the 30 hour cure for Sample C.
The 30 hour cure for Sample C returned another anomaly, this time in layer 4, where the Tg increased by 1.5 °C and Er more than doubled from the values of layer 3. Aside from layer 4, the other layers progressed as expected where the highest Tg and Er were recorded for layer 1 and these values decreased over each subsequent layer. The Tg is almost 3 times higher in layer 1 than layer 3 and Er is 11 times greater, showing that in the final cure for the bulk epoxy samples the crosslinking density is greater at the top of the sample, where it’s directly exposed to the heating environment. Both the 25 and 30 hour cures began with a lower Tg than either of the cures at 200 °C which also enforces the concept that higher curing temperatures create higher repolymerization rates, which was noted in the cures for the thin film sample.
(a) (b)
Figure 24: (a) Rubbery modulus for Sample C at 25 and 30 hour cure times for all 4 layers, (b) Glass transition temperature for Sample C at 25 and 30 hour cure times for all 4 layers.
39


CHAPTER V CONCLUSION
Solvent-assisted decomposition and repolymerization of covalent adaptable network polymers can be utilized to fully recycle thermosets. In this thesis, we investigated and predicted the kinetics and mechanics of repolymerization of CANs. To link the property changes with the chemical reaction, the degree of conversion of monomers was used as the internal variable to describe the curing system. The glass transition temperature and the viscosity evolution were linked with the degree of conversion of monomers. A multi-branch model was used to capture the viscoelastic properties of the re-cured polymer at different curing times. The re-cured polymer at different curing states were made by controlling the heating time. The degree of monomer conversion, viscosity, glass transition temperature, and dynamic mechanical properties were measured for different re-curing states. The results indicated that the model could capture these properties’ changes during curing. The findings in this thesis provide fundamental understanding of the repolymerization technique of CANs, which will promote its immediate engineering application in thermosets and composites recycling.
40


BIBLIOGRAPHY
1. Conroy, A., S. Halliwell, and T. Reynolds, Composite recycling in the construction industry. Composites Part A: Applied Science and Manufacturing, 2006. 37(8): p. 1216-1222.
2. Hartt, G.N. and D.P. Carey, Economics of recycling thermosets. 1992, SAE Technical Paper.
3. Kouparitsas, C., et al., Recycling of the fibrous fraction of reinforced thermoset composites. Polymer composites, 2002. 23(4): p. 682-689.
4. Ogi, K., et al., Mechanical properties ofABS resin reinforced with recycled CFRP. Advanced Composite Materials, 2007. 16(2): p. 181-194.
5. Palmer, J., et al., Successful closed-loop recycling of thermoset composites. Composites Part A: Applied Science and Manufacturing, 2009. 40(4): p. 490-498.
6. Pickering, S.J., Recycling technologies for thermoset composite materials—current status. Composites Part A: applied science and manufacturing, 2006. 37(8): p. 1206-1215.
7. Adzima, B. J., et al., Rheological and chemical analysis of reverse gelation in a covalently cross-linked Diels- Alder polymer network. Macromolecules, 2008. 41(23): p. 9112-9117.
8. Bowman, C.N. and C. J. Kloxin, Covalent adaptable networks: reversible bond structures incorporated in polymer networks. Angewandte Chemie International Edition, 2012. 51(18): p. 4272-4274.
9. Kloxin, C. J., et al., Covalent adaptable networks (CANs): a unique paradigm in cross-linked polymers. Macromolecules, 2010. 43(6): p. 2643-2653.
10. Tasdelen, M.A., Diels-Alder “click” reactions: recent applications in polymer and material science. Polymer Chemistry, 2011. 2(10): p. 2133-2145.
11. Wojtecki, R.J., M.A. Meador, and S.J. Rowan, Using the dynamic bond to access macroscopically responsive structurally dynamic polymers. Nature materials, 2011.
10(1): p. 14.
12. Chen, X., et al., New thermally remendable highly cross-linked polymeric materials. Macromolecules, 2003. 36(6): p. 1802-1807.
13. Zhang, Y., A. A. Broekhuis, and E Picchioni, Thermally self-healing polymeric materials: the next step to recycling thermoset polymers? Macromolecules, 2009. 42(6): p. 1906-1912.
41


14. Taynton, P., et al., Repairable woven carbon fiber composites with full recyclability enabled by malleable polyimine networks. Advanced Materials, 2016. 28(15): p. 2904-2909.
15. Yu, K., et al., Carbon fiber reinforced thermoset composite with near 100% recyclability. Advanced Functional Materials, 2016. 26(33): p. 6098-6106.
16. Buback, M., et al., Kinetics and modeling of free-radical batch polymerization of nonionized methacrylic acid in aqueous solution. Industrial & Engineering Chemistry Research, 2008. 47(21): p. 8197-8204.
17. Zhao, Z., et al., Desolvation induced origami of photocurable polymers by digit light processing. Macromolecular rapid communications, 2017. 38(13): p. 1600625.
18. Zarrelli, M., A. A. Skordos, and I.K. Partridge, Toward a constitutive model for cure-dependent modulus of a high temperature epoxy during the cure. European Polymer Journal, 2010. 46(8): p. 1705-1712.
19. Gan, S., J. Seferis, and R. Prime, A viscoelastic description of the glass transition-conversion relationship for reactive polymers. Journal of thermal analysis, 1991. 37(3): p. 463-478.
20. Dewaele, M., et al., Volume contraction inphotocured dental resins: the shrinkage-conversion relationship revisited. Dental Materials, 2006. 22(4): p. 359-365.
21. Wu, J., et al., Evolution of material properties during free radical photopolymerization. Journal of the Mechanics and Physics of Solids, 2018. 112: p. 25-49.
22. O'Brien, D.J., PT. Mather, and S.R. White, Viscoelastic properties of an epoxy resin during cure. Journal of composite materials, 2001. 35(10): p. 883-904.
23. Long, K.N., M.L. Dunn, and H. J. Qi, Mechanics of soft active materials with phase evolution. International Journal of Plasticity, 2010. 26(4): p. 603-616.
24. Parameswaranpillai, J., et al., Investigation of Cure Reaction, Rheology, Volume Shrinkage and Thermomechanical Properties ofNano-Ti02 Filled Epoxy/DDS Composites. Journal of Polymers, 2013. 2013: p. 183463.
25. Vijayan, PP, J. Pionteck, and S. Thomas, Volume Shrinkage and Cure Kinetics in Carboxyl-Terminated Poly (butadiene-co-acrylonitrile) (CTBN) Modified Epoxy/Clay Nanocomposites. Journal of Macromolecular Science Part a-Pure and Applied Chemistry, 2015. 52(5): p. 353-359.
26. Daoud, M., Viscoelasticity near the sol-gel transition. Macromolecules, 2000. 33(8): p. 3019-3022.
42


27. Rubinstein, M. and R.H. Colby, Polymer physics. 2003, New York: Oxford University Press, Oxford.
28. Yu, K., et al., Dissolution of covalent adaptable network polymers in organic solvent. Journal of the Mechanics and Physics of Solids, 2017. 109: p. 78-94.
29. Rubinstein, M. and R.H. Colby, Polymer physics. Vol. 23. 2003: Oxford University Press New York.
30. Di Marzio, E. A. and A. J. Yang, Configurational entropy approach to the kinetics of glasses. Journal of research of the National Institute of Standards and Technology, 1997. 102(2): p. 135.
31. Williams, M.L., R.F. Landel, and J.D. Ferry, The temperature dependence of relaxation mechanisms in amorphous polymers and other glass-forming liquids. Journal of the American Chemical society, 1955. 77(14): p. 3701-3707.
32. DiBenedetto, A., Prediction of the glass transition temperature ofpolymers: a model based on the principle of corresponding states. Journal of Polymer Science Part B: Polymer Physics, 1987. 25(9): p. 1949-1969.
33. Nielsen, L.E., Cross-linking-effect on physical properties ofpolymers. Journal of Macromolecular Science, Part C, 1969. 3(1): p. 69-103.
34. Fick, A., V. On liquid diffusion. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 1855. 10(63): p. 30-39.
43


