Nitrogen-doped carbon fibers and membranes by carbonization of electrospun poly(ionic liquid)s

Abstract

Poly(ionic liquid)s with well-defined chemical structure were prepared and applied as precursors for nitrogen-doped carbon fibers and membranesvia the electrospinning technique. The nitrogen contents of two prepared carbon membranes were found to be 6.3% and 8.0% respectively. Their electrical conductivity was measured to be ca. 200 S cm⁻¹.

---

Nitrogen-doped carbons have triggered considerable research interest in the past several years due to their unique properties and potentials, such as high conductivity and catalytic activity.^1–4 For example, it has been reported that nitrogen-doped carbons prepared at 1000 °C possessed graphitic structures and were highly conductive;^2,3 in addition, the active sites of N–C–N bonding within the nitrogen-doped carbons have been proven as efficient metal-free catalysts for oxygen reduction reaction.⁴ Generally doping carbons with nitrogen was performed using two strategies: (1) in situ doping via carbonization of nitrogen rich precursors and (2) post-treatment of carbons with nitrogen rich reagents. For the latter, the carbon product is subjected to a nitrogen containing atmosphere at high temperatures or room temperature chemical reactions. A main drawback of the latter strategy is the inhomogeneous distribution of nitrogen (mainly on the surface) and therefore ineffective improvement of the bulk properties. In contrast, the precursor approach allows for a homogeneous incorporation of nitrogen into the carbon matrix in a controlled fashion. Several reports have adopted this strategy to prepare nitrogen-doped carbons of different morphologies. For example, nitrogen-doped carbon nanotubes (CNTs) and nitrogen-doped carbon nanofibers were synthesized by methods similar to the catalytic growth of bulk CNTs using precursors such as melamine,^5,6benzylamine,⁷acetonitrile,^8,9N-heterocycles,^10,11 or phthalocyanines.¹² Other structures of nitrogen-doped carbons have been prepared using mainly hard templates, such as porous anodic alumina or mesoporous silica.^13–21

Ionic liquids (ILs) are materials consisting of ions and are liquid below 100 °C. They possess negligible vapor pressure and some are highly thermostable up to 400 °C. It has been reported that the carbonization of imidazolium- and pyridinium-based ILs with dicyanamide as nitrogen rich anion formed carbon materials of high nitrogen content. The advantages of being in a liquid state and high thermostability facilitate easy incorporation of ILs into porous templates without the assistance of high pressure techniques. On the other hand, in the absence of templates it is difficult to obtain nitrogen-doped carbon materials of designed shape from ILs.^22,23

Poly(ionic liquid)s (PILs) are the polymeric form of ionic liquids, which are commonly prepared by polymerization of a monomeric ionic liquid.^24–28 Although structurally PILs resemble a linear connection of ILs, the polymer nature enables them to be readily processed. Well defined morphologies that are inaccessible by ILs can be realized from PILs, e.g. thin films, fibers, etc.Electrospinning is a well-known powerful tool to prepare fibers from viscous solutions of polymers or their composites.^29,30Electrospinning of PILs has been recently reported by Chen and Elabd³¹ and is promoted by the ionic character of these polymers. Here, we demonstrate the preparation of nitrogen-doped carbon fibers and membranes using electrospun imidazolium- and pyridinium-based PIL fibers as precursors. This demonstrates that the superior processability of PILs and the inherent incorporation of the IL moieties in the macromolecular structure can be combined to generate advanced materials of desired morphology and complex architecture. The general synthetic route is illustrated in Scheme 1.

Scheme 1 Synthetic route to nitrogen-doped carbon fibers and membranes by carbonizing electrospun poly(ionic liquid)s.

Two types of PILs, poly(3-allyl-1-vinylimidazolium dicyanamide) (PAVIm-DCA) and poly(1-allyl-4-vinylpyridinium dicyanamide) (PAVP-DCA), were employed in this work. They are specially designed to solve various problems to encounter within the synthetic route. First, the consideration of dicyanamide as anion, apart from its high nitrogen content, stems from the fact that it undergoes condensation at 280–300 °C to firmly fix the fiber morphology. Second, an allyl group instead of normal alkyl chains was attached to the cation rings. Our initial trials with saturated PILs, such as poly(3-ethyl-1-vinylimidazolium dicyanamide), failed to generate carbon fibers. We found that electrospun PIL fibers melted into spheres upon heating. One obvious reason for this inconvenience is a low glass transition temperature (T_g) of PILs. For example, poly(3-ethyl-1-vinylimidazolium dicyanamide) has a T_g of only −15 °C (ESI†). This enhances the motion of polymer chains at elevated temperature and destroys the anisotropic shape. The introduction of allyl functions allows the PILs to cross-link and to maintain the fiber morphology at high temperature before the IL-typical condensation reactions take place.

