Thermo-oxidative stabilization of polyacrylonitrile and its copolymers: Effect of molecular weight, dispersity, and polymerization pathway

Accepted Manuscript Thermo-oxidative stabilization of polyacrylonitrile and its copolymers: Effect of molecular weight, dispersity, and polymerization pathway Jeremy D. Moskowitz, Jeffrey S. Wiggins PII:

S0141-3910(15)30175-0

DOI:

10.1016/j.polymdegradstab.2015.12.025

Reference:

PDST 7832

To appear in:

Polymer Degradation and Stability

Received Date: 12 August 2015 Revised Date:

13 October 2015

Accepted Date: 29 December 2015

Please cite this article as: Moskowitz JD, Wiggins JS, Thermo-oxidative stabilization of polyacrylonitrile and its copolymers: Effect of molecular weight, dispersity, and polymerization pathway, Polymer Degradation and Stability (2016), doi: 10.1016/j.polymdegradstab.2015.12.025. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

THERMO-OXIDATIVE STABILIZATION OF POLYACRYLONITRILE AND ITS COPOLYMERS: EFFECT OF MOLECULAR WEIGHT, DISPERSITY, AND POLYMERIZATION PATHWAY

RI PT

Jeremy D. Moskowitz1, Jeffrey S. Wiggins1 1

School of Polymers and High Performance Materials, The University of Southern Mississippi, 118 College Drive #5050, Hattiesburg, MS 39401, United States

Abstract

SC

Correspondence to: Jeffrey S. Wiggins ([email protected]), P: 601-266-4869, F: 601-266-5635

M AN U

Careful control of polyacrylonitrile (PAN) precursor properties and, in particular, the polymerization method and ensuing molecular weight and molecular weight distribution are important considerations for mitigating defects and enhancing processability of carbon fibers. Herein, a comprehensive study was performed to understand the influence of molecular weight,

TE D

molecular weight distribution, and polymerization method between reversible additionfragmentation chain transfer (RAFT) and conventional free radical (FR) solution polymerization on the cyclization behavior and structural evolution of stabilized PAN. The kinetic parameters of

EP

activation energy (Ea) and pre-exponential factor (A) were determined along with the cyclization index (CI) by differential scanning calorimetry (DSC) and the extent of stabilization (Es) was

AC C

measured via fourier transform infrared spectroscopy (FTIR). Thermogravimetric analysis (TGA) was used to determine the degradation differences in the polymers. Structural characterization was performed by wide-angle X-ray scattering (WAXS). DSC and FTIR analysis indicate that cyclization initiates at lower temperature for FR polymers. Significantly, PAN-based RAFT copolymers show greater mass retention post thermo-oxidative degradation as compared to analogous FR copolymers.

ACCEPTED MANUSCRIPT

Keywords: Polyacrylonitrile, cyclization, reversible addition-fragmentation chain transfer (RAFT), thermos-oxidative stability 1. Introduction

RI PT

Polyacrylonitrile (PAN) is the leading precursor material for carbon fiber production in industry and, in particular, yields the highest tensile strength in comparison to pitch and

cellulose-based carbon fibers [1,2]. Traditionally, PAN is synthesized via uncontrolled free

SC

radical polymerization in an emulsion or solution leading to atactic polymers with broad

dispersities and a small degree of branching. Typical carbon fiber precursors range between

M AN U

70,000−200,000 g/mol with dispersities (ÐM) (Mw/Mn) between 2−3 [2]. High molecular weight fractions can cause issues during fiber spinning and filtration such as gelation and insolubility [3]. In the carbon fiber manufacturing process, molecular weight and its distribution govern final carbon fiber properties [4–6]. Increasing molecular weight leads to increased crystallite size and

TE D

smaller void sizes as well as changes in the fiber morphology [4,7]. Moreover, increasing molecular weight has been shown to influence which small volatiles are released during thermooxidative stabilization of PAN-based precursors [8].

EP

The cyclization of PAN into ring-closed ladder structures at temperatures above 200 °C is recognized as one of the most important prerequisites for high performing carbon fiber (Scheme

AC C

1) [1], wherein the homopolymer cyclizes according to a well-established free radical mechanism while addition of acidic comonomers introduces an ionic route for cyclization [9,10]. However, the effect of molecular weight and its distribution on the cyclization of PAN remain unresolved in the literature. Tsai et al. claim that cyclization onset temperature increases with higher molecular weight, but samples in their study had high dispersity (ÐM) (>1.8) [4]. Chernikova et al. asserted the same finding for a series of low molecular weight samples (<

ACCEPTED MANUSCRIPT

3,000 g/mol), yet cyclization exotherms yielded several unresolved and overlapping peaks [6]. Other reports in the literature make similar claims [7], but lack hard evidence due to unreported

M AN U

SC

RI PT

dispersities or conflicts within the data.

Scheme 1. Cyclization of PAN (a) homopolymer and (b) copolymer with acrylic acid. Equally important to molecular weight is the polymerization mechanism. Solution

TE D

polymerization is favorable due to immediate manufacture of spinning dopes [11,12]. The two main drawbacks to solution polymerized precursors are that monomer conversion is typically only 50−70%, and that conventional solvent systems are characterized by high transfer constants

EP

leading to undesirable side reactions during polymerization [11]. It has been demonstrated that an emulsion polymerization at low temperature (22 °C) led to a smaller proportion of long chain

AC C

branches than solution polymerization at higher temperature (50 – 60 °C), even though the solution-formed polymer had a narrower dispersity [13]. Long chain branches present in conventional free radical solution polymers hinders reptational motion of the macromolecule. Therefore, polymerization mechanism must be taken into consideration when evaluating precursor properties.

ACCEPTED MANUSCRIPT

Although studies have shown that there is a molecular weight dependence on the formation PAN-based carbon fibers [6–8], it is not clear how dispersity and molecular weight affect the cyclization temperature, the extent of cyclization, and the structural changes that occur

RI PT

through the stabilization phase of PAN-based carbon fibers. This prompted our investigation into the thermo-oxidative stabilization of PAN of varying molecular weights and dispersities.

Recently, we have demonstrated the synthesis of controlled PAN with absolute number average

SC

molecular weights exceeding 130,000 g/mol and dispersities below 1.30 via reversible additionfragmentation chain transfer (RAFT) polymerization [14]. Herein, we compare the cyclization

M AN U

behavior of RAFT polymers with narrow molecular weight distributions to conventional free radical polymers with broad dispersities. Significantly, PAN-based RAFT copolymers demonstrate promising characteristics such as increased thermo-oxidative stability, which could lead to the development of carbon fibers based upon controlled polymerization techniques.

