rheological and laser additives for higher efficiency in

Christian Herbert

Rheological and LASER additives for higher efficiency in

producing poly(acrylonitrile)-based carbon fibers

Rheological and LASER additives for higher

efficiency in producing poly(acrylonitrile)-based

carbon fibers

Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der

RWTH Aachen University zur Erlangung des akademischen Grades eines

Doktors der Naturwissenschaften genehmigte Dissertation

vorgelegt von

Diplom-Chemiker

Christian Herbert

aus Heltau, Rumänien

Berichter: Universitätsprofessor Dr. rer. nat. Martin Möller

Universitätsprofessor Dr. rer. nat. Andrij Pich

Tag der mündlichen Prüfung: 05. April 2016

Diese Dissertation ist auf den Internetseiten der Universitätsbibliothek online verfügbar.

1

Table of Contents

List of Abbreviations 2

Summary 4

Zusammenfassung 5

Chapter 1 Introduction 6

Chapter 2 Development of a poly(acrylonitrile) copolymer

based carbon fiber precursor 24

Chapter 3 Influence of branched poly(acrylonitrile) additives

on the viscosity of poly(acrylonitrile) based

carbon fiber precursor solutions 47

Chapter 4 Characterization of absorption additives for

direct energy transfer by LASER light into

poly(acrylonitrile) based carbon fiber precursor 79

Chapter 5 Wet spinning of PAN-based fiber precursor with

hyperbranched PAN as rheological additive 114

List of Publications 130

Acknowledgment 131

Curriculum Vitae 132

2

List of Abbreviations

Å Angstrom

AFM atom force microscopy

AIBN azobisisobutyronitrile

CF carbon fiber

CNT carbon nano tubes

Đ dispersity of molar mass distribution

DMF N,N’-dimethylformamide

DSC differential scanning calorimetry

EC ethylene carbonate

η dynamic viscosity

[η] intrinsic viscosity

ηr relative viscosity

ηc specific viscosity

shear rate

FRP free radical polymerization

FTIR fourier transform infrared

IA itaconic acid

IR infrared

LASER light amplification by stimulated emission of radiation

MA methyl acrylate

MAA methacrylic acid

Mn number average molecular weight

Mw weight average molecular weight

PAN poly (acrylonitrile)

PMMA poly (methyl methacrylate)

RAFT reversible addition fragmentation chain transfer polymerization

3

SEC size exclusion chromatography

t time

T temperature

Tp peak temperature

TEM transmission electron microscopy

TGA thermogravimetric analysis

VA vinyl acetate

wt.-% weight percent

4

Summary

This work is based on the NRW Ziel2 ‘Megacarbon’ project which aims for the more resource

efficient production of carbon fibers (CF) for the automotive market. In cooperation with the Dralon

GmbH in Dormagen a CF precursor with properties at least equal to the industry reference fiber

Bluestar was developed and used in fiber spinning experiments.

For the improvement of the spinning process a hyperbranched, rheological additive was synthesized

for the decrease of dynamic viscosity over a broad sheer rate range. The decreased dynamic viscosity

would allow for a higher amount of polymer percentage in the spinning solution resulting in a higher

density of the fiber and less usage of the solvent DMF.

For the energy cost reduction of CF production thermal treatment with near infra-red (NIR) LASER

for direct energy transfer into the precursor fiber has been examined. To achieve the required control

of the thermal treatment a defined amount of NIR absorbing carbon black particles was

homogeneously dispersed in a polymer solution.

The synthesized hyperbranched, rheological additive was successfully wet-spun with a PAN-based

polymer in DMF at the laboratory spinning setup at Dralon.

5

Zusammenfassung

Diese Arbeit wurde im Rahmen des NRW Ziel2 Projektes „Megacarbon“ durchgeführt, welches die

ressourcen-effizientere Herstellung von Carbon-Fasern für den Automotive Markt zum Ziel hat. In

Zusammenarbeit mit der Dralon GmbH in Dormagen wurde ein Carbon-Faser Precursor hergestellt

mit mindestens vergleichbaren Eigenschaften zu der Industrie-Faser Bluestar.

Für die Verbesserung des Spinnprozesses wurde ein stark verzweigtes, rheologisch wirksames Additiv

synthetisiert für die Reduktion der dynamischen Viskosität über einen breiten Scherratenbereich. Die

verringerte dynamische Viskosität würde einen höheren Polymeranteil in der Spinnlösung erlauben,

wodurch weniger Lösungsmittel DMF notwendig wäre und eine dichtere Faser entstehen würde.

Der direkte Energieeintrag mittels Nah-Infrarot-LASER in die Precursorfaser wurde untersucht zur

Senkung der Heizkosten in der Carbon-Faser-Herstellung. Zur Untersuchung der LASER-Absorption

wurden kleinen Mengen von Nah-Infrarot absorbierenden Rußpartikel homogen in einer

Polymerlösung dispergiert.

Das synthetisierte, stark verzweigte, rheologisch wirksame Additiv wurde als Zusatz in einer PAN-

Copolymer Spinnlösung in der Laborspinnanlage von Dralon erfolgreich naß versponnen.

CHAPTER 1

6

Introduction

1.1 Carbon fibers

1.1.1 Overview

The first commercially successful fibers based on carbon were applied by Edison and Swan in 1881

for the invention of carbon filament light bulb. The fibers from Japanese bamboo were pyrolyzed to

form the carbon filament.1 But the carbon filament was outperformed by metal filaments in the early

20th century and fibers based on carbon (carbon fibers, CF) only became of interest again to the

aerospace industry in the 1950s.2 Carbon fibers can have better mechanical properties compared to

other materials like steel. The CFs offer higher tensile strength, higher Young’s modulus, higher

thermal stability and higher pressure resistance at lower density than steel (density ρCF(1.8 g/cm3) <

ρsteel(7.9 g/cm3). In the form of composite materials CF are essential for the aerospace market. Through

the reduction in weight fuel consumption can be reduced and the resulting carbon dioxide emission.

Due to the relatively high price of CFs manufacturers in the automotive market are still reluctant to

use CF composites on a big scale. As of 2014 only the car BMW i3 has been produced where some

parts the chassis and the passenger cell are made of CF composite materials.3

The carbon fibers on the market are based on the three starting materials poly(acrylonitrile), cellulose

and pitch. Poly(acrylonitrile) (PAN) is with 86% market share by far the most used precursor

material.4 Fibers made from pitch are today only used for CF with very high Young’s modulus due to

the high price of 100 – 3,000 $/kg.5 Although the carbon yield of pitch-based fibers is nearly double

than the carbon yield of PAN-based CF the process management of PAN-based CF costs less than for

pitch-based CF.6

The PAN-based carbon fibers can be categorized through their mechanical properties into three groups

with different applications and prices (see Table 1). 2,7,8

Depending on the production process specific

types of CF can be obtained. The standard modulus CF with a tensile strength of 3,500 – 5,000 MPa

costs between 15 – 30 $/kg and is primarily applied in aerospace applications and in high performance

CHAPTER 1

7

sporting goods. Intermediate modulus CF with tensile strengths of 4,800 – 6,400 MPa are more

expensive than standard modulus CF with 30 – 60 $/kg. The application is in a broad range of

industries where lightweight and tensile strength-critical properties are required such as pressure

vessels for aerospace vehicles, and satellites, as well as hydrogen storage tanks for fuel cell vehicles.

The high modulus CF with a Young’s modulus of 340 – 590 GPa is the most expensive category with

50 – 120 $/kg and is often used in stiffness and coefficient of thermal expansion-critical applications

such as premium sporting goods, industrial rollers and in spacecraft.

Table 1 PAN-based carbon fibers grouped by mechanical properties and price.2,7,8

Category Tensile strength [MPa] Young’s modulus [GPa] Price [$/kg]

Standard modulus (SM) 3,000 – 5,000 200 – 250 15 – 30

Intermediate modulus (IM) 4,500 – 6,000 250 – 350 30 – 60

High modulus (HM) 2,500 – 4,700 340 – 590 50 – 120

Even with all the advantages it is necessary to recognize that CFs are an expensive product. When the

production of CF in 2012 with 67,000 t/year is compared to the yearly production of glass fibers in

2012 with 4,330,000 t/year the niche roll of CF becomes obvious. Even with the estimated demand of

CF in 2020 with 320,000 t the price is still the essential factor.9

Recent research focuses on the starting material lignin from plant material10,11,12

and melt-spinning of

precursors13,14

. The mechanical properties of lignin-based CF have yet to be improved significantly for

market competition with PAN-based CFs.15

1.1.2 Production of PAN-based carbon fibers

Production of PAN-based CF starts with the preparation of a PAN copolymer, its dissolution in a

suitable solvent, the spinning of a precursor fiber under stretching, the heat treatment under stretching

and the finishing of the fibers (see Figure 1).

CHAPTER 1

8

Figure 1 Process steps in the production of PAN-based CF.

1.1.2.1 Polymerization of acrylonitrile

Technical polymerization of acrylonitrile (AN) and copolymerization with other comonomers is done

by free radical polymerization (FRP).16

FRP offers control on the polymerization rate (by adjusting the

temperature), a rough control of the molecular weight (MW) and the incorporation of comonomers.

The typical suspension polymerization is performed in water. The resulting polymer with a molecular

weight as high as Mw = 350,000 g/mol for the AN homopolymer is insoluble in water and therefore

precipitates allowing for simple separation from the reaction medium by filtering or skimming. The

polymerization initiators can be (i) azo-based organic compounds such as azobisisobutyronitrile

(AIBN) and 2,2-azo-2,4-dimethylvaleronitrile, (ii) peroxo compounds such as dibenzyl peroxide

(DPO), (iii) redox pair systems composing a persulfate and a bisulfite, (iv) electromagnetic γ-radiation

from radioactive decay. Generally low MW PAN with Mw = 50,000 – 80,000 g/mol is employed for

textile grade fibers whereas high MW PAN with Mw >120,000 g/mol is used for precursor fibers for

CF with high tensile strength and Young’s modulus.

An acrylic fiber consists at least of 85% acrylonitrile (AN) and with a content of AN between 85% and

35% the fibers are called modacrylic fibers.16

CF precursor fibers consisting of AN homopolymer

Copolymerization of acrylonitrile with specific comonomers

Dissolution of PAN copolymer in suitable solvent

Spinning of the polymer solution under stretching

Heat treatment under stretching

Fiber finishing

CHAPTER 1

9

result in CF with poorer mechanical properties than AN copolymer precursors having a low

percentage of comonomer.17

The comonomers have an influence on the solubility of the copolymer, its

spinnability and thermal as well as mechanical properties. Typically vinyl comonomers are used

having acidic side groups (see Table 2). Methyl acrylate (MA) and methyl methacrylate (MMA)

improve solubility in organic solvents such as N,N’-dimethylformamide (DMF), dimethyl sulfoxide

(DMSO) and N,N’-dimethyl acetamide (DMAc) and influence the resulting fiber morphology of the

spun precursor. Itaconic acid (IA) and methacrylic acid (MAA) reduce the initiation temperature of the

chemical reactions during heating as well as lower the exothermic enthalpy of the reactions. With its

two carboxylic groups IA has the strongest effect on the initiation temperature. Typical formulations

of copolymers consist of 90-94% AN, 6-9% MA and around 1% IA.18,19

Table 2 Different comonomers for the CF precursor material.

Comonomer Chemical structure

Methyl acrylate (MA)

Methyl methacrylate (MMA)

Itaconic acid (IA)

Methacrylic acid (MAA)

CHAPTER 1

10

1.1.2.2 Spinning under stretching

The mechanical properties of the CFs are dependent on the precursor fiber and the way it is produced.

Acrylic fibers can be spun using different techniques such as wet, dry and dry-jet wet spinning where

wet spinning is the most widely used technique.20

The wet spinning of fibers can be described by an extrusion of polymer solution through a multi-hole

spinneret which is submerged in a coagulation bath. It is common that multiple coagulation baths are

successively used with decreasing solvent concentration. By adjusting the coagulant/solvent ratio the

intensity of coagulation processes such as solvent exchange by diffusion can be controlled, resulting in

the intended pore sizes and distributions. The fiber orientation and crystallinity are influenced by the

stretching of the fiber, defined through the draw ratio. The shape of the fiber cross-section is

determined by the coagulation process, too. Residual solvents or salts that remain in the fiber

contribute to the discontinuity in the ladder structure during thermal treatment, causing thread

breakage. The comparison of wet-spun acrylic fibers with dry-spun acrylic fibers shows that the

former have a circular cross section whereas the latter have a bone shaped cross section. CF with

round cross section offer no weak spot to mechanical stresses as a bone shaped CF would. This

explains the preference of the wet spinning technique for the carbon fiber production. Wet spinning of

the dope is usually done with ~ 15 – 20 wt.-% polymer content and dry spinning with higher solid

contents of ~ 25 – 30 wt.-%.21

The dry spinning technique utilizes no coagulation bath but evaporates the solvent by passing the as-

spun fiber through hot chambers. This allows higher spinning rates and is often applied in the

production of textile fibers.

Dry-jet wet spinning combines the advantages of the before mentioned techniques. Here an air gap

between the multi-hole spinneret and the coagulation bath is applied. The separation through the gap

allows the extruded polymer solution to cool and relax the high stresses developed inside the multi-

hole spinneret before entering into the coagulation bath. Therefore, higher spinning speed and higher

stretching ratios can be applied compared to the wet spinning process.22

CHAPTER 1

11

In summary, with the concentration of solvent in coagulation bath, the temperature of the bath, fiber

residence time in bath (by controlling extrusion rate, length of bath and winding speed) and the nozzle

diameter the morphology of the resulting fiber is influenced and in conclusion its properties.

1.1.2.3 Heat treatment under stretching

The spooled PAN precursor is subjected to a specifically tailored heating program under stretching.

The heating process can be separated into three distinct heating steps, namely (i) stabilization, (ii)

carbonization and (iii) graphitization.

The first step is the stabilization of the PAN-based precursor taking place in normal air atmosphere

between 200 and 300 °C. Time of residence can take up to two hours making the stabilization the most

time-consuming step in the CF production. Although not all details of the stabilization have been

completely understood, the existence of three exothermic reactions is accepted: cyclization,

dehydrogenation and oxidation (see Figure 2).23

In 1986 Fitzer and coworker described the cyclization

and dehydrogenation as two competing reactions.24

They proceed parallel but are not finished before

the oxidation reactions starts. Further reactions lead to the evolution of gaseous products such as

carbon dioxide, hydrogen cyanide, water, carbon monoxide and ammonia. In the cyclization reaction

the nitrile group reacts with the neighboring nitrile group and forms a hexagonal ring structure with

five carbon atoms and one nitrogen atom. The resulting structure is stable against heat and a template

for the final CF. For the cyclization reaction no oxygen is needed. Irregularities in the polymer chains

such as comonomers, initiator residues and branching lead to un-cyclized parts which can degrade into

products such as hydrogen cyanide and ammonia.

During the dehydrogenation reaction oxygen reacts with the hydrogen of the ethylene polymer

backbone to form water. The resulting double bond in the polymer backbone improves the stability of

the chain.25

In the oxidation reaction oxygen is incorporated into the cyclized structure which can lead to a

significant mass loss of carbon by evolution of carbon dioxide and monoxide. But the presence of

oxygen is required for the dehydrogenation reaction. The oxygen content in a precursor fiber after the

CHAPTER 1

12

stabilization process should be about 10%. High oxygen content leads to CF with inferior mechanical

properties.21

Figure 2 Model of chemical reactions of PAN in the temperature range of 200 - 300°C.28

CHAPTER 1

13

The stabilization reactions are highly exothermic. Most efficient in lowering the released reaction heat

is methyl acrylate through broadening the temperature range of the reaction and for lowering the

starting temperature of the cyclization reaction is itaconic acid through the acidic catalyzation of the

cyclization reactions.26

Through lower temperatures the heating costs of the stabilization process can

be reduced and lower reaction heat release decreases the danger of fiber fusion and ignition.

The carbonization follows the stabilization step. Taking place at temperatures of 600 – 1,200°C an

inert gas atmosphere is required so the fibers do not start burning. Often nitrogen is applied as an inert

gas. The heating in the carbonization step should (i) eliminate the non-carbon elements and therefore

increase the percentage of carbon in the fiber and (ii) form a turbostratic graphite structure (see Figure

3).27

Figure 3 Graphite (a) and turbostratic carbon structure (b) in PAN-based CF.27

CHAPTER 1

14

The carbonization process lasts only a few minutes but the mass loss in this step is ~ 40% due to the

formation of gaseous products.28

For the production of CF with very high Young’s modulus the graphitization process can follow the

carbonization. In inert gas atmosphere of argon the CF is heated over 2,000°C to remove even more

non-carbon elements. The process is usually applied to pitch based fibers to achieve a Young’s

modulus over 800 GPa, which is very close to the theoretical Young’s modulus of 1,050 GPa of the C-

C bonding.29

1.2 Carbon based functional additives

1.2.1 Overview

Polymers are applied in a wide field such as adhesion, lubrication, friction and wear, composites,

microelectronics and biotechnology.30,31,32,33,34

Through the modification of the surface of materials

their properties are changed and the interaction between the material and its environment can be

controlled.35,36,37

Achieving this modification with polymers is possible through physisorption and

covalent attachment. The covalent binding connects the polymer firmly with the material surface and

keeps the modification permanent, whereas physisorption can be reversed through desorption.

Many of the industrial polymers have linear polymer chains, but only with branching of the polymer

chains new physicochemical properties appear and allow innovative uses in coatings or resin

formulations e.g. as viscosity reducing agent. Branched polymers especially offer application as melt

modifiers for thermoplastic polyolefines, polyamides, or polyesters, as compatibilizers or as blend

components.38,39,40,41,42

The synthesis of these polymers can be achieved through controlled

polymerization techniques including cationic, anionic, ring-opening metathesis and controlled radical

polymerization.

1.2.2 Improved CF formation by specific additives

In material science polymer nano composites (PNCs) represent a key role for the development of

advanced materials.43,44,45

The incorporation of small amounts of PNCs into the polymer matrix yields

CHAPTER 1

15

exceptional changes of mechanical46

, optical47

, electrical48

and thermal49

properties. The large surface-

area-to-volume ratio of nano fillers is responsible for the changes of the properties. The earliest

example is the clay reinforced resin Bakelite from the beginning of the 20th century. In recent research

carbon nanotubes (CNT) and graphene have been functionalized with different polymers.50,51

With the

appropriate functionalities the graphene tailored PNC can be mixed as filler into the polymer matrix.

The PNC can be synthesized through e.g. covalent binding of the polymer to the graphene via the

“grafting-to” or “grafting-from” method. In the grafting-to method first the polymer chains are

prepared and then appended by direct covalent linkage through esterification, amidation, click

chemistry, radical addition, etc. to graphene or graphene modifications. The grafting-from method

starts with the polymerization of monomers from initiator sites on the surface of graphene. The

initiator sites consist of functionalities such as hydroxyl or carboxylic acid groups covalently bond on

the graphene. Since the diffusion of monomer is not strongly hindered by the grafted polymer chains,

the grafting-from process is often used with different types of controlled radical polymerization

procedures such as RAFT polymerization to achieve high grafting densities.52,53,54

For example the

preparation of PNIPAM/graphene nano composites starts with the diazotization of aryl diazonium salts

containing alkyne groups with graphene and is followed by the esterification reaction between 3-

azido-1-propanol and S-1-dodecyl-S’-(α,α’-dimethyl-α’’-acetic acid) trithiocarbonate producing azido-

terminated RAFT agent. Through click chemistry the RAFT agent is immobilized on the alkyne

derivative of graphene and in the last step NIPAm is polymerized from graphene sheets using the

RAFT technique (see Figure 4). The RAFT polymerization allows the controlled polymerization of

various comonomers under a wide range of conditions. The resulting polymers feature predetermined

MW, narrow dispersity and advanced architectures.55,56

The RAFT technique does not require a

catalyst and is performed under mild conditions.57

CHAPTER 1

16

Figure 4 Outline of the preparation of PNIPAM/graphene nano composites through click

chemistry and RAFT polymerization by grafting-from method.51

1.2.3 Carbon black

Produced from incomplete combustion of petroleum compounds carbon black has a black color and

consists of mainly carbon. In comparison with soot, carbon black offers a remarkably higher surface-

area-to-volume ratio and significantly lower content of polycyclic aromatic hydrocarbon.58

Carbon black as nano particle filler in polymer or rubber material has a broad field of application such

as (i) improving the strength and stiffness of tennis rackets or hockey sticks; (ii) improving on

strength, wear and abrasion ability in tires; (iii) improving electrical conductivity for e.g. packaging

with electrostatic dissipation.59

1.2.3.1 CNT

Carbon nano tubes (CNT) were first described by Iijima in 1991 in form of the multi-walled nano

tubes (MWCNT).60

In 1993 nearly at the same time Iijima and coworker61

and Bethune and

coworkers62

discovered single-walled nano tubes (SWCNT) during the experiment of filling MWCNT

with metal atoms. CNT are typically in the dimensions of 5-20 nm diameters and a few µm in length

(see Figure 5). The shape of the CNT is a cylinder where the carbon atoms are attached to each other

CHAPTER 1

17

through sp2 hybridized bonding. The CNT can be synthesized e.g. through chemical vapor disposition,

arc discharge and LASER light ablation. CNT are incorporated into fibers to improve the mechanical

properties and the electrical conductivity.63,64

a) b)

Figure 5 a) Structure of SWCNT; b) representative electron-microscopic image of MWCNT

Baytubes C150P.65

1.2.3.2 Graphene

Graphene is a class of two-dimensional carbon nanostructure (see Figure 6). The source material is

graphite, a stack of flat graphene sheets, and is highly available from both natural and synthetic

sources.

a) b)

Figure 6 a) Structure of graphene sheet; b) representative TEM image of graphene sheet with

average surface area of 750 m2/g.

66

CHAPTER 1

18

Graphene can be applied in many different fields such as nano electronics, sensors, nano composites,

batteries, super capacitors and hydrogen storage.67,68,69,70

The main obstacle for the realization of the

applications is the inefficient production of processible graphene sheets in large quantities. Graphene,

like CNT and other nano materials tend to aggregate making them difficult to process in bulk

quantities. The stacking of the graphene sheets through van-der-Waals interaction leads to stable

aggregates, which can only be broken up through ultrasonic treatment or mechanical dispersion.

1.2.3.3 Asphaltene

A common definition for asphaltene is the n-heptane-insoluble, toluene-soluble component of a crude

oil or carbonaceous material.71

The material can be produced in high quantities and low price. The

modified Yen model describes asphaltene as single, polycyclic aromatic hydrocarbon ring system with

peripheral alkyl substituents (see Figure 7). Through π-π-stacking these structures can form nano

aggregates where a stack consists of ~ six ring systems. The exterior of the nano aggregate presents

the alkyl substituents and therefore influences the interaction with solvents. Containing a high number

of alkyl substituents allows for a good dispersibility of asphaltene in organic solvents with no

additional modification necessary.

CHAPTER 1

19

Figure 7 Proposed asphaltene structures by the modified Yen model. Reprinted with permission

from Mullins, O. C. Energy and Fuels 2010, 24, 2179. Copyright 2010 American

Chemical Society.71

1.3 Content of chapters 2-5

This dissertation has been performed within the interdisciplinary NRW Ziel 2 project “Megacarbon”

with the aim of reducing carbon fiber prices for the automotive sector by a more efficient and highly

productive manufacturing. The project started in 2010 and is scheduled until spring of 2015.