Full Text

PAGE 1

EVOLUTION OF MATERIAL PROPERTIES DURING THE SOLVENT ASSISTED RECYCLING OF COVALENT ADAPTABLE NETWORK POLYMERS by DRAKE ASHTON SOULE B.S., University of Colorado Denver, 2015 A thesis submitted to the Faculty of the Mechanical Engineering Department of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Mechanical Engineering Program 2019

PAGE 2

! ii © 2019 DRAKE ASHTON SOULE ALL RIGHTS RESERVED

PAGE 3

! iii This thesis for the Master of Science Mechanical Engineering degree by Drake Ashton Soule h as been approved for the Mechanical Engineering Program b y Sam Welch , Chair Dana Carpenter Christopher Yakacki Kai Yu, Advisor Date: May 1 8 , 2019

PAGE 4

! iv Soule, Drake Ashton (M.S. Mechanical Engineering Program) Evolution of Material Properties During the Solvent Assisted Recycling of Covalent Adaptable Network Polymers Thesis directed by Assistant Professor Kai Yu ABSTRACT Research in methods to recycle polymers has recently gained momentum, driven by economic concerns and pressing demand. This study explored an epoxy polymer with covalent adaptable network and its recyclability by decomposing it at high temperatures in an e thylene glycol solvent using a closed environment. This method creates a decomposed polymer solution and once decomposed, this epoxy can again be repolymerized by heating in an open environment and allowing the ethylene glycol to volatilize, leaving only t he repolymerized epoxy. By using mass measurements of the decomposed epoxy solutions throughout the repolymerization process, the evaporation rates of ethylene glycol solvent were achieved over a range of temperatures and heating times. The evolution of so lution viscosity was also characterized using rheometer. A dynamic mechanical analysis was performed on thin films of repolymerized epoxy to determine the storage modulus, glass transition temperature, and rubbery modulus of the material at a range of temp eratures and heating times. Bulk epoxy samples were produced, where the material was allowed to repolymerize to make samples significantly thicker than the thin samples (thicknesses of approximately 8 millimeters for the bulk sample and 0.7 millimeters for the thin sample). This bulk sample was cut into slices along its thickness and the slices were subjected to the same dynamic mechanical analysis procedure as the thin samples to identify how the storage modulus, glass transition temperature, and glassy mo dulus behaved at the different

PAGE 5

! v layers of the bulk sample. A multiscale constitutive model was established to study the repolymerization process. The form and content of this abstract are approved. I recommend its publication. Approved: Kai Yu

PAGE 6

! vi TABLE OF CONTENTS CHAPTER I INTRODUCTION ................................ ................................ .............................. 1 CHAPTER II MATERIALS AND EXPERIMENTS ................................ .............................. 4 2.1 Material Synthesis ................................ ................................ ................................ ........... 4 2.2 Decomposing and Repolymerizing the Networks ................................ ........................... 5 2.4 Solvent Evaporation Rate Measurement ................................ ................................ ........ 6 2.3 Viscosity Measurement ................................ ................................ ................................ ... 7 2.6 Dynamic Mechanical Analysis on Bulk Sample ................................ ............................. 7 CHAPTER III A THERMOVISCOELASTIC CONSTITUTIVE MODEL .......................... 8 3.1 Degree of Conversion Ratio ................................ ................................ ............................ 9 3.2 The Evolution of Equilibrium Modulus ................................ ................................ ......... 9 3.3 Evolution of Relaxation Time ................................ ................................ ....................... 10 3.4 The Evolution of Network T g ................................ ................................ ........................ 10 CHAPTER IV RESULTS AND DISCUSSIONS ................................ ................................ ... 11 4.1 Material Parameters Identification ................................ ................................ .............. 11 4.5 Results of Solvent Evaporation Rate ................................ ................................ ............ 16 4.4 Prediction of the Evolution of Viscosity During the CAN Repolymerization ............. 17 4.3 Prediction of the Glass Transition Behaviors Durin g the CAN Repolymerization .... 18 4.7 Results of Dynamic Mechanical Analysis on Bulk Sample ................................ .......... 27

PAGE 7

! vii CHAPTER V CONCLUSION ................................ ................................ ................................ 40 BIBLIOGRAPHY ................................ ................................ ................................ ................... 41

PAGE 8

! viii LIST OF FIGURES ! Figure 1: The repolymerization of thermosets ................................ ................................ ............. 2 ! Figure 2: Detailed structure information ................................ ................................ ...................... 5 ! Figure 3: Decomposition and repolymerization of epoxy with fatty acid linker ........................... 6 ! Figure 4: Bulk sample layers and direction of spatial network structure ................................ ...... 7 ! Figure 5: The one dimensional rheological model for the viscoelastic mechanical properties ...... 8 ! Figure 6: N ormalized mass increment of non catalyst epoxy samples with different diffusion temperatures ................................ ................................ ................................ ............................. 12 ! Figure 7: Stress relaxation behavior ................................ ................................ .......................... 13 ! Figure 8: Time temperature super position of the epoxy ................................ ............................ 14 ! Figure 9: Master curve fitting with 11 viscoelastic branches added into the model .................... 15 ! Figure 10: C oncentration of ethylene glycol solvent over time in the decomposed polymer solution ................................ ................................ ................................ ................................ ..... 17 ! Figure 11: Predictions of the viscosity during the CAN repolyme rization before gelation .......... 18 ! Figure 12: Prediction of the storage modulus and tan! curve ................................ ..................... 19 ! Figure 13: Predictions on the tan! and equilibrium modulus curve for different repolymerization times at 160 ¡C ................................ ................................ ................................ ......................... 21 ! Figure 14: The evolution of storage modulus and tan delta over time at 180 ¡ C ......................... 24 ! Figure 15: The evolution of storage modulus and tan delta over time at 200 ¡ C ......................... 26 ! Figure 16: Glass transition temperature and rubbery modulus for 160 ¡C, 180 ¡C, and 200 ¡C cures of thin film ................................ ................................ ................................ ....................... 27 ! Figure 17: Storage modulus and ta n delta for Bulk Sample A, 15 hour cure at 200 ¡C ............... 29 ! Figure 18: Rubbery modulus and glass transition temperature for Bulk Sample A ..................... 30 ! Figure 19: S torage modulus and tan delta for Bulk Sample B, 18 hour cure at 200 ¡C ............... 31 !

PAGE 9

! ix Figure 20: S torage modulus and tan delta f or Bulk Sample B, 21 hour cure at 200 ¡C ............... 33 ! Figure 21: Rubbery modulus and glass transition temperature for Bulk Sample B ..................... 34 ! Figure 22: S torage modulus and tan delta for Bulk Sample C, 25 hour cure at 180 ¡C ............... 36 ! Figure 23: S torage modulus and tan delta for Bulk Sample C, 30 hour cure at 180 ¡C ............... 38 ! Figure 24: Rubbery modulus and glass transition temperature for Bulk Sample C ..................... 39

PAGE 10

! x LIST OF TABLES Table 1: Material Parameters ................................ ................................ ................................ .... 15 Table 2: Material properties for 17 21 hour cures ................................ ................................ ...... 21 Table 3: Material properties for 9 13 hour cures ................................ ................................ ........ 22 Table 4: Material properties for 3 7 hour cures ................................ ................................ .......... 24 Table 5: Material properties for Sample A, layers 1 and 2, 15 hour cure ................................ .... 28 Table 6: Material properties for Sample B, layers 1 4, 18 hour cure ................................ .......... 30 Table 7: Material properties for Sample B, layers 1 4, 21 hour cure ................................ .......... 32 Table 8: Material properties for Sample C, layers 1 4, 25 hour cure ................................ .......... 35 Table 9: Material properties for Sample C, layers 1 4, 30 hour cure ................................ .......... 37

PAGE 11

! 1 CHAPTER I INTRODUCTION Conventional thermoset polymers cannot be reshaped or recycled once hardened because of their cross linked networks, which cause a large amount of plastic waste that ends up in landfill [1 6]. Recently, with the advent of covalent adaptable networks (CANs), the recycling of covalently crosslinked thermoset materials has become possible [7 11]. Their reversible bonds allo w the polymer network to be continuously rearranged through the bond exchange reaction (BER), thus giving thermoset polymers the properties such as surface welding, self healing and malleability [12, 13]. CANs can be completely decomposed in suitable solvents through BERs between polymer network and solvent molecules. Excitingly, repolymerization can occur by heating the decomposed polymer solution in an open environment. The repolymerization process is shown in Figure 1. When heated in an open environ ment, BERs can occur between the reversible groups at each end of each segment in the polymer solution, thereby connecting the different polymer chains. As the solvent molecules evaporate, the polymer network is gradually formed. The repolymerized material has almost the same network structure and thermomechanical properties as the fresh sample ; for example, the elastic modulus and strength of the recycled materials are about 98.3% and 93.3%, respectively, of the fresh ones [14, 15].