2.1 Materials

TE D

2. Materials and Methods

EP

Acrylonitrile (AN, 99%, 35-45 ppm monomethyl ether hydroquinone inhibitor, Sigma-

AC C

Aldrich Chemical Co.) was passed through a neutral aluminum oxide (50-200 µm) column to remove the inhibitor. Ethylene carbonate (EC, 98%, Sigma-Aldrich Chemical Co.), 2-cyano-2propyl dodecyl trithiocarbonate (CPDT, 97%, Sigma Aldrich Chemical Co.), Nisopropylacrylamide (NIPAM, 99%, Sigma Aldrich Chemical Co.), 2,2'-Azobis(4-methoxy-2.4dimethyl valeronitrile) (V-70, Wako Pure Chemical Industries, Ltd.), 2,2’-asobis(isobutyronitrile) (AIBN, 98%, Sigma Aldrich Chemical Co.), and N,N-Dimethylformamide (DMF, 99%, Sigma

ACCEPTED MANUSCRIPT

Aldrich Chemical Co.) were used as received. Acrylic acid (AA, 99%, 180-200 ppm MEHQ

RI PT

inhibitor) was vacuum distilled to remove inhibitor.

2.2 Conventional Free Radical Polymerization

SC

A typical solution polymerization was as follows: ethylene carbonate (32.0 g),

acrylonitrile (7.99 g, 151 mmol, 20 wt. %), and V-70 (9.5 mg, 0.031 mmol, stock solution in

M AN U

tetrahydrofuran (THF) 20 mg/mL) were charged to a 100 mL round bottom flask equipped with a magnetic stir bar. The flask was capped with a rubber stopper and placed in an ice bath and subsequently purged for 1 hour under nitrogen gas before being transferred to an oil bath set to 30 °C for 8 hours. When AIBN was used as the initiator the oil bath was set to 65 °C (the equivalent

TE D

initiator 10 h half-life temperature). The polymer was precipitated into a 1 L bath of approximately 70:30 (v:v) deionized water and methanol followed by isolation via filtration and

EP

allowed to dry for 30 minutes under nitrogen. The polymer was then re-dissolved in dimethylformamide (DMF) at 20 to 25 weight percent polymer, and the solution poured over a

AC C

flat surface and allowed to spread into a film. The film was coagulated by soaking in deionized water for 3-4 hours followed by a 24 h Soxhlet extraction in methanol to remove residual solvent. The polymer was then dried in a vacuum oven at 60 °C overnight.

2.3 RAFT Polymer Synthesis

ACCEPTED MANUSCRIPT

A typical RAFT polymerization for a molar ratio of [AN]0:[CPDT]0:[V-70]0 = 10,000:1:0.67 was as follows: ethylene carbonate (32.0 g) , acrylonitrile (7.99 g, 151 mmol, 20 wt.%), CPDT (5.2 mg, 0.015 mmol, stock solution in dimethylformamide (DMF) 20 mg/mL),

RI PT

and V-70 (3.1 mg, 0.010 mmol, stock solution in tetrahydrofuran (THF) 20 mg/mL) were

charged into a 100 mL round bottom flask equipped with a magnetic stir bar. The flask was capped with a rubber stopper and placed in an ice bath and subsequently purged for 1 hour under

SC

nitrogen gas before being transferred to an oil bath set to 30 °C for 24 or 48 hours. Polymers were precipitated in and dried in the same manner as the uncontrolled free radical polymers.

M n ,theor =

M AN U

Theoretical number average molecular weights, Mn,theor, were determined by Equation 1 [15]:

M MW • [ M ]0 • conversion + CTAMW [CTA]0

(1)

where [M] 0 and [CTA]0 correspond to the initial monomer and chain-transfer agent

TE D

concentrations, respectively. MMW and CTAMW correspond to the molecular weights of the monomer and chain-transfer agent, respectively.

EP

2.4 Oxidation Treatment

AC C

To simulate industrial processing techniques, polymer films were subjected to an oxidation treatment in a programmable oven. About 50 mg of each polymer was ramped to 225 °C at 5 °C/min air and held at 225 °C. The oxidation temperature of 225 °C was predetermined to be below the rapid exotherm temperature and at which observable differences could be demonstrated for short times (30 min) and for long reaction times (5 hours).

ACCEPTED MANUSCRIPT

2.5 Size Exclusion Chromatography (SEC) Number average molecular weight (Mn), weight average molecular weight (Mw), and

RI PT

dispersity of the polymer samples were measured using a SEC system consisting of a Waters Alliance 2695 separation module, online multi-angle laser light scattering (MALLS) detector fitted with a gallium arsenide laser (20 mW power) operating at 690 nm (MiniDAWN Wyatt

SC

Technology Inc.), an interferometric refractometer (Optilab DSP, Wyatt Technology Inc.), and two Agilent PLgel mixed-C columns connected in series. HPLC grade dimethylformamide

M AN U

(DMF) with 0.02 M LiBr served as the mobile phase and was delivered at a flow rate of 0.5 mL/min while operating at 60 °C. Sample concentrations were prepared at about 5 mg/mL with an injection volume of 100 µL. The detector signals were simultaneously recorded using ASTRA software (Wyatt Technology Inc.). The dn/dc for PAN (0.084 mL/g) was measured from the refractive index given at 5 polymer concentrations between 1 and 10 mg/mL. Molecular weights

TE D

were also determined relative to polymethylmethacrylate (PMMA) standards.

1

EP

2.6 Nuclear Magnetic Resonance (NMR) Spectroscopy H-NMR experiments were recorded with a Bruker Acend™ 600 MHz spectrometer. 1H-

AC C

NMR experiments were conducted at 30 °C with 5 wt. % solutions in DMSO-d6 using 32 scans and a 1 second relaxation delay. Conversion was determined by the monomer concentration (δ (ppm): d, 6.22; d, 6.36; quad, 6.00) with respect to the ethylene carbonate (δ (ppm): s, 4.50).