In chapter 2 the development of a CF precursor with properties similar to a commercial CF precursor

with high tensile strength is described. For that the polymerization technique has been varied as well

as the formulation of the PAN-based precursor material. Polymer analysis was performed by size

exclusion chromatography and thermal analysis.

CHAPTER 1

20

Chapter 3 describes a concept and presents experimental results for polymers that reduce the dynamic

viscosity of concentrated polymer spinning solutions over a broad shear rate range. The rheologically

active polymers have been synthesized by RAFT polymerization and analyzed by capillary rheometry.

In chapter 4 results on carbon black modifications used as absorption pigments for the direct energy

transfer of LASER light into a precursor fiber are described. Solutions are presented that lead to

homogeneous dispersions of carbon black pigments in polymer solutions. The effect of the carbon

pigments during the heat treatment on the reaction enthalpy has been examined by thermal analysis.

In chapter 5 the first results of applicability of the newly synthesized rheological additive in a wet

spinning process with a highly concentrated polymer solution is presented. The ternary phase diagram

of the testing system of a commercial PAN-based textile precursor in a mixture of DMF and water has

been determined and the influence of the rheological additive on the cloud-point formation of the

ternary system has been estimated.

1.4 References

(1) Edison, T. A. Carbon for Electric Lamps. Patent No. 251540, 1881.

(2) Koch, P. A. Faserstoff-Tabellen nach P.-A. Koch; 1989.

(3) Sloan, J. , BMW Leipzig: the epicenter of i3 production,

http://www.compositesworld.com/articles/bmw-leipzig-the-epicenter-of-i3-production-

(accessed Jan 2, 2015).

(4) AVK - Industrievereinigung Verstärkte Kunststoffe e.V. Handbuch Faserverbundkunststoffe /

Composites; 4th ed.; Springer Vieweg, 2013.

(5) Schürmann, H. Konstruieren mit Faser-Kunststoff-Verbunden; Heidelberg: Springer: Berlin,

2005.

(6) Spengler, H.; van Galen, J. Herstellung von Kohlenstofffasern aus Steinkohlenteerpech; p.

1990.

(7) Jäger, H.; Hauke, T. Carbonfasern und ihre Verbundwerkstoffe; verlag moderne instudrie,

2010.

CHAPTER 1

21

(8) Toray Carbon Fibers, http://www.toraycfa.com/product.html (accessed Jan 2, 2015).

(9) Sloan, J. , Market Outlook: Surplus in Carbon Fiber’s Future?,

http://www.compositesworld.com/articles/market-outlook-surplus-in-carbon-fibers-future

(accessed Nov 20, 2014).

(10) Kadla, J. F.; Kubo, S.; Venditti, R. A.; Gilbert, R. D.; Compere, A. L.; Griffith, W. Carbon N.

Y. 2002, 40, 2913.

(11) Baker, D. a.; Rials, T. G. J. Appl. Polym. Sci. 2013, n/a.

(12) Braun, J. L.; Holtman, K. M.; Kadla, J. F. Carbon N. Y. 2005, 43, 385.

(13) Bortner, M. J. Melt processing of metastable acrylic copolymer carbon precursors, Blackburg,

VA, Virginia Polytechnic Institute and State University, 2003.

(14) Bajaj, P.; Sreekumar, T. V.; Sen, K. Polymer (Guildf). 2001, 42, 1707.

(15) Frank, E.; Steudle, L. M.; Ingildeev, D.; Spörl, J. M.; Buchmeiser, M. R. Angew. Chem. Int.

Ed. Engl. 2014, 2.

(16) Wade, B.; Knorr, R. Acrylic Fiber Technology and Applications; Masson, J. C., Ed.; Marcel

Dekker, Inc.: New York, 1995; pp. 37–67.

(17) Muller, T. Int. Fiber J. 1988, 46.

(18) Morgan, P. Carbon Fibers and their Composites; Taylor and Francis: London; New York;

Singapore, 2005.

(19) Grassie, N.; McGuchan, R. Eur. Polym. J. 1972, 8, 257.

(20) Capone, G. J. In Acrylic Fiber Tech. and Appl.; Masson, J. C., Ed.; Marcel Dekker, Inc.: New

York, 1994; p. 69.

(21) Gupta, A. K.; Paliwal, D. K.; Bajaj, P. Polym. Rev. 1991, 31, 1.

(22) Zhang, J.; Zhang, Y.; Zhang, D.; Zhao, J. J. Appl. Polym. Sci. 2011.

(23) Fitzer, E. Acta Polym. 1990, 41, 381.

(24) Fitzer, E.; Frohs, W.; Heine, M. Carbon N. Y. 1986, 24, 387.

(25) Donnet, J. B.; Wang, T. K.; Peng, J. C. M. Carbon fibers; 3. Ed.; Dekker: New York; Basel;

Hong Kong, 1998.

(26) Bajaj, P.; Sreekumar, T. .; Sen, K. Polymer (Guildf). 2001, 42, 1707.

(27) Hoffman, W. P.; Hurley, W. C.; Liu, P. M.; Owens, T. W. The surface topography of non-shear

treated pitch and PAN carbon fibers as viewed by the STM. Journal of Materials Research,

1991, 6, 1685–1694.

(28) Bajaj, P.; Roopanwal, a. K. J. Macromol. Sci. Part C Polym. Rev. 1997, 37, 97.

CHAPTER 1

22

(29) Flemming, M.; Ziegmann, G.; Roth, S. Faserverbundbauweisen: Fasern und Matrices;

Springer: Berlin; Heidelberg, 1995.

(30) Ikada, Y. Adv. Polym. Sci. 1984, 57, 103.

(31) Raphael, E.; De Gennes, P. G. J. Phys. Chem. 1992, 96, 4002.

(32) Kaplan, S. L.; Lopata, E. S.; Smith, J. Surf. Interface Anal. 1993, 20, 331.

(33) Parnas, R. S.; Cohen, Y. Rheol. Acta 1994, 33, 485.

(34) Klein, J. Shear, Friction, and Lubrication Forces Between Polymer-Bearing Surfaces. Annual

Review of Materials Science, 1996, 26, 581–612.

(35) Yan, D.; Gao, C.; Frey, H. Hyperbranched Polymers: Synthesis, Properties, and Applications;

Yan, D.; Gao, C.; Frey, H., Eds.; 2011.

(36) Tsujii, Y.; Ohno, K.; Yamamoto, S.; Goto, A.; Fukuda, T. Structure and properties of high-

density polymer brushes prepared by surface-initiated living radical polymerization. Advances

in Polymer Science, 2006, 197, 1–45.

(37) Zhao, B.; Brittain, W. J. Prog. Polym. Sci. 2000, 25, 677.

(38) Braig, T.; Joachimi, D.; Persigehl, P.; Van, M. R. High flow polymer compositions comprising

branched flow agents. EP1424360, 2003.

(39) Owens, B. A.; Jackson, N. E.; Collias, D. I. No Title. WO 03/097740, 2002.

(40) Grutke, S.; Gruber, F.; Voit, B.; Huber, T. Fließfähige Polyamide. DE 19953950, 1999.

(41) Jannerfeldt, G.; Booch, L.; Månson, J. A. E. J. Polym. Sci. Part B Polym. Phys. 1999, 37, 2069.

(42) Kim, Y. H. J. Polym. Sci. Part A Polym. Chem. 1998, 36, 1685.

(43) Ajayan, P. M.; Schadler, L. S.; Giannaris, C.; Rubio, A. Adv. Mater. 2000, 12, 750.

(44) Thostenson, E. T.; Ren, Z.; Chou, T.-W. Advances in the science and technology of carbon

nanotubes and their composites: a review. Composites Science and Technology, 2001, 61,

1899–1912.

(45) Sinha Ray, S.; Okamoto, M. Polymer/layered silicate nanocomposites: A review from

preparation to processing. Progress in Polymer Science (Oxford), 2003, 28, 1539–1641.

(46) Potts, J. R.; Dreyer, D. R.; Bielawski, C. W.; Ruoff, R. S. Polymer (Guildf). 2011, 52, 5.

(47) Matras-Postolek, M.; Bogdal, D. Adv. Polym. Sci. 2010, 230, 221.

(48) Sengupta, R.; Bhattacharya, M.; Bandyopadhyay, S.; Bhowmick, A. K. A review on the

mechanical and electrical properties of graphite and modified graphite reinforced polymer

composites. Progress in Polymer Science (Oxford), 2011, 36, 638–670.

(49) Han, Z.; Fina, A. Prog. Polym. Sci. 2011, 36, 914.

CHAPTER 1

23

(50) Salavagione, H. J.; Martínez, G.; Ellis, G. Macromol. Rapid Commun. 2011, 32, 1771.

(51) Layek, R. K.; Nandi, A. K. Polym. (United Kingdom) 2013, 54, 5087. http://creativecommons.org/licenses/by-nc-nd/4.0/

(52) Layek, R. K.; Samanta, S.; Chatterjee, D. P.; Nandi, A. K. Polymer (Guildf). 2010, 51, 5846.

(53) Deng, Y.; Li, Y.; Dai, J.; Lang, M.; Huang, X. J. Polym. Sci. Part A Polym. Chem. 2011, 49,

4747.

(54) Yang, Y.; Song, X.; Yuan, L.; Li, M.; Liu, J.; Ji, R.; Zhao, H. J. Polym. Sci. Part A Polym.

Chem. 2012, 50, 329.

(55) Moad, G.; Rizzardo, E.; Thang, S. H. Aust. J. Chem. 2006, 59, 669.


(57) Barner-kowollik, C. Handbook of RAFT polymerization; 2008.

(58) Tsubokawa, N. Preparation and Properties of Polymer-grafted Carbon Nanotubes and

Nanofibers. Polymer Journal, 2005, 37, 637–655.

(59) Paul, D. R.; Robeson, L. M. Polymer (Guildf). 2008, 49, 3187.

(60) Iijima, S. Helical microtubules of graphitic carbon. Nature, 1991, 354, 56–58.

(61) Iijima, S.; Ichihashi, T. Nature 1993, 363, 603.

(62) Bethune, D. S.; Klang, C. H.; de Vries, M. S.; Gorman, G.; Savoy, R.; Vazquez, J.; Beyers, R.

Cobalt-catalysed growth of carbon nanotubes with single-atomic-layer walls. Nature, 1993,

363, 605–607.

(63) Choi, Y. H. Polyacrylonitrile/Carbon Nanotube Composite Fibers: Effect Of Various

Processing Parameters On Fiber Structure And Properties, Georgia Institute of Technology,

USA, 2010.

(64) Chae, H. G.; Sreekumar, T. V.; Uchida, T.; Kumar, S. Polymer (Guildf). 2005, 46, 10925.

(65) Bayer MaterialScience AG. Baytubes C 150 P - Product Datasheet; 2010; pp. 1–3.

(66) Sciences, X. xGnP Graphene Nanoplatelets - Grade C; 2012.

(67) Geim, A. K.; Novoselov, K. S. Nat. Mater. 2007, 6, 183.

(68) Ramanathan, T.; Abdala, A. A.; Stankovich, S.; Dikin, D. A.; Herrera-Alonso, M.; Piner, R. D.;

Adamson, D. H.; Schniepp, H. C.; Chen, X.; Ruoff, R. S.; Nguyen, S. T.; Aksay, I. A.;

Prud’Homme, R. K.; Brinson, L. C. Nat. Nanotechnol. 2008, 3, 327.

(69) Wu, S.; He, Q.; Tan, C.; Wang, Y.; Zhang, H. Small 2013, 9, 1160.

(70) Liu, C.; Yu, Z.; Neff, D.; Zhamu, A.; Jang, B. Z. Nano Lett. 2010, 10, 4863.

(71) Mullins, O. C. Energy and Fuels 2010, 24, 2179.

http://creativecommons.org/licenses/by-nc-nd/4.0/

CHAPTER 2

24

Development of a poly(acrylonitrile) copolymer based

carbon fiber precursor

2.1 Introduction

In carbon fiber (CF) precursor production poly(acrylonitrile) (PAN) is the most important precursor

apart from other starting materials, with the market share of 86% next to cellulose with 10% and pitch

with 4%.1 This high preference towards PAN is due to its synthetic nature which allows for chemical

modification to influence the further steps in CF production such as spinning, stabilization and

carbonization, as well as various CF properties like flame retardancy, crystallite structure and

mechanical strength, by adjusting the polymerization conditions and copolymerization with other

monomers.2,3,4

Chemical synthesis of CF precursors is done industrially through free radical

polymerization in solution of acrylonitrile (AN) with methyl acrylate (MA), methyl methacrylate

(MMA), acrylic acid (AA), itaconic acid (IA), vinyl acetate (VA) and sodium methallyl sulfonates

(SMS) as comonomers.5,6

Typical solvents for the polymerization and spinning are N,N’-

dimethylformamide (DMF), dimethylacetamide (DMAc), dimethylsulfoxide (DMSO) and aqueous

thiocyanate solutions.7 For continuous processing the aqueous precipitation polymerization enables

skimming the polymer, then washing and drying it.

After the spinning, washing and drying of the CF precursor fibers starts the thermal treatment under

stretching consisting of the stabilization (200-400°C) in air, the carbonization (800-1600°C) in

nitrogen atmosphere and for further increase of the E-modulus the graphitization (2000-2500°C) in

inert gas atmosphere respectively.8 In this work we focus on the stabilization reactions where the

stretched linear polymer chains of PAN are oriented parallel to each other and constitute the so-called

“ladder polymer”. This ladder structure helps achieving the following step of carbonization to

CHAPTER 2

25

turbostratic carbon in nitrogen atmosphere without combustion, as well as with minimal shape change

and mass loss (see Figure 1).7

Figure 1 Ladder structure of stabilized PAN-based CF precursor.

The stabilization process consists of many different chemical reactions which can be summarized into

three major groups of energetic reactions namely: dehydrogenation, cyclization and oxidation.9,10,11,12

The dehydrogenation and cyclization reactions form the ladder structure and during the oxidation

reaction the oxygen promotes crosslinking for structural stability.13

The idealized thermogram below

shows the heat flow vs. temperature of the PAN-based CF precursor in nitrogen and air atmosphere

with overlapping chemical reactions (see Figure 2). The enthalpy values and peak temperatures of the

three stabilization reactions can be determined via peak deconvolution.14

Figure 2 DSC thermogram of PAN-based CF precursor in nitrogen and air atmosphere.

CHAPTER 2

26

In our work we aimed for the development of a CF precursor with properties similar to a commercial

CF precursor with high tensile strength. We changed the polymerization conditions of the FRP

synthesis by varying the ratios of the redox initiator pair persulfate/bisulfite. The characterization of

the polymers with size exclusion chromatography (SEC) probes the best synthesis conditions. Four CF

precursor candidates were synthesized composed of the following four monomers with their beneficial

properties: a) acrylonitrile (AN) which allows for cyclization of linear chains; b) methyl acrylate (MA)

which facilitates the oxidation and improves on final CF tensile properties; c) itaconic acid (IA) which

drops the starting temperature of the stabilization reactions through acidic catalysis; and d) ionic

comonomer (IC) is used coloring with ionic dyes in textile polymers.15

One of the candidates is the

commercial, PAN-based textile polymer DralonX which would represent the cheapest option if its

thermal properties prove to be suitable for high tensile strength CF. Thermal analysis in nitrogen and

air atmosphere is utilized to describe the influence of the chosen monomers and determine a

composition that approximates the thermal stability of the commercial CF precursor Bluestar.

2.2 Experimental part

2.2.1 Materials

The polymer DralonX (composed of AN, MA, IC) and all the CF precursor candidates (Precursor-M,

Precursor-B, Precursor-T composed of varying amounts of AN, MA, IA) were supplied by Dralon

GmbH (Dormagen, Germany); the Bluestar CF precursor fiber was supplied by Bluestar Fibres Co.

Ltd. Thermally treated CF precursor fibers were supplied by the Institut für Industrieofenbau und

Wärmetechnik of the RWTH Aachen. Synthetic air (80% nitrogen, 20% oxygen) was bought from Air

Products.

CHAPTER 2

27

2.2.2 Methods

Molecular weights (number average Mn and weight average Mw) were determined by size exclusion

chromatography (SEC), combining a pump (ERC 6420) and a refractive index detector (WGE Dr.

Bures ETA 2020, running at 30°C); the flow rate of the eluting solvent N,N’-dimethylformamide

(DMF) being 1.0 mL/min. Four columns with PSS GRAM gel were applied: the length of each

column was 300 mm, the diameter 8 mm, the diameter of gel particles 10 µm, the nominal pore widths

of 30, 100, 1000, 10000 Å. Calibration was achieved using poly(methyl methacrylate) (PMMA)

standards. Results were evaluated using WinGPC Unity software.

Differential Scanning Calorimetry (DSC) measurements were carried out on a Netzsch DSC 204

differential scanning calorimeter in nitrogen or in synthetic air atmosphere. Samples were measured in

perforated, closed aluminum pans using approximately 4 ± 0.05 mg of sample with a heating rate of 10

K/min unless otherwise indicated. The samples were measured in the temperature range of 30 – 400

°C. The heat flow was measured as a function of temperature. A sigmoidal baseline was applied for

the integration of the signal area. The resulting heat flow with the unit J/g was multiplied with the

medium molar mass of the repeating units leading to the enthalpy of the reaction.

Thermo gravimetric analysis (TGA) measurements were performed using a Perkin Elmer STA 6000

operating in nitrogen or synthetic air atmosphere. Samples of 4 ± 0.05 mg were placed in standard

Perkin Elmer alumina 100 µL crucibles and heated at 10 K/min, unless otherwise indicated, to 400 °C.

Before each measurement baseline correction was performed with an empty crucible.

Infrared (IR)-Spectroscopy was performed using a Nexus 470 FT-IR from ThermoNicolet with a

spectral resolution of 4 cm-1

in KBr.

The 13

C solid-state NMR spectra were measured using cross-polarization/magic angle spinning

(CP/MAS) method with 1H power decoupling by two-pulse phase modulation method at rotor

frequency of 5 kHz. A 13

C resonance frequency of 176.08 MHz and a 1H resonance frequency of

700.24 MHz and a number of scans to 2048 were set as the operating parameters. All measurements

were performed at room temperature.

CHAPTER 2

28

2.3 Results and discussion

2.3.1 Dralon precursor candidates prepared with different redox initiator ratios

The production of the CF precursors starts with the free radical polymerization (FRP) of AN with the

corresponding comonomers. FRP in water using a persulfates/bisulfite redox pair is industrially

important due to the low price of radical initiators compared to azo- or photoinitiators.16

The

precipitation polymerization of AN, MA and IA was done in water at 60°C as a continuous process in

a 5 L reactor at Dralon GmbH. There are two paths generating radicals (see Scheme 1). One path is the

reaction of hydrogen sulfite (generated through reaction of bisulfite with water) with persulfate to

sulfate and hydrogen sulfite radicals and the other path is the homolytic dissociation of persulfate to

sulfate radicals. Water is an important solvent due to its zero chain transfer coefficient leading to high

molecular weight PAN copolymers which are important for producing high tensile strength CF.17

Modern 1H NMR spectrometer have an accuracy of ±5%, therefore small amounts of comonomer

cannot be determined precisely and most PAN copolymer spectra look like the DralonX polymer

spectrum below (see Figure 3).

Scheme 1 FRP of PAN with redox system persulfate/bisulfite as initiator.

CHAPTER 2

29

Figure 3 1H NMR spectrum of DralonX copolymer in DMSO-d6 with two prominent polymer

backbone signals a, c, e (CH2 at 2.0 ppm) and b, d (CH at 3.1 ppm). The triplett at 3.7

ppm is caused by unknown additive in the industrial production.

Depending on the molar ratio of persulfate to bisulfite the molecular weight can be adjusted (see Table

1). Excess of bisulfite leads to higher amount of hydrogen sulfite which upon reaction with persulfate

results in a higher concentration of free radicals leading in turn to more growing polymer chains. This

excess of radicals explains the low Mn value of 28.000 g/mol for the 1:50 molar ratio of persulfate to

bisulfite. The Mn value with 48.000 g/mol of the inverse molar ratio 50:1 persulfate to bisulfite is

significantly higher due to the homolytic dissociation of the persulfate18

into sulfate radicals. The

resulting increase of radicals starting polymer chains constitutes the major change in dispersity Đ in

this chain growth polymerization from 2.7-3.1 to 4.4. The molar ratio of 1:1 generates radicals at the

slowest rate because for each radical generating redox reaction the hydrogen sulfite has to be first

generated. This prolonged presence of relatively low concentrations of polymer chain initiating

radicals leads to the high Mn value of 222,000 g/mol.

ppm (t1) 1.02.03.04.05.0

3,7

07

3,6

91

3,6

73

3,3

32

3,1

19

2,5

00

2,0

32

2.1

8

1.0

0

0.1

2

Water

DMSO

a, c, e

b, d

IC

CHAPTER 2

30

Table 1 Influence of persulfate to bisulfite molar ratio on the molecular weight of the

synthesized copolymers (AN, MA, IA) with constant reaction time. a

SEC with

PMMA as calibration reference.

With increasing molecular weight the viscosity of the polymer solutions experiences a major increase,

resulting in the inability to use the existing spinning setup-ups in the Dralon labs. Therefore the redox

initiator molar ratio of 1:5-10 of persulfate/bisulfite is applied for the CF precursor synthesis with Mn

= 140.000 g/mol.

2.3.2 Thermal analysis of Dralon precursor candidates with different comonomers measured

in nitrogen atmosphere

Previously we determined the FRP redox initiator conditions and continue with the selection of the

relevant comonomers to achieve fiber properties similar to the industrial CF precursor reference

Bluestar. Four PAN-based CF precursor candidates are selected with the comonomers MA, IA and IC

(see Table 2). The exact chemical composition of the Bluestar fiber is unknown but literature suggests

a content of at least 90% AN.5

Molar ratio Mn [g/mol]a Mw [g/mol]

a Đ

a

1 Persulfate : 1 Bisulfite 222,000 591,000 2.7





CF precursor

(1: 5 – 10 persulfate/bisulfite)

140,000 386,000 2.8

CHAPTER 2

31

Table 2 Name and comonomers of CF precursor candidates.

Bluestar (industry reference) At least 90% AN, MA, IA

DralonX (textile polymer) AN, MA, IC

Precursor-T (Dralon product) AN, MA, IA

Precursor-B (Dralon product) AN, MA

Precursor-M (Dralon product) AN

The CF precursor candidates are analyzed by differential scanning calorimetry (DSC) and thermo

gravimetric analysis (TGA) in nitrogen atmosphere with non-isothermal heating. Literature shows the

independence of molecular weight and molar mass distribution of polymers from the onset or the peak

temperatures under non-isothermal heating.19

The resulting thermograms and the exact chemical

composition are shown below (see Figure 4 and Table 3).

Figure 4 Thermal effects upon heating of DralonX, Bluestar, Precursor-T, Precursor-B and

Precursor-M powder samples in nitrogen atmosphere.