PAGE 12

! 2 Figure 1 : The repolymerization of thermosets can be realized via the transesterification type BERs. The repolymerization of thermoset polymers offers a green and sustainable recycle method for thermoset materials and composites which simply involves he ating the polymer solution in an open environment. However, since its existing research is limited to conceptual arguments, no systematic experimental or theoretical work has been proposed to reveal the relationship between polymer repolymerization and net work dynamics or to examine the spatial network structure and mechanical properties of the repolymerizing network which limits its engineering applications. Repolymerization is a complex physical process: The polymer solution undergoes a liquid to solid phase transition; the polymer chain length and crosslinking density evolve with the conversion ratio; repolymerization of decomposed polymer solution starts from the surface and involves non uniform solvent concentration and structure developed in the thic kness direction. These changes at the microscopic scale during repolymerization affect their macroscopic properties of the materials, such as Young's Modulus, glass transition temperature and mechanical properties. Therefore, it is critical to study the ev olution of material properties to achieve better performance control. The degree of conversion (DoC) is commonly used to

PAGE 13

! 3 characterize and simulate the evolution of polymer systems during curing of thermoset epoxy resins [16 18]. As the repolymerization pro ceeds, the monomers gradually grow into polymer chains. In this process, the number distribution of polymer chains with different chain lengths is described by DoC, and the material properties can be calculated by a superposition method, and then the repol ymerization process is associated with the evolution of material properties [19 21]. The effect of temperature on the time scale of mechanical behavior is described by the time temperature superposition principle to study the viscoelastic behavior of therm osetting polymer during repolymerization processes [22]. In addition, the glass transition temperature (Tg) of the curing system increases as the repolymerization proceeds. Non linear viscoelastic behavior may be exhibited when the cured material enters th e glassy state [19, 22], so there is a need for a model that can describe the stress strain behavior of a cured glassy polymer upon large deformations. In this thesis, we propose a multi scale modeling framework combining the macro scale mechanical propert ies with micro scale spatial network structure to describe the evolution of material property during the repolymerization process. However, there are some unique polymerization phenomena for the repolymerization of CANs. First, the reversible groups at the end of the segment in the polymer solution are combined through BERs to form a polymer chain and its time scale depends on the kinetics of chemical reaction. Second, the repolymerization of decomposed polymer solution always starts from the surface as the solvent evaporates. A higher temperature usually leads to the quick formation of a stiff film on the surface of the solution, which will suppress the evaporation of the ethylene glycol (EG), and consequently inhibit the repolymerization of epoxy beneath t he surface. But, reducing the temperature will lead to a lower repolymerization rate and a longer repolymerization time, which means that the optimum

PAGE 14

! 4 repolymerization temperature of the epoxy needs to be found. And the most challenging task is understandin g how to describe the length evolution of chain segments during the repolymerization . In this thesis, a micro scale model is first established based on the kinetics of BERs to describe the develop of length of chain segments during the repolymerization, an d consequently determine the evolution of DoC. The DoC is then used as an internal variable to characterize the reaction system. The time temperature DoC superposition is used to capture the thermomechanical behaviors of the cured polymers. The model shows good prediction on the conversion rate of functional groups, mechanical properties, and repolymerization rate of polymer network. It also helps to reveal the influence of different material and processing variables, such as time and temperature. CHAPTER II MATERIALS AND EXPERIMENTS 2.1 Material Synthesis The material used in this study was the epoxy thermosetting polymer developed by Leibler and coworkers [16]. The epoxy samples were synthesized from a catalyst (metal catalyst Zn(Ac)2 (Sigma Aldrich, St. Louis, MO)), monomers diglycidyl ether of bisphenol A (DGEBA, Sigma Aldrich), and crosslinkers (fatty acids Pripol 1040 (Croda, Houston, TX)). Detailed chemical structures of each reagent are shown in Figure 2. For polymer synthesis, the catalyst was first mixed with a fatty acid (10 mol% to COOH groups) in a beaker. The temperature was gradually increased from 100 ¡C to 150 ¡C while maintaining the mixture under vacuum until no gas evaporation was observed and the catalyst particles were completely decompo sed (3 hours). DGEBA was then added to the previous fatty acid mixture containing solubilized catalyst (stoichiometry between COOH and epoxy is 1:1) in a round bottom flask and manually stirred at

PAGE 15

! 5 130 ¡C until the mixture became homogeneous. Finally, the m ixture was poured into a mold and placed in an oven for 6 hours at 130 ¡C. Figure 2 : Detailed structure information about the chemical reagents used in this study. 2.2 Decomposing and Repolymerizing the Networks For the decomposition, epoxy samples were immersed in ethylene glycol solvent (volume ratio between polymer and solvent was 1:2). At high temperature, solvent molecules diffuse into the network and participate in the transesterification type BERs with poly mer chains, which effectively break the long chains into short segments at the position of ester groups on the backbone. The container was sealed to avoid solvent evaporation. After being heated in 180 ¡C for 6 hours, the polymer was fully decomposed. For the repolymerization, the decomposed polymer solution was poured into a mold and heated at a given temperature (160 ¡C, 180 ¡C, and 200 ¡C) in an open environment. The BER proceeded stepwise between short segments by combining their functional groups to fo rm a repolymerized polymer network. The schematic view of the decomposition and repolymerization of epoxy with fatty acid linker was shown in Figure 3.

PAGE 16

! 6 Figure 3 : Decomposition and repolymerization of epoxy with fatty acid linker. 2.4 Solvent Evaporation Rate Measurement Masses for epoxy samples and ethylene glycol were measured before the samples were decomposed in the ethylene glycol solution to create a decomposed polymer solution using the same method discussed in section 2.2. Once the epoxy mixture was completely decomposed it was poured into a mold with a known mass. The mass of the mold containing the decomposed polymer solution was measured and the known mass of the empty mold was subtracted from the total mass, leaving the mass of the decomposed polymer solution. Assuming a uniform ratio of epoxy to ethylene glycol solution, the mass of the decomposed polymer solution was used to calculate the masses of the epoxy sample and the ethylene glycol solution. Once these masses wer e calculated, the mold containing the decomposed polymer solution was again heated at a given temperature (160 ¡C, 180 ¡C, and 200 ¡C) in an open environment. The mold allowed the ethylene glycol to evaporate from the decomposed polymer solution and the ma ss of the mold was measured and recorded after each hour it was exposed to the heated environment. The difference in mass after each hour was due to the evaporating ethylene glycol and after each interval the new percent by mass of ethylene glycol in the d ecomposed polymer solution was calculated as the solution repolymerized.

PAGE 17

! 7 2.3 Viscosity Measurement In the repolymerization, the resin viscosity was measured before gelation to identify the evolution of chain segment length. After being heated for differe nt lengths of time, the viscosity was measured by using a rheometer (TA Instruments, AR G2, New Castle, DE, USA) with parallel plate geometry (20 mm diameter, 1,000 µm gap). During the measurements, temperature was maintained constant at room temperature w hile the shear rate was increased from 10 3 to 200 s 1 . 2.6 Dynamic Mechanical Analysis on Bulk Sample A silicone mold was used to cure the polymer solution into a bulk sample which was placed in an oven to cure at temperatures of 180 and 200 ¡C at different heating times. Once cured, the sample was sliced in layers perpendicular to the bulk sample's thickness, as shown in Figure 4. The same Dynamic Mechanical Analysis (DMA, Model Q800, TA Instruments, New Castle, DE, USA) process described in sectio n 2.4 was then conducted on these layers to determine the spatial network structure and mechanical properties of bulk thermoset during the repolymerization. ! Figure 4 : Bulk sample layers and direction of spatial network structure.