2.7 Differential Scanning Calorimetry (DSC)

ACCEPTED MANUSCRIPT

DSC was performed on a TA Instruments DSC Q200 equipped with a nitrogen purge gas. Cyclization activation energy was determined from the Flynn-Wall-Ozawa (FWO) and Kissinger

RI PT

methods, Equations 2 and 3, respectively, where Ea is the activation energy, R is the universal gas constant, Tp is the peak exotherm temperature in Kelvin, and φ is the temperature ramp rate in

Equation 4 [10]. Ea d (ln φ ) = 0.951 • R  1  d  T   p

M AN U

−

SC

Kelvin [10,16–19]. The pre-exponential ‘collision frequency’ factor, A, is determined from

TE D

  φ  d ln  2    E   T p  − a =  R  1  d  T   p

A=

φ Ea RT

2 p

e

Ea

(2)

(3)

RTp

(4)

EP

The cyclization index (CI) was calculated using heat released related to exotherm DSC peaks by Equation 5,[20–23] where Hv and Ho are the heat released for the virgin and stabilized

AC C

polymers, respectively:

CI =

H v − Ho • 100% Hv

(5)

The Hv and Ho were calculated by integrating the total area under the exothermic peak related to cyclization. 2.8 Fourier Transform Infrared Spectroscopy (FTIR)

ACCEPTED MANUSCRIPT

FTIR spectra in transmission mode were recorded using a Thermo Scientific Nicolet 6700 FTIR in the range of 4000-400 cm-1. An infrared source was used in conjunction with a KBr beam

RI PT

splitter and a DTGS KBr detector. Samples were prepared by dissolving about 5-10 wt.% polymer in DMF and then cast onto a polished 25mm x 4mm NaCl salt disc. The samples were then placed in a vacuum oven at 60 °C overnight to remove the DMF. The samples were monitored in real-

SC

time in a Simplex Scientific Heating Cell where 32 scans at 4 cm-1 resolution were acquired every

M AN U

5 minutes during the cyclization reaction. The sample chamber was purged with nitrogen throughout the duration of the reaction as the temperature was raised to 225 °C at 5 °C/min and then held at 225 °C for 5 hours. The unreacted nitrile fraction was determined by Equation 6 [9], while the extent of stabilization, Es, was determined by Equation 7 [10], where A2240cm-1 is the

TE D

absorption intensity of the nitrile (-C≡N), A1590cm-1 is the absorption intensity of the imine (-C=N), and f is the ratio of the absorptivity constants (0.29) [24]:

Es =

A2240cm−1

A2240cm−1 +( f • A1590cm−1 )

(6)

A1590cm−1 A2240cm−1

(7)

AC C

EP

− CN unreacted =

2.9 Thermogravimetric Analysis (TGA)

TGA results were recorded on a TA Instruments Discovery Thermogravimetric Analyzer. TGA was utilized to determine mass loss of the PAN samples in a nitrogen atmosphere ramped at 10 °C/min to 700 °C. The TGA was also programmed to pre-oxidize the polymers by heating the

ACCEPTED MANUSCRIPT

polymers in air at 10 °C/min and holding at 230, 245, 253, and 265 °C, for 15 minutes each,

2.10

RI PT

before ramping to 600 °C in nitrogen.

Wide Angle X-Ray Scattering (WAXS)

SC

Wide angle X-ray scattering measurements were performed on a Brüker NanoSTAR system with an Incoatec IµS microfocus X-ray source operating at 45 kV and 650 mA with a

M AN U

wavelength of Cu Kα = 0.1542 nm. The primary beam was collimated with cross-coupled Göbel mirrors and a pinhole of 0.1 mm in diameter. The sample to detector distance was 5.2 cm and a 20 min exposure time was used to collect the WAXS pattern on a Fuji Photo Film image plate with a Fuji FLA-7000 Scanner. Diffraction peaks and integral areas were acquired via a peak fitting

TE D

method by the software Origin 8.0. The lateral crystallite thickness (Lc) of the pseudohexagonal crystal lattice [25] was determined by the Scherrer equation (Eqn. 8) for the (100) reflection at 2θ

EP

= 17°, where λ = 0.154 nm is the wavelength of CuKα X-ray, θ is the Bragg diffraction angle, B is

AC C

the full width at half maximum intensity in radians, and K is a constant and is equal to 0.9 [26]:

Lc =

Kλ B cos(θ )

(8)

The percent crystallinity (χc) was determined by the Bell and Dumbleton method [27] (Eqn. 9), where Ac and Aa are the integral areas of the crystalline zone c.a. 2θ = 17° and the amorphous zones, respectively:

ACCEPTED MANUSCRIPT

χc =

Ac • 100% Ac + Aa

(9)

The stabilization index (SI) was used to evaluate the quantitative changes in structural order

RI PT

related to the formation of cyclic structure by Equation 10, where I0 and Ii are the intensities of the 2θ = 17° peak for the virgin and heat treated polymers, respectively [28]:

I0 − Ii • 100% I0

( 10 )

SC

SI =

M AN U

3. Results and Discussion 3.1 RAFT and FR Polymerization of Acrylonitrile

Six PAN samples were selected of varying molecular weight and dispersity. The PAN samples are subdivided into two groups: samples prepared by conventional uncontrolled free-

TE D

radical solution polymerization (FR-1, FR-2, FR-3), and those prepared via RAFT polymerization (RAFT-1, RAFT-2, RAFT-3). The results are summarized in Table 1. The molecular weights of the FR samples range between 25,000 – 260,000 g/mol and ÐM ranges

EP

from 1.59 − 2.09. FR-2 was synthesized utilizing the initiator, AIBN, at 65 °C to give the same end groups (2-cyano-2-propyl) as the RAFT polymers to mitigate end group effects [8,29]. The

AC C

RAFT samples have molecular weights (MALLS) ranging between 42,000 – 135,000 g/mol and ÐM from 1.12 – 1.27. The molecular weights are confined within the lower and upper limits of the FR polymers.

Molecular weights were also reported relative to PMMA standards for

comparison with literature reported values due to the issues of performing SEC on PAN [30].

Table 1. FR and RAFT samples of PAN.

ACCEPTED MANUSCRIPT

Entrya

T (°C)b Time

[M]0:[CTA]0:[I]0

Conv.

