Synthesis of polyacrylonitrile copolymers as potential carbon fibre precursors in CO2 (09.02.2015)

Andrei V. Shlyahtin,a,b,c Ilya E. Nifant’ev,*a,b,c,d Vladimir V. Bagrov,a Dmitrii A. Lemenovskii,a,b Aleksander N. Tavtorkinc,d and Peter S. Timasheve

Binary and ternary acrylonitrile (AN) copolymers with methyl acrylate (MA) and either itaconic acid (IA, 1) or IA derivatives (monomethyl itaconate (2), monoethyl itaconate (3), IA monoamide (4), IA mono-n-octylamide (5)) were prepared in CO2. The obtained copolymers have good morphological and molecular weight characteristics. They demonstrate broad heat evolution during thermal cyclisation according to DSC measurements; therefore, they may be used to produce carbon fibres. The polymer yield per reaction volume is 2 to 3 times higher relative to that produced in solution (in DMSO or DMF) or an aqueous suspension.


Polyacrylonitrile-copolymers (PAN-copolymers) are macromolecules that contain predominantly acrylonitrile; the AN content exceeds 85%. Although these copolymers are extensively used as textile fibres, there is extensive interest in applying them in high-quality carbon fibre (CF) production because approximately 90% of the world CF is produced from a PANcopolymer that has been drawn into a filament (PAN precursor). Often, up to 5 mol% MA is introduced to the PAN precursor as an internal plasticiser that improves filament production. Additionally, the commercial PAN precursors contain approximately 0.4–1 mol% itaconic or acrylic acid; these additives facilitate the stabilisation step during the precursor’s transformation into CF.1,2 The need to introduce the acidic comonomers complicates the polymer preparation process.

PAN copolymers are prepared via radical polymerisation because catalytic and anionic polymerisation processes do not produce linear polymers.3 Typical approaches for preparing PAN copolymers include copolymerising AN in organic (dimethyl sulfoxide (DMSO), dimethylformamide (DMF)) or inorganic solvents (aqueous solutions of ZnCl2 or NaSCN), in addition to aqueous dispersion polymerisation.1 Both approaches have considerable economic and ecological disadvantages because they require malodorous and toxic solvents; regenerating the unreacted monomers and solvents is also expensive.

To protect the environment, nowadays the industry is looking for alternative polymerisation media; sub- or supercritical CO2 is one of the most attractive solvents.4 CO2 can be easily regenerated and removed; it is also non-combustible and non-toxic.5

The disclosure of AN polymerisation in CO2 appeared approximately 15 years ago: DeSimone and co-workers reported homopolymerising AN in a CO2 medium, demonstrating that the resulting polymer’s morphology was unsatisfactory. Narrowly dispersed micron-sized PAN particles may be prepared using AN polymerisation with stabilisers. For example, polystyrene-poly(1,1-dihydroperfluorooctyl acrylate (PS-b-PFOA) is a block copolymer composed of CO2-philic and lipophilic blocks.6 Later studies described the effects of different parameters, such as the initiator concentration and monomer concentrations, as well as pressure, on the yield and molecular weight of the homopolymer.7–9 AN was also copolymerised in CO2 with vinyl acetate, methyl acrylate, and 2-chlorostyrene; the resulting copolymers were prepared in moderate yields.10,11 Subsequently, Teng processed the polyacrylonitrile homopolymer obtained in CO2, while Shen et al. studied its surface properties, including the total surface free energy, in addition to the related Lifshitz–van der Waals and Lewis acid– base components.8,12 In contrast to PAN prepared by conventional methods, the surface free energy of the PAN prepared in supercritical CO2 increased as the molecular weight decreased. Researchers have demonstrated that the sequence distributions of PAN obtained via precipitation polymerisation in supercritical CO2 obeyed the Bernoulli statistics, while its stereo-regularity was completely random; the isotacticity of CO2-PAN was lower than that of PAN prepared in water because CO2 is non-polar.13

In 2003 Okubo took an AN–CO2 mixture with a high AN content (up to 25% v/v and more) to perform a precipitation polymerisation in CO2. A few experiments with high AN concentrations were carried out; uniform particles were obtained, and the homopolymer formed in high yield.7

Despite the wealth of published information about AN polymerisation in CO2, PAN precursors (that are AN copolymers containing MA and IA) able to convert efficiently into CF have not yet been prepared in CO2.