As mentioned earlier, PILs are commonly prepared by polymerization of corresponding IL monomers. In our attempt, polymer modification of poly(1-vinylimidazole) (PVIm) and poly(4-vinylpyridine) (P4VP) as an alternative approach was conducted. It allows for the synthesis of PILs with high molecular weight, which are difficult to access by direct polymerization of IL monomers. PVIm with an apparent M_n ≈ 85 000 g mol⁻¹ (ESI†) was obtained by free radical polymerization of 1-vinylimidazole in DMF. P4VP of M_w ≈ 160 000 g mol⁻¹ is commercially available. The chemical modification of polymers, here quaternization of PVIm and P4VP is a key step, as a non-quantitative reaction results in a random copolymer rather than a PIL homopolymer. The reactions were performed in DMF at 100 °C with excessive 1-allyl bromide. Fig. 1A shows the ¹H-NMR spectra of PVIm before and after modification. The proton signals of the imidazole ring in PVIm (6.4–7.1 ppm) completely vanished after the reaction and shifted to 7.5–10.0 ppm. Indeed, a quantitative quaternization was achieved for PVIm. In the ¹H-NMR spectra of P4VP in Fig. 1B, the proton signals overlap with the quaternized product; however, the integration ratio of allyl protons (5.0–6.5 ppm) to pyridinium ring ones (7.6–9.3 ppm) is found to be 5.0/4.0, exactly matching the calculated value. A second proof comes from their ¹³C-NMR spectra. The signals of pyridine rings in P4VP entirely shifted after the quaternization, indicating a complete quaternization of P4VP (ESI†). After modification, an anion exchange process viametathesis was conducted by reacting PILs (Br as anion) with excessive silver dicyanamide. Since this reaction involved no change in hydrogen atoms, ¹³C-NMR measurements instead of ¹H-NMR were used for characterization. As shown in Fig. 1C and D, a new peak at 119 ppm appeared in both PILs, attributed to the dicyanamide anion (ESI†). In addition, FTIR spectra of both PILs after anion exchange display characteristic peaks of the dicyanamide function between 2100 and 2300 cm⁻¹ (ESI†). Quantitative evidence can be provided by elemental analysis. A dramatic increase in nitrogen content was observed in both PILs after reaction (ESI†). Although the found nitrogen contents were a little lower than calculated because of a tiny portion of water absorbed by the hydrophilic PILs, the C/N ratios fitted the calculated ones well. Indeed, all characterizations support that polymer modifications end up with desired PILs.

Fig. 1 (A and B) ¹H-NMR spectra of PVIm/poly(3-allyl-1-vinylimidazolium bromide) and P4VP/poly(1-allyl-4-vinylpyridinium bromide). (C and D) ¹³C-NMR spectra of poly(3-allyl-1-vinylimidazolium bromide)/PAVIm-DCA and poly(1-allkyl-4-vinylpyridinium bromide)/PAVP-DCA.

The electrospinning of PILs proceeded in a nitrogen chamber with controlled humidity. To achieve a high degree of crosslinking, a co-crosslinking agent trimethylolpropane tris(3-mercaptopropionate) (TRIS) and a radical initiator 4,4′-azobis(4-cyanovaleric acid) (AVCA) were dissolved in the PIL solution for electrospinning. Early reports have demonstrated that the combination of TRIS and ACVA was effective to achieve a high crosslinking density and hold the polymeric anisotropic superstructures.^32,33 The electrospun PIL fibers were crosslinked at 80 °C under high vacuum for 2 days. The carbonization started with a pre-stabilization step at 280 °C in air, followed by heating to 1000 °C at a rate of 100 °C h⁻¹ under nitrogen. Fig. 2A shows an overview of a monolayer of carbon fibers prepared on a silicon wafer. The dark linear objects with smooth contour, i.e., the obtained carbon fibers are 0.2–3 μm in diameter. A close view in Fig. 2B shows some strongly curved fibers, presumably caused by a strong interaction of the charged PIL jets with the electric field. Thus the fiber morphology was indeed well-kept after crosslinking and carbonization. Via the same process, freestanding carbon membranes consisting of overlapped multilayers of carbon fibers were prepared by using an aluminium collector with an empty hole in the center. Electrospun PIL mats accumulated in the central hole were dried, crosslinked and carbonized at 1000 °C. Fig. 2C shows a SEM image of a carbon membrane. It appears very homogeneous over a large size scale. An enlarged view at the edge (Fig. 2D) displays a cross-section of the membrane structure. Its thickness is 3–6 μm, and in fact is tunable in terms of electrospinning time to collect the fibers. A top view in Fig. 2E indicates that most carbon fibers maintain the original morphology of the PIL fibers. Random mesh holes are observed all over the membrane, which allows for a mass transport from one side of the membrane to the other. Fig. 2F shows a close view of the fibers. While the big fibers remain straightforward and rigid, fibers of very small diameters collapse on the large ones but without forfeiting their fiber morphology. Evidently the skeleton of the web structure is built from the large fibers and the grid size is further tuned by the small ones. The inhomogeneity in the fiber size is an important factor to keep the 3-dimensional membrane structure. Elemental analysis and powder X-ray diffraction (XRD) of the carbon membranes further indicate that the carbons are graphitic and doped with 6.33% of nitrogen.