200 220 240 260 280 300 320 340 360Temperatur /°C

-20

-15

-10

-5

0

DSC /(mW/mg)

BluestarDralonXPrecursor-T

Precursor-M

Precursor-B

[1] [2][3][4][5] Exo

CHAPTER 2

32

The thermograms can be interpreted by the enthalpy of the cyclization reaction being the peak area

integral as well as the peak maximum and temperature range of the cyclization reaction. The heat flow

of the reaction in the units J/g is multiplied with the averaged comonomer molar mass to yield the

enthalpy in kJ/mol. The enthalpy of Bluestar is calculated based on the estimated composition (see

Table 3). The data is supplemented by the relative weight loss from the TGA experiments (see Table

3). An ideal CF precursor candidate should fulfill simultaneously three conditions: a) good control of

the stabilization reactions over a broad temperature range so that the released heat can dissipate and

does not lead to fusion or combustion of the material; b) having a low peak maximum of the

temperature thus requiring less overall heating energy; c) high percentage of cyclization to achieve a

stable ladder polymer.

The PAN homopolymer Precursor-M shows a high exothermic peak maximum at 291.4°C within a

narrow 35 degrees temperature range. It has the lowest reaction enthalpy with 25.6 kJ/mol and highest

relative weight loss with 42% of all tested candidates and therefore disqualifies as a possible CF

precursor. CF precursor candidate Precursor-B combines the monomer AN and comonomer MA

which broadens the range of reaction temperature to 63 degrees but increases the peak maximum

temperature to 297.7°C. The relative weight loss is reduced to 34% and the cyclization enthalpy

increased to 28.7 kJ/mol. The introduction of MA affects the control of the stabilization reactions

positively making it a good choice as comonomer. Adding to AN and MA a third ionic comonomer

(DralonX) reduces the relative weight loss again, down to 27% and increases the cyclization enthalpy

to 36.7 kJ/mol. The temperature range of reaction and the peak maximum vary only within a few

degrees Celsius compared to the polymer poly(AN-MA). To decrease the peak temperature of the

cyclization reaction the comonomer IA is introduced in the candidate Precursor-T (AN, MA, IA).

Compared to DralonX this composition yields not only a wider range of temperature with 91degree

Celsius but additionally the peak maximum decreases with 11°C. The industry benchmark Bluestar

shows the lowest relative weight loss and broadest temperature range of all candidates. The peak

maximum of the temperature (Tpeak) of Bluestar is equal to Tpeak of Precursor-T but the cyclization

CHAPTER 2

33

enthalpy of Bluestar of 33 kJ/mol is lower than the enthalpy of Precursor-T (34.3 kJ/mol). The higher

enthalpy of Precursor-T suggests a more complete conversion during the stabilization reactions.

Table 3 Thermal properties of different precursor candidates in nitrogen atmosphere depending

on their composition. The exact composition of the Dralon precursors will not be

mentioned in this work.

Precursor

candidates

Composition

Enthalpy of

cyclization [-kJ/mol]

Tpeak

[°C]

Range of

reaction

[°C]

relative

weight

loss [%]

Bluestar

> 90% AN

< 8% MA

< 2% IA

33 289.1 100 23

DralonX

> 90% AN

< 8% MA

< 2% IC

36.7 299.4 56 27

Precursor-T

> 90% AN

< 8% MA

< 2% IA

34.3 288.9 91 26

Precursor-B

> 90% AN

< 10% MA

28.7 297.7 63 34

Precursor-M 100% AN 25.6 291.4 35 42

Based on the thermal analysis in nitrogen atmosphere the candidate Precursor-T with the comonomers

AN, MA and IA shows the best fit to the industry reference Bluestar in terms of Tpeak, relative weight

loss and range of reaction interval, with a higher cyclization enthalpy equaling improved conversion.

Therefore the composition of Bluestar must be similar to Precursor-T.

CHAPTER 2

34

2.3.3 Thermal analysis of Dralon precursor candidates with different comonomers measured

in oxygen atmosphere

The thermal analysis of the precursor candidates in powder form in nitrogen atmosphere solely

describes the cyclization reaction out of the three stabilization reactions. The measured DSC

thermogram shows three temperature maxima which can be attributed to the (1) ionic cyclization, (2)

radical cyclization and dehydrogenation and (3) oxidation (see Figure 5).20

Due to the occurring

overlapping of the stabilization reactions, the determination of the enthalpies was done through peak

deconvolution with OriginLab OriginPro 8.6. All applied fits have a coefficient of determination R2 >

0.997 (see Appendix A).

200 250 300 350 400

-16

-14

-12

-10

-8

-6

-4

-2

0

2

he

at

flo

w [

mW

/mg

]

temperature [°C]

1

2

3

Figure 5 DSC measurement of Precursor-T powder in synthetic air atmosphere. 1) Tpeak of ionic

cyclization reaction; 2) Tpeak of radical cyclization and dehydrogenation reaction; 3)

Tpeak of oxidation reaction.

CHAPTER 2

35

The powdered CF precursor candidates are analyzed by differential scanning calorimetry (DSC) in

synthetic air atmosphere with non-isothermal heating. The resulting thermograms are shown below

(see Figure 6).

Figure 6 Exothermic effects on heating curves of DralonX, Bluestar, Precursor-T, Precursor-B

and Precursor-M powder in synthetic air atmosphere.

The influence of the comonomers is described by the enthalpy of stabilization consisting of

cyclization, dehydrogenation and oxidation (see Table 4). The CF precursor candidate with the highest

reaction enthalpy is the PAN homopolymer Precursor-M with 313.8 kJ/mol. The oxidation reaction

accounts for 79% (246.6 kJ/mol) of the enthalpy of all stabilization reactions. The cyclization and

dehydrogenation reactions play a minor role with 8% and 13% respectively. Its Tpeak has the highest

value of all candidates with 312.1 °C. Adding the comonomer MA (Precursor-B) results in a similar

enthalpy value of stabilization as with Precursor-M but with 10% lower enthalpy for the oxidation

reaction (now at 69% of total). The Tpeak value is reduced to 308.4 °C. From the addition of MA the

dehydrogenation reaction benefitted most, increasing from 13% to 22% contribution to the total

enthalpy. DralonX with the comonomers AN, MA and IC shows the lowest overall reaction enthalpy

and a high Tpeak value of 312.0 °C. The low enthalpy value points at a low conversion of the polymer

150 200 250 300 350 400Temperatur /°C

-30

-25

-20

-15

-10

-5

0

DSC /(mW/mg)

Bluestar

Precursor-B

Precursor-M

DralonXPrecursor-T

[1]

[2]

[3][4][5]

Exo

CHAPTER 2

36

which is detrimental to the energy and time efficiency of the process. The ionic comonomer helps to

increase the contribution of the enthalpy of cyclization to 20%, the highest value for the enthalpy of

cyclization. This increase in cyclization can be explained by the acidic nature of the comonomers

catalyzing the reaction.21

The higher percentage of cyclization and dehydrogenation contributions of

Precursor-T vs. Bluestar could be explained by a higher content of MA and IA in Precursor-T. In

return the Tpeak of Bluestar is 12 degrees lower than Tpeak of Precursor-T and the stabilization enthalpy

of Bluestar is 14 kJ/mol higher than the stabilization enthalpy of Precursor-T.

CHAPTER 2

37

Tab

le 4

M

easu

red

an

d c

alcu

late

d e

nth

alpie

s of

the

stab

iliz

atio

n r

eact

ions

of

the

hea

t tr

eate

d p

recu

rsor

can

did

ates

.

Pre

curs

or

can

did

ates

Co

mp

osi

tio

n

Tp

eak

[°C

]

Enth

alpy o

f

stab

iliz

atio

n [

kJ/

mol]

Enth

alpy o

f

cycl

izat

ion [

kJ/

mol]

En

thal

py o

f

deh

yd

rogen

atio

n [

kJ/

mo

l]

En

thal

py o

f

ox

idat

ion

[kJ/

mo

l]

Blu

esta

r

> 9

3%

AN

< 6

% M

A

< 1

% I

A

286.7

276

33

(12%

)

71

(26

%)

17

2

(62

%)

Dra

lonX

> 9

0%

AN

< 8

% M

A

< 2

% I

C

312.0

181.3

36.7

(20

%)

46

.9

(26%

)

97

.7

(54%

)

Pre

curs

or-

T

> 9

0%

AN

< 8

% M

A

< 2

% I

A

298.6

262.3

34.3

(13%

)

10

0.1

(38

%)

12

7.9

(49

%)

Pre

curs

or-

B

> 9

0%

AN

< 1

0%

MA

308.4

314.1

28.7

(9%

)

69

.2

(22

%)

21

6.2

(69

%)

Pre

curs

or-

M

10

0%

AN

312.1

313.8

25.6

(8%

)

41

.6

(13

%)

24

6.6

(79

%)

CHAPTER 2

38

Based on the thermal analysis in synthetic air atmosphere the candidate Precursor-T with the

comonomers AN, MA and IA shows the best fit to the industry reference Bluestar in the enthalpy of

stabilization.

2.3.4 Analysis of heat treated Bluestar and Precursor-T fibers

After characterizing the non-isothermal behavior of Precursor-T powder, looking at the isothermal

behavior of the fiber is also relevant for the industrial process of stabilization treatment. We received

from the Institut für Industrieofenbau und Wärmetechnik of the RWTH Aachen a 3000 filament tow

from the candidates Bluestar and Precursor-T which were treated in four isothermal heating steps. The

eight fiber samples were taken from a continuous heating treatment, starting with points 1 and 2

(220°C after 15 and 30 min, respectively), following with points 3 and 4 (240°C after 15 and 30 min,

respectively), continuing at points 5 and 6 (260°C after 15 and 30 min, respectively) and ending at

points 7 and 8 (280°C after 15 and 30 min, respectively) (see Figure 7).

0 40 80 120 1600

100

200

300

86

75

tem

pe

ratu

re [

°C]

reaction time [min]

1 23 4

Figure 7 Temperature profile for precursor fibers consisting of four heating steps and eight

sampling points.

The fibers were measured with non-isothermal DSC in synthetic air atmosphere. The total reaction

enthalpy was measured and the enthalpies of dehydrogenation, cyclization and oxidation were

CHAPTER 2

39

determined by peak deconvolution (see Table 5 and Appendix A). The fiber samples were taken at

points 1, 2 and 3. The data of the other points had a high margin of error due to the far progression of

the stabilization reaction towards carbon. The reaction enthalpies of stabilization of both fibers are

similar, around 65 kJ/mol. The cyclization reactions show similar values around 30 kJ/mol with a drop

to 25 kJ/mol at point 3. The dehydrogenation reactions values at Point 1 and 2 are around 11 kJ/mol

and drop to 2 kJ/mol at point 3. The oxidation reactions of Bluestar and Precursor-T show the biggest

difference at point 2 with only 5%. All in all the fibers of Bluestar and Precursor-T show similar

thermal behavior.

Table 5 Calculation of reaction enthalpies of the heat treated precursor fibers in air

atmosphere.

Fibers Treatment condition Point 1

(220°C, 15 min)

Point 2

(220°C, 30 min)

Point 3

(240 °C, 15 min)

Blu

esta

r

Enthalpy [kJ/mol] 67.0 72.2 68.8

Cyclization [kJ/mol] 28.6 29.1 24.7

Dehydrogenation

[kJ/mol]

11.5 11.4 3.4

Oxidation [kJ/mol] 27.0 31.7 40.8

Pre

curs

or-

T

Enthalpy [kJ/mol] 62.3 68.6 63.7

Cyclization [kJ/mol] 27.0 29.0 24.7

Dehydrogenation

[kJ/mol]

11.0 12.8 1.8

Oxidation [kJ/mol] 24.3 26.8 37.2

CHAPTER 2

40

The FT-IR spectra at point 8 of the conversion of the CF precursor are similar. The progress of the

stabilization is followed by the decrease of the nitrile peak νs-CN at 2243 cm-1

and the increase of the

cyclized nitrile peak νs=CN at 1585 cm-1

. The virgin fibers of Bluestar and Precursor-T and samples

from point 8 are compared qualitatively in transmission vs. wave number graph (see Figure 8). No

significant differences could be observed in the quantitative analysis, pointing again at the similar

nature of the stabilization reactions.

4000 3000 2000 1000

s-CN

sC=N

Bluestar

point 8

Precursor-T

Precursor-T

point 8

transm

issio

n [%

]

wave number [cm-1]

Bluestar

2243

1585

Figure 8 Comparison of FT-IR spectra of Bluestar and Precursor-T fibers.

For a higher focus on the conversion of polymer functionalities solid-state 13

C CP/MAS NMR

spectroscopy was applied (see Figure 9). The signal decrease of aliphatic 13

C from the polymer

backbone was plotted against the four isothermal heating steps, at 220, 240, 260 and 280°C (see

Figure 10). Precursor-T showed at 220°C and 240°C a relatively stronger decrease in aliphatic 13

C

signal intensity meaning a faster conversion at the lower temperature of the applied heating profile.

Reactions at higher heating rates proceed with the similar speed.

; t = 0

; t = 0

CHAPTER 2

41

Figure 9 Solid-State 13

C CP/MAS NMR spectrum of Bluestar copolymer at 700 MHz.

BS step 1 step 2 step 3 step 4

1

2

3

4

5

6

7

8 aliphatic 13

C

signal intensity

Bluestar

13C

no

rm.

inte

g.

int. [

10

-2 m

g-1]

15 min.

30 min

Dralon step 1 step2 step 3 step 40

2

4

6

8

10

12

14 aliphatic 13

C

signal intensity

Precursor-T

13C

no

rm.

inte

g.

int. [

10

-2 m

g-1]

15 min.

30 min.

Figure 10 Normalized aliphatic 13

C intensity diagram of Bluestar (left) and Precursor-T (right)

fibers at the four isothermal heating steps.

ppm (t1)

-300-200-1000100200300

20

6.6

63

17

7.4

18

14

8.0

54

12

2.3

44

29

.44

8

13C CPMAS spectrum at 700 MHz

Dralon films 0% wt. Rüß

CH2, CH

C=O

C≡N

CHAPTER 2

42

2.4 Conclusion

Using different analytical methods such as SEC, solid and liquid state NMR and thermal analysis

(DSC and TGA) this work presents the characterization of several materials for being utilized as viable

CF precursor. By optimizing free radical polymerization conditions via radical initiator systems and

selection of the monomer AN and comonomers MA and IA, Precursor-T was yielded and showed to

be on par with the industry fiber reference Bluestar. The non-isothermal treatment in nitrogen

atmosphere showed similar thermal behavior of both precursors; in air atmosphere Precursor-T

displayed a preference for dehydrogenation reactions whereas Bluestar precursor presented a higher

degree of oxidation reactions during the stabilization process. Achieving a high tensile strength CF

requires the identification of proper thermal and mechanical treatment of the Precursor-T material

taking into consideration all these gained insights.

2.5 References

(1) AVK - Industrievereinigung Verstärkte Kunststoffe e.V. Handbuch Faserverbundkunststoffe /

Composites; 4th ed.; Springer Vieweg, 2013.

(2) Frank, E.; Steudle, L. M.; Ingildeev, D.; Spörl, J. M.; Buchmeiser, M. R. Angew. Chem. Int.

Ed. Engl. 2014, 2.

(3) Horrocks, A. R.; Zhang, J.; Hall, M. E. Polym. Int. 1994, 33, 303.

(4) Bang, Y. H.; Lee, S.; Cho, H. H. J. Appl. Polym. Sci. 1998, 68, 2205.


(6) Wangxi, Z.; Jie, L. J. Wuhan Univ. Technol. - Mater. Sci. Ed. 2006, 21, 2004.


(8) Rahaman, M. S. A.; Ismail, A. F.; Mustafa, A. Polym. Degrad. Stab. 2007, 92, 1421.

(9) Nguyen-Thai, N. U.; Hong, S. C. Macromolecules 2013, 46, 5882.

(10) Xue, Y.; Liu, J.; Liang, J. Polym. Degrad. Stab. 2013, 98, 219.

(11) Arbab, S.; Zeinolebadi, A. Polym. Degrad. Stab. 2013, 98, 2537.

CHAPTER 2

43

(12) Ju, A.; Guang, S.; Xu, H. Carbon N. Y. 2013, 54, 323.

(13) Jain, M. K.; Abhiraman, A. S. J. Mater. Sci. 1987, 22, 278.

(14) Belyaev, S. S.; Arkhangelsky, I. V.; Makarenko, I. V. Thermochim. Acta 2010, 507-508, 9.

(15) Rajalingam, P.; Radhakrishxan, G. Polym. Rev. 1991, 31, 301.

(16) Nesvadba, P. Radical Polymerization in Industry. Encyclopedia of Radicals in Chemistry,

Biology and Materials, 2012.

(17) Zhao, Y.; Wang, C.; Wang, Y.; Zhu, B. J. Appl. Polym. Sci. 2009, 111, 3163.

(18) Ebdon, J. R.; Huckerby, T. N.; Hunter, T. C. Polymer (Guildf). 1994, 35, 250.

(19) Ahn, C. H.; Park, C. R.; Park, Y. H.; Min, B. G.; Son, T. W. Han’guk Somyu Konghakhoechi

1994, 31, 983.

(20) Ouyang, Q.; Cheng, L.; Wang, H.; Li, K. Polym. Degrad. Stab. 2008, 93, 1415.

(21) Tsai, J.-S.; Lin, C.-H. J. Appl. Polym. Sci. 1991, 43, 679.

CHAPTER 2

44

Appendix A

a) DSC thermograms

Peak deconvolution of baseline corrected thermogram of Bluestar powder

100 200 300 400

-16

-14

-12

-10

-8

-6

-4

-2

0

2

hea

t flo

w (

mW

/mg)

Bluestar powder

heat flow

Peak anpassen 1

Peak anpassen 2

Peak anpassen 3

Kumulativer Impulsfit

Modell Gauss

Gleichung y=y0 + (A/(w*sqrt(PI/2)))*exp(-2*((x-xc)/w)^2)

Chi-Quadr Reduziert

0,0075

Kor. R-Quadra 0,99942

Wert Standardfehler

Impuls1(C) y0 0,03887 0,00254

Impuls1(C) xc 263,38742 0,30586

Impuls1(C) w 43,07631 0,36194

Impuls1(C) A -114,64829 1,73027

Impuls1(C) Sigma 21,53816 0,18097

Impuls1(C) Halbwertsbreit 50,71848 0,42616

Impuls1(C) Höhe -2,12358 0,01629

Impuls2(C) y0 0,03887 0,00254

Impuls2(C) xc 285,98499 0,00804

Impuls2(C) w 14,24849 0,02446

Impuls2(C) A -166,92986 0,54669

Impuls2(C) Sigma 7,12425 0,01223


Impuls2(C) Höhe -9,34771 0,01784

Impuls3(C) y0 0,03887 0,00254

Impuls3(C) xc 318,35738 0,08652

Impuls3(C) w 53,82942 0,10766

Impuls3(C) A -469,09906 1,40635

Impuls3(C) Sigma 26,91471 0,05383


Impuls3(C) Höhe -6,9532 0,00954

Peak deconvolution of baseline corrected thermogram of DralonX powder

200 300 400

-14

-12

-10

-8

-6

-4

-2

0

2

heat flow

Fit Peak 1

Fit Peak 2

Fit Peak 3

Cumulative Fit Peak

hea

t flo

w [

mW

/mg

]

DralonX powder

Model Gauss

Equation y=y0 + (A/(w*sqrt(PI/2)))*exp(-2*((x-xc)/w)^2)

Reduced Chi-Sqr 0,00209

Adj. R-Square 0,99983

Value Standard Error

Peak1(heat flow) y0 0 0

Peak1(heat flow) xc 312,43765 0,01264

Peak1(heat flow) w 12,39838 0,03609

Peak1(heat flow) A -80,49471 0,67156

Peak1(heat flow) sigma 6,19919 0,01804

Peak1(heat flow) FWHM 14,59797 0,04249

Peak1(heat flow) Height -5,18015 0,02926


Peak2(heat flow) xc 329,45972 0,07011

Peak2(heat flow) w 49,26938 0,06376

Peak2(heat flow) A -271,9099 0,72345





Peak3(heat flow) xc 304,5005 0,05326

Peak3(heat flow) w 22,95831 0,06022

Peak3(heat flow) A -151,9917 1,05612




CHAPTER 2

45

Peak deconvolution of baseline corrected thermogram of Precursor-T powder

200 300 400

-14

-12

-10

-8

-6

-4

-2

0

heat flow

[m

W/m

g]

Precursor-T powder

heat flow

Fit Peak 1

Fit Peak 2

Fit Peak 3

Cumulative Fit Peak

Model Gauss


Reduced Chi-Sqr

0,03665



Peak1(C) y0 0 0

Peak1(C) xc 260,48565 0,04875

Peak1(C) w 15,75713 0,09689

Peak1(C) A -69,34671 0,4021

Peak1(C) sigma 7,87857 0,04844

Peak1(C) FWHM 18,55261 0,11407

Peak1(C) Height -3,51147 0,01527

Peak2(C) y0 0 0

Peak2(C) xc 295,16535 0,03401

Peak2(C) w 25,01385 0,08343

Peak2(C) A -318,25376 2,10489

Peak2(C) sigma 12,50693 0,04171

Peak2(C) FWHM 29,45156 0,09823

Peak2(C) Height -10,15157 0,04017

Peak3(C) y0 0 0

Peak3(C) xc 332,10429 0,15397

Peak3(C) w 46,12522 0,21981

Peak3(C) A -368,79109 2,14094

Peak3(C) sigma 23,06261 0,1099

Peak3(C) FWHM 54,3083 0,2588

Peak3(C) Height -6,37943 0,01291

Peak deconvolution of baseline corrected thermogram of Precursor-B powder

200 300 400

-30

-28

-26

-24

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

2

hea

t flo

w [

mW

/mg

]

Precursor-B

heat flow

Peak anpassen 1

Peak anpassen 2


Modell Gauss


Chi-Quadr Reduziert

0,07907

Kor. R-Quadrat 0,99854

Wert Standardfehler

Impuls1(C) y0 -0,06901 0,00842

Impuls1(C) xc 308,10629 0,01122

Impuls1(C) w 12,54947 0,02825

Impuls1(C) A -290,91304 0,93571

Impuls1(C) Sigma 6,27474 0,01413


Impuls1(C) Höhe -18,496 0,03466

Impuls2(C) y0 -0,06901 0,00842

Impuls2(C) xc 323,83725 0,04832

Impuls2(C) w 42,34256 0,07656

Impuls2(C) A -641,8509 1,64498

Impuls2(C) Sigma 21,17128 0,03828


Impuls2(C) Höhe -12,09476 0,02378

CHAPTER 2

46

Peak deconvolution of baseline corrected thermogram of Precursor-M powder

200 300 400

-26

-24

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

2

hea

t flo

w [

mW

/mg

]

Precursor-M

heat flow

Peak anpassen 1

Peak anpassen 2


Modell Gauss


Chi-Quadr Reduziert

0,05796

Kor. R-Quadra 0,99883

Wert Standardfehler

Impuls1(C) y0 -0,15326 0,0071

Impuls1(C) xc 308,66327 0,01372

Impuls1(C) w 12,37451 0,03422

Impuls1(C) A -200,99727 0,79387

Impuls1(C) Sigma 6,18726 0,01711


Impuls1(C) Höhe -12,95992 0,02984

Impuls2(C) y0 -0,15326 0,0071

Impuls2(C) xc 321,27178 0,03245

Impuls2(C) w 41,86196 0,05578

Impuls2(C) A -737,75568 1,3055

Impuls2(C) Sigma 20,93098 0,02789


Impuls2(C) Höhe -14,06154 0,02289

CHAPTER 3

47

Influence of branched poly(acrylonitrile) additives on the

viscosity of poly(acrylonitrile) based carbon fiber

precursor solutions

3.1 Introduction

Branched polymers with their large number of chain ends per molecule have attracted much attention

due to their physicochemical properties and are applied in various fields including for example

additives, coating, gene/drug carriers and nanoreactors.1,2,3,4,5

Many approaches for the preparation of

hyperbranched polymers are possible, such as step-growth polymerization via polycondensation or

addition polymerization of multifunctional monomers,6 copolymerization of monomers via self-

condensing vinyl polymerization (SCVP),7 or copolymerization of vinyl monomers with

multifunctional vinyl comonomers.8 Generally, “living”/ controlled radical polymerization such as

iniferter-mediated polymerization,9 nitroxide-mediated polymerization,

10 atom transfer radical

polymerization,11,12

and reversible addition-fragmentation chain transfer (RAFT) polymerization13,14,15

have been applied to synthesize star and hyperbranched polymers with controlled compositions and

functionality. RAFT polymerization is a powerful tool for the synthesis of star and hyperbranched

polymers due to many advantages such as mild reaction conditions, applicability of different

monomers, tolerance of various functionalities, and lack of metal catalysts.