Mn (Theor.)d Mn (MALLS)e Mn (PMMA)f ÐMe

(h)

(%)c

(g/mol)

(g/mol)

130:0:1

30

8

-

-

25,900

FR-2

3,260:0:1

65

8

-

-

173,400

447,000

1.59

FR-3

4,900:0:1

30

8

-

-

257,800

546,300

2.09

RAFT-1

1,333:1:0.22

30

24

59

42,100

42,000

116,500

1.12

RAFT-2

6,667:1:0.67

30

48

67

236,700

90,000

281,000

1.23

RAFT-3

10,000:1:0.67

30

48

71

377,100

131,400

423,100

1.27

SC

M AN U

a

71,400

RI PT

FR-1

1.88

Polyacrylonitrile synthesized in ethylene carbonate at 20 wt.% monomer concentration under a nitrogen atmosphere. bReactions at 30 °C used V70 as the initiator and reaction at 65 °C used AIBN as the initiator. cConversions were determined by 1H-NMR spectroscopy recorded in DMSOd6. dTheoretical number average molecular weights were calculated according to the following equation: Mn=(ρ·MWmon·[M]/[CTA]) + MWCTA where ρ is the fractional monomer conversion, MWmon is the molecular weight of the monomer, and MWCTA is the molecular weight of the CTA. e Number average molecular weight (Mn) as determined by SEC (0.5 mL/min, 60 °C, Polymer Labs PL gel 5 µm mixed-C column, DMF with 0.02 M LiBr eluent) equipped with RI and MALLS detectors. fMn relative to PMMA standards.

TE D

3.2 Influence of Molecular Weight and Dispersity on the Cyclization of PAN

The DSC method is one of the most prominent methods for characterizing the event of

EP

cyclization [20,22,31]. DSC gives a very narrow peak for the exothermic cyclization reaction for homopolymer PAN due to the rapid free radical ring-closing mechanism of adjacent nitriles along

AC C

the polymer chain [10,31]. Figure 1 shows the overlaid DSC curves for the PAN samples for 5 °C/min ramps in nitrogen. The RAFT homopolymers exhibit very narrow exotherms with peaks centered between 274−279 °C, while the FR homopolymers show exothermic peaks in the range of 260−265 °C. Additionally, the RAFT polymer peaks are sharper and have heat flows c.a. 30 W/g, which is high compared to values for the FR polymers with heat flows c.a. 10−15 W/g. This

ACCEPTED MANUSCRIPT

massive amount of energy that is released over a narrow temperature range would be detrimental

M AN U

SC

RI PT

to carbon fiber fabrication and it could prematurely degrade the polymer backbone.

TE D

Figure 1. DSC curves of RAFT and FR PAN homopolymer samples at 5 °C/min.

Interestingly, both FR-1, possessing the lowest Mn, and FR-3, owning the highest Mn, have lower peak cyclization temperatures than any of the RAFT polymers as seen in Figure 1. Because

EP

the end groups of the FR-2 polymer are the same as the RAFT polymers (2-cyano-2-propyl),

AC C

which still has a lower peak cyclization temperature, end group effects are expected to be negligible. This indicates that dispersity and polymerization mechanism could have a profound effect on the cyclization exotherm. Moreover, RAFT-1 has a higher peak cyclization temperature than the RAFT-2 and RAFT-3 polymers, which contradicts literature claims that higher Mn leads to higher Tp [4,6,7]. The peak exotherm temperature for several ramp rates (2, 5, 10, and 20 °C/min) are listed in Table 2. The exotherm temperature increases with increasing ramp rate.

ACCEPTED MANUSCRIPT

Clearly, all peak exotherms are higher for the RAFT samples than the analogous FR samples for a given ramp rate.

RI PT

Table 2. DSC peak maxima for RAFT and FR PAN Homopolymer samples

Peak Temperature (°C) Entry

FR-1

252

265

FR-2

249

260

FR-3

257

261

RAFT-1

261

RAFT-2

262

RAFT-3

261

10 °C/min

20 °C/min

SC

5 °C/min

272

286

268

276

278

289

M AN U

2 °C/min

279

291

312

274

288

310

275

290

307

TE D

Applying the Ozawa and Kissinger equations, we can obtain activation energy and preexponential factors, which are compiled in Table 3 along with the results from the 5 °C/min ramp

EP

rate.

Table 3. Activation energies of RAFT and FR PAN homopolymers Kissinger Methodb

AC C

Ozawa Methoda

Entry

Ea (kJ/mol)

A (sec-1)

5 °C/min

Ea (kJ/mol) A (sec-1)

13

FR-1

158.3

1.32x10

FR-2

196.0

1.10x10

FR-3

150.6

2.76x10

RAFT-1

114.8

2.78x10

17 12 8

Breadthd Intensitye

Heat Flowf

(°C)

(W/g)

Tpc (°C)

13

157.6

1.11x10

197.3

1.47x10

149.3

2.07x10

111.5

1.31x10

17 12 8

(J/g)

265

15

512

13.5

260

19

671

9.6

261

13

589

13.7

279

7

563

29.4

ACCEPTED MANUSCRIPT

RAFT-2

115.9

RAFT-3

121.1

8

112.7

4.59x10 9

118.2

1.39x10

a

8

2.18x10 8

7.10x10

274

6

591

31.9

275

9

531

32.1

b

c

RI PT

Activation energy determined from Ozawa method. Activation energy determined by Kissinger method. The peak exotherm temperature was determined as the temperature with the maximum heat flow via DSC. dThe peak exotherm breadth was determined by the onset and end of the peak exotherm temperature. eThe intensity was calculated by the integral area under the peak exotherm. fThe heat flow was taken as the maximum heat flow at the peak temperature.