The present work studies the synthesis of binary and ternary PAN-copolymers containing IA or its derivatives (2–5) in CO2 media. Initially, the AN homopolymerisation experiments were performed to optimise the reaction conditions by varying the monomer’s charge, the amount of the initiator, the temperature, and the reaction time. We charged the reactor with more than 25 vol.% monomers, similar to Okubo, and demonstrated that these conditions generate excellent polymer yields. After the conditions were selected, binary and ternary copolymers were prepared. We studied the morphology and composition of the polymer particles, as well as their molecular weight characteristics and thermal behaviour at 20–300 °C. The characteristics of the copolymers obtained using the CO2 medium are similar to those of the commercial PAN-copolymers produced using conventional non-green methods.



Commercial CO2 (99.999%, OJSC Linde Gas Rus) and IA (Aldrich, 99%) were used as received.

AN (Acros, 99%) and MA (Acros, 99%) were distilled prior to use.

AIBN (CJSC VEKTON) was recrystallized from methanol before use.

Monomethyl itaconate 2, monoethyl itaconate 3, and IA monoamide 4 were synthesized from itaconic anhydride by reported procedures.14–16

IA n-octylamide 5 was synthesized by the procedure described for itaconic acid alkylamides.16 1H NMR (400 MHz, CDCl3, δ): 6.41 (s, 1H, CH2=), 5.98 (br.s, 1H, NH), 5.87 (s, 1H, CH2=), 3.24 (q, 2H, 3JHH = 6.9 Hz, –NHCH2CH2–), 3.23 (s, 2H, =C–CH2–), 1.49 (quint, 2H, 3JHH = 7.36 Hz, –NHCH2CH2CH2–), 1.27 (br.m, 10H, –(CH2)5CH3), 0.87 (t, 3H, 3JHH = 6.9 Hz, –CH3); the peaks were assigned relative to the CHCl3 residual signal at δ = 7.26 ppm. 13C NMR (400 MHz, CD3OD, δ): 172.91, 169.59, 136.87, 128.84, 40.56, 40.18, 33.00, 30.41, 30.36 (2C), 27.96, 23.72, 14.43; the signal assignment was based on the CD3OD solvent signal at δ = 49.00 ppm.

Polymerization procedure

The polymerisation reactions were performed in a reactor with an 80 ml internal volume constructed from non-magnetic stainless steel. A schematic drawing of the set-up is displayed in Fig. 1. The reactor was purged with CO2 before polymerisation. A degassed mixture of the monomer(s) and the initiator were placed inside with a magnetic stir bar (Tables 1 and 2). The reactor was tightly closed and 80 bar CO2 was pumped into the reactor while stirring the mixture. The temperature of the reactor was maintained at 22 ± 1 °C during CO2 addition. Subsequently, the reactor was heated at 10 °C min−1, and the selected temperature was maintained for the required amount of time (Tables 1 and 2). The system’s pressure at the beginning of the polymerisation was 370 ± 10 bar at 65 °C. After cooling the reactor and removing the CO2, the polymer was collected as a dense, crumbly powder. The polymer was washed twice with water and dried in vacuo.

Fig. 1 Schematic of the set-up for polymerization in the CO2 medium.


NMR analysis. The 1H and 13C NMR analyses were carried out with a Bruker AVANCE 400 spectrometer operating at 400 MHz. The samples were prepared by dissolving 14 to 18 mg of the polymer in 600 μl DMSO-d6 at 50 °C. The spectra were recorded at 60 °C, while the sample spun at 20 Hz. The MA and IA monoester contents were determined by dividing the integrals of proton signals for the ester groups (δ 3.69 ppm for MA, δ 3.59 ppm for 2, δ 4.06 ppm for CH3CH2O– 3) by the integrated methylene signals from the polymer chain (broad δ 2.23–1.82 ppm). To determine the IA content, the polymer was dissolved in DMSO-d6, treated with excess CH2N2 in diethyl ether, and evaporated.17 Afterwards, the methylated carboxyl group and MA contents were determined in a similar manner.