Branched polymers with a core-arm structure, so called star-shaped polymers, have been well-studied.

They can be synthesized by an “arm first” 16,17

or “core first” 18,19

approach or by a combination of

both.20,21

The arm first approach starts with the synthesis of the polymer arms, followed by reaction

with a multifunctional core reagent. Not only it is a facile reaction, but it yields star polymers with low

dispersity. Problems of this method are (i) a varying number of arms, (ii) a high synthetic demand of

the complex core structure and (iii) the removal of unreacted linear chains. The core first approach

CHAPTER 3

48

applies a chain transfer agent (CTA), the functionality of which sets the arm number of the star

polymer.22,23,24

The rheological properties of branched polymers are very different from their linear counterparts.25

The relationship between molecular weight and intrinsic viscosity is strongly dependent on the

polymer topology (see Figure 1). For a specific molecular weight the intrinsic viscosity of a

hyperbranched polymer is lower than of its linear analog.26

Although the intrinsic viscosities of

dendrimers are even lower, hyperbranched polymers have comparable properties such as low

viscosity, high solubility and weak intermolecular entanglement. The cost-effective and large-scale

manufacturing of hyperbranched polymers makes them more preferable in industrial applications than

dendrimers.

Figure 1 Schematic plots for the relationship between intrinsic viscosity (log [η]) and molecular

weight (log [M]) for various polymer topologies.26

The aim of this work is the development of branched polymers which can be added to a concentrated

polymer solution with the result of a decreased solution viscosity, more precisely, the solution

viscosity of poly(acrylonitrile)-based carbon fiber precursor spinning dope. A lower viscosity of the

spinning solution would allow for a higher spinning dope mass concentration, being beneficial for the

precipitation process and the production rate. In this work the commercial polymer DralonX and a

log [η]

log [M]

CHAPTER 3

49

newly developed product Precursor-T from Dralon GmbH are used for testing experiments. Graessley

and coworkers have shown that solutions of high molecular weight four- and six-arm star-shaped

poly(isoprene)s present a lower zero shear viscosity than their linear counterparts.27

Based on this

reference we developed new rheological additives based on linear, three-, four- and six-arm star-

shaped polymers and on hyperbranched polymers.


3.2.1 Materials

Acrylonitrile (AN, ≥99%, Aldrich) and methyl acrylate (MA, 99%, Acros Organics) was purified by

column chromatography on Al2O3 (activated, neutral, Brockmann I, Aldrich) before use.

Azobisisobutyronitrile (AIBN, 98+%, Fluka) was recrystallized twice from methanol before use. N,N’-

dimethylformamide (DMF, HPLC grade, VWR), benzyl chloride (Aldrich), methanol (dried, Merck),

ethylene carbonate (EC, 99%, ABCR), sodium methoxide (≥97%, Fluka), trimethylolpropan-tris(3-

mercaptopropionate), pentaerythritol-tetrakis(3-mercaptopropionate), dipentaerythritolhexakis(3-

mercaptopropionate), carbon disulfide (≥99%, Aldrich), 1-propane thiol (Merck) and 4-vinylbenzyl

chloride (90%, Aldrich) were used without further purification.

DralonX polymer (prepared from AN, MA, ionic comonomer) and Precursor-T polymer (prepared

from AN, MA, IA) were supplied by Dralon GmbH, Dormagen.

All reactions were carried out in nitrogen atmosphere. Nitrogen (Linde, 5.0) was passed over

molecular sieves (4Å) and finely distributed potassium on aluminum oxide.

3.2.2 Methods

1H NMR spectra were recorded on a Bruker AV 400 spectrometer at 400 MHz. Deuterated chloroform

(CDCl3) or deuterated dimethylsulfoxide (DMSO-d6) were used as solvent, tetramethyl silane or

CHAPTER 3

50

residual non deuterated solvent peaks were used as internal standards. Peak assignments in 1H and

13C

spectra were made by first order analysis of the spectra.

Molecular weights (number average Mn and weight average Mw) were determined by size exclusion

chromatography (SEC) using a high pressure liquid chromatography pump (ERC 6420) and a

refractive index detector (WGE Dr. Bures ETA 2020) at 30°C; the flow rate of the eluting solvent

being 1.0 mL/min. Four columns with PSS GRAM gel were applied: the length of each column 300

mm, the diameter 8 mm, the diameter of gel particles 10 µm, the nominal pore widths of the gel

particles were 30, 100, 1000 and 10,000 Å. Calibration was achieved using poly(methyl methacrylate)

(PMMA) standards. Results were evaluated using WinGPC Unity software.

Rotational rheology was measured with a Discovery Hybrid Rheometer DHR2 from TA Instruments

in a 4 cm plate-plate setup at 30°C. Viscosity was determined in a shear rate range from 0.1 – 1,000 s-1

of a 10 wt.-% additive solution in DMF. Zero shear viscosity has been determined from the Newtonian

region at low shear rates.

Capillary rheology was measured with a capillary rheometer Rheograph 25 from Göttfert Werkstoff

Prüfmaschinen GmbH with a 100 bar pressure transducer. The round shaped capillary was 0.3 mm

long and 0.1 mm thick. Each run at 11 different shear rates was repeated three times using a 30 wt.-%

DralonX polymer solution at 80°C. All apparent shear rates and apparent viscosities were corrected

according to the Rabinowitsch-Weissenberg method.


differential scanning calorimeter in nitrogen or synthetic air atmosphere. Samples were measured in

perforated closed aluminum pans using approximately 4 ± 0.05 mg of sample with a heating rate of 10


°C. The heat flow was determined as a function of temperature.

CHAPTER 3

51

3.2.3 Syntheses

Dibenzyl trithiocarbonate (1)

1

The synthesis of dibenzyl trithiocarbonate (DBTTC) 1 was performed according to Endo and

coworkers.28

Carbon disulfide (4.0 g, 52.5 mmol) was added to a solution of benzyl chloride (6.33 g,

50.0 mmol) in N,N’-dimethylformamide (DMF) (50 mL), and to the resulting mixture was added

potassium carbonate (K2CO3) (6.91 g, 50.0 mmol) at 25 °C. After the mixture was stirred at 40 °C for

24 h, the reaction was quenched by pouring into ice-water. The product was extracted with ethyl

acetate, dried over anhydrous sodium sulfate, filtered, and evaporated to give NMR-pure DBTTC 1

(100%) as yellow oil. Further purification can be achieved by silica gel column chromatography with

hexane as eluent to yield the pure 1 (99%) as pale yellow crystals. 1H NMR (CDCl3): δ = 4.62 (s, 4 H,

SCH2), 7.24–7.34 (m, 10 H, ArH).

Linear PAN polymer (2)

2

Purified acrylonitrile (8.1 g, 152.7 mmol), dibenzyl trithiocarbonate 1 (0.056 g, 0.192 mmol), AIBN

(0.033 g, 0.192 mmol) and ethylene carbonate (26.4 g) were added in a Schlenk flask. The degassing

was done by three freeze-pump-thaw cycles, afterwards charged with nitrogen and sealed. The

polymerization tube was heated to 80 °C in a thermostatic oil bath. After a predetermined time, the

content was poured into methanol, filtered, dissolved in DMF and precipitated in methanol again.

CHAPTER 3

52

After filtration the polymer 2 was dried in vacuo. 1H NMR (DMSO-d6): δ = 1.90–2.20 (m, 2 H, CH2),

3.00–3.25 (m, 1 H, CH), 7.20–7.40 (m, 10 H, ArH).

Trimethylolpropane-tris-3-(S-benzyl-trithiocarbonyl)propionate) (3)

3

The synthesis was performed according to Boschmann29

. To a solution of trimethylolpropan-tris(3-

mercaptopropionate) (1.21 g, 3.04 mmol) in chloroform (50 mL) was added triethylamine (1.11 g,

10.9 mmol) and after stirring for one hour carbon disulfide (5 mL) was added drop-wise first and then

benzyl bromide (1.87 g, 10.9 mmol). The solution was stirred for 16.5h at room temperature and

quenched with 10 % hydrochloric acid (50 mL). The organic phase was separated and washed twice

with water. After drying the solution with sodium sulfate the solvent was removed and 3 dried in

vacuo to yield orange-yellow oil. 1H NMR (CDCl3): δ = 0.88 (t, 3 H, CH3), 1.47 (q, 2 H, CH2), 2.79 (t,

6 H, CH2), 3.61 (t, 6 H, CH2), 4.04 (s, 6 H, CH2), 4.60(s, 6 H, CH), 7.32 (m, 15 H, ArH).

3-Arm star PAN copolymer (4)

4

Trimethylolpropane-tris-3-(S-benzyl-trithiocarbonyl)propionate) 3 (0.172 g, 0.192 mmol), purified

acrylonitrile (8.1 g, 152.7 mmol), AIBN (0.033 g, 0.192 mmol) and ethylene carbonate (26.4 g) were

added in a Schlenk flask. The Schlenk flask was degassed by three freeze-pump-thaw cycles, charged

with nitrogen and sealed. The reaction mixture was heated to 80 °C in a thermostatic oil bath. After a

predetermined time, the content was precipitated in methanol, filtered, dissolved in DMF and

CHAPTER 3

53

precipitated in methanol again. After filtration the polymer 4 was dried in vacuo. 1

H NMR (DMSO-

d6): δ = 1.90–2.20 (m, 2 H, CH2), 3.00–3.25 (m, 1 H, CH), 7.20–7.40 (m, 15 H, ArH).

Pentaerythritol-tetrakis(3-mercaptopropionate) (5)

5


. A solution of pentaerythritol-tetrakis(3-

mercaptopropionate) (4.80 g, 9.82 mmol) and carbon disulfide (5 mL) in dichloromethane (30 mL)

was combined with a solution of triethylamine (8.10 g, 80.0 mmol) in dichloromethane (20 mL).

After stirring for 1.5 h at room temperature, benzyl bromide (6.61 g, 38.5 mmol) was added drop-wise.

The solution was stirred for 3 h at room temperature and then quenched with 10 % hydrochloric acid

(50 mL). The organic phase was separated and washed twice with water. After drying the solution

with sodium sulfate the solvent was removed and 5 dried in vacuo to yield orange oil. 1H NMR

(CDCl3): δ = 2.72 (d, 8 H, CH2), 3.53 (d, 8 H, CH2), 4.07 (s, 8 H, CH2), 4.53 (s, 8 H, CH2), 7.23 (m, 20

H, ArH).


6

Pentaerythritoltetrakis(3-(S-benzyltrithiocarbonyl)propionate) 5 (0.222 g, 0.192 mmol), purified





CHAPTER 3

54


H NMR (DMSO-


Dipentaerythritolhexakis(3-mercaptopropionate) (7)

7


. A solution of dipentaerythritolhexakis(3-

mercaptopropionate) (3.92 g, 5.0 mmol) and carbon disulfide (5 mL) in dichloromethane (30 mL) was

combined with a solution of triethylamine (6.01 g, 60.0 mmol) in dichloromethane (20 mL). After

stirring for 1.5 h at room temperature, benzyl bromide (5.99 g, 35.0 mmol) was added drop-wise. The

solution was stirred for 3 h at room temperature and then quenched with 10 % hydrochloric acid (50

mL). The organic phase was separated and washed twice with water. After drying the solution with

sodium sulfate the solvent was removed and 7 dried in vacuo to yield orange oil. 1H NMR (CDCl3): δ

= 2.70 (d, 12 H, CH2), 3.52 (d, 12 H, CH2), 4.04 (s, 12 H, CH2), 4.59 (s, 12 H, CH2), 7.22–7.24 (m, 30

H, ArH).


8

Pentaerythritoltetrakis(3-(S-benzyltrithiocarbonyl)propionate) 7 (0.342 g, 0.192 mmol), purified





CHAPTER 3

55


H NMR (DMSO-


S-(4-vinyl)benzyl S’-propyltrithiocarbonate (VBPT) (9)

9

The synthesis was performed according to Zhao and coworkers.30

To a stirred solution of 1-

propanethiol (1.54 g, 20.0 mmol) in anhydrous methanol (16 mL) was slowly added a solution of

sodium methoxide (1.15 g, 20.2 mmol) in methanol (20 mL) under nitrogen atmosphere. After 2h

stirring carbon disulfide (1.90 g, 25.0 mmol) was added drop-wise, and the mixture was further stirred

at room temperature for 5 h. To the yellow solution obtained was added slowly 4-vinylbenzyl chloride

(3.72 g, 21.9 mmol) and the mixture was stirred overnight in nitrogen atmosphere. The mixture was

poured into water from which the organic product was extracted with dichloromethane. The separated

yellow dichloromethane solution was dried with sodium sulfate. The solution was concentrated to a

few milliliters and purified by flash column chromatography eluting with hexane. VBPT 9 was

isolated as yellow viscous oil. 1H NMR (CDCl3): δ = 1.04 (t, 3H, CH3CH2), 1.76 (m, 2H, CH3CH2),

3.37 (t, 2H, SCH2), 4.59 (s, 2H, ArCH2), 5.27 (d, 1H, CH2), 5.76 (d, 1H, CH2), 6.67 and 6.70 (ABq,

1H, CH), 7.30 and 7.34 (ABq, 4H, ArH).

CHAPTER 3

56

Hyperbranched PAN polymer (10)

10

VBPT 9 (7.32 g, 27.0 mmol), purified acrylonitrile (7.17 g, 135.0 mmol), AIBN (0.440 g, 2.7 mmol)

and dimethylformamide (75 mL) were added in a Schlenk flask, degassed by four freeze-pump-thaw

cycles, charged with nitrogen and sealed. The reaction mixture was heated to 80 °C in a thermostatic

oil bath. After a predetermined time, the content was precipitated in methanol, centrifuged at 5,000

rpm for 15 min, redissolved in DMF and precipitated in methanol again. After centrifugation the

solvent was removed and the polymer 10 was dried in vacuo. 1

H NMR (DMSO-d6): δ = 0.94

(CH3CH2CH2S, terminal trithiocarbonate functionality), 1.3–3.1 (CH2 and CH, VBPT and AN units),

3.17 and 3.36 (CH3CH2CH2S, terminal trithiocarbonate functionality), 4.11 (CH2(Ar)CHS, styryl unit

originated from reacted VBPT), 4.66 (CH(Ar)CH2S, VBPT unit), 4.74 (NCCHS, terminal AN unit),

6.7-7.5 (ArH, VBPT unit).

CHAPTER 3

57


In this work we synthesized polymers with linear, three-arm, four-arm, six-arm and hyperbranched

architecture by core-first Z-group approach RAFT polymerization.31

The kinetics of the RAFT

polymerization was examined to verify the controlled nature of the polymerization reactions.

Rheological experiments were performed with selected star-shaped and hyperbranched polymers to

prove the expected viscosity decrease of a polymer solution when a branched polymeric additive is

present. The intrinsic viscosity of branched poly(acrylonitrile)s in DMF solution were determined and

revealed a lower viscosity than linear PAN.32

Furthermore the influence of branched PANs on the flow

properties of DralonX (a commercial PAN) solutions in DMF were determined. Finally the kinetics

for the thermal conversion of PAN in the presence of hyperbranched additives is reported.

3.3.1 Synthesis of star-shaped and hyperbranched poly(acrylonitrile)s

Starting with the bifunctional initator dibenzyl trithiocarbonate 1, the trifunctional initiator

trimethylolpropane-tris-3-(S-benzyl-trithiocarbonyl)propionate) 3, the tetrafunctional initiator

pentaerythritol-tetrakis(3-mercaptopropionate) 5, the hexafunctional initiator

dipentaerythritolhexakis(3-mercaptopropionate) 7 and the inimer S-(4-vinyl)benzyl S’-

propyltrithiocarbonate 9 linear PAN 2, three-arm star-shaped PAN 4, four-arm star-shaped PAN 6,

six-arm star-shaped PAN 8 and hyperbranched PAN 10 were prepared using RAFT polymerization.

The molecular characteristics of the star-shaped polymers as determined with size exclusion

chromatography (SEC) in DMF all showing a molecular weight of Mn = 45,000 g/mol with dispersity

indices of 1.4. Three different hyperbranched polymers with molecular weights of 2,200 – 17,600

were prepared with polydispersity indices of 2.2 – 2.6 (see Table 1).

CHAPTER 3

58

Table 1 Molecular characteristics of synthesized PAN based additives. [M] = molar

concentration of monomer; [R] = molar concentration of RAFT agent; [I] = molar

concentration of radical initiator (SEC in DMF at 30°C against PMMA standards).

Name Architecture Mn [g/mol] Mw [g/mol] Đ [M]/[R]/[I] t [min]

2a Linear 44,000 62,000 1.4 800 / 1 / 1 110

4a 3-arm star 45,000 59,000 1.3 800 / 1 / 1 120

6a 4-arm star 45,000 63,000 1.4 800 / 1 / 1 180

8a 6-arm star 45,000 63,000 1.4 800 / 1 / 1 360

10a hyperbranched 2,200 5,700 2.6 50 / 10 / 1 120

10b hyperbranched 12,000 26,400 2.2 50 / 10 / 1 270

10c hyperbranched 17,600 38,000 2.2 50 / 10 / 1 470

The kinetic data for the polymerization of acrylonitrile with bi-, tri-, tetra-, and hexafunctional

initiators (1, 3, 5, 7) were determined. The conversion was calculated from the 1H NMR spectra as

shown exemplary for the polymerization of acrylonitrile with the bifunctional initiator 1 (see Table 2).

Noticeable is the difference in molecular weight Mn determined by two different methods, SEC and

NMR spectroscopy. The SEC is a relative method relating the hydrodynamic radius of the polymer to

a PMMA standard whereas interpretation of NMR spectra relies on the quantitative integration of

known proton signals. The kinetic data for the trifunctional initiator 3, the tetrafunctional initiator 5

and the hexafunctional initiator 7 are displayed in Appendix B.

CHAPTER 3

59

Table 2 Kinetic data for the polymerization of acrylonitrile with the bifunctional initiator 1.

time [min] Mn (SEC) [g/mol] Mn (NMR) [g/mol] aMn,th (NMR) [g/mol] conversion [%]

30 26,000 963 952 1.9

60 32,000 4,641 4,911 13.4

110 40,000 9,771 9,651 27.1

180 51,000 16,332 15,886 45.1

300 53,000 25,529 24,325 69.5

445 56,000 30,818 28,589 81.8

630 68,000 32,551 32,047 91.8

1455 70,000 35,027 32,369 92.8

a Mn,th = MW(1) + MW(AN) · ([AN]0 /[1]0) · conversion.

Plotting the molecular weight Mn for the linear PAN 2, the three-arm star PAN 4, the four-arm star

PAN 6 and the six-arm star PAN 8 versus monomer conversion, in all cases a linear increase in the

molecular weight up to 80% AN conversion was observed (see Figure 2).

CHAPTER 3

60

0 10 20 30 40 50 60 70 80 90 100

0

10000

20000

30000

40000

50000

60000 linear PAN 2

3-arm star-shaped PAN 4



Mn (

NM

R)

[g/m

ol]

conversion [%]

Figure 2 Kinetic data of RAFT polymerization: Mn (NMR) versus conversion for linear PAN 2,

three-arm star-shaped PAN 4, four-arm star-shaped 6 and six-arm star-shaped 8

polymers.

3.3.2 Comparison of the intrinsic viscosity of branched poly(acrylonitrile)s obtained by RAFT

polymerization

The viscosity decreasing effect of the synthesized additives can be expressed by the intrinsic viscosity

[η] which is defined as the specific volume of a polymer at infinite dilution. Primary data were

determined by a SEC setup with DMF as eluent using an on-line viscosity detector. When the intrinsic

viscosity [η] is plotted against the molecular weight Mn the comparison of the oligo-arm star polymers

4, 6, 8 and the hyperbranched additive 10a with the linear polymer 2 confirms the theory of higher

segment density in branched molecules than found with linear chains (see Figure 3).25

The viscosity of

the polymers is defined by their intermolecular entanglement. The reduced entanglement overlap is a

result of a more compact polymer coil of branched polymers which can be described by the

contraction factor g’ = [η]b / [η]l, where [η]b and [η]l are the intrinsic viscosities of branched and linear

CHAPTER 3

61

polymers with the same molecular weight, respectively.32

The intrinsic viscosities of polymers with

different architectures and with a molecular weight of Mn = 45,000 g/mol are compared and g’ is

calculated (see Table 3). The linear polymer showed, as expected, the highest value for intrinsic

viscosity with 5.28 L/g, but the values of the oligo-armed additives were not that much lower with

4.03 – 4.14 L/g and contraction factor values of g’ = 0.76 – 0.78. The hyperbranched polymer

displayed the lowest intrinsic viscosities with only 1.35 L/g and a low value of g’ = 0.26. The values

of the contraction factors with 0.76 – 0.78 fit reasonably well with literature values of 3- to 6-armed

star-shaped polymers33

expressing their shape similarity to linear polymers, whereas the contraction

factor of 0.26 fits reasonably well with literature values of hyperbranched polymers34

expressing their

shape similarity to hyperbranched polymers. Capillary rheology must be applied to verify if the

decrease in polymer entanglement of the star-shaped polymers also helps decrease further the viscosity

of the DralonX polymer solution with added star-shaped and hyperbranched polymers.

1000 10000 100000 1000000

1

10

[] [L

/g]

Mn [g/mol]

linear PAN 2




hyperbranched PAN 10

Figure 3 Mark-Houwink-Sakurada plots of the rheological additives 2, 4, 6, 8 and 10 obtained

by RAFT polymerization (determined by SEC on-line viscosity detector).

CHAPTER 3

62

Table 3 Intrinsic viscosity and contraction factor g’ of PAN polymers of Mn = 45,000 g/mol

measured with SEC in DMF at 30°C against PMMA standards.