As shown in Table 3, the activation energies of the three polymers prepared by RAFT polymerization are lower than the respective FR polymers. The exotherm intensities for all six

SC

samples are similar. Interestingly, the collision frequency (A) is 4−9 orders of magnitude higher in

M AN U

the case of the FR polymers. We can rationalize that FR polymers have more defect sites and impurities from residual polymerization products that provide more opportunities for cyclization to occur [8]. Also, the frequency factor increases in order of molecular weight for the RAFT polymers, which intuitively suggests that high molecular weight PAN has an increased statistical

TE D

probability of nitriles sterically orienting. Moreover, the RAFT polymers exhibit peak breadths ranging from 6−9 °C compared to the FR polymers with peak breadths between 13−19 °C. This could be directly related to the lower activation energy of the RAFT polymers. We can reason that

EP

a low ÐM with fewer impurities leads to chains initiating cyclization simultaneously. In general,

AC C

the higher the A-value the lower the peak exotherm temperature, but the higher the Ea the broader the exotherm for homopolymer PAN. 3.3 In-situ FTIR Analysis of Extent of Stabilization

Real-time FTIR spectra were recorded in a heating cell at 225 °C for 5 hours as shown in Figure 2 (FR-2 shown, see Supplemental Information for other samples). In general, stabilization leads to a decreased intensity of the narrow band at 2240 cm-1 related to the nitrile stretching and

ACCEPTED MANUSCRIPT

an increased intensity of the broad band centered around 1590 cm-1 corresponding to vibrations of –C=C– and –C=N– linkages in the aromatic rings [10,22,32]. The other band assignments are as

RI PT

follows: 1452 cm-1 from the polymer backbone (–CH2– bend), 1672 cm-1 due to oxygen uptake (– C=O– stretch) or termination by disproportionation (–C=C– stretch), 2939 cm-1 from the polymer backbone (–CH2– stretch), 3212 and 3360 cm-1 due to tautomerization and isomerization of the

SC

imine into enamine and hanging amines[8] (–NH– stretch). Additionally, a small peak at about

M AN U

2220 cm-1 forms, which could be due to the isomerization of the ladder structure or

AC C

EP

TE D

dehydrogenation resulting in α,β-unsaturated nitrile groups [10,32].

Figure 2. FTIR spectra of FR-2 heated at 225 °C for 5 h in nitrogen (left) and zoomed in around 2240 and 1590 cm-1 (right). Times shown are with respect to post ramp to 225 °C at 5 °C/min. The plots for unreacted nitrile content and Es are shown in Figure 3. The stabilization

temperature of 225 °C was selected to be below the peak exotherm temperature in order to observe quantitative differences in cyclization rate. In accordance with the DSC results, the FR polymers begin cyclization prior to the RAFT polymers and reach a greater extent of

ACCEPTED MANUSCRIPT

stabilization at 225 °C over the course of 5 hours as seen in Figure 3. Specifically, RAFT-1 does not begin conversion of nitrile groups until about 45 min of heat treatment (Fig. 3a), while the FR polymers exhibit the greatest nitrile conversions. Interestingly, FR-1 shows that Es increases

RI PT

quickly, but then seems to level off over time (Fig. 3b). This could be due to the low molecular weight, but high ÐM, of this polymer, which could cause a slope change due to unequal reactivity

TE D

M AN U

SC

of the varying polymer chain sizes.

Figure 3. (a) Real-time unreacted nitrile content and (b) extent of stabilization as a function of reaction time at 225 °C in N2.

EP

3.4 TGA Analysis of Untreated and Pre-oxidized Polymers

AC C

Mass loss post thermo-oxidative stabilization is an indicator for carbon content as having a greater mass retention would suggest a more complete ladder structure formation. The TGA plots for untreated and the simulated pre-oxidation stages are shown in Figure 4. The TGA plots can be divided into several parts:

(1) The plateau between 200−270 °C represents the cyclization period which forms the ladder structure and should have minimal weight loss [31]

ACCEPTED MANUSCRIPT

(2) Rapid weight loss between 270−300 °C due to dehydrogenation and the heat liberated

RI PT

from the cyclization, which is accompanied by release of small volatiles [11]

(3) Steady weight loss between 300−450 °C due to removal of small impurities from the cyclized ladder structure such as methane, ammonia, hydrogen cyanide, water, and

SC

carbon dioxide [1]

M AN U

(4) The plateau between 450−600 °C, which encompasses the stabilized structure before

EP

TE D

further release of hydrogen cyanide, nitrogen, and hydrogen through carbonization [8].

AC C

Figure 4. TGA plots of (a) untreated PAN polymers ramped at 10 °C/min to 600 °C in N2 and (b) polymers ramped at 10 °C/min in air and held at isotherms of 230, 245, 253, and 265 °C for 15 min each and then ramped at 10 °C/min to 600 °C in N2.

Apparent in Figure 4a, the FR polymers begin weight loss (c.a. 270 °C) before the RAFT

polymers (c.a. 280 °C). Furthermore, the rapid weight loss between 270−300 °C of the RAFT polymers is steeper (> 20% mass loss) due to the greater heat flow generated during cyclization as

ACCEPTED MANUSCRIPT

compared to the more gradual slope change of the FR polymers, especially FR-3. The final plateaus (450−600 °C) coincide with the order of molecular weight, which may suggest that

RI PT

higher molecular weights lead to greater residual masses. Figure 4b demonstrates that a preoxidation treatment greatly increases stability of the ladder structures and minimizes rapid mass

SC

loss post cyclization, while nearly eliminating the effect of molecular weight.

M AN U

3.5 Cyclization Index (CI) via DSC

Polymer films were then treated in an oven at 225 °C for 30 min and for 5 h in air and then characterized by DSC. Examples of the DSC curves for oxidized samples of FR-2 and RAFT-2 are demonstrated in Figure 5. The exotherm peaks shift to a higher temperature indicating a higher

TE D

degree of stability with longer pre-oxidation time. The cyclization reaction is mostly complete by 5 h at 225 °C noted by the significant decline in exotherm intensity. The results are summarized in

EP

Table 4. The CI is consistent with the previous results showing that cyclization progresses more rapidly at 225 °C for the FR polymers. After 30 min of oxidation at 225 °C, FR polymers reach

AC C

15-20% completion of cyclization, while RAFT polymers reach only 5-10% completion and after 5 h the spread increases.

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 5. DSC curves of pre-oxidized PAN treated at 225 °C in air for 30 min and 5 h (a) FR-2 and (b) RAFT-2.

M AN U

Table 4. Cyclization Index (CI) of RAFT and FR homopolymers at 225 °C in air.

CI (%)

30 min

5h

18.5

88.6

20.1

81.5

TE D

Entry

16.0

87.8

6.7

62.2

8.1

63.1

8.5

75.0

FR-1 FR-2 FR-3 RAFT-1 RAFT-2

EP

RAFT-3

AC C

3.6 WAXS Diffraction Patterns and Crystallite Properties

The structural differences were characterized by WAXS as shown in Figure 6.