The incorporation of amides 4 and 5 was quantified by integrating the signals of the amide group –NH2 (δ 7.44, 6.91 ppm) and the methyl group CH3– (δ 0.85 ppm), respectively.

GPC analysis. GPC analysis was carried out using a PL-GPC 220 chromatograph (Agilent) equipped with a Styrogel HR 5E column at 50 °C. A 0.1% solution of LiBr in DMF flowing at 1 ml min−1 was used as the mobile phase. The samples were prepared by dissolving 2 mg of the polymer sample in 4 ml of eluent. Polymethylmethacrylate standards were used for calibration.

DSC. DSC was carried out under N2 on a DSC-823e instrument (Mettler Toledo) heating at 5 or 20 K min−1.

SEM. The morphology of the obtained polymers was examined using a LEO 1450 (Carl Zeiss) scanning electron microscope (SEM).


Miscibility of the components

A phase diagram of the AN–CO2 system has already been experimentally determined using a variable volume cell.18 The molar AN content in the AN–CO2 mixture used for most of the experiments described in this study was approximately 30% (determined by weighing the cell before and after filling with CO2). According to the AN–CO2 phase diagram, a system with the given composition was not in the supercritical state at the temperature used for polymerisation (65–80 °C); even for 17 mol% AN, the Tcr of the mixture is 105 °C, and if the AN content in the AN–CO2 mixture increases, then this value should be closer to the Tcr of pure AN (263 °C).

Study of monomer solubility in the initial reaction mixture

The solubility of the itaconic comonomers in the initial reaction mixture was not obvious, especially considering that every IA derivative used in this study was crystalline at room temperature. To visualise the comonomers’ dissolution, we used a viewing cell (5.0 ml internal volume, Thar Instruments, Inc., USA). The initial comonomer mixture containing AN/MA/ 1(2,3,4,5) = 97/2/1 mol% (1.81 ml AN, corresponding to 29 ml AN in the 80 ml cell) was placed in the cell, and the cell was filled with 80 bar CO2. Fig. 2a and 2c display the photographs of the cell after filling it with AN/MA/1 = 97/2/1 mol% and carbon dioxide. The IA was partially dissolved when the cell was filled with CO2 at room temperature (cf. Fig. 2a and 2b);

Fig. 2 (a) Filling of the cell with a magnetic spin bar containing an AN/MA/1 (white crystals) mixture with CO2. (b) Mixture of AN/MA/1/CO2 at RT in the cell with a magnetic spin bar. (c) Mixture of AN/MA/1/CO2 at 65 °C in the cell with a rotating magnetic spin bar.

Fig. 3 (a) Mixture of AN/MA/2/CO2 at RT. (b) Mixture of AN/MA/3/CO2 at RT.

Fig. 4 (a) Mixture of AN/MA/4/CO2 at 65 °C. (b) Mixture of AN/MA/5/ CO2 at 65 °C.

heating the mixture to 65 °C did not completely dissolve IA (Fig. 2с). In contrast, methyl itaconate (2) and ethyl itaconate (3) were completely soluble in the reaction mixture at room temperature (Fig. 3а and b). The itaconic acid amides (4 and 5) presented in Fig. 4а and b were poorly soluble in the initial reaction mixture even at 65 °C. Presumably, the incomplete dissolution of monomers 1, 4, and 5 may decrease their reactivity and hinder their incorporation into the polymer chain.

Homopolymerisation of AN

We began our investigation with an AN homopolymerisation to select the process parameters optimal for providing an acceptable polymer yield without worsening its molecular weight characteristics. Initially, we utilised the reaction conditions reported by Okubo (monomer concentration = ∼25% v/v, AN/ AIBN ratio = ∼400 w/w, temperature = 65 °C), and subsequently varied the parameters (Table 1) to demonstrate how the polymerisation conditions influence the yield and molar weight characteristics of the polymer:

  • temperatures from 65 to 80 °C,
  • monomer loadings from 20 to 40 ml (25–50% v/v),
  • AIBN loadings from 40 to 300 mg (an AN/AIBN ratio of ∼80 to 600),
  • reaction times from 4 to 14 hours.