Name Architecture [η] [L/g] g’

2a Linear 5.28 /

4a 3-arm 4.05 0.77

6a 4-arm 4.14 0.78

8a 6-arm 4.03 0.76

10a hyperbranched 1.35 0.26

3.3.3 Influence of branched PANs on the flow properties of DralonX solutions in DMF

The process of carbon fiber precursor spinning utilizes solutions of 25-30 wt.-% DralonX polymer in

DMF at temperatures of 90°C.35

The spinning dope is pumped from a continuous stirred-tank reactor

through tubes of different diameters until the arrival at the spinning nozzle. Such a highly concentrated

solution exhibits shear rate dependent dynamic viscosities dependent on the tube diameter during the

flow: (i) low shear rates from laminar flow in the tube (~ 1,000 s-1

) and (ii) high shear rates at the

spinning nozzle (~ 40,000 s-1

). For the determination of the effect of the synthesized additives on the

solution viscosity during the flow, capillary rheology is applied to measure the dynamic viscosity over

the shear rate range of 50 – 100,000 s-1

. The viscosity curves of 30 wt.-% DralonX solutions in DMF

with 3 wt.-% linear 2a, three-arm star-shaped 4a, four-arm star-shaped 6a and six-arm star-shaped 8a

additives showed a non-Newtonian viscoelastic behavior in the measured shear rate range (see Figure

4). The drop in dynamic viscosity with increasing shear rate is called shear thinning and is a result of

decreasing intermolecular polymer entanglement through shape deformation.

CHAPTER 3

63

10 100 1000 10000 100000

10

100

1000

DralonX

DralonX with 3 wt.-% 2a




[P

a·s

]

shear rate [s-1]

110 s-1

590 s-1

3300 s-1

30000 s-1

Figure 4 Dynamic viscosity curves of 3 wt.-% additives in 30 wt.-% DralonX solutions in DMF

at 80°C.

The dynamic viscosities of the 30 wt.-% DralonX solution with and without additives are similar (see

Table 4). Adding the additives in 3 wt.-% showed mostly a slight increase of solution viscosity except

of six-arm star-shaped additive 8a which showed no significant influence. The expectation of a

viscosity decreasing behavior of the oligo-arm star shaped additives was not fulfilled. Most likely this

is a result of the only slightly lower values of the intrinsic viscosities of the star additives compared to

the linear additive (see Figure 3).

CHAPTER 3

64

Table 4 Dynamic viscosities of 3 wt.-% additives in 30 wt.-% DralonX solutions in DMF at

80°C.

Shear rate [s-1

] 110 590 3,300 30,000

DralonX [Pa·s] 1,420 790 320 76

DralonX with linear additive 2a [Pa·s] 1,540 840 340 82

DralonX with three-arm additive 4a [Pa·s] 1,450 810 330 81

DralonX with four-arm additive 6a [Pa·s] 1,450 810 330 80

DralonX with six-arm additive 8a [Pa·s] 1,390 790 320 76

The hyperbranched additives show considerably lower intrinsic viscosities than the star-shaped

additives. Capillary rheology of 30 wt.-% polymer solutions in DMF with the hyperbranched additives

10a, 10b and 10c of molecular weights from Mn 2,200 – 17,600 g/mol showed a significant reduction

in viscosity, the strongest effect appearing in the shear rate range of 280 – 3,200 s-1

from the additive

10a with Mn 2,200 g/mol with a 15% reduction of viscosity at 1,200 s-1

from 550 Pa·s to 470 Pa·s (see

Figure 5 and Table 5). The shear thinning behavior of the DralonX polymer solution becomes

pronounced by the hyperbranched additive which probably acts as a spacer between polymer coils,

reducing the entanglement and therefore decreasing the viscosity. The RAFT inimer VBPT 9 as

additive shows less viscosity decrease than the hyperbranched additives proving that viscosity

decrease by the additives 10a-c is a result of the polymer chain interactions.

CHAPTER 3

65

10 100 1000 10000 100000

10

100

1000

[

Pa

·s]

shear rate [s-1]

DralonX

DralonX with 3 wt.-% 9


DralonX with 3 wt.-% 10b

DralonX with 3 wt.-% 10c

280 s-1

590 s-1

1200 s-1

3200 s-1

Figure 5 Dynamic viscosity curves of 30 wt.-% DralonX solutions in DMF at 80°C with 3 wt.-

% additives of VBPT 9, hyperbranched 10a, 10b and 10c.

Table 5 Dynamic viscosities of 3 wt.-% hyperbranched additives in 30 wt.-% DralonX

solutions in DMF at 80°C displayed at four different shear rates.

Shear rate [s-1

] 280 590 1,200 3,200

DralonX [Pa·s] 1,045 780 550 330

DralonX with VBPT 9 [Pa·s] 1,000 760 540 315

DralonX with hyperbranched additive 10a [Pa·s] 930 680 470 290

DralonX with hyperbranched additive 10b [Pa·s] 975 685 490 305

DralonX with hyperbranched additive 10c [Pa·s] 930 690 500 310

CHAPTER 3

66

The aim of the addition of rheological additives is a lower or at least unchanged dynamic viscosity of

solutions with a polymer mass concentration higher than 30 wt.-%. Dynamic viscosities of DralonX

polymer mass concentrations of 31 and 32 wt.-% with and without the hyperbranched additive 10a

have been compared (see Figure 6). The additive 10a was chosen due to its strongest viscosity

decreasing effect of all synthesized hyperbranched additives. The measured viscosities at five selected

shear rates of 280, 590, 1200, 3200 and 29,000 s-1

are compared (see Table 6). For the 30 wt.-%

DralonX solution in DMF without additive the viscosity decreases in an exponential way as expected

by the shear thinning behavior of polymeric solutions, starting at a shear rate of 280 s-1

from η = 1,045

Pa·s to a shear rate of 29,000 s-1

with η = 78 Pa·s. At this high shear rates the shear deformation and

resulting disentanglement of the polymer coils has progressed considerably. For the further increase of

the shear rate a theoretical plateau of a constant viscosity is expected where a maximal

disentanglement of polymer coils exists.

With 3 wt.-% additive 10a in the solution the viscosity decreases at the shear rate 280 s-1

down to 930

Pa·s, an 11% decrease. At the high shear rate of 29,000 s-1

the viscosity is η = 76 Pa·s, showing only a

decrease of 3%. When increasing the mass concentration of DralonX polymer the same viscosity

decreasing behavior is displayed. The interesting observation is now how strongly the viscosity

decrease is for the 31 or 32 wt.-% DralonX solutions with 3 wt.-% 10a additive. The experiments

showed that for the 31 wt.-% DralonX solution with 10a from a shear rate range of 280 – 1,200 s-1

the

dynamic viscosities remained below the dynamic viscosities of a 30 wt.-% DralonX solution without

additive.

The hyperbranched additives with their relatively low values of intrinsic viscosity displayed the

expected effect of decreasing the viscosity of a polymer spinning solution, at least for a low shear rate

range. A decrease of viscosity at higher shear rates should be possible by a further increase of the

degree of branching in the additive while keeping the dispersity Đ low.

CHAPTER 3

67

10 100 1000 10000 100000

10

100

1000

[

Pa

·s]

shear rate [s-1]

30% DralonX without 10a

30% DralonX with 3 wt.-% 10a





280 s-1

590 s-1

1200 s-1

3200 s-1

29000 s-1

Figure 6 Viscosity curves of 30 wt.-% DralonX solutions in DMF at 80°C with and without 3

wt.-% hyperbranched additives.

Table 6 Dynamic viscosities of DralonX solutions in DMF at 80°C with and without

hyperbranched additive at five different shear rates.

Shear rate [s-1

] 280 590 1,200 3,200 29,000

30% DralonX without 10a [Pa·s] 1,045 780 550 330 78

30% DralonX with 10a [Pa·s] 930 680 470 290 76

31% DralonX without 10a [Pa·s] 1,210 890 630 370 86

31% DralonX with 10a [Pa·s] 1,020 735 540 345 85

32% DralonX without 10a [Pa·s] 1,440 1,035 720 420 96

32% DralonX with 10a [Pa·s] 1,160 835 620 385 95

CHAPTER 3

68

3.3.4 Kinetic parameters for the thermal conversion of PAN in the presence of hyperbranched

PAN additives

For this work the hyperbranched PAN 10a was applied in the spinning of a PAN-based carbon fiber

precursor. For control over the resulting carbon fiber it is important to know the thermal behavior of

the PAN-based precursor with the additive. In DSC thermograms these processes can be assigned to

the following two steps: the stabilization process, in air atmosphere between 200 – 400°C and the

carbonization process, in nitrogen atmosphere between 500 – 1200°C. The stabilization process in

PAN-based carbon fibers consists of many different chemical reactions which can be represented by

three major energetic reactions: dehydrogenation, cyclization and oxidation (see Figure 7).36,37,38,39

These reactions are important for the mechanical properties of the resulting carbon fiber and will be

described with kinetic parameters. Every additive which is not of the same chemical composition as

the precursor material will influence the reactions and will make modifications to the heating profile

necessary. Ideally the additive either will not influence the kinetic parameters of the stabilization

process of the precursor fibers or have a beneficial catalytic influence by lowering the required energy

cost for the reaction.

Figure 7 DSC thermogram of PAN-based CF precursor in air atmosphere.

CHAPTER 3

69

For the evaluation of the kinetic parameters of the stabilization reactions the apparent activation

energy Ea was determined by the Kissinger40

method. The other method mostly used in literature is the

Ozawa41

method but for the PAN copolymer system the two methods yield similar results.42

Four DSC

experiments were carried out with DralonX and Precursor-T films with and without 3 wt.-%

hyperbranched additive (Mn 2200 g/mol) at five different heating rates (β = 2, 5, 10, 15 and 20 K/min)

in synthetic air atmosphere (see Figure 8, Appendix B). The influence on the stabilization reactions

was determined by the shift of the first exothermic peak temperature Tp of the

dehydrogenation/cyclization reaction (see Table 7).

Figure 8 DSC thermograms of Precursor-T PAN copolymer without additive at five different

heating rates (β = 2, 5, 10, 15 and 20 K/min).

150 200 250 300 350 400Temperatur /°C

-10

-8

-6

-4

-2

0

DSC /(mW/mg)

[1]

[2][3]

[4]

[5]

Exo

2 K/min 5 K/min 10 K/min 15 K/min 20 K/min

CHAPTER 3

70

Table 7 Peak temperatures for the powdered PAN copolymers determined by DSC in synthetic

air.

Heating rate

[K/min]

Tp of DralonX

Tp of DralonX with 3

wt.-% additive

Tp of

Precursor-T

Tp of Precursor-T with

3 wt.-% additive

2 278.6 278.4 263.8 264.2

5 295.2 294.0 276.7 276.1

10 312.0 308.4 288.9 286.2

15 323.1 318.0 296.5 294.1

20 329.1 322.9 302.9 299.8

The mathematical expression for the Kissinger method is as follows:

where Tp is the peak temperature; β is the heating rate; Ea is the apparent activation energy; A is the

pre-exponential factor, and R is the molar gas constant. According to this equation, Ea and A can be

calculated from a linear fitted plot of ln(β/Tp2) against 1/RTp (slope and intercept, respectively). The

plotted fits of the DralonX and Precursor-T with and without hyperbranched additive are displayed

below (see Figure 9).

CHAPTER 3

71

0,00012 0,00015 0,00018 0,00021 0,00024 0,0018

-12,0

-11,5

-11,0

-10,5

-10,0

-9,5

ln(

/T2 p)

1/RTp

DralonX

DralonX + additive

0,0002025 0,0002100 0,0002175 0,0002250

-12,0

-11,5

-11,0

-10,5

-10,0

-9,5

ln(

/T2 p)

1/RTp

Precursor-T

Precursor-T + additive

Figure 9 Linear fitted plots from the kinetics for the cyclization/dehydrogenation reaction

during the stabilization of PAN-based precursors by the Kissinger method. Left:

DralonX and DralonX with hyperbranched additive; right: Precursor-T and Precursor-

T with hyperbranched additive.

The calculated apparent activation energies from the slope and the pre-exponential factor A from the

intercept are listed for the different copolymers (see Table 8). In literature, Devasia and coworkers

found the apparent activation energy of a PAN copolymer with itaconic acid to be 153.2 kJ/mol which

is comparable to Precursor-T in chemical composition and activation energy Ea of 142.0 kJ/mol.43

The

addition of the additive increases the apparent activation energy by roughly 11%. The increase can be

explained by the introduction of the chemically different RAFT inimer 9 with its sulfur groups and

aromatic vinylbenzyl systems.

The thermal analysis shows that the rheological additives have an influence on the thermal behavior of

the precursor. The lower peak temperatures of the dehydrogenation/cyclization reactions are beneficial

because the required heating temperature can be lowered and therefore energy can be saved. The

apparent activation energies increase slightly due to a small amount of non-carbon compounds in the

precursor. The pre-exponential factor A describes a relationship between temperature and rate

coefficient. The increased values represent a beneficial increase of the conversion rate of the reaction.

CHAPTER 3

72

Table 8 Calculated apparent activation energies and measured reaction enthalpies of DralonX

and Precursor-T with and without hyperbranched additive in synthetic air atmosphere.

Sample Apparent activation energy Ea

[kJ/mol]

Pre-exponential factor A

[min-1

]

DralonX 113.6 5.44 x 109

DralonX with 3 wt.-% additive 129.1 1.82 x 1011

Precursor-T 142.0 8.48 x 1012

Precursor-T with 3 wt.-% additive 156.5 2.34 x 1014

3.4 Conclusion

The synthesized star polymers with Mn of 45,000 g/mol showed no beneficial entanglement disruption

under the applied conditions due to its low branching factor. Only with the increase of the branching in

the hyperbranched additive, the desired effect of decreased solution viscosity could be observed,

although only within a range of small shear rates. A relative increase of 3% of the mass concentration

in the carbon fiber precursor spinning dope was shown to be possible for low shear rates. Further

optimization of the branching and molecular weight could lead to a more pronounced effect of shear

thinning over a broader shear rate range. The introduction of the PAN rheological additive with sulfur

functionality and aromatic benzyl groups increased the apparent activation energy. With this

knowledge improved thermal treatment profiles can be developed to achieve the optimal properties of

the resulting carbon fibers.

3.5 References

(1) Hadjichristidis, N.; Pitsikalis, M.; Pispas, S.; Iatrou, H. Chem. Rev. 2001, 101, 3747.

CHAPTER 3

73

(2) Gao, C.; Yan, D. Y. Prog. Polym. Sci. 2004, 29, 183.

(3) Zhou, L.; Gao, C.; Xu, W.; Wang, X.; Xu, Y. H. Biomacromolecules 2009, 10, 1865.

(4) Stiriba, S. E.; Kautz, H.; Frey, H. J. Am. Chem. Soc. 2002, 124, 9698.

(5) Gao, H. F.; Matyjaszewski, K. Prog. Polym. Sci. 2009, 34, 317.

(6) Voit, B. I.; Lederer, A. Chem. Rev. 2009, 109, 5924.

(7) Simon, P. F. W.; Müller, A. H. E.; Pakula, T. Macromolecules 2001, 34, 1677.

(8) Liu, B.; Kazlauciunas, A.; Guthrie, J. T.; Perrier, S. Macromolecules 2005, 38, 2131.

(9) Otsu, T. J. Polym. Sci. Part A Polym. Chem. 2000, 38, 2121.

(10) Hawker, C. J.; Bosman, A. W.; Harth, E. Chem. Rev. 2001, 101, 3661.

(11) Matyjaszewski, K.; Xia, J. Chem. Rev. 2001, 101, 2921.

(12) Kamigaito, M.; Ando, T.; Sawamoto, M. Chem. Rev. 2001, 101, 3689.


(14) Barner-kowollik, C.; Perrier, S. J. Polym. Sci. Part A Polym. Chem. 2008, 46, 5715.

(15) Chaffey-Millar, H.; Stenzel, M. H.; Davis, T. P.; Coote, M. L.; Barner-Kowollik, C.

Macromolecules 2006, 39, 6406.

(16) Gao, H. F.; Min, K.; Matyjaszewski, K. Macromolecules 2009, 42, 8039.

(17) Baek, K. Y.; Kamigaito, M.; Sawamoto, M. Macromolecules 2001, 34, 7629.

(18) Gao, H. F.; Matyjaszewski, K. Macromolecules 2008, 41, 1118.

(19) Hovestad, N. J.; van Koten, G.; Bon, S. A. F.; Haddleton, D. M. Macromolecules 2000, 33,

4048.

(20) Whittaker, M. R.; Urbani, C. N.; Monteiro, M. J. J. Am. Chem. Soc. 2006, 128, 11360.

(21) Altintas, O.; Hizal, G.; Tunca, U. J. Polym. Sci. Part A Polym. Chem. 2006, 44, 5699.

(22) Zhao, Y. L.; Chen, Y. M.; Chen, C. F.; Xi, F. Polymer (Guildf). 2005, 46, 5808.

(23) Zhao, Y. L.; Shuai, X. T.; Chen, C. F.; Xi, F. Macromolecules 2004, 37, 8854.

(24) Kasko, A. M.; Heintz, A. M.; Pugh, C. Macromolecules 1998, 31, 256.

(25) Burchard, W. In Advances in Polymer Science; 1999; Vol. 143.

(26) Tomalia, D. A.; Fréchet, J. M. J. Dendrimers and other Dendritic Polymers; Tomalia, D. A.;

Fréchet, J. M. J., Eds.; John Wiley & Sons, Ltd.: West Sussex, UK, 2001.

CHAPTER 3

74

(27) Graessley, W. W.; Masuda, T.; Roovers, J. E. L.; Hadjichristidis, N. Macromolecules 1976, 9,

127.

(28) Aoyagi, N.; Endo, T. J. Polym. Sci. Part A Polym. Chem. 2009, 47, 3702.

(29) Boschmann, D. Sternpolymere mittels RAFT- Polymerisation, 2008.

(30) Zhang, C.; Zhou, Y.; Liu, Q.; Li, S.; Perrier, S.; Zhao, Y. Macromolecules 2011, 44, 2034.

(31) Barner-Kowollik, C.; Davis, T. P.; Stenzel, M. H. Aust. J. Chem. 2006, 59, 719.

(32) McKee, M. G.; Unal, S.; Wilkes, G. L.; Long, T. E. Prog. Polym. Sci. 2005, 30, 507.

(33) Balke, S. T.; Mourey, T. H.; Robello, D. R.; Davis, T. a.; Kraus, A.; Skonieczny, K. J. Appl.

Polym. Sci. 2002, 85, 552.

(34) Wang, W.-J.; Wang, D.; Li, B.-G.; Zhu, S. Macromolecules 2010, 43, 4062.






(40) Kissinger, H. E. Anal. Chem. 1957, 29, 1702.

(41) Ozawa, T. Bull. Chem. Soc. Jpn. 1965, 38, 1881.


(43) Devasia, R.; Nair, C. P. R.; Sivadasan, P.; Katherine, B. K.; Ninan, K. N. J. Appl. Polym. Sci.

2003, 88, 915.

CHAPTER 3

75

Appendix B

a) DSC thermograms

Precursor-T without additive

Precursor-T with hyperbranched additive 10a

150 200 250 300 350 400Temperatur /°C

-10

-8

-6

-4

-2

0

DSC /(mW/mg)

[1]

[2][3]

[4]

[5]

Exo

150 200 250 300 350 400Temperatur /°C

-8.00

-7.00

-6.00

-5.00

-4.00

-3.00

-2.00

-1.00

0

1.00

DSC /(mW/mg)

[1]

[2]

[3]

[4][5]

Exo

CHAPTER 3

76

DralonX without additive

DralonX with hyperbranched additive 10a

150 200 250 300 350 400Temperatur /°C

-25

-20

-15

-10

-5

0

DSC /(mW/mg)

[1]

[2]

[3]

[4]

[5]

Exo

150 200 250 300 350 400Temperatur /°C

-16

-14

-12

-10

-8

-6

-4

-2

0

DSC /(mW/mg)

[1]

[2]

[3][4][5]

Exo

CHAPTER 3

77

b) Kinetic data of RAFT polymerization

Kinetic data for the polymerization of AN with the trifunctional initiator 3.

time

[min]

Mn (SEC)

[g/mol]

Mn (NMR)

[g/mol]

aMn,th (NMR)

[g/mol]

conversion [%]

20 / 1,251 1,748 3.0

40 32,000 1,605 3,550 9.5

60 36,000 1,959 5,587 16.7

90 41,000 2,489 8,406 26.8

120 46,000 3,020 10,970 35.9

180 53,000 4,081 14,861 49.8

300 58,000 6,203 19,526 66.4

420 61,000 8,326 23,015 78.9

610 61,000 11,686 24,994 85.9

1325 63,000 24,332 27,534 95,0


Kinetic data for the polymerization of AN with the tetrafunctional initiator 5.

time

[min]

Mn (SEC)

[g/mol]

Mn (NMR)

[g/mol]

aMn,th (NMR)

[g/mol]

conversion [%]

5 / 1,242 1,154 0,0

10 / 1,331 1,154 0,0

20 28,000 1,507 1,463 0,7

35 28,000 1,773 2,738 3,4

60 32,000 2,215 6,642 11,8

CHAPTER 3

78

113 39,000 3,152 15,798 31,5

180 44,000 4,337 24,882 51,0

247 48,000 5,522 31,432 65,1

1449 53,000 26,782 45,307 95,0


Kinetic data for the polymerization of AN with the hexafunctional initiator 7.

time

[min]

Mn (SEC)

[g/mol]

Mn (NMR)

[g/mol]

aMn,th (NMR)

[g/mol]

conversion [%]

15 30,000 2,046 2,266 0,8

30 28,000 2,311 4,448 4,4

75 32,000 3,107 14,023 20,3

130 37,000 4,080 25,041 38,6

200 41,000 5,318 36,851 58,2

270 45,000 6,556 43,756 69,7

362 46,000 8,183 47,685 76,2

1318 48,000 25,092 58,224 93,7


CHAPTER 4

79

Characterization of absorption additives for direct energy

transfer by LASER light into poly(acrylonitrile) based

carbon fiber precursor

4.1 Introduction

Absorption additives are compounds which absorb electromagnetic radiation. They are divided into

dyes and pigments where the difference lies in the solubility in a vehicle (or binder). Dyes are soluble

and pigments are insoluble. The absorption additives are used for different applications such as

promoting the coloration and brilliance of textiles, they can be photosensitizers in photographic films,

thermal printing, optical brighteners, LASER absorption promoters for resin curing and much more.1,2

Applying pigments in dispersion require a fine powder for stable dispersion. If this dispersion will be

processed by spraying, it must be able to flow unhindered through micrometer sized nozzles.

Therefore the size of the pigment particles or aggregates in the dispersion must be smaller in size than

the nozzle diameter or the nozzle will be blocked.