Fig. 2 SEM images of nitrogen-doped carbon fibers and membranes prepared from electrospun PAVIm-DCA fibers. (A and B) A monolayer of carbon fibers on a silicon wafer; (C–F) free-standing nitrogen-doped membranes: (C) overview; (D) side view; (E) top view; and (F) enlarged view.

The synthetic strategy to nitrogen-doped carbon materials is in fact not an individual case but applicable to other PILs, provided proper chemistry and electrospinning process is accompanied. Here, as an example, pyridinium-type PIL PAVP-DCA was processed in the same way to prepare nitrogen-doped carbon membranes. Fig. 3A shows a large-scale SEM image of a dense carbon membrane prepared from electrospun PAVP-DCA fibers. The majority of the fibers are here 0.4–2 μm in diameter. A less compact membrane structure is shown in Fig. 3B. As can be clearly seen, the membrane displays structural characters different to those of PAVIm-DCA. The membranes derived from PAVIm-DCA (in Fig. 2) are made up from the overlapping of individual carbon fibers. Here, a single intact fiber does not exist and all fibers fuse into neighboring ones at the junctions. This morphology formed actually not due to the carbonization but during the electrospinning when the PIL jets landed on the collector. Fig. 3C shows an image of the uncrosslinked PIL fibers. Obviously the fibers fused into each other already at this stage. It is worth noting that though a single fiber shape was not maintained, the formation of the carbon membranes was not disturbed at all. Fig. 3D displays a photograph of such membranes taken from a digital camera. The membrane is centimetre-sized and macroscopically homogeneous. It can be freely cut, similar to a thin film. The nitrogen content is determined to be 8.0%. To characterize the local atomic structure, the membranes were subjected to XRD characterization. It showed the existence of graphitic phase (Fig. 3E). Characteristic graphitic bands at 25.2°, 44.1°, and 80.2° were clearly observed.

Fig. 3 (A and B) SEM images of nitrogen-doped carbon membranes prepared from electrospun PAVP-DCA fibers. (C) An image of PAVP-DCA membranes before crosslinking taken from optical microscopy. (D) A photograph of the as-synthesized centimetre-sized membrane. (E) The XRD pattern of the carbon membrane.

The conductivity of nitrogen-doped carbon fibers and membranes is of great interest due to their potential applications in electrochemistry. Here 2-point potentiostatic DC-resistivity measurements as well as AC-impedance measurements were performed (ESI†). The as-synthesized carbon membranes were found to be easily broken when applying the electrode on them. To provide a realistic value of the conductivity, thin carbon films of 150 nm in thickness on quartz were prepared in the same procedure except using spin-coating instead of electrospinning to process the PIL solution. The conductivity is determined to be 200 ± 60 S cm⁻¹, which is fairly satisfactory compared to carbons prepared from many other organic polymers and materials at the same temperature.

In conclusion, nitrogen-doped carbon fibers and membranes were prepared via the carbonization of electrospun poly(ionic liquid)s, which were prepared from chemical modification of existing polymers to achieve a high molecular weight. They contain 6.33% and 8.00% of nitrogen, respectively, and show a conductivity of ca. 200 S cm⁻¹. This synthetic approach not only expands the applicative spectrum of poly(ionic liquid)s as a polymeric form of ionic liquids, but also points out a facile route to obtain nitrogen-doped carbon materials of morphologies not limited to fibers but also others that are provided by polymer processing techniques.

Acknowledgements

The authors deeply appreciate Prof. Bernd Smarsly and his research group for the helpful discussions on the conductivity measurements. J.Y. and A.G.M. acknowledge postdoc scholarships received from the Max Planck Society. Dr Z. Schnepp is thankful for the SEM imaging. B. Dai is acknowledged for the synthesis of PAVP-DCA in a lab course.