PAGE 18

! 8 CHAPTER III A THERMOVISCOELASTIC CONSTITUTIVE MODEL A multi branch constitutive model is applied to study the viscoelastic behaviors of repolymerized epoxy CANs. Figure 5 shows the one dimensional rheological analogy of this model consist ing of two par ts: the equilibrium branch and the N non equilibrium branches that describe the relaxation of viscoelastic polymer chains. As the number of crosslinks increases, the number of balanced branches increases, which can be described by a phase evolution model [ 23 ] . The newly added cross linking will prevent the old cross linking from relaxing at the rate of the old relaxation rate [ 21 ] ; therefore, according to the time t emperature DoC stack, the relaxation time will shift to the new relaxation time. Following Figure 5, The total stress " eq can be calculated by summing the stresses in the equilibrium branch and in the non equilibrium branches. ! Figure 5 : The one dimensional rheological model for the viscoelastic mechanical properties modeling. Red and blue parts of these branches respectively describe the equilibrium branches and viscoelastic branches. The true stress " of the m odel is the summation of each paralleled components. ! " ! #$ % & ! '( ) ( * + ( 1 ) where " eq and " vj are respectively the true stresses of the equilibrium branch and viscoelastic branches. Each branch has the same stretch, which equals to the total stretch of the multi branched model #.

PAGE 19

! 9 For viscoelastic branches, the stress can be calculated by: ! '( " , '( -. / 0 '( # 1 " , '( 2 '( 3 0 '( ' 4 0 '( ' 45 ( 2 ) where E vj and # vj e are respectively the elastic modulus and stretch ratio of the spring. $ bi and # vj v are respectively the relaxation time and stretch ratio of the dashpot in the j th viscoelastic branch. 3.1 Degree of Conversion Ratio The evolution of network structure and thermomechanical properties are commonly linked to the degree of conversion ( DoC ) ratio of functional groups [ 24 , 25 ] . During the re polymerization, each chain connection reaction gene rates a solvent molecule that evaporates out of the system. The chain segment is increased correspondingly. In this thesis, we take conversion ratio degree to be linear with the normalized solvent evaporation rate, namely: 6 7 5 8 " 3 9 : 7 5 8 : ; < , ( 3 ) where : 7 5 8 is the solvent evaporation during the re polymerization, and : ; is the initial solvent content. T hus, t he solvent evaporation rate and degree of conversion ratio can be experimentally measured. 3.2 The Evolution of Equilibrium Modulus The dependence of the viscoelastic properties of thermosetting systems on the degree of cure has been widely investigated, both in terms of experimental results and modelling approach. The rubbery modulus during the re polymerization is scaled to the correlation length of chain segments l 0 by , ; = > ? @ ; A [ 26 ] . l 0 is further scaled to the D O C as @ ; = 7 3 9 6 B 8 C D E [ 27 , 28 ] , with ! v being a correlation exponent. Therefore, the evolution of rubbery modulus can be written as: , ; 7 5 8 " , F G 3 9 6 B 7 5 8 H A I E . ( 4 )

PAGE 20

! 10 3.3 Evolution of Relaxation Time The relaxation time of each Maxwell element is a function of temperature. The time temperature relaxation at temperature T in the viscoelastic branches also follow the time temperature superposition principle (TTSP), which can be represented by [ 29 ] : 2 '( 7 > 8 " 2 '( ; J ' 7 > 8 ( 5 ) where $ vj0 is the relaxation time of the reference sample at the reference temperature T g . Here we choose the polymer cured at XX hours as the reference state; % v (T) is the temperature shift factor which can be described by WLF equation ( 6 a) above the reference temperature T g and by Arrhenius equation ( 6 b) below the T g [ 30 , 31 ] . -. G K ' 7 > 8 H " 9 L M N O P Q 3 > % RST U 3V 9 3 > W % RST U 3V X Y K5 Y > Z > W ( 6 a ) -[\ G K ' 7 > 8 H " 9 ] + / > 9 > W 1 ] B % > 9 > W Y K5 Y > ^ > W (6b) where A and F c are material specific constants. C 1 and C 2 are fitting parameters. T s is a reference temperature with ! v ( T s )=1. It can be taken to be the crossing point of two curves represented on a ! vs T plot. 3.4 The Evolution of Network T g The glass transition temperature is an important indication of the extent of curing and of the viscous property. As the curing progresses, the small monomers grow into large molecules and are connected to each other. The entire system has widely distribute d polymer chains of different molecular weights. As mentioned above, the macroscopic properties measured in the experiments are the average behavior of these chains and the DoC is a suitable evaluation parameter to characterize the average properties of th e overall system. The relationship between the T g and the DoC has been widely studied. Gan et al. [ 19 ] developed a viscoelastic type mo del

PAGE 21

! 11 for the curing polymer by using the empirical Dibenedetto equation [ 32 , 33 ] . The mode l considers the effect of cross linking on the mobility of the curing system and can successfully capture the T g evolution of various curing systems. Here, we use this viscoelastic model to consider the T g changes during the curing: > W " , _ ` -. G a + 7 3 9 6 8 b % a B H ( 7 ) where E r is the activation energy of transition from the glassy to the rubbery state, R is the gas constant, " is the parameter for accounting the effects of chain entanglement. g 1 , g 2 are two material constants. CHAPTER IV RESULTS AND DISCUSSIONS 4.1 Material Parameters Identification The diffusivity of EG molecules in epoxy samples were determined by diffusion data of non catalyst epoxy and the transportation of EG solvent can be described by using Fick's laws of diffusion [ 34 ] : cd c5 " e 7 > 8 c B d c f B ( 8 ) where D (T) is the molecular diffusivity and its temperature dependency follows the Arrhenius equation: e 7 > 8 " e ; ghi Y j 9 , Ik `> l ( 9 ) where D 0 is the reference diffusivity, E as is the activation energy, R is the gas constant R = 8.314& J& mol ' 1 & K ' 1 . The Eq. (8) can be solved numerically by using the finite difference method, where the time and space derivatives are respectively calculated using forward difference and central difference. The normalized mass increment of non catalyst epoxy samples at different

PAGE 22

! 12 temperature is shown in Figure 6. When the diffusion coefficient E as = 34.2kJ / mol and the pre factor D 0 = 3.23 ( 10 6 m 2 /s, the model predictions agree well with the experimental data. Figure 6 : Normalized mass increment of non catalyst epoxy samples with different diffusion temperatures. Note: experimental results are plotted in dots and solid lines plot the model predictions. We can use the method developed by Yu et al. to determine the BER energy barrier E ab by performing a series of stress relaxation tests on the fresh epoxy sample at different temperatures. During the stress relaxation tests, the normalized relaxation modulus , m _ can be characterized by the following exponential function: , m _ " nf6 j 9 5 2 l ( 10 ) where the relaxation time 2 can be measured from the experimental relaxation curves when normalized relaxation modulus declines to 1/e (~36.8%). The relationship between relaxation time and the BER energy barrier is: 2 " O ; nf6 o , IP ` 7 > % RST 8 p ( 11 ) where the k 0 is a correlation exponent and the coefficient determine the energy barrier E ab . Figure 7a shows the stress relaxation curves of fresh epoxy CANs at different temperatures. The dots in Figure 7b show the measured relaxation times as a function of temperature. By fitting the

PAGE 23

! 13 relaxation time, the energy barrier for the fresh epoxy samp le is determined to be E ab =8.15KJ/mol. (a) (b) Figure 7 : (a) Stress relaxation behavior of fresh epoxy CAN. The relaxation time is determined to be the timing point when the relaxation modulus is reduced 63.2% (the dash line). (b) The relationship between relaxation time and temperature. The relaxation curves which represent the contribution of viscoelastic branches are plotted in Figure 8a. Based on the TTSP, each curve is shifted to the reference temperature Ts (17 ¡C) to form a master relaxation curve (Figure 8b). The corresponding shifting factors are plotted in Figure 8c as a function of temperature. At temperatures above and below T s , the shift factors are respectively captured by the WLF and Arrhenius equations (shown in 3. 3 ). The corresponding parameters were determined to be AFc/kb= 40070, C1=8.658, and C2=15.76.