In recent years carbon black pigments like carbon nanotubes (CNT), soot, graphene and asphaltene

(isolated from bitumen) have been added to polymeric solutions or paints to achieve better product

properties such as improved mechanical strength, higher conductivity or stronger coloring effect.3,4,5

The hydrophobic nature of carbon necessitates surfactants such as cetrimonium bromide (CTAB) to

reduce the aggregation by lowering the surface tension of the typically aqueous solvent. To avoid the

use of surfactants, the carbon black pigments have to be treated (i) physically by extensive ultrasonic

or mechanical dispersion or (ii) chemically by a coupling reaction or grafting to yield

“dyes”.6,7,8,9,10,11,12,13

Combining the concept of dyes and carbon black pigments novel applications open up due to the

strong light absorption of the black particles over a broad range of the electromagnetic spectrum. The

CHAPTER 4

80

absorption allows to directly transfer energy into the target, for example carbon black additives are

applied for LASER assisted sintering of poly(ether ether ketone) (PEEK).14

The carbon fiber

manufacturer Toho Tenax Co., Ltd. holds a patent about carbon black absorption additives for the

microwave-assisted thermal treatment of poly(acrylonitrile) (PAN) based precursor fibers.15

The aim of this work is the research on energy efficient alternatives of CF precursor heating by the

application of LASER light absorption additives. In contrast to the conventional heating by convection

of the CF precursor fiber in an oven, the LASER light adsorption would transfer the energy directly

into the material and would allow precise control over the thermal reaction for the formation of the

CF. Through the application of low-priced LASER light diodes with a typical NIR wavelength of 980

nm the energy costs could be lowered significantly and the price of the resulting CF, therefore making

CFs more interesting for the automotive industry. With a tailored heating program CFs with optimal

mechanical properties could be produced, the CF consisting theoretically of pure carbon. The main

problem in this approach of alternative heating is the non-existent NIR absorption of PAN. To achieve

the absorption of LASER light in the PAN-based precursor fibers, two routes are thinkable: (i) the

polymerization of comonomer with NIR light absorbing functionalities or (ii) absorption additives

have to be dispersed into the CF precursor. The latter approach was chosen to realize a low-price

solution through absorption additives. This addition will introduce impurities (elements other than

carbon) as well into the fiber and will have a detrimental effect on the mechanical properties of the

resulting CF. To minimize the introduction of impurities, the examined absorption additives are

selected from the class of carbon black materials which are made of over 90% carbon.

In this work we examine several carbon black candidates through their NIR absorption in N,N’-

dimethylformamide (DMF). Routes of physical and chemical modification of the carbon black

pigments have been explored for a homogeneous and stable dispersion in a spinning dope – a PAN-

based polymer solution in DMF. For the research on the effect of the carbon black additive on the

kinetics and enthalpy of the thermal reactions the experiments were done with a typical plate heating

DSC setup since a DSC setup with LASER light heating was not available for use.

CHAPTER 4

81


4.2.1 Materials

Acrylonitrile (AN, ≥99%, Aldrich) was purified by column chromatography on Al2O3 (activated,

neutral, Brockmann I, Aldrich) before use. Azobisisobutyronitrile (AIBN, 98+%, Fluka) was

recrystallized twice from methanol before use. The graphene source was xGnP graphene nano platelets

grade C with an average surface area of 750 m2/g by XG Sciences. The asphaltene source was

heptane-washed bitumen (Nynas) residue. Multi-walled carbon nanotubes (MW-CNT) Baytubes

C150P (Bayer MaterialScience), N,N’-dimethylformamide (DMF, HPLC grade, VWR), methanol

(dried, Merck), toluene (98%, Merck), sulfuric acid (98%, Merck) and potassium permanganate

(KMnO4, ≥99%, Aldrich) were used without further purification.

DralonX copolymer (prepared from acrylonitrile, methyl acrylate and ionic comonomer) and

Precursor-T copolymer (prepared from acrylonitrile, methyl acrylate and itaconic acid) were supplied

by Dralon GmbH, Dormagen. Graphene oxide was prepared from a natural graphite source according

to a modified Hummers method3 and reduced graphene oxide (reduction by hydrazine) was

synthesized by Manuel Noack from our institute.



4.2.2 Methods




13C

NMR spectra were made by first order analysis of the spectra.

CHAPTER 4

82

Near infrared (NIR) spectroscopy was performed using a Bruker FT-NIR Matrix-F with a submersible

probe head. The spectrum consists of an average of 16 scans with a spectral resolution of 8 cm-1

in the

range of 15700 – 4000 cm-1

.

Analysis of the optical microscopy images was performed using the software ImageJ v1.43u by W.

Rasband with a circularity parameter between 0 and 1.

Polymeric films were pulverized in a ball mill from Perkin Elmer for 4 min at swinging frequency of

30 Hz.


differential scanning calorimeter in nitrogen or synthetic air atmosphere. Samples were measured in

perforated closed aluminum pans using approximately 4 ± 0.05 mg of sample with a heating rate of 10


°C. The heat flow was determined as a function of temperature. A sigmoidal baseline was applied for

the integration of the signal area.

Transmission electron microscopy (TEM) was measured on Zeiss LibraTM 120 (Carl Zeiss,

Oberkochen, Germany). The electron beam accelerating voltage was set at 100 kV. A drop of the

sample was trickled on a silica grid. Before being placed into the TEM specimen holder, the copper

grid was air-dried under ambient conditions.

The height of the graphene flakes was observed with AFM (Bruker Dimension Icon with OMCL tips;

spring constant 44.2-49.9 N/m, resonant frequency 304.3-316.5 kHz) at room temperature. For the

experiments 20 μL of DMF dispersions were spin coated at 1500 rpm for 30 s onto a silica wafer,

which was activated by plasma treatment for 60 s (Plasma Activate Flecto 10 USB, 100 W, 0.2 mbar).

CHAPTER 4

83

4.2.3 Preparation and synthesis

Asphaltene (1)

1

Nynas bitumen (1 g) in heptane (40 mL) was sonicated for 30 min at room temperature. The black

mixture was centrifuged at 6000 rpm for 10 min and the supernatant liquid was discarded. The black

residue was again sonicated in heptane and centrifuged. After drying a black powder with the

approximate structure of 1 was obtained. 1H NMR (CDCl3): δ = 0.87–0.90 (t, CH3), 1.27 (m, CH).

Asphaltene-g-PAN (5)

5

CHAPTER 4

84

Purified acrylonitrile (1.1 g, 20.7 mmol), asphaltene 1 (15 mg), AIBN (0.060 g, 0.4 mmol) and toluene

(7 g) were added in a Schlenk flask. The degassing was done by three freeze-pump-thaw cycles,

afterwards the Schlenk flask was charged with nitrogen and sealed. The polymerization tube was

heated to 80 °C in a thermostatic oil bath for 30 min. Afterwards the content was first precipitated

twice in methanol and centrifuged at 5000 rpm for 2 min, then precipitated twice with toluene and

centrifuged again. The grafted polymer with the approximate structure of 5 was dried in vacuo before

use. 1H NMR (DMSO-d6): δ = 1.90–2.20 (m, 2 H, CH2), 3.00–3.25 (m, 1 H, CH), 7.10–7.40 (m, ArH).

SEC (DMF): Mn = 28,000 g/mol Mw = 135,000 g/mol, Đ = 4.8.

Furthermore multi-walled carbon nano tubes 2 with an outer mean diameter of 13 nm, inner mean

diameter of 4 nm, length greater than 1 µm and a bulk density of 130-150 kg/m3, xGnP graphene nano

platelets 3 grade C with an average surface area of 750 m2/g and graphene oxide 4 as well as reduced

graphene oxide 4a were used as received.


In this work we used four carbon black candidates (asphaltene 1, MW-CNT 2, graphene nano platelets

3 and graphene oxide 4) and synthesized a DMF-soluble carbon black pigment through “grafting-

from” polymerization (polymer 5).12

For the four carbon black candidates the NIR LASER light

absorbance at 980 nm was determined in DMF. For the applicability in a spinning process the

dispersion of the four carbon black candidates in a DralonX (a commercial PAN) film was examined

by optical microscopy and the size of the PAN grafted carbon black pigment by electron microscopy.

Finally the kinetics for the thermal conversion of PAN in the presence of graphene is presented.

CHAPTER 4

85

4.3.1 NIR LASER light absorption of carbon black candidates

Diode LASERs are cost efficient sources for energy transfer in the form of thermic radiation compared

to the energy inefficient oven heating by convection. The poly(acrylonitrile) based DralonX polymer

has no functionalities which can absorb infrared LASER light, the percentage of absorption in the

range of 15,000 to 5,000 cm-1

is effectively zero (see Figure 1).

500 1000 1500 2000 2500

0

10

20

20000 10000 6667 5000 4000

wave number [cm-1]

ab

so

rptio

n [

%]

wave length [nm]

Figure 1 Absorption spectrum of poly(acrylonitrile) based DralonX polymer film.

During heat treatment up to 300 °C the linear PAN chains oxidize and cyclize to aromatic ring systems

and the polymer changes color from white over yellow to orange brown ending with black and with

this change the absorption ability improves significantly. A dark DralonX polymer film heated up to

260 °C absorbed 40% of the NIR LASER light at 980 nm. Therefore, an increase in absorption of the

white precursor fiber is necessary, leading to the idea of introducing black absorption additives. The

absorption additives are selected from carbon black modifications and hence, being made out of

carbon - the same material as the resulting carbon fiber - structure defects from the additives

CHAPTER 4

86

detrimental to fiber strength are expected to be minimal. The applied carbon black candidates consist

of asphaltene 1 isolated from bitumen, multi-walled carbon nano tubes (MW-CNT) 2 with an average

outer diameter of 13-15 nm, graphene nano platelets 3 and graphene oxide (GO) 4 from a natural

graphite source (see Figure 2). All candidates contain alkyl side chains and oxygen functionalities,

where 2 and 3 represent the graphene in a purer form with relatively low content of oxygen (< 2%)

whereas 1 and 4 have a relatively high number of alkyl side chains and oxygen functionalities (> 2%).

The conjugated graphene systems allow absorption of a broad wavelength range of the

electromagnetic spectrum. On the other hand the aromatic nature results in low solubility in many

solvents such as DMF. By oxidation of the graphene to GO the solubility in DMF increases and

enables a stable dispersion. But the oxidation decreases the absorption ability of the graphene oxide

shown as well in the change of the bulk material color from black graphene to beige graphene oxide.

Reduction of GO by aqueous hydrazine solution leads to reduced graphene oxide (rGO) 4a in a fine

dispersion resulting in increased absorption ability through higher aromaticity.

Figure 2 Structure of carbon black candidates examined: asphaltene 1, MW-CNT 2, graphene

nano platelets 3 and graphene oxide 4.

The absorbance of the carbon black candidates dispersed in N,N’-dimethylformamide (DMF) having a

concentration of 0.111 mg/mL was measured in the range of 15,000 to 4,000 cm-1

(see Figure 3). For

CHAPTER 4

87

reference the commercial, metal-free NIR absorber Epolight 1117 was applied. The absorbance A,

being the negative logarithm of the transmittance (Iout/Iin), for Epolight 1117 with the value AEpolight 1117

= 1.266 at 980 nm resulted in a transmittance of less than 0.1, meaning less than 10% of the radiation

made it through the sample. Reduced graphene oxide and graphene nano platelets with an absorbance

of ArGO = 0.148 and Agraphene = 0.092 respectively, allow roughly 80% of the radiation to pass through

the sample. The graphene oxide and asphaltene are practically as unabsorbing as the PAN-based

polymer DralonX and the solvent DMF (see Table 1).

14000 12000 10000 8000 6000

0,0

0,3

0,6

0,9

1,2

1,5

714 833 1000 1250 1667

wave length [nm]

abso

rban

ce

wave number [cm-1]

Epolight 1117

rGO 4a

graphene 3

GO 4

asphaltene 1

DralonX

DMF

Figure 3 NIR absorption spectra of different carbon black candidates in a concentration of

0.111 mg/mL in DMF. Arrow marks 980 nm / 10,204 cm-1

.

The reduced graphene oxide (rGO) 4a candidate shows the highest absorption of the carbon black

candidates but has the crucial drawback of having residual traces of the reducing agent, hydrazine or

CHAPTER 4

88

its product of decomposition ammonia, in the solution. These basic nitrogen compounds will react

with the poly(acrylonitrile) and will catalyze chain scission of the polymer backbone into shorter

chains resulting in an uncontrolled reduction of the dynamic viscosity of the solution as well as

orange-brown discoloration of the solution.16

Ultimately the spinning of a CF precursor fiber from a

polymer solution of uncontrolled viscosity and the shorter polymer chains will lead to a CF with poor

mechanical properties. Alternatives for reducing agents have to meet the requirement of leaving no

basic residue in solution, ideally by decomposing quantitatively into escaping gases, leaving no

impurities in the solution. Additionally, the reducing agent has to be applied in a time efficient way

since DMF slowly decomposes at room temperature into the basic compound dimethyl amine. In the

industrial application DMF is therefore freshly distilled before use to keep control of a) the viscosity

of the spinning polymer solution and b) the color of the polymer in the case of textile fibers.

Table 1 Absorbance of different compounds in mass concentration of 0.111 mg/mL in NIR

LASER light of 980 nm / 10,204 cm-1

.

Sample name Absorbance

DralonX 0.001

Epolight 1117 1.266

Graphene 3 0.092

Reduced graphene oxide 4a 0.148

Graphene oxide 4 0.029

Asphaltene 1 0.005

The commercial dye Epolight 1117 yields the highest mass specific absorption of NIR LASER light in

DMF solution compared to the other carbon black candidates. Out of the carbon black candidates with

their big advantage being made out of over 90% carbon, graphene 3 combines the best applicability

with the highest absorbance. With an increase in graphene mass concentration by factor 10 similar

absorbance to Epolight 1117 could be achieved. For laboratory use the price of 1 gram of Epolight

CHAPTER 4

89

1117 is $ 37 and 1 g of graphene 3 is $ 0.13 making the carbon black additive the price efficient

option.

4.3.2 Dispersion and microscopy of carbon black candidates in polymeric films

For the examination of a dispersion of micro and nano particles different microscopy methods can be

applied such as optical microscopy, transmission electron microscopy (TEM), scanning electron

microscopy (SEM) and atomic force microscopy (AFM). The advantage of TEM, SEM and AFM over

optical microscopy is the detailed information about the shape and arrangement of the dispersed

particles but only for a relatively small investigated area. Optical microscopy in turn covers a larger

area of observation with less detailed information. To make assumptions from examination of a small

area in the nanometer to micrometer scale can a) introduce a big error by generalizing the observed

dispersion for the whole system in e.g. a laboratory/industrial mixing equipment and b) lead to

subjective visual analysis in picking an area guided by personal preferences. For a meaningful

statistical evaluation of the overall system usually a few images of small areas are not sufficient17

,

therefore the optical microscopy is applied for the evaluation.

For the efficient use of the carbon black additives homogeneous dispersion is required for two reasons:

(i) controlled absorption of NIR LASER light in the fiber material. An inhomogeneous dispersion

would yield irregular particle sizes and the absorption of these particles would cause locally irregular

heating resulting in structure defects leading to poor mechanical fiber properties; (ii) aggregates bigger

than the spinning nozzle would block the flow immediately. Efficient dispersion in the spinning

solution is only possible before the spinning of the precursor fiber. The spinning of the precursor fiber

with spinning nozzle diameters of typically as small as 40 µm leads to specific demands on the size of

the dispersed additive.

In this work the dispersions are prepared from DralonX (commercial PAN) polymer solution in DMF

with 0.1 wt.-% carbon black candidate and the homogeneity of examined samples will be described by

the dispersion parameter D, as defined by Glaskova and coworkers.18

The assumptions start with an

CHAPTER 4

90

ideal homogeneous dispersion consisting of a uniform distribution of round particles of equal size. The

real dispersion deviates from the ideal because carbon black particles exhibit π- π interaction, therefore

additionally formations of clusters occur (see Figure 4).

a . . .

. .. . .

. . . .

. . . .

. . . . . .

. . .

. . . .

. . . .

. . . . . .

. . . .

b . . ::: .

. . . . .

. … . .

:. . .:::.. .

. .. . . . .

. . .

. .. . . ..

. .:. . . .

: . . ::.

. .. . . .

Figure 4 Ideal uniform (a) distribution and distribution with clusters (b) of round particles of

equal size.

For a quantitative estimation of the dispersion the area of the individual particles has to be defined.

The analysis is done by the software ImageJ19

through binarizing the images obtained by optical

microscopy and separating the particles from the background by setting a brightness threshold

between particles and background. The area of particles and clusters is represented by the number of

pixels in the image. Therefore, an association of the dispersion of the particles with their area is

assumed. If two or more neighbor particles are in contact a cluster results. Then the more

homogeneous a system is the more equal are its particle areas. For that, the average area of a particle µ

has to be determined from the area of the individual particle areas a1, a2, a3, …, ai, …, aN with N

particles by

∑

and its standard deviation

CHAPTER 4

91

√∑

.

With the assumption of a normal (Gaussian) distribution the section under the curve of a probability

density function in the range of 0.9µ to 1.1µ will be expressed by the dispersion parameter D through

the term

√

.

Through the dispersion parameter D systems with the same amount of carbon additives can be

quantitatively compared and the degree of homogeneity described. This means with more particles

having equal size (area), the larger is the section under the probability density function, and

consequently the more homogeneous is a system yielding a higher value of the dispersion parameter

D. Glaskova and coworkers examined a system of epoxy resin EPON 828 with 0.2 wt.-% ultra-

sonically dispersed MW-CNT particles (average diameter of 9.5 nm and average length 1.5 µm)

yielding a dispersion parameter of D = 0.035. All prepared samples in this work have 0.1 wt.-%

carbon black particles added, therefore yielded dispersion parameters should be at least D > 0.035.

For the comparison of the different systems it is important to keep the thickness of the examined

sections constant. The films were cast with a doctor blade. The solvent was evaporated with the film

being in a level position and afterwards the film was washed with distilled water. All films are 25 ± 1

µm thick.

The image analysis with a typical resolution of 3 pixel/µm yielded the number of counted particles N,

the average area of particles µ, the standard deviation σ and the dispersion parameter D of the carbon

black candidates MW-CNT 2, graphene 3, asphaltene 1 and asphaltene-g-PAN 5 (see Table 2).

CHAPTER 4

92

Table 2 Image analysis results of the different optical microscopy images done with the

software ImageJ.

Candidate

Counted

particles N

Average area of

particles µ [µm2]

Standard deviation

σ [µm2]

Dispersion

parameter D

MW-CNT 2 3,474 5.2 12.9 0.032

Graphene 3 744 11.2 13.0 0.069

Asphaltene 1 1,741 9.0 23.4 0.031

Asphaltene-g-PAN 5 1,385 1.3 1.3 0.081

The first approach is the dispersion of nano scale additives in DralonX polymer solution in DMF.

MW-CNT 2 was sonicated in DMF for 1h and mixed with the DralonX polymer. After casting, drying

and washing the film was examined by optical microscopy and showed highly inhomogeneous particle

dispersion (Figure 5). With counted areas of N = 3,474 the dispersion parameter was determined with

D = 0.032.

Figure 5 Left: optical microscopy image of 25 µm thin DralonX copolymer film with dispersed

MW-CNT 2; right: optical microscopy image after binarization and brightness

adjustment.

100 µm

CHAPTER 4

93

Commercial graphene 3 was dispersed by the previously used protocol and examined by optical

microscopy (see Figure 6). Analysis of the image with N = 744 counted particles yielded a dispersion

parameter of D = 0.069 (see Table 2). Compared to D of the MW-CNT 2 the distribution parameter of

the commercial graphene 3 is higher equaling a more homogeneous dispersion.


graphene 3; right: optical microscopy image after binarization and brightness

adjustment.

One way to increase the solubility of graphene is to change its hydrophilicity by chemical oxidation.

Applying the modified Hummers3 method yields sheets of graphene oxide 4. The introduction of

oxygen breaks up the aromatic systems resulting in a visible color shift from black to brown.

Examples in literature show good solubility of graphene oxide in various solvents as well as in

DMF.8,20,21

To remove non-oxidized particles the GO-DMF mixture was treated with a Microfluidizer

for 15 min at 1000 bar, then centrifuged four times at 11,000 rpm for 10 min each and finally filtered

through a PTFE membrane with a cutoff at 200 nm. The resulting solution was examined by TEM and

AFM to determine the sheet dimensions (see Figure 7). The 20 counted sheets show an average area µ

= 0.056 µm2 and the standard deviation of σ = 0.036 µm

2. The AFM image of the sheets display an

average height of 1.3 nm, indicating monolayers which have a typical height of 1.0 nm.22

The lateral

100 µm

CHAPTER 4

94

dimensions are on the micrometer scale and could lead to the formation of larger aggregates if the

concentration in the TEM sample of 0.111 mg/mL is increased. The size of aggregates must be under

control during the spinning process to minimize the risk of blocked spinning nozzles.

a) b)

c)

Figure 7 a) TEM image of graphene oxide 4 (GO) sheets; b) AFM image of GO 4 with three

height profile lines; c) Height profile of the sheets measured with AFM.

Although the GO sheets can be produced in nano sizes, for using them as NIR LASER light absorber a

way of reducing GO to graphene has to be found which a) increases the NIR LASER light absorption

significantly, b) keeps the particles dispersed in nano size scale and c) leaves no basic compounds in

solution behind.

An option with a lower production price could be the asphaltene 1, extracted from bitumen. The

dispersion of asphaltene 1 in a DralonX copolymer film was examined by optical microscopy (see

CHAPTER 4

95

Figure 8). Analysis of the image with N = 1,741 counted particles yielded a dispersion parameter of D

= 0.031 (see Table 2). This poor homogeneity is caused by the low solubility of carbon based additives

in DMF. Therefore, to increase solubility alternatively to oxidation, “grafting-from” polymerization

with the monomer acrylonitrile was performed to synthesize asphaltene-g-PAN 5.


asphaltene 1; right: optical microscopy image after binarization and brightness

adjustment.

A DralonX copolymer film with the dispersed asphaltene-g-PAN 5 was examined by optical

microscopy (see Figure 9). Analysis of the image with N = 1,385 counted particles yielded a

dispersion parameter of D = 0.081 (see Table 2).

100 µm

CHAPTER 4

96


asphaltene-g-PAN 5; right: optical microscopy image after binarization and brightness

adjustment.

TEM images of the asphaltene-g-PAN 5 showed aggregation behavior with sizes of several hundred

nanometers (see Figure 10). The aggregates appear to consist of smaller particles of 40 - 50 nm. The

aggregation could be a result from the sample preparation. A too high concentration of asphaltene-g-

PAN 5 during evaporation of the solvent is a possible reason.

Figure 10 TEM images of asphaltene-g-PAN 5 at two different regions of the sample.

100 µm

CHAPTER 4

97

Compared to the other dispersion parameters of the carbon black candidates, asphaltene-g-PAN 5

yields the highest value and therefore the most homogeneous dispersion of carbon black in a PAN

copolymer solution in DMF.

With light transmission experiments the effectiveness of the dispersed carbon black additives can be

examined. The transmittance T will be defined as

T = Iout/Iin

where Iout and Iin represent, respectively the intensities of the transmitted and incident light. The

transmittance is measured instead of the absorbance to account for intensity loss through reflection.

These parameters follow the relationship: transmittance = incident intensity – absorbance – reflection.

For the same carbon black additive concentrations of 0.1 wt.-% a lower value of the transmittance T

would mean higher absorption at a constant value for reflection. The transmittance spectra of DralonX

polymer, asphaltene-g-PAN 5, graphene 3, graphene oxide 4 and Epolight 1117 are presented (see

Figure 11 and Table 3). The transmittance T of DralonX polymer without additive at 980 nm is T980nm

= 92.8 %. As reference the commercial NIR light absorber Epolight 1117 was applied, yielding the

lowest value of transmittance with T980nm = 84.4 %. Graphene 3 displays a lower transmittance as well,

T980nm = 87.3 %, making it a valid candidate for an effective dispersant in polymer films. The low

transmittance difference of 3.0% of the asphaltene-g-PAN 5 can be resolved due to its good

dispersibility by increasing the concentration higher than 0.1 wt.-%. The high transmittance of

graphene oxide 4 is expected and has been shown in solution as well in the previous section 4.3.1.