Remarkably, there are no evident structural differences between the untreated polymers despite the vastly different stabilization characteristics. Apparent in Figure 6, after 30 min of oxidation treatment, the peak c.a. 2θ = 17° (100) becomes more pronounced in all samples as compared to

ACCEPTED MANUSCRIPT

the untreated polymers. In addition a new band appears, which is attributed to the (110) planes [33]. Sharper bands and the formation of new bands with heat treatment are not uncommon in the

RI PT

literature [23,28,34]. This phenomenon is a result of thermal relaxation and an increase in flexibility of the molecular chain and a consequent orientation of polymer chains into a higher crystalline order [23]. Most obvious from Figure 6 are the differences between the samples at 5 h

EP

TE D

M AN U

much more diffuse, especially for FR-1 and FR-2.

SC

of pre-oxidation treatment, which show that the scattering patterns in the FR samples become

AC C

Figure 6. WAXS patterns for PAN samples at given heat treatment conditions in air.

Examples of the 2θ vs. intensity plots for FR-2 and RAFT-2 are demonstrated in Figure 7.

From Figure 7, the formation of the new (110) band appears c.a. 2θ = 29° peak after 30 min Furthermore, the small shift in the (110) band to a higher diffraction angle (Figure 7a) indicates a smaller d-spacing, which could also suggest structural relaxation. After 5 h, a broad halo appears

ACCEPTED MANUSCRIPT

between 15−30°, especially for the FR-2 polymer (Figure 7a) indicating amorphitization of the

M AN U

SC

RI PT

crystal structure.

Figure 7. 2θ vs. intensity of (a) FR-2 and (b) RAFT-2 at 225 °C in air for various times.

TE D

The crystallite thickness, Lc, stabilization index, SI, and crystallinity, χc, of each polymer sample are compiled in Table 5. The Lc for all samples increases after 30 min of oxidation

EP

treatment, while the SI is shown to decrease after 30 min of heat treatment. After 5 h of oxidation at 225 °C, the Lc then decreases and the SI increases dramatically, especially for FR polymers.

AC C

Noted in Table 5, the FR polymers reach a higher SI than the analogous RAFT polymers, consistent with the thermal characterization. Table 5. WAXS data summarized for RAFT and FR PAN Homopolymer samples

Untreated

30 min at 225 °C in air

5 h at 225 °C in air

Entry χc (%)

Lc (nm)

χc (%)

Lc (nm)

SI (%)

χc (%)

Lc (nm)

SI (%)

ACCEPTED MANUSCRIPT

37.6

3.58

24.2

5.97

-27.4

10.5

2.48

53.3

FR-2

44.4

3.91

28.6

5.89

-33.1

18.2

3.80

39.8

FR-3

32.3

4.41

32.7

6.45

-4.64

21.1

5.28

27.9

RAFT-1

37.6

4.82

38.1

6.95

-25.7

RAFT-2

44.0

3.83

43.9

6.46

-28.1

RAFT-3

34.3

3.67

33.0

6.26

-26.8

RI PT

FR-1

5.66

19.4

29.0

5.66

-6.2

30.4

4.46

25.7

SC

23.4

M AN U

3.7 PAN-based Copolymers

In practice, PAN homopolymers are seldom used as carbon fiber precursors due to their insolubility and poor processability. Typically, 3-10 wt.% of various comonomers are added to enhance solubility, spinnability, drawability, hydrophilicity, and assist in the stabilization of

TE D

carbon fibers during pyrolysis [35,36]. Therefore, copolymers with 5 mol% AA and NIPAM were prepared by RAFT and FR methods analogous to the homopolymers. The molecular weights and dispersities are reported in Table 6, which show that the RAFT copolymers have

EP

higher molecular weights and narrower dispersities. Significantly, this demonstrates that the low temperature RAFT polymerization utilized [14] yields high molecular weight and narrow

AC C

dispersity copolymers with no signs of reduction in conversion often reported in previous RAFT studies of acrylonitrile [3,37,38].

ACCEPTED MANUSCRIPT

Table 6. Molecular weight, conversion, and dispersities of 5 mol% RAFT and FR Copolymers (AA and NIPAM) Mn (SEC-MALLS)c (g/mol)

Conversion (%)b

5 mol% AA RAFT

197 500

78

5 mol% NIPAM RAFT

154 700

64

5 mol% AA FR

89 500

-

5 mol% NIPAM FR

78 900

-

ÐMc

SC

RI PT

Entrya

1.23

1.40

1.73

2.10

M AN U

a RAFT copolymers synthesized in ethylene carbonate at 20 wt.% monomer concentration under a nitrogen atmosphere at 30 °C and V-70 as the initiator for 48 h and FR copolymers synthesized at 65 °C with 0.02 mol% AIBN as the initiator for 8 h. bConversions were determined by 1HNMR spectroscopy recorded in DMSO-d6. cNumber average molecular weight (Mn) as determined by SEC (0.5 mL/min, 60 °C, Polymer Labs PL gel 5 µm mixed-C column, DMF with 0.02 M LiBr eluent) equipped with RI and MALLS detectors.

The PAN-based copolymers were studied by DSC as shown in Figure 8. Consistent with homopolymers, the onset of the cyclization occurs at a lower temperature for the FR copolymers. The FR copolymers should have more chain ends and more defects to initiate cyclization as

TE D

compared to RAFT copolymers. Additionally, the cyclization exotherms of the RAFT copolymers are narrower, which was shown in the case of homopolymers. The exotherm properties are summarized in Table 7. Unique from the homopolymers, the RAFT copolymers

EP

display much greater exotherm intensities than the analogous FR copolymers. The increased intensity could be related to a higher extent of reaction, which would suggest a more complete

AC C

ladder structure. Moreover, the change in cyclization mechanism as demonstrated in Scheme 1, which shows that comonomers cyclize via an ionic pathway, has significantly reduced the max heat flows in the copolymers as compared to homopolymers, which is commonly reported in the literature [9,10,39].

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 8. DSC curves of 5 mol% RAFT and FR copolymers at 5 °C/min (AA and NIPAM).

Polymer

Breadthb

a

Tp (°C)

AA_RAFT

AA_FR a

Intensityc Max Heat Flowd (W/g) (J/g)

287

18

475

4.8

241

35

461

2.5

230

24

320

1.7

52

251

2.3

AC C

NIPAM_FR

5 °C/min

(°C)

EP

NIPAM_RAFT

TE D

Table 7. Activation Energy and DSC Exotherm Behavior of 5 mol% RAFT and FR Copolymers (AA and NIPAM)

233

The peak exotherm temperature was determined as the temperature with the maximum heat flow via DSC. bThe peak exotherm breadth was determined by the onset and end of the peak exotherm temperature. cThe intensity was calculated by the integral area under the peak exotherm. d The heat flow was taken as the maximum heat flow at the peak temperature.