The results of our experiments are presented in Table 1. We concluded the following:

1. Generally, our results reproduced the published results for AN polymerisation in CO2 media. In particular, increasing the monomer concentration improves monomer conversion and increases the molecular weight of the polymer (cf. entries 1, 7, and 8); increasing the initiator concentration leads to higher conversion but generates lower molecular weight polymers (cf. entries 1, 9, 10, and 11).7–9 Increasing the monomer concentration raises the dispersity Đ (IUPAC recommended term, Đ = Mw/Mn 19) because the mixture becomes more difficult to stir. Higher AN content causes the polymer particles to coagulate earlier, increasing the temperature and concentration inhomogeneities of the process and, therefore, raising the dispersity (cf. entries 1, 7 and 8).

2. Homopolymerising AN is exothermic. When we used a high monomer concentration, controlling the temperature was important. During our preliminary experiments, we could not accurately maintain a constant reaction temperature, especially when the specified temperature exceeded 70 °C. Moreover, the conversion and molecular weight characteristics of PAN were temperature dependent. For example, when the temperature was raised from 65 °C to 80 °C, the conversion increased due to the heightened polymerisation rate. However, raising the temperature caused the polymer’s molecular weight to drop because the rate of chain termination by recombination and disproportionation was also enhanced.20 The dispersity was not substantially affected by the temperature (cf. entries 1, 2, and 3).

3. The polymer’s molecular weight reaches a maximum as the reaction continues (cf. entries 1, 4, 5 and 6) because AN polymerisation is exothermic and self-accelerating.9 Consequently, we selected the following reaction parameters for the copolymerisation experiments:

  • 65 °C (raising the temperature decreases the molecular weight, while slightly increasing the yield),
  • 30 ml (37.5% v/v) monomer (increasing the monomer content from 20 to 30 ml did not significantly increase the dispersity, but the yield per reaction volume increased),
  • 10 h reaction time (increasing the time to 14 h did not considerably increase the yield),
  • 300 mg AIBN (increasing the amount of AIBN increases the yield and decreases the dispersity).

Copolymerisation of AN with MA

We studied copolymerisation of AN with 2% MA in CO2 (Table 2, entry 1). MA readily copolymerised with AN; the copolymer contained only slightly less MA than was put in the initial reaction mixture. Notably, introducing MA did not hinder the polymerisation process; therefore, thorough temperature control was also needed during the copolymerisation of AN and MA, like the AN homopolymerisation.

Ternary copolymerisation of AN with MA and itaconates

We studied the ternary polymerisation of AN with 2 mol% MA and 1 mol% IA (1) or its derivatives (2–5) in CO2 (Table 2). We demonstrated that, similar to the studies with MA, the itaconic monomers were readily incorporated into the polymer chain; the terpolymer yield was high. As noted above, IA was not completely soluble in the initial reaction mixture; therefore, terpolymerisation involving crystalline itaconic acid produces a polymer containing relatively little itaconic monomer (Table 2, entry 2). However, if IA was ground (Table 2, entry 3), the IA was incorporated in amounts comparable to monoesters 2 and 3. The incorporation of amide 4 was imprecisely determined because the amide group signals in the 1H NMR-spectrum were broad. Amide 5 was efficiently incorporated into the polymer chain without pre-treatment, even though it was not completely soluble in the initial reaction mixture.

After comparing the AN homopolymerisation to the copolymerisation with itaconates, we concluded that the incorporation of itaconates significantly slows the process. Consequently, heat evolves more smoothly, and the reaction temperature may be controlled more easily. Finally, the dispersity of the ternary copolymers obtained was lower than that of the homopolymer synthesised under similar conditions (Table 2, entries 3, 4, 5 and 6).