CHAPTER 4

98

500 1000 1500 2000

70

80

90

100

20000 10000 6667 5000

no additive

Epolight 1117

Asphaltene-g-PAN 5

Graphene 3

Graphene oxide 4

wave number [cm-1]

tran

sm

itta

nce

[%

]

wave length [nm]

980 nm

10204 cm-1

Figure 11 Transmittance spectra of DralonX polymer films with different additives in 0.1 wt-%

concentration.

Table 3 Transmittances at 980 nm of the DralonX polymer films with different additives in 0.1

wt-% concentration.

0.1 wt.-% additive in the

DralonX polymer film

Transmittance T980nm [%] Transmittance difference [%]

No additive 92.8 0

Epolight 1117 84.4 8.4

Asphaltene-g-PAN 5 89.8 3.0

Graphene 3 87.3 5.5

Graphene oxide 4 92.3 0.5

CHAPTER 4

99

4.3.3 Kinetic parameters for the thermal conversion of PAN in the presence of carbon black

candidates

Each precursor formulation or different spinning technique requires a tailored heating and fiber

stretching program for yielding a CF with the intended mechanical properties. Every impurity changes

the demands on the program. Therefore, although the carbon black additives are mainly made of

carbon, they will have an effect on the chemical reactions during the heating. In this work we

examined the reaction enthalpies ΔH of the precursor copolymer Precursor-T with and without

graphene additive by classical differential scanning calorimetry (DSC) to determine the nature of the

effect. In future experiments the measurements of reaction enthalpies ΔH should be performed by

LASER heated DSC to simulate the intended heating by LASER light. In DSC thermograms the

reaction processes can be assigned to the following two steps: the stabilization process, in air

atmosphere between 200 – 400°C and the carbonization process, in nitrogen atmosphere between 500

– 1200°C. The stabilization process in PAN-based carbon fibers consists of many different chemical

reactions which can be represented by three major energetic reactions: dehydrogenation, cyclization

and oxidation (see Figure 12).23,24,25,26

These reactions are important for the mechanical properties of

the resulting carbon fiber. Every additive will influence the reaction enthalpies and will make

modifications to the heating program necessary. Therefore, two possibilities arise for an influence of

the carbon black additive on the reaction enthalpy of stabilization: an increase or a decrease. Literature

shows that with the addition of MW-CNT to homopolymer of acrylonitrile the stabilization enthalpy

decreases whereas the addition of MW-CNT to a copolymer of acrylonitrile and methyl acrylate or

methyl acrylic acid the reaction enthalpy increases.27,28,29

By applying carbon black additive to the

copolymer Precursor-T an increase will be expected.

CHAPTER 4

100

Figure 12 DSC thermogram of PAN-based CF precursor in air atmosphere.

The DSC experiments are performed with Precursor-T and the carbon black candidate graphene 3 in

0.1 wt.-% addition. The reaction enthalpy at the heating rate of 10 K/min of Precursor-T yields ΔH =

85.1 kJ/mol whereas the addition of graphene 3 yields the higher value with ΔH = 104.1 kJ/mol, as

expected (see Table 4 and Figure 13). Variation of the heating rate shows that the increase of enthalpy

is strongest with low heating rates, e.g. at 2 K/min the difference between the presence and absence of

graphene is 29.7 kJ/mol whereas at high heating rates the difference is quite small, e.g. at 20 K/min

1.7 kJ/mol.

Table 4 Reaction enthalpies for the powdered PAN copolymers with and without 0.1 wt.-%

graphene 3 at 10 K/min determined by DSC in synthetic air atmosphere.

Sample

Apparent reaction enthalpy [kJ/mol] at heating rate


Precursor-T 118.0 111.0 85.1 78.5 73.8

Precursor-T with graphene 3 147.7 134.3 104.1 94.2 75.5

CHAPTER 4

101

Figure 13 DSC thermograms of Precursor-T PAN copolymer with 0.1 wt-% graphene 3 at five

different heating rates (β = 2, 5, 10, 15 and 20 K/min).

The proposed explanation for the increase in enthalpy which is stronger with lower heating rates,

involves the diffusion ability of oxygen through the sample. The presence of graphene sheets allows

for more adsorption of oxygen in the sample, thus making oxygen more available for the different

chemical reactions – summarized in dehydration and oxidation – starting at different temperatures

yielding a higher conversion (see Figure 14).

150 200 250 300 350 400Temperatur /°C

-12

-10

-8

-6

-4

-2

0

DSC /(mW/mg)

[1][2]

[3]

[4]

[5]

Exo


CHAPTER 4

102

a)

b)

Figure 14 Oxygen consuming chemical reactions of the a) dehydration and b) oxidation

processes.30

With low heating rates the sample is longer exposed to heat and the adsorption effects play a more

important role. The increase in enthalpy should therefore show an increase of the oxygen consuming

reaction, especially the oxidation, of the stabilization process. Through the process of peak

deconvolution the DSC thermograms are analyzed and the oxidation process enthalpy is calculated for

2 K/min and 10 K/min (see Table 5 and Appendix C). The differences of the stabilization enthalpies

are compared with the differences of the oxidation enthalpies for 2 K/min and 10 K/min (see Table 6).

The comparison of the differences shows that the change in oxidation enthalpy makes up for 81% in

the change of the stabilization enthalpy when adding 0.1 wt.-% graphene 3 at a heating rate of 2

K/min. At a heating rate of 10 K/min the percentage of change in the oxidation enthalpy only amounts

for 45% in the change of the stabilization enthalpy, proving that the influence of oxidation reaction is

strongest at low heating rates.

CHAPTER 4

103

Table 5 Oxidation enthalpies for the powdered PAN copolymers with and without 0.1 wt.-%

graphene 3 at 2 and 10 K/min determined by DSC in synthetic air atmosphere.

Sample

Apparent reaction enthalpy [kJ/mol]

2 K/min 10 K/min

Precursor-T 55.8 24.8

Precursor-T with graphene 3 79.8 33.3

Table 6 Differences of stabilization enthalpies and oxidation enthalpies for the powdered PAN

copolymers with and without 0.1 wt.-% graphene 3 for 2 K/min and 10 K/min.

Heating rate 2 K/min 10 K/min

Stabilization enthalpy difference 29.7 kJ/mol 18.9 kJ/mol

Oxidation enthalpy difference 24.0 kJ/mol 8.5 kJ/mol

Percentage 81 % 45 %

If the increase in stabilization enthalpy really relies on the adsorption of oxygen it should be reflected

in the kinetics of the stabilization. For the evaluation of the kinetic parameters of the stabilization

reactions the apparent activation energy Ea was determined by the Kissinger31

method. The other

method mostly used in literature is the Ozawa32

method but for the PAN copolymer system the two

methods yield similar results.33

The influence on the stabilization reactions was determined by the shift

of the first exothermic peak temperature Tp of the dehydrogenation/cyclization reaction (see Table 7).

CHAPTER 4

104

Table 7 Peak temperatures Tp for the powdered PAN copolymers with and without 0.1 wt.-%

graphene 3 determined by DSC in synthetic air atmosphere.

Heating rate [K/min] Tp (Precursor-T) Tp (Precursor-T + 3)

2 263.8 267.4

5 279.9 280.9

10 288.9 291.7

15 296.5 299.3

20 302.9 303.9

The mathematical expression for the Kissinger method is as follows:

where Tp is the peak temperature; β is the heating rate; Ea is the apparent activation energy; A is the

pre-exponential factor, and R is the molar gas constant. According to this equation, Ea and A can be

calculated from a linear fitted plot of ln(β/Tp2) against 1/RTp (slope and intercept, respectively). The

plotted fits of Precursor-T with and without graphene 3 are displayed (see Figure 15).

CHAPTER 4

105

0,000203 0,000210 0,000217 0,000224

-12,0

-11,5

-11,0

-10,5

-10,0

-9,5

ln(

/T2 p)

1/RTp

Precursor-T

Precursor-T + additive

Figure 15 Linear fitted plots from the kinetics for the cyclization/dehydrogenation reaction

during the stabilization of Precursor-T with and without graphene additive by the

Kissinger method.

The calculated apparent activation energies determined from the slope and the pre-exponential factor

A determined from the intercept are listed below (see Table 8). In literature, Devasia and coworkers

found the apparent activation energy of a PAN copolymer with itaconic acid to be 153.2 kJ/mol which

is comparable to Precursor-T in chemical composition and activation energy Ea of 142.0 kJ/mol.34

The

presence of the additive increases the apparent activation energy of Precursor-T by 11.5 kJ/mol,

roughly 10%. This energy value can be converted for a single oxygen molecule with 1 mol =

6.022·1023

particles and 1 eV = 1.602·10-19

J resulting in 1.19 eV. The work of Mehmood and

coworkers35

examined the absorption and diffusion of oxygen through DFT calculations36

and

experiments on graphene without defects (pristine graphene) and with defects. The study shows that

the absorption energy of molecular oxygen on pristine graphene requires 0.11 eV and the activation

energy barrier of oxygen diffusion on the pristine graphene requires 0.9 eV. Absorption of oxygen on

graphene with defects requires more energy. Assuming graphene 3 consists of mainly pristine

graphene the values of adsorption and diffusion with 1.01 eV agree very well with the 1.2 eV increase

in calculated activation energy by the Kissinger method. The increased pre-exponential factor A -

CHAPTER 4

106

describing a relationship between temperature and rate coefficient - represents a beneficial increase of

the conversion rate of the reaction, in agreement with the previous results of higher stabilization

enthalpies.

Table 8 Calculated apparent activation energies of Precursor-T with and without 0.1wt.-%

graphene 3 in synthetic air atmosphere.

Sample Apparent activation energy Ea

[kJ/mol] Pre-exponential factor A [min

-1]

Precursor-T 142.0 8.48 x 10

12

Precursor-T with graphene 3 153.5 9.01 x 10

13

4.4 Conclusion

For the more energy efficient production of PAN-based carbon fibers, an alternative for energy

transfer to the precursor material other than the conventional oven heating by convection has been

presented. The examination of NIR LASER light absorption carbon black particles at 980 nm showed

nano platelet graphene having the best trade-off between absorbance and price compared to a

commercial NIR LASER light absorber.

The dispersion of the graphene in the PAN-based DralonX solution in DMF yielded a better dispersion

parameter than the examined alternatives MW-CNT and asphaltene. Through the “grafting-from”

polymerization of acrylonitrile with asphaltene an even better dispersion was accessible. While

graphene oxide can be dispersed in a nano scale as monolayer sheets, for the application as NIR

LASER light absorber an efficient pathway for reduction of GO has to be determined. The

introduction of carbon black particles has an influence on the viscosity of the polymer solution. In

regard to the application of the polymer solution in fiber spinning future studies have to examine the

rheological behavior of the system with the dispersed particles.

CHAPTER 4

107

The heat treatment of precursor polymer with dispersed graphene showed a beneficial increase in

enthalpy and apparent activation energy of the system. It was explained that for low heating rates the

graphene works as oxygen adsorber supplying chemical reactions more efficiently with oxygen thus

improving the conversion. Future works should involve DSC experiments with a LASER light heating

source to gain knowledge about the intended industrial application of NIR diode LASER elements.

Future works should include basic research about the incorporation of NIR LASER light absorbing

functionalities in copolymers. Such functionalities could be detrimental to the mechanical properties

such as tensile strength of the resulting CF but low tensile strength CF could still be applied as

composite for low strain components when the price is low enough.

4.5 References

(1) Industrial Dyes; Hunger, K., Ed.; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, FRG,

2002.

(2) Zollinger, H. Color Chemistry: Syntheses, Properties, and Applications of Organic Dyes and

Pigments; Wiley-VCH, 2003.

(3) Xu, Z.; Gao, C. Macromolecules 2010, 43, 6716.

(4) Behabtu, N.; Young, C. C.; Tsentalovich, D. E.; Kleinerman, O.; Wang, X.; Ma, A. W. K.;

Bengio, E. A.; ter Waarbeek, R. F.; de Jong, J. J.; Hoogerwerf, R. E.; Fairchild, S. B.;

Ferguson, J. B.; Maruyama, B.; Kono, J.; Talmon, Y.; Cohen, Y.; Otto, M. J.; Pasquali, M.

Science 2013, 339, 182.

(5) Wang, Q.; Du, Y.; Feng, Q.; Huang, F.; Lu, K.; Liu, J.; Wei, Q. J. Appl. Polym. Sci. 2012, n/a.

(6) Tölle, F. J.; Fabritius, M.; Mülhaupt, R. Adv. Funct. Mater. 2012, 22, 1136.

(7) Luo, S.; Liu, T.; Wang, B. Carbon N. Y. 2010, 48, 2992.

(8) Paredes, J. I.; Villar-Rodil, S.; Martínez-Alonso, a; Tascón, J. M. D. Langmuir 2008, 24,

10560.

(9) Li, D.; Müller, M. B.; Gilje, S.; Kaner, R. B.; Wallace, G. G. Nat. Nanotechnol. 2008, 3, 101.

(10) You, Y.-Z.; Hong, C.-Y.; Pan, C.-Y. Macromol. Rapid Commun. 2006, 27, 2001.

CHAPTER 4

108

(11) Kan, L.; Xu, Z.; Gao, C. Macromolecules 2011, 44, 444.

(12) Salavagione, H. J.; Martínez, G.; Ellis, G. Macromol. Rapid Commun. 2011, 1.

(13) Layek, R. K.; Nandi, A. K. Polymer (Guildf). 2013.

(14) Kroh, M.; Bonten, C.; Eyerer, P. AIP Conf. Proc. 2013, 1593, 724.

(15) Suzuki, T.; Zushi, H.; Woelki, M.; Hunyar, C. Carbon fiber precursor fibers, their manufacture,

and manufacture of flame-retardant fibers and carbon fibers, 2011.

(16) Hamada, F.; Takahashi, G. Kobunshi Kagaku 1961, 18, 715.

(17) Paul, D. R.; Robeson, L. M. Polymer (Guildf). 2008, 49, 3187.

(18) Glaskova, T.; Zarrelli, M.; Borisova, a.; Timchenko, K.; Aniskevich, a.; Giordano, M. Compos.

Sci. Technol. 2011, 71, 1543.

(19) Rasband, W. ImageJ, 1997, <http://imagej.nih.gov/ij/> [10.01.2015].

(20) Botas, C.; Álvarez, P.; Blanco, P.; Granda, M.; Blanco, C.; Santamaría, R.; Romasanta, L. J.;

Verdejo, R.; López-Manchado, M. a.; Menéndez, R. Carbon N. Y. 2013, 65, 156.

(21) Park, S.; An, J.; Piner, R. D.; Jung, I.; Yang, D.; Velamakanni, A.; Nguyen, S. T.; Ruoff, R. S.

Chem. Mater. 2008, 20, 6592.

(22) Botas, C.; Álvarez, P.; Blanco, C.; Santamaría, R.; Granda, M.; Ares, P.; Rodríguez-Reinoso,

F.; Menéndez, R. Carbon N. Y. 2012, 50, 275.





(27) Schäfer, R. C. Herstellung und Charakterisierung von Fasern aus Polymer-Komposits mit

Metall-haltigen Kohlenstoffnanoröhrchen, Universität Stuttgart, 2010.

(28) Choi, Y. H. Polyacrylonitrile/Carbon Nanotube Composite Fibers: Effect Of Various

Processing Parameters On Fiber Structure And Properties, Georgia Institute of Technology,

USA, 2010.

(29) Jain, R. Carbon Nanotube Reinforced Polyacrylonitrile And Poly(etherketone) Fibers, Georgia

Institute of Technology, USA, 2009.


(31) Kissinger, H. E. Anal. Chem. 1957, 29, 1702.

(32) Ozawa, T. Bull. Chem. Soc. Jpn. 1965, 38, 1881.

CHAPTER 4

109


(34) Devasia, R.; Nair, C. P. R.; Sivadasan, P.; Katherine, B. K.; Ninan, K. N. J. Appl. Polym. Sci.

2003, 88, 915.

(35) Mehmood, F.; Pachter, R.; Lu, W.; Boeckl, J. J. J. Phys. Chem. C 2013, 117, 10366.

(36) Dion, M.; Rydberg, H.; Schröder, E.; Langreth, D. C.; Lundqvist, B. I. Phys. Rev. Lett. 2004,

92, 246401.

Appendix C

b) DSC thermograms

Precursor-T without additive

150 200 250 300 350 400Temperatur /°C

-10

-8

-6

-4

-2

0

DSC /(mW/mg)

[1]

[2]

[3]

[4]

[5]

Exo

CHAPTER 4

110

Precursor-T with graphene additive 3

c) NMR spectra

Asphaltene in CDCl3

150 200 250 300 350 400Temperatur /°C

-12

-10

-8

-6

-4

-2

0

DSC /(mW/mg)

[1][2]

[3]

[4]

[5]

Exo

ppm (t1) 1.02.03.04.05.06.07.08.0

7,2

60

1,2

68

0,9

02

0,8

86

0,8

68

CHAPTER 4

111

Asphaltene-g-PAN 5 in DMSO-d6

d) Peak deconvolution of DSC thermograms

Precursor-T without graphene 3 at 2 K/min

50 100 150 200 250 300 350 400

-1,4

-1,2

-1,0

-0,8

-0,6

-0,4

-0,2

0,0

0,2

heat flow

[m

W/m

g]

temperature [°C]

Subtracted from C

Fit Peak 1

Fit Peak 2

Fit Peak 3

Cumulative Fit Peak

Model Gauss


Reduced Chi-Sqr

4,20866E-5



Peak1(C) y0 0 0

Peak1(C) xc 261,50872 0,0163

Peak1(C) w 23,03616 0,0274

Peak1(C) A -24,39326 0,0686

Peak1(C) sigma 11,51808 0,0137

Peak1(C) FWHM 27,12301 0,03226

Peak1(C) Height -0,84489 0,00143

Peak2(C) y0 0 0

Peak2(C) xc 296,46607 0,02291

Peak2(C) w 42,53714 0,03224

Peak2(C) A -34,13507 0,03195

Peak2(C) sigma 21,26857 0,01612

Peak2(C) FWHM 50,08365 0,03796

Peak2(C) Height -0,64028 2,12375E-4

Peak3(C) y0 0 0

Peak3(C) xc 235,40673 0,05356

Peak3(C) w 27,69159 0,05716

Peak3(C) A -13,58756 0,05009

Peak3(C) sigma 13,8458 0,02858

Peak3(C) FWHM 32,60436 0,0673

Peak3(C) Height -0,3915 7,0446E-4

ppm (t1) 1.02.03.04.05.06.07.0

7,3

15

7,2

48

7,1

80

7,1

39

3,1

36

2,5

00

2,0

34

CHAPTER 4

112

Precursor-T without graphene 3 at 10 K/min

50 100 150 200 250 300 350 400

-7

-6

-5

-4

-3

-2

-1

0

1h

ea

t flo

w [

mW

/mg

]

temperature [°C]

Subtracted from C

Fit Peak 1

Fit Peak 2

Fit Peak 3

Cumulative Fit Peak

Model Gauss


Reduced Chi-Sqr

0,00121



Peak1(C) y0 0 0

Peak1(C) xc 288,51192 0,01716

Peak1(C) w 23,9255 0,04994

Peak1(C) A -153,65913 0,55956

Peak1(C) sigma 11,96275 0,02497

Peak1(C) FWHM 28,17013 0,0588

Peak1(C) Height -5,12433 0,00917

Peak2(C) y0 0 0

Peak2(C) xc 323,36431 0,13992

Peak2(C) w 43,12873 0,18958

Peak2(C) A -77,42932 0,4395

Peak2(C) sigma 21,56437 0,09479

Peak2(C) FWHM 50,7802 0,22321

Peak2(C) Height -1,43245 0,0029

Peak3(C) y0 0 0

Peak3(C) xc 260,05687 0,06725

Peak3(C) w 20,5258 0,09165

Peak3(C) A -34,58134 0,23069

Peak3(C) sigma 10,2629 0,04582

Peak3(C) FWHM 24,16728 0,1079

Peak3(C) Height -1,34426 0,00433

Precursor-T with graphene 3 at 2 K/min

100 150 200 250 300 350 400

-1,8

-1,6

-1,4

-1,2

-1,0

-0,8

-0,6

-0,4

-0,2

0,0

0,2

hea

t flo

w [

mW

/mg

]

temperature [°C]

Subtracted from C

Fit Peak 1

Fit Peak 2

Fit Peak 3

Cumulative Fit Peak

Model Gauss


Reduced Chi-Sqr

1,2114E-4



Peak1(C) y0 0 0

Peak1(C) xc 264,57208 0,02331

Peak1(C) w 21,64324 0,03607

Peak1(C) A -25,97358 0,10113

Peak1(C) sigma 10,82162 0,01803

Peak1(C) FWHM 25,48297 0,04247

Peak1(C) Height -0,95752 0,00224

Peak2(C) y0 0 0

Peak2(C) xc 300,63488 0,02245

Peak2(C) w 43,18548 0,03475

Peak2(C) A -47,95916 0,04312

Peak2(C) sigma 21,59274 0,01737

Peak2(C) FWHM 50,84702 0,04091

Peak2(C) Height -0,88608 2,87697E-4

Peak3(C) y0 0 0

Peak3(C) xc 239,96728 0,07354

Peak3(C) w 26,15074 0,0796

Peak3(C) A -14,83393 0,07908

Peak3(C) sigma 13,07537 0,0398

Peak3(C) FWHM 30,79015 0,09372

Peak3(C) Height -0,4526 0,00116

CHAPTER 4

113

Precursor-T with graphene 3 at 10 K/min

50 100 150 200 250 300 350 400

-8

-7

-6

-5

-4

-3

-2

-1

0

1heat flow

[m

W/m

g]

temperature [°C]

Subtracted from C

Fit Peak 1

Fit Peak 2

Fit Peak 3

Cumulative Fit Peak

Model Gauss


Reduced Chi-Sqr

7,00444E-4



Peak1(C) y0 0 0

Peak1(C) xc 290,16869 0,01357

Peak1(C) w 25,63512 0,03924

Peak1(C) A -180,79595 0,44063

Peak1(C) sigma 12,81756 0,01962

Peak1(C) FWHM 30,18305 0,0462

Peak1(C) Height -5,62721 0,00595

Peak2(C) y0 0 0

Peak2(C) xc 328,25451 0,07557

Peak2(C) w 43,43282 0,10509

Peak2(C) A -102,30837 0,31199

Peak2(C) sigma 21,71641 0,05254

Peak2(C) FWHM 51,13824 0,12373

Peak2(C) Height -1,87946 0,00203

Peak3(C) y0 0 0

Peak3(C) xc 262,01269 0,05036

Peak3(C) w 20,09719 0,06443

Peak3(C) A -36,28126 0,20755

Peak3(C) sigma 10,04859 0,03221

Peak3(C) FWHM 23,66263 0,07586

Peak3(C) Height -1,44041 0,00457

CHAPTER 5

114

Wet spinning of PAN-based fiber precursor with

hyperbranched PAN as rheological additive

5.1 Introduction

Acrylic fibers can be spun using different techniques such as wet, dry and dry-jet wet spinning where

wet spinning is the most widely used technique.1 The synthesized copolymer is dissolved in a solvent.