PAGE 24

! 14 (a) (b) (c) Figure 8 : (a) The relaxation moduli of viscoelastic branches; (b) Master curve after shifting each temperature to reference temperature (17 ¡C). (c) Shift factor in each temperature. Red circles are the experimental data and the black solid line is the predictions of Arrhenius and WLF equation. The capability of the applied multi branch model in capturing the complicated stress relaxation process originates from the additionally introduced nonequilibrium branches. By adding nonequilibrium branches gradually, the sim ulated curve is gradually approaching the master curve. As shown in Figure 9, when adding 11 nonequilibrium branches, the model prediction curve rejoins with the experimental relaxation master curve. The corresponding E bi and $ bi values are listed in Table 1 .

PAGE 25

! 15 ! Figure 9 : Master curve fitting with 11 viscoelastic branches added into the model. Red circle line is model prediction and the black solid line is the experimental master curve. ! Table 1 : Material Parameter . Description Symbol Value Known constants Gas constant R 8.31J& mol ' 1 & K ' 1 Boltzmann constant k b 1.38 ( 10 ' 23 & m 2 & kg & s ' 2 & K ' 1 Avogadro constant N A 6.02(10 23 Directly measured parameters Energy barrier for diffusivity E as 34.2kJ / mol Energy b arrier for BER E ab 8.15kJ / mol Evaporation rate of EG at 160 Evaporation rate of EG at 180 Evaporation rate of EG at 200 5.122 ( 10 5 mol/m 2 /s 7.488 ( 10 5 mol/m 2 /s 9.750 ( 10 5 mol/m 2 /s Calculated parameters Number of viscoelastic branches N 11 Parameter of Arrhenius equation AF c /k b 40070 Elastic moduli in each viscoelastic branche E vn 171.591,180,210,160,30,11,5,1 .1,1,0.801,1.099 Relaxation times at T s in each viscoelastic branche # vn 1 (10 7 , 1 (10 6 , 1 .5(10 5 , 6 (10 5 , 6 (10 4 , 3 (10 3 , 2.5(10 2 , 2.5(10 1 , 0,2.5,25,600 MPa Parameter of WLF equation C 1 8.658 Parameter of WLF equation C 2 15.76

PAGE 26

! 16 4.5 Results of Solvent Evaporation Rate The mass of epoxy and ethylene glycol in the decomposed polymer solution was calculated based on the ratio of epoxy and ethylene glycol masses in the mixture. For the sample cured at 160 ¡C, 24.46g of epoxy and 47.01g of ethylene glycol was used for a total mixture mass of 71.47g. 34.2% of the decomposed polymer solution's mass was epoxy whil e the remaining 65.8% was the ethylene glycol. The mass of the decomposed polymer solution in the mold was measured to be 1.56g before it began exposure to heat. The epoxy, 0.53g of the decomposed polymer solution's mass of 1.56g, was expected to remain co nstant during the evaporation process while the 1.05g of ethylene glycol decreased as it evaporated. The evaporation rate can be seen in Figure 10 which compares the concentration of ethylene glycol solvent over time between samples heated at 160 ¡C, 180 ¡ C, and 200 ¡C. The second sample was cured at 180 ¡C and had a decomposed polymer solution with a ratio of 15.0g of epoxy sample to 28.49g of ethylene glycol solution which combined for a total mass of 43.49g. The mass of the decomposed polymer solution i n the mold was 2.07g. The concentration by mass of epoxy in the mold is 34.4% to make 0.71g while the remaining 65.6% of the mass was ethylene glycol for 1.35g. The evaporation rate of the 180 ¡C sample can be seen in Figure 10. The decomposed polymer sol ution used to cure at 160 ¡C was also used to cure at 200 ¡C (the mixture was 24.46g of epoxy and 47.01g of ethylene glycol). The starting mass of the 200 ¡C sample was 1.53g, meaning 34.2% of the decomposed polymer solution's mass in the mold was 0.52g of epoxy. The remaining 65.8% of the decomposed polymer solution's mass in the mold was 1.01g of ethylene glycol. The evaporation rate of the 200 ¡C can be seen in Figure 10.

PAGE 27

! 17 ! Figure 10 : A plot of the concentration of ethylene glyco l solvent over time in the decomposed polymer solution at 160 ¡C, 180 ¡C, and 200 ¡C. The curve fitting tool in MATLAB was used to determine the fitted curves for the recorded data using the format: : " Y : ; n C q r ( 12 ) where : is the percent of ethylene glycol solvent by mass, : ; is some initial value of ethylene glycol solvent, 5 is the cure time, and 2 is the relaxation time of the system. These relaxation times were found to be 5.998 hours, 4.542 hours, and 2.361 hours fo r the 160 ¡C, 180 ¡C, and 200 ¡C cures respectively and it is at these times that the viscoelastic nature of the polymer pulls the system back into a solid state. The jump from 160 ¡C to 200 ¡C cuts this relaxation time by more than half, and is the first physical evidence showing that as curing temperatures increase so does the repolymerization rate, both driving the material back toward the mechanical properties it demonstrates before decomposition in the ethylene glycol solvent. 4.4 Prediction of the Evo lution of Viscosity During the CAN Repolymerization The evolution of viscosity during the CAN repolymerization is predicted by our model. The epoxy solution was subject to heating at different temperature (160 ¡C, 180 ¡C, 200 ¡C) for repolymerization. As t he solvent volatilizes, the segments in the solution repolymerize by BER

PAGE 28

! 18 and the solution viscosity is gradually increased which is shown in Figure 11. Changes in the viscosity can be taken to be an indicator of the repolymerization degree. It can be seen that our model fits the experimental results well. ! Figure 11 : Predictions of the viscosity during the CAN repolymerization before gelation at 160 ¡C 180 ¡C and 200 ¡C respectively. 4.3 Prediction of the Glass Transition Behaviors During the CAN Repolymerization The thin samples tested were created from the same decomposed polymer solutions that were used for the ethylene glycol solvent evaporation rate measurement in section 4.5. The thin samples were tested as soon as they were cured enough to remove them from the glass plate they were cured on without damaging the sample. If removed from the plate too soon, the samples would break or only portions of the sample would separate from the plate. The samples had to cure lon ger at lower temperatures to achieve enough stability to be removed from the curing plate. An initial DMA test was performed on a thin film sample of fresh epoxy that had not yet been decomposed and repolymerized. This test returned a tan delta (tan!) of 1 .176508, a glass transition temperature (T g ) of 27.52858 ¡C, and a rubbery modulus (E r , the value of the storage modulus when the sample reached 80 ¡C during the DMA test) of 4.350665 MPa.

PAGE 29

! 19 The multi branched model was used to predict the storage modulus an d tan $ values during the glass transition process by DMA testing on the samples. For the 1D multi branched model, the temperature dependent storage modulus E s , loss modulus E l , and tan! are respectively expressed as: , k " & , Ps t B 2 Ps B 3 % t B 2 Ps B u s * + % & , '( t B 2 '( B 3 % t B 2 '( B ) ( * + ( 13 a) , v " & , Ps t 2 Ps 3 % t B 2 Ps B u s * + % & , '( t 2 '( 3 % t B 2 '( B ) ( * + ( 13b ) wx. y " , v , k (13c) where % is the testing frequency (1Hz in the experiments). From Figure 12, it can be seen that the predictions of storage modulus and tan! agree with experimental data ranging from 20 to 80 ¡C . ! Figure 12 : Prediction of the storage modulus and tan$ curve. Red circles indicate the predicted behavior while the black solid lines are experimental data. The equilibrium modulus and T g of CANs for different repolymerization time s at 160 ¡C were predicted by our model and presented in Figure s 1 3 a and 13 b. T he corresponding storage modulus and tan! curve were predicted by the multi branched model and presented in Figures 13c through 13g . The mechanical properties of the 160 ¡C cure can be seen in Table 2.