We have managed to prepare the terpolymers in satisfactory yields (75–80% and higher). The polymeric yield per reaction volume was at least 2–3 times higher than that obtained during the solution or suspension polymerisation of AN. Using CO2 allows the product to be separated easily from the reaction medium, demonstrating an advantage of using CO2 instead of conventional media.5 It is also important that the molecular weight characteristics of the polymers thus obtained are close to the corresponding parameters of the copolymers used in the industrial synthesis of carbon fibres.1 Moreover, our method allows to tune the molar mass of the PAN-copolymer by varying the temperature, monomer concentrations, and AN/ AIBN ratio to fit the special requirements.

DSC study of binary and ternary copolymers

The efficiency of incorporating comonomers into the PAN copolymers was analysed based on their effects on the copolymers’ thermal behaviour. Fig. 5 and 6 present the DSC profiles of the PAN homopolymer (Table 1, entry 11), binary copolymer (Table 2, entry 1), and ternary copolymers (Table 2, entries 3 to 7) measured at 5 and 20 K min−1, respectively. The homopolymer exhibits a narrow heat evolution peak attributed to the radically initiated cyclisation of nitrile groups.1,21–24 Introducing MA shifts the onset of the heat evolution toward lower temperatures; however, the peak remains relatively sharp, and its maximum shifts toward higher temperatures (Fig. 5 and 6). Ternary copolymers containing 1, 2, 3, 4, and 5 have similar smooth DSC curves. The onset of the heat evolution occurs at lower temperatures in comparison with the AN–MA binary copolymer, and the peaks of the DSC curves for ternary copolymers are much broader than the corresponding peaks of the AN–MA binary copolymer and the AN-homopolymer (Fig. 5 and 6). The broadening of the DSC curve is attributed to the change in the mechanism of the cyclisation from the radical mechanism to the ionic one.1,21–24

Fig. 5 DSC curves of homoPAN and PAN-copolymers measured under
N2, heating rate 5 K min−1.
Fig. 6 DSC curves of homoPAN and PAN-copolymers measured under
N2, heating rate 20 K min−1.

The DSC data obtained at 5 and 20 K min−1 were used to estimate the activation energy Ea of the polymeric cyclisations via the Kissinger (eqn (1)) and Ozawa methods (eqn (2)).

where Tm,i is the maximum temperature at heating rate βi (K min−1).24–26

The data used for the calculations and the obtained results are summarised in Table 3.

Both methods generated similar results for the binary and ternary copolymers. Introducing both MA and MA/comonomer 1–5 substantially decreased the cyclisation’s activation energy relative to that of the homopolymer; comonomer 2 caused the best effects.

Fig. 7 (a) SEM image of PAN homopolymer (Table 1, entry 12). (b) SEM image of P(AN–MA) binary copolymer (Table 2, entry 1). (c) SEM image of P(AN–MA-1) ternary copolymer (Table 2, entry 3). (d) SEM image of P(AN–MA-2) ternary copolymer (Table 2, entry 4). (e) SEM image of P(AN–MA-3) ternary copolymer (Table 2, entry 5).

SEM study of the copolymers

We used scanning electron microscopy (SEM) to study the polymer particles’ morphology. Fig. 7a–e display SEM images of the homopolymer (Fig. 7a), the binary copolymer (Fig. 7b), and the three ternary copolymers containing 1, 2, and 3 (Fig. 7c and 7e). The figures reveal that the obtained polymers contain regular and narrowly dispersed particles approximately 1 μm in diameter, agreeing with Okubo’s results for a homopolymer obtained using more than 25% v/v monomer.7 Each particle consists of smaller coagulated particles, except the ternary copolymer with IA; this exception may be caused by the insufficient resolution of the SEM image.


Copolymerisations of acrylonitrile with methyl acrylate and itaconic acid, monoesters of itaconic acid, and monoamides of itaconic acid were studied for the first time in CO2 media. The polymers were obtained in satisfactory yields (approximately 80%). The comonomeric contents of the copolymers were practically identical to the initial reaction mixtures. The itaconic comonomers were soluble in the initial reaction mixture, and therefore the reaction progress was uniform during the early stages. The large monomer volume fraction allowed more efficient use of the reaction volume (2–3 times) than observed during solution or dispersion polymerisation. The copolymerisations were performed without surfactant stabilisers and resulted in regularly shaped particles, making the purification and subsequent handling of the polymer very simple.