The resulting polymer solution is known as dope. Due to high polarity of the nitrile group, solid PAN

orders itself in densely packed crystalline structures. The strong intermolecular forces cause the

melting point (Tm = 320 °C) to be higher than the degradation temperature (Td = 250°C).2 In

consequence, highly polar solvents are required to dissolve PAN. The solvents can be organic solvents

with highly polar molecules (N,N’-dimethylformamide, dimethyl sulfoxide, N,N’-dimethylacetamide)

and inorganic solvents with strong ionic bonding (nitric acid, aqueous sodium thiocyanate, aqueous

zinc chloride). The solubility of PAN is higher with lower molecular weights but the resulting tensile

strength of the fiber would be lower. Therefore the lowest limit of MW is defined by the lowest limit

of required tensile strength. Wet spinning of the dope is usually done with ~ 15 – 20 wt.-% solid

content and dry spinning with higher solid contents of ~ 25 – 30 wt.-%.3

After thorough mixing of the spinning dope, in the wet spinning process the fibers are pressed out of a

multi-hole spinneret, an end-piece with small nozzles (~ 40 µm), in a coagulation bath filled with

water, a non-solvent for PAN (see Figure 1). The process of coagulation typically consists of leading

the fiber under stretching through a series of 3-4 washing baths having a fixed decreasing ratio of

water/polymer solvent. The solvent is washed out of the fiber and afterwards the fiber is dried, treated

with a finisher and spooled. Depending on the chemical composition of the acrylonitrile-based

copolymer either textile fibers or carbon fiber (CF) precursors can be spun. CF precursors receive a

tailored heat treatment for pyrolysis to a CF.

CHAPTER 5

115

Figure 1 Schematic process of wet spinning of precursor fibers.

During this process the fiber morphology is determined by two important processes: (i) the

displacement of the solvent through water by diffusion defines the pore structure of the fiber. When

the ratio of water/solvent is high, a so called core-shell structure will form (see Figure 2). In the outer

cross-section area big pores would form (macro voids) and in the inner part smaller pores (micro

voids). The resulting inhomogeneity of the pores distribution over the cross-section would result in a

CF with lower tensile strength and stiffness. (ii) Essential for the formation of a homogeneous and

optimal precursor fiber is the continuous stretching of the fiber starting from the exit of the spinning

nozzle until the last step of the heating process. The stretching allows for the polymer chains to orient

themselves in a parallel orientation. For the thermal reaction of cyclization a highly parallel orientation

is beneficial as well as for the resulting CF in achieving a high tensile strength.

CHAPTER 5

116

Figure 2 Core-shell structure of a precursor fiber precipitated with in bath of high non-solvent

content.

The three components polymer (PAN copolymer), solvent (DMF) and non-solvent (water) are not

soluble in each other in every concentrations and therefore this ternary phase system is divided in

mono-phase and multi-phase areas with phase boundaries (bimodal and spinodal curves) where phase

transition occurs. The relationship can be displayed in a phase diagram (see Figure 3). The binodal

curve is separating the homogeneous mono-phase area I from the inhomogeneous, metastable two

phase area II. And the two phase II is separated by the spinodal curve from the instable multi-phase

area III. In the metastable area II the ternary polymer system is thermodynamically unstable but phase

transition occurs delayed due to slow nucleation reactions. In the instable phase area III the dissolved

polymer is separated spontaneously into two phases. The tangential point of the binodal and spinodal

curve is called the critical point.

With the knowledge of the phase diagram at specific temperatures the optimal concentration of the

components can be selected to achieve the fastest transition in the coagulating process of the spun

fiber while minimizing unwanted macro voids (core-shell structure).

CHAPTER 5

117

0,00 0,25 0,50 0,75 1,00

0,00

0,25

0,50

0,75

1,000,00

0,25

0,50

0,75

1,00

weight-%

PAN

WaterDMF

I

II

III

spinodal curvebinodal curve

critical point

Figure 3 Phase diagram of the ternary system PAN/DMF/water at T = 25 °C.4

Therefore, with the concentration of solvent in coagulation bath, the temperature of the bath, fiber

residence time in bath (by controlling extrusion rate, length of bath and winding speed) and the nozzle

diameter the morphology of the resulting fiber is determined and consequently the properties.

The project Megacarbon aims to increase the polymer percentage of the spinning dope from 30 wt.-%

upwards by a few percent while counteracting the increasing viscosity through a rheological additive.

Preliminary experiments were undertaken from an industrial standpoint to prove the feasibility of

precursor fiber spinning with the rheological additive. Future works should include further analysis of

the ternary system without and with rheological additive.


5.2.1 Materials

Acrylonitrile (AN, ≥99%, Aldrich) was purified by column chromatography on Al2O3 (activated,

neutral, Brockmann I, Aldrich) before use. Azobisisobutyronitrile (AIBN, 98+%, Fluka) was

CHAPTER 5

118

recrystallized twice from methanol before use. N,N’-dimethylformamide (DMF, HPLC grade, VWR),

methanol (dried, Merck), carbon disulfide (≥99%, Aldrich), 1-propane thiol (Merck) and 4-

vinylbenzyl chloride (90%, Aldrich) were used without further purification.

DralonX polymer (prepared from acrylonitrile, methyl acrylate, ionic comonomer) was supplied by

Dralon GmbH, Dormagen.



The cloud point determinations and the fiber spinning as well as the electron microscopy were carried

out by Tobias Lülf from our institute.

5.2.2 Methods

A multi-hole spinneret and a heated coagulation bath were used as spinning device.

For the determination of cloud point temperature a turbidimeter from the University of Twente with a

four channel sensor was used. The solutions were heated to 95°C and cooled to determine the loss of

light by formation of cloud points.

The electron microscopy images were taken by a Hitachi field emission scanning electron microscope

FESEM S4800. The samples were sputtered with gold.




13C

spectra were made by first order analysis of the spectra.

CHAPTER 5

119

5.2.3 Syntheses

S-(4-vinyl)benzyl S’-propyltrithiocarbonate (VBPT) (1)

1

The synthesis was performed according to Zhao and coworkers.5 To a stirred solution of 1-

propanethiol (1.54 g, 20.0 mmol) in anhydrous methanol (16 mL) was slowly added a solution of

sodium methoxide (1.15 g, 20.2 mmol) in methanol (20 mL) under nitrogen atmosphere. After 2h

stirring carbon disulfide (1.90 g, 25.0 mmol) was added drop-wise, and the mixture was further stirred

at room temperature for 5 h. To the yellow solution obtained was added slowly 4-vinylbenzyl chloride

(3.72 g, 21.9 mmol) and the mixture was stirred overnight in nitrogen atmosphere. The mixture was

poured into water from which the organic product was extracted with dichloromethane. The separated

yellow dichloromethane solution was dried with sodium sulfate. The solution was concentrated to a

few milliliters and purified by flash column chromatography eluting with hexane. VBPT 1 was

isolated as yellow viscous oil. 1H NMR (CDCl3): δ = 1.04 (t, 3H, CH3CH2), 1.76 (m, 2H, CH3CH2),

3.37 (t, 2H, SCH2), 4.59 (s, 2H, ArCH2), 5.27 (d, 1H, CH2), 5.76 (d, 1H, CH2), 6.67 and 6.70 (ABq,

1H, CH), 7.30 and 7.34 (ABq, 4H, ArH).

CHAPTER 5

120

Hyperbranched PAN polymer (2)

2

VBPT 1 (7.32 g, 27.0 mmol), purified acrylonitrile (7.17 g, 135.0 mmol), AIBN (0.440 g, 2.7 mmol)

and dimethylformamide (75 mL) were added in a Schlenk flask, degassed by four freeze-pump-thaw

cycles, charged with nitrogen and sealed. The reaction mixture was heated to 80 °C in a thermostatic

oil bath. After a predetermined time, the polymer was precipitated in methanol, centrifuged at 5,000

rpm for 15 min, the solid was redissolved in DMF and precipitated in methanol again. After

centrifugation the solvent was removed and the polymer 2 was dried in vacuo. 1H NMR (DMSO-d6): δ

= 0.94 (CH3CH2CH2S, terminal trithiocarbonate functionality), 1.3–3.1 (CH2 and CH, VBPT and AN

units), 3.17 and 3.36 (CH3CH2CH2S, terminal trithiocarbonate functionality), 4.11 (CH2(Ar)CHS,

styryl unit originated from reacted VBPT), 4.66 (CH(Ar)CH2S, VBPT unit), 4.74 (NCCHS, terminal

AN unit), 6.7-7.5 (ArH, VBPT unit).


In this work we examined the cloud point temperatures of the ternary phase system of the PAN-based

DralonX copolymer in the solvent DMF and coagulant water. Additionally the influence of addition of

the rheological additive 2 on the cloud point temperature in a spinning solution was tested by

CHAPTER 5

121

turbidimetry. Fibers of DralonX polymer solution with and without rheological additive 2 were spun

and analyzed through field emission scanning electron microscopy.

5.3.1 Advantages of a ternary phase diagram for the spinning process

The spinning of PAN-based fibers typically involves a concentrated polymer solution of PAN in DMF

and the coagulant water. In the coagulation bath the three components are mixed and their interaction

determines the resulting fiber properties. The diffusion processes of DMF and water taking place

simultaneously make the coagulation mechanism of the spinning dope complicated. The phase

behavior of this ternary system comprising PAN-based polymer DralonX in DMF as solvent and the

non-solvent water was examined. For low polymer concentrations under 8 wt.-% the phase diagram

can be obtained by cloud point (CP) titration.6,7

The cloud point data represent the binodal curve. For

polymer concentrations over 8 wt.-% it has been shown that the interactions between polymer chains

become so strong that polymer solutions exhibit signs of crystallization or become gel-like, therefore

making the cloud point titration inapplicable.8,9,10

For the ternary system Boom and coworkers

suggested an empirical linearized cloud point (LCP) correlation11

:

where w is the mass fraction of the component with the subscripts 1, 2, 3 referring to the non-solvent,

solvent and polymer, respectively. In the equation b is the slope and a is the intercept. With this LCP

correlation the composition at the cloud point can be determined for the ternary system with polymer

concentrations over 8 wt.-% (see Figure 4).

CHAPTER 5

122

Figure 4 The ternary phase diagram with cloud point data and theoretical binodal curve based

on cloud points determined by the LCP correlation.4

With the help of the ternary phase diagram procedures of coagulation for good fiber properties could

be identified. Tan and coworkers differentiate between three possible routes the coagulation process

can proceed (see Figure 5).4 In route 1 and 3 the diffusion processes in the fiber lead to skin-core

structures where the outer part of the fiber solidifies more quickly than the inner part. This leads to a

polymer-rich fiber skin. The slower diffusion rate in the inner part of the fiber leads to a polymer-poor

core with inhomogeneous pore structures. PAN fibers with a skin-core structure have poor mechanical

properties due to the inhomogeneity of the whole fiber. With route 2, crossing through the critical

point, an unstable region will be entered where liquid-liquid demixing will occur spontaneously. The

resulting structure will be homogeneous with the consequence of good mechanical properties.

CHAPTER 5

123

Figure 5 Schematic phase diagram of a ternary system consisting of a nonsolvent (NS), a

solvent (S), and a polymer.4

For the Megacarbon project first measurements of the cloud points were undertaken (see Figure 6).

The determination was done by cooling a mixture of PAN/DMF/H2O down from 95°C. The resulting

cloud point temperatures give an indication for the coagulation temperatures. In agreement with

literature the temperature of the cloud points increases with higher water content. This is due to the

increased dissolving ability of DMF at higher temperatures necessitating higher amount of water to

precipitate the polymer. Future works should focus on a titration of water into an isothermal system

polymer/DMF to gain phase diagrams at fixed temperatures leading to the binodal and spinodal curves

for the determination of the critical point.

CHAPTER 5

124

Figure 6 Diagram of cloud points in the ternary system PAN/DMF/H2O at different

temperatures.

For the purpose of determining the spinnability of DralonX polymer with rheological additive

measurements of cloud point temperatures have been carried out (see Figure 7). The 30 wt.-%

DralonX solution in DMF with 6, 8, 10 and 12 wt.-% water was mixed with 1 wt.-% additive 2 and

cooled down from 95°C. The cloud point temperatures of the ternary mixture with additive are higher

than the CP temperatures of ternary mixtures without additive. Addition of the rheological additive 2

increases the CP temperature by roughly 20°C. In turn the increased CP temperature caused by the

additive results in a quicker onset of coagulation when cooling. Wang and coworkers showed for an

acrylonitrile copolymer/solvent/water system that a high temperature of the coagulation bath on one

hand, leads to beneficial round fiber cross-sections with a higher solid content in the dope but on the

other hand, comes with a relatively higher number of macro voids.12

Lowering the bath temperature

will increase the fiber bulk density but will also decrease the elongation at break and the fiber tenacity.

For a carbon fiber with good mechanical properties round cross-section, high bulk density and high

tenacity are necessary for the precursor fiber. Therefore the fiber cross-sections should be examined to

determine the scope of the macro voids and the influence of additive on pore structure of the precursor

fiber.

DMF

PAN

H2O

CHAPTER 5

125

6 8 10 12 1460

65

70

75

80

85

90

0% additive 2

1% additive 2

clo

ud

po

int

tem

pe

ratu

re [

°C]

water concentration [%]

Figure 7 Cloud point measurements of a 30 wt.-% DralonX solution in DMF with 6 wt.-%

water with 1 wt.-% rheological additive 2 (circle); without rheological additive 2

(square).

5.3.2 SEM images of spun poly(acrylonitrile) based precursor fibers and the influence of

rheological additive on the fiber morphology

Results of the spinning process can only be compared with a fixed set of parameters. Changing a

parameter such as the coagulation bath temperature would have a significant impact on the diffusion

rates of the solvent and non-solvent and consequently on the fiber morphology. The phase diagram of

the ternary system as described in the previous subchapter would help to identify the optimal

combination of the parameters.

First preliminary tests in a laboratory setup were made with the aim of showing the feasibility of

spinning fibers containing the synthesized rheological additive 2 in the amount of 3 wt.-%. The

precursor fibers were spun from a 30 wt.-% DralonX copolymer in DMF without fiber stretching.

CHAPTER 5

126

The DralonX fiber without rheological additive was wet spun at a laboratory setup with a feed

temperature of 60°C into a pure water coagulation bath at temperatures of 30°C and 60°C. After

drying, the fiber cross-sections were analyzed by electron microscopy (see Figure 8). The cross-

section of a fiber spun in a water coagulation bath at 30°C shows a high homogeneity with no

observable macro voids. The structural pore sizes are below 200 nm. The cross-section suggests at a

bone shaped fiber profile. Spinning with an increased coagulation bath temperature of 60°C leads to

finger-formed macro voids bigger than 10 µm. The structural pore sizes are below 200 nm as well.

The cross-section indicated an ideally circular fiber profile which is preferred for carbon fibers due to

ideal stress stability.

Figure 8 DralonX fibers without additive spun into a pure water coagulation bath at 30°C (a)

with an enlargement on the pore structure (b) and at 60°C (c) with an enlargement (d).

b a

c d

CHAPTER 5

127

To the 30 wt.-% DralonX spinning solution 3 wt.-% of rheological additive 2 has been added and spun

into a water-DMF coagulation bath of the ratio of 40:60 and a temperature of 60°C without stretching.

The fiber cross-sections have been analyzed through electron microscopy (see Figure 9). The images

show fine structural pores of sizes of 200 to 500 nm as well as macro voids with sizes between 5 – 20

µm. Deshmukh and coworker have shown for dry-jet wet spinning that a hydrophilic additive

increases the precipitation rate and leads to the formation of larger macro voids in the fiber.13

The

hydrophilic trithiocarbonate group of the RAFT agent could have a similar effect, explaining the

bigger structural pore sizes. The number and size of the unwanted macro voids can be minimized by

the stretching process.

Figure 9 DralonX fiber spun with 3 wt.-% rheological additive (a); magnification of the pore

structure of the DralonX fiber (b).

After showing the feasibility of the spinning process with the rheological additive and the influence on

the fiber morphology a spinning experiment has been undertaken with the lab spinning equipment at

Dralon. Stretching to the spun fiber was applied by an increase of roller speed from 4 m/min to 36

m/min, equaling a stretching factor of 9. The resulted precursor fiber exhibited a tensile strength of 49

cN/tex which lies in the tensile strength range of PAN-based CF precursor fibers with 27 – 80

cN/tex.14

The fiber cross-sections have been analyzed through electron microscopy (see Figure 10). The cross-

section shows homogeneous structural pores with sizes of 50 nm and scattered pores with sizes of 200

a b

CHAPTER 5

128

nm. The spinning of a 30 wt.-% DralonX fiber without rheological additive was not possible due to

safety reasons of the laboratory spinning equipment.

Figure 10 SEM image of DralonX fiber with 3 wt.-% rheological additive 2 spun at

Dralon with stretching.

5.4 Conclusion

Every increase in the concentration of the polymer spinning solution can lower the production price of

the fiber as well as improve the fiber morphology. For managing the resulting increase in viscosity a

viscosity-reducing additive can be applied. In the experiments the feasibility of spinning with the

rheological additive under near industrial conditions was shown. The influence on the fiber

morphology through formation of macro voids could be counteracted with stretching. In future works

the spinning conditions have to be optimized to reduce the formation of macro voids and improve the

stretching parameters for further reduction in number and size of macro voids, therefore improving the

mechanical fiber properties. Experiments should include the analysis of fibers with and without

additive prepared under the same spinning conditions.

CHAPTER 5

129

5.5 References

(1) Capone, G. J. In Acrylic Fiber Tech. and Appl.; Masson, J. C., Ed.; Marcel Dekker, Inc.: New

York, 1994; p. 69.

(2) Koch, P. A. Faserstoff-Tabellen nach P.-A. Koch; 1989.


(4) Tan, L.; Pan, D.; Pan, N. J. Appl. Polym. Sci. 2008, 110, 3439.

(5) Zhang, C.; Zhou, Y.; Liu, Q.; Li, S.; Perrier, S.; Zhao, Y. Macromolecules 2011, 44, 2034.

(6) Wei, Y. M.; Xu, Z. L.; Yang, X. T.; Liu, H. L. Desalination 2006, 192, 91.

(7) Haghtalab, A.; Mokhtarani, B. Fluid Phase Equilib. 2004, 215, 151.

(8) Sadeghi, R.; Ziamajidi, F. Fluid Phase Equilib. 2007, 255, 46.

(9) García-Lopera, R.; Monzó, I. S.; Abad, C.; Campos, A. Eur. Polym. J. 2007, 43, 231.

(10) Tan, L.; Liu, S.; Pan, D. Colloids Surfaces A Physicochem. Eng. Asp. 2009, 340, 168.

(11) Boom, R. M.; van den Boomgaard, T.; van den Berg, J. W. A.; Smolders, C. A. Polymer

(Guildf). 1993, 34, 2348.

(12) Wang, Y. X.; Wang, C. G.; Yu, M. J. J. Appl. Polym. Sci. 2007, 104, 3723.

(13) Deshmukh, S. P.; Li, K. J. Memb. Sci. 1998, 150, 75.

(14) Wade, B.; Knorr, R. Acrylic Fiber Technology and Applications; Masson, J. C., Ed.; Marcel

Dekker, Inc.: New York, 1995.

List of Publications

130

List of Publications

1. Thavanesan, Thaanuskah; Herbert, Christian; Plamper, Felix A., Insight in the LCST

Pecularities of Poly(dialkylaminoethyl methacrylate)s, Langmuir 2014, 30(19), 5609-

5619.

2. Steinschulte, Alexander A.; Schulte, Bjoern; Drude, Natascha; Erberich, Michael; Herbert,

Christian; Okuda, Jun; Moeller, Martin; Plamper, Felix A., A nondestructive, statistical

method for determination of initiation efficiency: dipentaerythriol-aided synthesis of

ternary ABC3 miktoarm stars using a combined „arm-first“ and „core-first“ approach,

Polymer Chemistry 2013, 4(13), 3885-3895.

3. Plamper, Felix A.; Steinschulte, Alexander A.; Hofmann, Christian H.; Drude, Natascha;

Mergel, Olga; Herbert, Christian; Erberich, Michael; Schulte, Bjoern; Winter, Roland;

Richtering, Walter, Toward Copolymers with Ideal Thermosensitivity: Solution

Properties of Linear, Well-Defined Polymers of N-Isopropyl Acrylamide and N,N-

Diethyl Acrylamide, Macromolecules 2012, 45(19), 8021-8026.

4. Tu, Tao; Mao, Han; Herbert, Christian; Xu, Mizhi; Dötz, Karl Heinz, A pyridine-bridged

bis-benzimidazolylidene pincer nickel(II) complex: synthesis and practical catalytic

application towards Suzuki-Miyaura coupling with less-activated electrophiles, Chemical

Communications 2010, 46(41), 7796-7798.

Acknowledgment

131

Acknowledgment

My dissertation was only possible because of the contributions of the many people I had the fortune to

work and talk with and due to the continuous support of my family. Therefore I would like to say my

thanks to these people.

Starting with Prof. Dr. Martin Möller, I would like to thank him for giving me the chance to work on

this interesting topic and the informative talks we had.

My thanks go to my supervisor Dr. Helmut Keul for his scientific support and never ending patience

despite all the questions I asked him. I’m thankful to him for teaching me how to focus on the relevant

points and become a better scientist.

Many thanks go to my co-supervisor Dr. Helga Thomas for supporting me and always having a

sympathetic ear for my questions and challenges.

Myself being a part of the Keul group, I would also like to thank all my colleagues and friends in the

DWI, especially my old lab mates Wiktor, Jörg and Thorsten for the welcoming reception and their

efforts in introducing me to the deeper mysteries of polymer chemistry. I want to thank my research

students Pascal Schmitz, Stefan Chang and Sascha Schriever for their very good work and making the

time a pleasant experience. I also want to thank all the other members of the Keul group for their

support and the friendly atmosphere during my PhD time. And of course, I want to thank Rainer Haas

for his tireless support and work to keep things running.

Many thanks go to Tobias Lülf for his contribution to my thesis and to the technical staff of the DWI

without whose help my thesis would have been impossible. And a special shout-out to the PhD

students of the Pich group for all the good times we had together.

Last but not least, I would like to thank my family and my friends for helping and supporting me over

the time.

Curriculum Vitae

132

Curriculum Vitae

Personal Details

Name Christian Herbert

Date of birth May 15th, 1981

Place of birth Heltau, Romania

Citizenship German

Education

August 1997 – June 2000 High school: Gymnasium Zitadelle Jülich, degree: Abitur.

October 2001 – March 2009 Study chemistry at the Rheinische Friedrich-Wilhelms-

Universität Bonn; degree: Diplom.

September 2009 – April 2014 PhD work under the supervision of Prof. Dr. Martin Möller at

the DWI Leibniz-Institut für Interaktive Materialien.

PhD thesis “Rheological and LASER additives for higher

efficiency in producing poly(acrylonitrile)-based carbon

fibers”

rheological and laser additives for higher efficiency in

Documents