Notes and references

1. K. K. R. Datta, V. V. Balasubramanian, K. Ariga, T. Mori and A. Vinu, Chem.–Eur. J., 2011, 17, 390–3397.
2. J. P. Paraknowitsch, A. Thomas and M. Antonietti, J. Mater. Chem., 2010, 20, 6746–6758.
3. J. P. Paraknowitsch, J. Zhang, D. Su, A. Thomas and M. Antonietti, Adv. Mater., 2010, 22, 87–92.
4. W. Yang, T.-P. Fellinger and M. Antonietti, J. Am. Chem. Soc., 2011, 133, 206–209.
5. M. Terrones, P. Redlich, N. Grobert, S. Trasobares, W.-K. Hsu, H. Terrones, Y.-Q. Zhu, J. P. Hare, C. L. Reeves, A. K. Cheetham, M. Rühle, H. W. Kroto and D. R. M. Walton, Adv. Mater., 1999, 11, 655–658.
6. M. Terrones, H. Terrones, N. Grobert, W. K. Hsu, Y. Q. Zhu, J. P. Hare, H. W. Kroto, D. R. M. Walton, P. Kohler-Redlich, M. Ruhle, J. P. Zhang and A. K. Cheetham, Appl. Phys. Lett., 1999, 75, 3932.
7. M. Terrones, R. Kamalakaran and T. M. R. Seeger, Chem. Commun., 2000, 2235–2236.
8. M. Glerup, M. Castignolles, M. Holzinger, G. Hug, A. Loiseau and P. Bernier, Chem. Commun., 2003, 2542–2543.
9. A. G. Kudashov, A. V. Okotrub, L. G. Bulusheva, I. P. Asanov, Y. V. Shubin, N. F. Yudanov, L. I. Yudanova, V. S. Danilovich and O. G. Abrosimov, J. Phys. Chem. B, 2004, 108, 9048–9053.
10. R. Sen, B. C. Satishkumar, A. Govindaraj, K. R. Harikumar, M. K. Renganathan and C. N. R. Rao, J. Mater. Chem., 1997, 7, 2335–2337.
11. J. Liu, R. Czerw and D. L. Carroll, J. Mater. Res., 2005, 20, 538–543.
12. L. Zhi, T. Gorelik, R. Friedlein, J. Wu, U. Kolb, W. R. Salaneck and K. Müllen, Small, 2005, 1, 798–801.
13. A. Vinu, K. Ariga, T. Mori, T. Nakanishi, S. Hishita, D. Golberg and Y. Bando, Adv. Mater., 2005, 17, 1648–1652.
14. R. V. Parthasarathy, K. L. N. Phani and C. R. Martin, Adv. Mater., 1995, 7, 896–897.
15. J. Jang and J. H. Oh, Chem. Commun., 2004, 882–883.
16. T. Maiyalagan and B. Viswanathan, Mater. Chem. Phys., 2005, 93, 291–295.
17. L. Zhi, T. Gorelik, J. Wu, U. Kolb and K. Müllen, J. Am. Chem. Soc., 2005, 127, 12792–12793.
18. Y. Xia and R. Mokaya, Chem. Mater., 2005, 17, 1553–1560.
19. Y. Xia and R. Mokaya, Adv. Mater., 2004, 16, 1553–1558.
20. F. Goettmann and M. Antonietti, Chem. Mater., 2008, 20, 738–755.
21. R. Liu, D. Wu, X. Feng and K. Müllen, Angew. Chem., Int. Ed., 2010, 122, 2619–2623.
22. J. S. Lee, X. Wang, H. Luo, G. A. Baker and S. Dai, J. Am. Chem. Soc., 2009, 131, 4596–4597.
23. P. F. Fulvio, J. S. Lee, R. T. Mayes, X. Wang, S. M. Mahurin and S. Dai, Phys. Chem. Chem. Phys., 2011 10.1039/c1cp20631a.
24. J. Lu, F. Yan and J. Texter, Prog. Polym. Sci., 2009, 34, 431–448.
25. O. Green, S. Grubjesic, S. Lee and M. A. Firestone, Polym. Rev., 2009, 49, 339–360.
26. J. Yuan and M. Antonietti, Polymer, 2011, 52, 1469–1482.
27. F. Yan and J. Texter, Chem. Commun., 2006, 2696–2698.
28. J. Yuan and M. Antonietti, Macromolecules, 2011, 44, 744–750.
29. M. Graeser, M. Bognitzki, W. Massa, C. Pietzonka, A. Greiner and J. H. Wendorff, Adv. Mater., 2007, 19, 4244–4247.
30. A. Greiner and J. H. Wendorff, Angew. Chem., Int. Ed., 2007, 46, 5670–5703.
31. H. Chen and Y. A. Elabd, Macromolecules, 2009, 42, 3368–3373.
32. A. Walther, A. Gröschel and A. H. E. Müller, Polymer, 2008, 49, 3217–3227.
33. A. Walther, J. Yuan, V. Abetz and A. H. E. Müller, Nano Lett., 2009, 9, 2026–2030.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details and some characterization of poly(ionic liquid)s. See DOI: 10.1039/c1py00196e

---