PAGE 30

! 20 (a) (b) (c) (d) (e) (f)

PAGE 31

! 21 (g) Figure 13 : Predictions on the (a) tan$ and (b) equilibrium modulus curve for different repolymerization times at 160 ¡C; (c) (g) Prediction on the storage modulus and tan$ curve of CANs at 17, 18, 19, 20, 21 hours' repolymerization respectively. Red circles are predicted behavior, black solid lines are experimental data. Table 2 : Material properties for 17 21 hour cures 17 Hour 18 Hour 19 Hour 20 Hour 21 Hour Tan Delta 1.103934 0.9125098 0.9078816 0.8486876 0.8736528 T g (¡C) 15.98789 11.47278 10.60691 11.36923 14.86556 E rubbery (MPa) 1.1855 0.7013098 0.8959832 0.9048311 1.065385 It can be seen that T g and E r trend upward to both increase as the sample cures longer (with the exception of the 17 hour data which stands out as an anomaly for this sample), indicating that as the solvent volatizes the network density increases. The glass transition temperature at t he 21 hour mark is at 54% of the fresh sample, meaning the crosslinking in this sample has not yet reached the density of the fresh sample. The rubbery modulus of the 21 hour cure stands at 24% of the strength of the fresh sample. After 9 hours of curing a t 180 ¡C the next epoxy sample was ready to be tested. This time the sample would be tested after 9, 10, 11, 12, and 13 hours of curing and the material properties recorded during the dynamic mechanical analysis can be seen in Table 3. The hour by hour

PAGE 32

! 22 com parison of the tan delta and storage modulus recorded by the DMA machine can be seen in Figures 14a through 14e. A comparison of all recorded tan delta and all recorded storage modulus for the 180 ¡C cure can be seen in Figures 14f and 14g respectively. Ta ble 3 : Material properties for 9 13 hour cures 9 Hour 10 Hour 11 Hour 12 Hour 13 Hour Tan Delta 1.125106 1.058598 0.98586 1.002632 0.9333222 T g (¡C) 6.234282 13.68666 18.80338 18.61292 21.67086 E rubbery (MPa) 1.3743 3.5375 3.5799 3.8911 3.5925 Note that at 180 ¡C the mechanical properties of the repolymerized sample are much stronger than those of the 160 ¡C . This shows that the polymerization rate significantly increases as the temperature of the process increases. The glass transition temperature and rubbery modulus more than double from the 9 hour cure to the 10 hour cure. Recall that the earliest this sam ple could safely be removed from the curing plate was at the 9 hour mark, when the sample was still relatively weak. The T g and E r values for this sample get much closer to the strength for the fresh sample at 79% and 83% respectively. This indicates the n etwork properties are increased with higher temperatures and faster curing times.

PAGE 33

! 23 (a) (b) (c) (d) (e) (f)

PAGE 34

! 24 (g) Figure 14 : The evolution of storage modulus and tan delta at each hour interval (a) Ð (e) and all tan delta (f) and all storage modulus (g) produced by dynamic mechanical analysis for the 180 ¡C cure. The final thin film sample cured at 200 ¡C and was ready to be tested after 3 hours of curing. The additional dynamic mechanical analysis testing also took p lace after 4, 5, 6, and 7 hours of curing. The material properties recorded during these tests can be seen in Table 4. The evolution of storage modulus and tan delta on an hour by hour basis can be seen in F igures 1 5 a though 1 5 e. The comparison of all tan delta and all storage modulus recorded for each cure time can be seen in Figures 1 5 f and 1 5 g respectively. Table 4 : Material properties for 3 7 hour cures 3 Hour 4 Hour 5 Hour 6 Hour 7 Hour Tan Delta 1.09131 0.996882 0.9243482 1.042304 0.9908801 T g (¡C) 14.33166 19.15065 30.60437 25.64553 33.00221 E rubbery (MPa) 1.1537 1.748 2.3108 2.924 1.8215 The T g for the sample cured at 200 ¡C ended up exceeding that of the fresh sample by

PAGE 35

! 25 20% which continues to show the progressive nature of this mechanical property with respect to increased curing temperatures and cure rates. Peculiarly, the rubbery modulus didn't continue to grow as it previously had with the other cure times. The stronge st E r returned from this sample, 2.924 MPa after 6 hours, only reached 67% of the rubbery modulus for the fresh sample. In the previous samples the polymerization rate increased as the curing temperature increased, leading to the rubbery modulus of each sa mple to increase and approach the E r of the fresh sample. Perhaps there is a threshold for this relationship to be studied in more depth. (a) (b) (c) (d)

PAGE 36

! 26 (e) (f) (g) Figure 15 : The evolution of storage modulus and tan delta at each hour interval (a) Ð (e) and all tan delta (f) and all storage modulus (g) produced by dynamic mechanical analysis for the 200 ¡C cure. Figure 16 shows the relationship between T g and E r by cure time for each of the thin film samples. Not ice that the glass transition temperature tends to be higher for samples cured at higher temperature and these samples also take less time to cure. The average glass transition temperature for the 160, 180, and 200 ¡C cures is 12.86047, 15.80162, and 24.54 69 ¡C respectively. This continues to show that the repolymerization rate increases with higher curing temperatures and during this process the mechanical property approaches that of the fresh sample. During the 200 ¡C cure the glass transition temperature approached and even surpassed

PAGE 37

! 27 the 27.52858 ¡C of the fresh sample and if another cure at 220 ¡C were to take place, the average T g of the sample may be at or higher than the fresh sample. The average rubbery modulus for the 160, 180, and 200 ¡C cures is 0 .94997, 3.18664, and 1.99779 MPa respectively. From the 160 ¡C to the 180 ¡C samples the rubbery modulus approached the fresh sample's E r of 4.35067 MPa before regressing at the final cure of 200 ¡C . Generally, the trend seen in this figure shows that T g a nd E r both simultaneously increase or decrease, with the exception of a few instances where one will increase while the other will decrease. These instances are isolated and may be considered anomalies along with the regression in the average rubbery modulus from the 180 ¡C to 200 ¡C cure. ! Figure 16 : Glass transition temperature and rubbery modulus for 160 ¡C, 180 ¡C, and 200 ¡C cures of thin film. 4.7 Results of Dynamic Mechanical Analy sis on Bulk Sample The dynamic mechanical analysis testing was done on three bulk samples, A, B, and C. Samples A and B were cured first at 200 ¡C before being tested , followed by sample C , which cured at 180 ¡C before testing. The goal of the bulk sample study is to show change in crosslinking density from the top layer (layer 1) , where the crosslinking density is highest, to the bottom of the bulk sample (layer 4), where the crosslinking is least dense . This is accomplished by recording the

PAGE 38

! 28 glass transiti on temperature and rubbery modulus of the epoxy at each layer. Sample A was created by decomposing 24.52g of epoxy in ethylene glycol and reducing the decomposed polymer solution until the epoxy constituted 59% of the mixture's mass (the mass of ethylene glycol in the solution was 17.13g) before the mixture was poured into the mold and cured for 15 hours at 200 ¡C. After 15 hours a piece of the bulk epoxy was removed from the mold but only two layers could be produced from the bulk sample because it was st ill too soft. Dynamic mechanical analysis testing was performed on the samples and the material properties from the two layers can be seen in Table 5. The tan delta and storage modulus for layer 1 and layer 2 can be seen in Figures 1 7 a and 1 7 b respectively . Both layers have their tan delta and storage modulus compared to each other's in Figures 1 7 c and 1 7 d respectively. Table 5 : Material properties for Sample A, layers 1 and 2, 15 hour cure. Layer 1 Layer 2 Tan Delta 1.03118 1.183961 T g (¡C) 25.10708 9.748312 E rubbery (MPa) 0.6183148 0.02265562 ! !

PAGE 39

! 29 ! (a) ! (b) ! (c) ! (d) Figure 17 : Storage modulus and tan delta for layer 1 (a) and layer 2 (b). The tan delta compared for both layers (c) and the storage modulus compared for both layers (d). Produced by dynamic mechanical analysis for Sample A, layers 1 and 2, 15 hour cure at 200 ¡C. While Sample A didn't produce the desired number of layers, the goal of studying it was still reached. It can be observed that the T g of layer 1 is more than twice that of layer 2, while the E r for layer 1 is 27 times greater than layer 2. This is the first physical evidence to demonstrate increased crosslinking density in the system at the top layer. Figure 18 shows the rubbery modulus and glass transition temperature by layer.