The differences in mass retention of RAFT and FR copolymers were characterized by

TGA as shown in Figure 9. The RAFT copolymers show a higher stability to degradation up to about 280 °C, likely because there are fewer low molecular weight impurities. Furthermore, the RAFT copolymers yield a higher mass retention at 600 °C. Unlike the RAFT homopolymers,

ACCEPTED MANUSCRIPT

which have extreme heat flows and rapid mass loss, the RAFT copolymers exhibit more gradual mass loss due to the alternative pathway for cyclization. Thus, the greater mass retention of the RAFT copolymers post pyrolysis supports the idea that RAFT copolymers lead to a greater

RI PT

extent of ladder structure formation as compared to FR copolymers. More intriguingly, Figure 9b shows that mass loss between the RAFT and FR copolymers is even more accentuated when the

TE D

M AN U

about 10−20% more mass than the RAFT copolymers.

SC

copolymers are pre-oxidized. Between 350−400 °C (post cyclization), the FR copolymers lose

AC C

EP

Figure 9. TGA plots of 5 mol% RAFT and FR copolymers (AA and NIPAM) (a) ramped at 10 °C/min to 700 °C in N2 and (b) ramped at 10 °C/min in air and held at isotherms of 230, 245, 253, and 265 °C for 15 min each and then ramped at 10 °C/min to 600 °C in N2. 4. Conclusions

We have demonstrated the differences between stabilization behavior of PAN polymers

produced by conventional free radical polymerization with broad dispersities and those by controlled RAFT polymerization with narrow dispersities. The initiation temperature of cyclization for homopolymers strongly correlates to collision frequency wherein FR homopolymers having higher A-values likely due to more defects and radical sources derived

ACCEPTED MANUSCRIPT

from short or long chain branches, small oligomer byproducts, or other residual polymerization impurities. Meanwhile, RAFT homopolymers have lower activation energies, which leads to narrower exotherm temperatures and greater evolution of heat that prematurely degrades the

RI PT

polymers. Although no signs of major structural differences, the FR homopolymers lead to a greater extent of stabilization due to the lower temperature initiation. When extended to PANbased copolymers, the RAFT copolymers exhibit (1) greater exotherm intensities, (2) fewer

SC

small volatiles produced below 280 °C, and (3) higher residual mass contents post thermo-

oxidative stabilization, which suggests that RAFT copolymers resulted in an improved ladder

M AN U

structure formation. The use of PAN-based copolymers with narrow dispersities via RAFT polymerization presents new opportunities for designing next generation carbon fiber precursors. 5. Acknowledgements

TE D

The authors would like to thank the Department of Education GAANN for financial support (Grant No. GR04690). The authors would also like to thank Dr. Richard Liang’s research group for their assistance with the work on WAXS at Florida State’s High Performance

[1]

[2] [3]

AC C

EP

Material Institute.

6. References and Notes

Rahaman M, Ismail A, Mustafa A. A review of heat treatment on polyacrylonitrile fiber. Polym Degrad Stab 2007;92:1421–32. doi:10.1016/j.polymdegradstab.2007.03.023.

Chung DDL. Carbon Fiber Composites. Newton, MA: Butterworth-Heinemann; 1994. Spörl JM, Ota A, Beyer R, Lehr T, Müller A, Hermanutz F, et al. Carbon fibers prepared from tailored reversible-addition-fragmentation transfer copolymerization-derived poly ( acrylonitrile ) - co - poly ( methylmethacrylate ). J Polym Sci Part A Polym Chem 2014;52:1322–33. doi:10.1002/pola.27121.

ACCEPTED MANUSCRIPT

Tsai JS, Lin CH. Effect of molecular weight on the cross section and properties of polyacrylonitrile precursor and resulting carbon fiber. J Appl Polym Sci 1991;42:3045–50. doi:10.1002/app.1991.070421124.

[5]

Devasia R, Reghunadhan Nair CP, Sadhana R, Babu NS, Ninan KN. Fourier transform infrared and wide-angle X-ray diffraction studies of the thermal cyclization reactions of high-molar-mass polyacrylonitrile-co-itaconic acid. J Appl Polym Sci 2006;100:3055–62. doi:10.1002/app.23705.

[6]

Chernikova E, Poteryaeva Z, Belyaev S, Nifant’ev I, Shlyakhtin A, Kostina Y, et al. Controlled synthesis of polyacrylonitrile via reversible addition-fragmentation chaintransfer pseudoliving radical polymerization and its thermal behavior. Polym Sci Ser B 2011;53:391–403. doi:10.1134/S1560090411070013.

[7]

Lin X, Wang C, Yu M, Lin Z. The effect of molecular weight on the structure and properties of PAN. Adv Mater Res 2013;781-784:2609–13.

[8]

Wilkie CA, Xue TJ, Mckinney MA. The thermal degradation of polyacrylonitrile. Polym Degrad Stab 1997;58:193–202.

[9]

Beltz LA, Gustafson RR. Cyclization Kinetics of Poly(acrylonitrile). Carbon N Y 1996;34:561–6.

[10]

Ouyang Q, Cheng L, Wang H, Li K. Mechanism and kinetics of the stabilization reactions of itaconic acid-modified polyacrylonitrile. Polym Degrad Stab 2008;93:1415–21. doi:10.1016/j.polymdegradstab.2008.05.021.

[11]

Frank E, Steudle LM, Ingildeev D, Spörl JM, Buchmeiser MR. Carbon fibers: precursor systems, processing, structure, and properties. Angew Chem Int Ed Engl 2014;53:5262– 98. doi:10.1002/anie.201306129.

[12]

Morgan P. Carbon Fiber and Their Composites. Boca Raton: CRC Press; 2005.

[13]

Bol’bit NM, Dubova EA, Duflot VR, Chevychelov VA. The effect of the polymerization mechanism on the microstructure of chains and the rheology of polyacrylonitrile solutions. Polym Sci Ser A 2011;53:289–95. doi:10.1134/S0965545X11040018.