Therefore, the synthesis of PAN-copolymers in CO2 is attractive from both the chemical and ecological points of view. This new technology makes it possible to produce PAN-copolymers of a given composition and to control the molar mass characteristics of the copolymers and, in this way, it is possible to meet all the criteria for the high-quality PAN-precursor. It is important that the synthesis of PAN-copolymers in CO2 media represents a truly green technology compared with synthesis in organic solvents, mainly due to its energy effectiveness and the total absence of sulphur-containing admixtures in the product.


We thank the Russian Federation for partial financial support of this work (SC #14.513.11.0033).

Notes and references

1 P. Morgan, Carbon fibers and their composites, Taylor & Francis Group, 2005.
2 X. Huang, Materials, 2009, 2, 2369.
3 F. Schaper, S. R. Foley and R. F. Jordan, J. Am. Chem. Soc., 2004, 126, 2114.
4 M. F. Kemmere and T. Meyer, Supercritical Carbon Dioxide in Polymer Reaction Engineering, WILEY-VCH Verlag GmbH & Co., Weinheim, Germany, 2005.
5 A. I. Cooper, J. Mater. Chem., 2000, 10, 207.
6 H. Shiho and J. M. DeSimone, Macromolecules, 2000, 33, 1565.
7 M. Okubo, S. Fujii, H. Maenaka and H. Minami, Colloid Polym. Sci., 2003, 281, 964.
8 X. R. Teng, J. Appl. Polym. Sci., 2003, 87, 1393.
9 X. R. Teng, H. L. Shao and X. C. Hu, J. Appl. Polym. Sci., 2002, 86, 2338.
10 Y. Yang, Q. Dong, Z. Wang, C. Shen, Z. Huang, H. Zhu, T. Liu and C. P. Hu, J. Appl. Polym. Sci., 2006, 102, 5640.
11 S. D. Yeo and E. Kiran, Macromolecules, 2004, 37, 8239.
12 Q. Shen, Q. F. Gu, J. F. Hu, X. R. Teng and Y. F. Zhu, J. Colloid Interface Sci., 2003, 267, 333.
13 X. R. Teng, X. C. Hu and H. L. Shao, Polym. J., 2002, 34, 534.
14 B. R. Baker, R. E. Schaub and J. H. Williams, J. Org. Chem., 1952, 17, 116.
15 J. Barnett and C. J. O. R. Morris, Biochem. J., 1946, 40, 450.
16 A. Zilkha and U. Golik, J. Org. Chem., 1963, 28, 2007.
17 M. J. C. Cudero, M. M. C. López-González and J. M. Barrales-Rienda, Polym. Int., 1997, 44, 61.
18 H. S. Byun, Y. S. Kim and J. K. Im, J. Korean Ind. Eng. Chem., 1998, 9, 924.
19 R. G. Gilbert, M. Hess, A. D. Jenkins, R. G. Jones, P. Kratochvíl and R. F. T. Stepto, Pure Appl. Chem., 2009, 81, 351.
20 B. C. Ji, H. S. Lee, N. S. Yoon, S. S. Han, W. S. Yoon, J. Lee and W. S. Lyoo, Int. J. Polym. Mater. Polym. Biomater., 2000, 46, 423.
21 Q. Ouyang, L. Cheng, H. Wang and K. Li, Polym. Degrad. Stab., 2008, 93, 1415.
22 T. Sun, Y. Hou and H. Wang, J. Appl. Polym. Sci., 2010, 118, 462.
23 A. V. Shlyakhtin, D. A. Lemenovskii and I. E. Nifant’ev, Mendeleev Commun., 2013, 23, 277.
24 A. Ju, S. Guang and H. Xu, Carbon, 2013, 54, 323.
25 H. E. Kissinger, Anal. Chem., 1957, 29, 1702.
26 T. Ozawa, Bull. Chem. Soc. Jpn., 1965, 38, 1881.

Все права защищены 2013 © Эльбрус

Адрес: г. Москва, ул. Барклая д.6, стр.5
+7 (495) 269-12-92 /

Карта проезда