PAGE 40

! 30 (a) (b) Figure 18 : (a) Rubbery modulus for layers 1 and 2 of Sample A. (b) Glass transition temperature for layers 1 and 2. Sample B was created with 18.63g of epoxy which was decomposed in ethylene glycol solvent and then reduced until th e decomposed polymer solution consisted of 59% decomposed epoxy by mass (the mass of the ethylene glycol in the solution was 12.88g). This time the sample was cured for 18 hours, again at 200 ¡C, before a portion of the bulk sample was removed and cut into 4 layers. The results of the dynamic mechanical analysis for Sample B after 18 hours of curing can be seen in Table 6. The tan delta and storage modulus behavior for each layer can be seen in Figures 19a through 19d. Figures 19e and 19f show comparisons o f tan delta for all layers and storage modulus for all layers respectively. Table 6 : Material properties for Sample B, layers 1 4, 18 hour cure. Layer 1 Layer 2 Layer 3 Layer 4 Tan Delta 1.185251 1.323853 1.007943 1.363466 T g (¡C) 34.22455 19.26777 21.34405 18.00127 E rubbery (MPa) 0.3022709 0.06155862 0.07846831 0.01455358

PAGE 41

! 31 (a) (b) (c) (d) (e) (f) Figure 19 : The evolution of storage modulus and tan delta from layers 1 (a) though 4 (e) and all tan delta (e) and all storage modulus (f). Produced by dynamic mechanical analysis for Sample B, layers 1 4, 18 hour cure at 200 ¡C.

PAGE 42

! 32 Sample B produced 4 more layers after curing an additional 3 hours for a total cure time of 21 hours. The DMA measurements from these layers can be seen in Table 7. Figures 20a though 20d indicate the evolving behavior for storage modulus and tan delta for each layer. Tan delta for all l ayers and the storage modulus for all layers can be seen in F igures 20e and 20f respectively. Table 7 : Material properties for Sample B, layers 1 4, 21 hour cure Layer 1 Layer 2 Layer 3 Layer 4 Tan Delta 1.123647 1.242308 1.163087 1.197856 T g (¡C) 36.27581 20.208 20.80494 18.40949 E rubbery (MPa) 2.023795 0.1287223 0.1154176 0.05926586 ! ! (a) ! (b)

PAGE 43

! 33 (c) (d) (e) (f) Figure 20 : The evolution of storage modulus and tan delta from layers 1 (a) though 4 (e) and all tan delta (e) and all storage modulus (f). Produced by dynamic mechanical analysis for Sample B, layers 1 4, 21 hour cure at 200 ¡C. The data from the 18 hour cure for Sample B also indicates that the goal of the bulk sample study was reached. With the exception of layer 3, the T g and E r decreased with each layer. Layer 3 produced a 2 ¡C increase in T g over layer 2 as well as a 17 kPa increase in E r . Layer 4 produced lo wer T g and E r values than layer 2 which suggests layer 3 was an anomaly to the study. The glass transition temperature of layer 1 is twice that of layer 4 while layer 1 also has an E r almost 21 times larger than layer 4. This shows that the rubbery modulus decreases at a significantly higher rate than the glass transition temperature over the whole depth of the bulk

PAGE 44

! 34 epoxy sample. The anomaly notwithstanding, Sample B continues to demonstrate increased crosslinking density at the higher layers. It should als o be noted that as T g decreases from layer to layer, so does E r , showing that these two mechanical properties for the epoxy are tied together. Figure 21 shows the T g and E r behavior for the 18 hour cure and compares it to the 21 hour cure. The 21 hour cure of Sample B goes further to illustrate the relationship between the layer of the bulk epoxy sample and the material's crosslinking density, where the higher layers in this cure displayed T g and E r values that were greater than the lower layers. With this cure the T g and E r both decrease from layer to layer (with the exception of a 0.6 ¡C increase in T g from layers 2 to 3 which is even less of an anomaly) and T g decreases by half from layer 1 to layer 4. E r in layer 1 is 34 times greater than the rubbery modulus in layer 4. The decrease in T g and E r from layer 1 to layer 4 is consistent with the findings in the 18 hour cure and continues to emphasize the difference in strength between the top of the bulk sam ple, where it's directly exposed to the heating environment, and the lower layers due to crosslinking density. This behavior can be seen in Figure 21. (a) (b) Figure 21 : (a) Rubbery modulus for Sample B at 18 and 21 hour cure times for all 4 layers. (b) Glass transition temperature for Sample B at 18 and 21 hour cure times for all 4 layers.

PAGE 45

! 35 Sample B was also cured at 180 ¡C but without success at the ratio of 59% epoxy to 41% ethylene glycol solvent by mass. Only the portion of decomposed polymer solution exposed to the air at the top of the mold would cure at this temperature, creating a lay er of repolymerized epoxy at the top of the mold while the decomposed epoxy solution under the cured portion remained a viscous liquid. To successfully repolymerize the decomposed epoxy solution in a bulk sample at 180 ¡C, it was decided to increase the co ncentration of the epoxy in the mixture, thus creating Sample C. Sample C contained 16.97g of epoxy that was decomposed in the ethylene glycol solvent and then reducing the mixture until it contained 73.75% epoxy by mass (leaving the remaining 26.25% of th e mixture to be 6.04g of solvent). This sample was first cured for 25 hours before a portion of the bulk sample could be removed and cut into 4 testable layers. The results of the dynamic mechanical analysis for Sample C after 25 hours of curing can be see n in Table 8. Figures 2 2 a through 2 2 d show the relationship between the storage modulus and tan delta in each of the four layers. The tan delta for all layers and storage modulus for all layers can be seen in Figures 2 2 e and 2 2 f. Table 8 : Material properties for Sample C, layers 1 4, 25 hour cur e Layer 1 Layer 2 Layer 3 Layer 4 Tan Delta 1.234807 1.179738 1.234899 1.27985 T g (¡C) 27.98791 25.42427 14.43008 9.267545 E rubebry (MPa) 1.002956 0.3832622 0.1780851 0.06788154

PAGE 46

! 36 (a) (b) (c) (d) (e) (f) Figure 22 : The evolution of storage modulus and tan delta from layers 1 (a) though 4 (e) and all tan delta (e) and all storage modulus (f). Produced by dynamic mechanical analysis for Sample C, layers 1 4, 25 hour cure at 180 ¡C.

PAGE 47

! 37 To show the development in the beha vior of the tan delta and storage modulus, Sample C was cured for another 5 hours for a total of 30 hours at 180 ¡C. Another portion of the bulk sample was removed and yielded 4 layers for dynamic mechanical analysis testing. The measurements recorded from the DMA can be seen in Table 9 while the relationship between the storage modulus and the tan delta for each la y er can be seen in Figures 23a though 23d. Comparisons of all tan delta and all storage modulus can be seen in Figures 23e and 23f. Table 9 : Material properties for Sample C, layers 1 4, 30 hour cure Layer 1 Layer 2 Layer 3 Layer 4 Tan Delta 1.212818 1.352075 1.448953 1.331817 T g (¡C) 28.32718 13.13534 10.54569 12.02929 E rubbery (MPa) 1.30335 0.2503388 0.1186616 0.2800871 (a) (b)

PAGE 48

! 38 (c) (d) (e) (f) Figure 23 : The evolution of storage modulus and tan delta from layers 1 (a) though 4 (e) and all tan delta (e) and all storage modulus (f). Produced by dynamic mechanical analysis for Sample C, layers 1 4, 30 hour cure at 180 ¡C. Sample C continued the trend of T g and E r values decreasing from the top layer to the subsequent layers. The 25 hour cure showed layer 1 with a T g 3 times higher than layer 4 while E r for layer 1 was almost 15 times larger than layer 4. The glass transition temperature showed a wider range for Sample C than Sample B while the rubbery modulus did not have a range as wide, suggesting that the decreased curing temperature of Sample C allowed a more even cure throughout the sample. This can be contributed to the top layer not repolymerizing befo re solvent in the lower layers could volatize from the overall bulk sample. In the curing of Sample B the top

PAGE 49

! 39 layer repolymerized quick enough that remaining solvent in the lower layers was trapped. Since both T g and E r decreased with each layer for the 25 hour cure of Sample C, this cure also shows that crosslinking density is greater at the top layers. Figure 24 shows the T g and E r behavior for the 25 hour cure and compares it to the 30 hour cure for Sample C. The 30 hour cure for Sample C returned anoth er anomaly, this time in layer 4, where the T g increased by 1.5 ¡C and E r more than doubled from the values of layer 3. Aside from layer 4, the other layers progressed as expected where the highest T g and E r were recorded for layer 1 and these values decre ased over each subsequent layer. The T g is almost 3 times higher in layer 1 than layer 3 and E r is 11 times greater, showing that in the final cure for the bulk epoxy samples the crosslinking density is greater at the top of the sample, where it's directly exposed to the heating environment. Both the 25 and 30 hour cures began with a lower T g than either of the cures at 200 ¡C which also enforces the concept that higher curing temperatures create higher repolymerization rates, which was noted in the cures f or the thin film sample. (a) (b) Figure 24 : (a) Rubbery modulus for Sample C at 25 and 30 hour cure times for all 4 layers. (b) Glass transition temperature for Sample C at 25 and 30 hour cure times for all 4 layers.