[15]

SC

M AN U

TE D

EP

AC C

[14]

RI PT

[4]

Moskowitz J, Abel B, McCormick C, Wiggins J. High molecular weight and low dispersity polyacrylonitrile by low temperature RAFT polymerization. J Polym Sci Part A Polym Chem 2015. Thomas DB, Convertine AJ, Myrick LJ, Scales CW, Smith AE, Lowe AB, et al. Kinetics and molecular weight control of the polymerization of acrylamide via RAFT. Macromolecules 2004;37:8941–50.

ACCEPTED MANUSCRIPT

Bajaj P, Sreekumar T V., Sen K, Kumar R, Brar AS. Structural investigations of acrylonitrile-vinyl acid copolymers by NMR spectroscopy. J Appl Polym Sci 2003;88:1211–7. doi:10.1002/app.11779.

[17]

Avilés MA, Ginés JM, Rio JC, Pascual J. Thermal analysis of acrylonitrile polymerization and cyclization in the presence of N,N-dimethylformamide. J Therm Anal Calorim 2002;67:177–88.

[18]

Lange J, Altmann N, Kelly CT, Halley PJ. Understanding vitrification during cure of epoxy resins using dynamic scanning calorimetry and rheological techniques. Polymer (Guildf) 2000;41:5949–55. doi:10.1016/S0032-3861(99)00758-2.

[19]

Kandelbauer A, Wuzella G, Mahendran A, Taudes I, Widsten P. Model-free kinetic analysis of melamine–formaldehyde resin cure. Chem Eng J 2009;152:556–65. doi:10.1016/j.cej.2009.05.027.

[20]

Tsai JS, Hsu H. Determination of the aromatization index for oxidized polyacrylonitrile fibre by the differential scanning calorimetry method. J Mater Sci Lett 1992;11:1403–5.

[21]

Qin XH. Structure and property of electrospinning PAN nanofibers by different preoxidation temperature. J Therm Anal Calorim 2009;99:571–5. doi:10.1007/s10973009-0126-0.

[22]

Arbab S, Mirbaha H, Zeinolebadi A, Nourpanah P. Indicators for evaluation of progress in thermal stabilization reactions of polyacrylonitrile fibers. J Appl Polym Sci 2014;131:40343. doi:10.1002/app.40343.

[23]

Bang YH, Lee S, Cho HH. Effect of methyl acrylate composition on the microstructure changes of high molecular weight polyacrylonitrile for heat treatment. J Appl Polym Sci 1998;68:2205–13.

[24]

Collins GL, Thomas NW, Williams GE. Kinetic relationships between heat generation and nitrile consumption in the reaction of poly(acrylonitrile) in air at 265°C. Carbon N Y 1988;26:671–9.

[26]

[27]

SC

M AN U

TE D

EP

AC C

[25]

RI PT

[16]

Gupta AK, Singhal RP. Effect of copolymerization and heat treatment on the structure and x-ray diffraction of polyacrylonitrile. J Polym Sci Polym Phys Ed 1983;21:2243–62.

Gupta VB, Kumar S. The effect of heat setting on the structure and mechanical properties of poly(ethylene terephthalate) fiber. III. Anelastic properties and their dependence on structure. J Appl Polym Sci 1981;26:1885–95. Hinrichsen G. Structural changes of drawn polyacrylonitrile during annealing. J Polym Sci Part C Polym Symp 1972;38:303–14.

ACCEPTED MANUSCRIPT

Yu MJ, Bai YJ, Wang CG, Xu Y, Guo PZ. A new method for the evaluation of stabilization index of polyacrylonitrile fibers. Mater Lett 2007;61:2292–4. doi:10.1016/j.matlet.2006.08.071.

[29]

Thompson E V. The thermal behavior of acrylonitrile polymers. I. On the decomposition of polyacrylonitrile between 250 and 325°C. J Polym Sci Part B Polym Lett 1966;4:361– 6.

[30]

Dong H, Tang W, Matyjaszewski K. Well-defined high-molecular-weight polyacrylonitrile via activators regenerated by electron transfer ATRP. Macromolecules 2007;40:2974–7. doi:10.1021/ma070424e.

[31]

Bajaj P, Sreekumar T V, Sen K. Thermal behaviour of acrylonitrile copolymers having methacrylic and itaconic acid comonomers. Polymer (Guildf) 2001;42:1707–18. doi:10.1016/S0032-3861(00)00583-8.

[32]

Dalton S, Heatley F, Budd PM. Thermal stabilization of polyacrylonitrile fibres. Polymer (Guildf) 1999;40:5531–43.

[33]

Liu J, He L, Ma S, Liang J, Zhao Y, Fong H. Effects of chemical composition and postspinning stretching process on the morphological, structural, and thermo-chemical properties of electrospun polyacrylonitrile copolymer precursor nanofibers. Polymer (Guildf) 2015;61:20–8. doi:10.1016/j.polymer.2015.01.063.

[34]

Yu M, Wang C, Bai Y, Wang Y, Xu Y. Influence of precursor properties on the thermal stabilization of polyacrylonitrile fibers. Polym Bull 2006;57:757–63. doi:10.1007/s00289006-0629-9.

[35]

Bajaj P, Paliwal DK, Gupta AK. Acrylonitrile - Acrylic Acids Copolymers. I. Synthesis and Characterization. J Appl Polym Sci 1993;49:823–33.

[36]

Bansal RC, Donnet JB. Pyrolytic formation of high-performance carbon fibers. Compr. Polym. Sci., Pergamom Press; 1990, p. 501–20.

[37]

Liu X, Li Y, Lin Y, Li Y. 2-cyanoprop-2-yl dithiobenzoate mediated reversible addition – fragmentation chain transfer polymerization of acrylonitrile targeting a polymer with a higher molecular weight. J Polym Sci Part A Polym Chem 2007;45:1272–81. doi:10.1002/pola.

SC

M AN U

TE D

EP

AC C

[38]

RI PT

[28]

Liu X, Zhang G, Li B, Bai Y, Pan D, Li Y. Well-defined higher-molecular-weight polyacrylonitrile via RAFT technique in the presence of disulfide compounds as a source of chain transfer agent. Eur Polym J 2008;44:1200–8. doi:10.1016/j.eurpolymj.2008.01.031.

ACCEPTED MANUSCRIPT

EP

TE D

M AN U

SC

RI PT

Ju A, Guang S, Xu H. Effect of comonomer structure on the stabilization and spinnability of polyacrylonitrile copolymers. Carbon N Y 2013;54:323–35. doi:10.1016/j.carbon.2012.11.044.

AC C

[39]