PAGE 50

! 40 CHAPTER V CONCLUSION Solvent assisted decomposition and repolymerization of covalent adaptable network polymers can be utilized to fully recycle thermosets. In this thesis, we investigated and predicted the kinetics and mechanics of repolymerization of CAN s. To link the property changes with the chemical reaction, the degree of conversion of monomers was used as the internal variable to describe the curing system. The glass transition temperature and the viscosity evolution were linked with the degree of co nversion of monomers. A multi branch model was used to capture the viscoelastic properties of the re cured polymer at different curing times. The re cured polymer at different curing states were made by controlling the heating time. The degree of monomer c onversion, viscosity, glass transition temperature, and dynamic mechanical properties were measured for different re curing states. The results indicated that the model could capture these properties' changes during curing. The findings in this thesis prov ide fundamental understanding of the repolymerization technique of CANs, which will promote its immediate engineering application in thermosets and composites recycling.

PAGE 51

! 41 BIBLIOGRAPHY ! 1. Conroy, A., S. Halliwell, and T. Reynolds, Composite recycling in the construction industry. Composites Part A: Applied Science and Manufacturing, 2006. 37 (8): p. 1216 1222. 2. Hartt, G.N. and D.P. Carey, Economics of recycling thermosets . 1992, SAE Technical Paper. 3. Kouparitsas, C., et al., Re cycling of the fibrous fraction of reinforced thermoset composites. Polymer composites, 2002. 23 (4): p. 682 689. 4. Ogi, K., et al., Mechanical properties of ABS resin reinforced with recycled CFRP. Advanced Composite Materials, 2007. 16 (2): p. 181 194. 5. Palmer, J., et al., Successful closed loop recycling of thermoset composites. Composites Part A: Applied Science and Manufacturing, 2009. 40 (4): p. 490 498. 6. Pickering, S.J., Recycling technologies for thermoset composite materials Ñ current status. Co mposites Part A: applied science and manufacturing, 2006. 37 (8): p. 1206 1215. 7. Adzima, B.J., et al., Rheological and chemical analysis of reverse gelation in a covalently cross linked Diels& Alder polymer network. Macromolecules, 2008. 41 (23): p. 9112 9117. 8. Bowman, C.N. and C.J. Kloxin, Covalent adaptable networks: reversible bond structures incorporated in polymer networks. Angewandte Chemie International Edition, 2012. 51 (18): p. 4272 4274. 9. Kloxin, C.J., et al., Covalent adaptable networks (CA Ns): a unique paradigm in cross linked polymers. Macromolecules, 2010. 43 (6): p. 2643 2653. 10. Tasdelen, M.A., Diels Ð Alder "click" reactions: recent applications in polymer and material science. Polymer Chemistry, 2011. 2 (10): p. 2133 2145. 11. Wojtecki , R.J., M.A. Meador, and S.J. Rowan, Using the dynamic bond to access macroscopically responsive structurally dynamic polymers. Nature materials, 2011. 10 (1): p. 14. 12. Chen, X., et al., New thermally remendable highly cross linked polymeric materials. Macromolecules, 2003. 36 (6): p. 1802 1807. 13. Zhang, Y., A.A. Broekhuis, and F. Picchioni, Thermally self healing polymeric materials: the next step to recycling thermoset polymers? Macromolecules, 2009. 42 (6): p. 1906 1912.

PAGE 52

! 42 14. Taynton, P., et al., Repairable woven carbon fiber composites with full recyclability enabled by malleable polyimine networks. Advanced Materials, 2016. 28 (15): p. 2904 2909. 15. Yu, K., et al., Carbon fiber reinforced thermoset composite with near 100% recyclability. Advance d Functional Materials, 2016. 26 (33): p. 6098 6106. 16. Buback, M., et al., Kinetics and modeling of free radical batch polymerization of nonionized methacrylic acid in aqueous solution. Industrial & Engineering Chemistry Research, 2008. 47 (21): p. 8197 8 204. 17. Zhao, Z., et al., Desolvation induced origami of photocurable polymers by digit light processing. Macromolecular rapid communications, 2017. 38 (13): p. 1600625. 18. Zarrelli, M., A.A. Skordos, and I.K. Partridge, Toward a constitutive model for cure dependent modulus of a high temperature epoxy during the cure. European Polymer Journal, 2010. 46 (8): p. 1705 1712. 19. Gan, S., J. Seferis, and R. Prime, A viscoelastic description of the glass transition conversion relationship for reactive polymer s. Journal of thermal analysis, 1991. 37 (3): p. 463 478. 20. Dewaele, M., et al., Volume contraction in photocured dental resins: the shrinkage conversion relationship revisited. Dental Materials, 2006. 22 (4): p. 359 365. 21. Wu, J., et al., Evolution of material properties during free radical photopolymerization. Journal of the Mechanics and Physics of Solids, 2018. 112 : p. 25 49. 22. O'Brien, D.J., P.T. Mather, and S.R. White, Viscoelastic properties of an epoxy resin during cure. Journal of composite materials, 2001. 35 (10): p. 883 904. 23. Long, K.N., M.L. Dunn, and H.J. Qi, Mechanics of soft active materials with phase evolution. International Journal of Plasticity, 2010. 26 (4): p. 603 616. 24. Parameswaranpillai, J., et al., Investigation of Cure Reaction, Rheology, Volume Shrinkage and Thermomechanical Properties of Nano TiO2 Filled Epoxy/DDS Composites. Journal of Polymers, 2013. 2013 : p. 183463. 25. Vijayan, P.P., J. Pionteck, and S. Thomas, Volume Shrinkage and Cure Kinetics in Carboxyl Termin ated Poly(butadiene co acrylonitrile) (CTBN) Modified Epoxy/Clay Nanocomposites. Journal of Macromolecular Science Part a Pure and Applied Chemistry, 2015. 52 (5): p. 353 359. 26. Daoud, M., Viscoelasticity near the sol gel transition. Macromolecules, 2000 . 33 (8): p. 3019 3022.

PAGE 53

! 43 27. Rubinstein, M. and R.H. Colby, Polymer physics . 2003, New York: Oxford University Press, Oxford. 28. Yu, K., et al., Dissolution of covalent adaptable network polymers in organic solvent. Journal of the Mechanics and Physics of Solids, 2017. 109 : p. 78 94. 29. Rubinstein, M. and R.H. Colby, Polymer physics . Vol. 23. 2003: Oxford University Press New York. 30. Di Marzio, E.A. and A.J. Yang, Configurational entropy approach to the kinetics of glasses. Journal of research of the National Institute of Standards and Technology, 1997. 102 (2): p. 135. 31. Williams, M.L., R.F. Landel, and J.D. Ferry, The temperature dependence of relaxation mechanisms in amorphous polymers and other glass formin g liquids. Journal of the American Chemical society, 1955. 77 (14): p. 3701 3707. 32. DiBenedetto, A., Prediction of the glass transition temperature of polymers: a model based on the principle of corresponding states. Journal of Polymer Science Part B: Po lymer Physics, 1987. 25 (9): p. 1949 1969. 33. Nielsen, L.E., Cross linking Ð effect on physical properties of polymers. Journal of Macromolecular Science, Part C, 1969. 3 (1): p. 69 103. 34. Fick, A., V. On liquid diffusion. The London, Edinburgh, and Dubli n Philosophical Magazine and Journal of Science, 1855. 10 (63): p. 30 39.