The importance of being porous: polysaccharide-derived mesoporous materials for use in dye adsorption

Abstract

The controlled pyrolysis of mesoporous polysaccharide-derived materials, from starch and alginic acid, formed carbonaceous materials (Starbons^®) and were demonstrated as efficient materials for the removal of dyes from wastewater. The resulting materials were characterised by solid-state NMR, N₂ adsorption porosimetry, FT-IR, scanning electron microscopy (SEM) and tunnelling electron microscopy (TEM). The material’s efficiency for dye adsorption was tested using methylene blue (MB) and acid blue 92 (AB) dyes. Adsorption data indicated that the mesoporosity of the material had a far greater influence on the adsorption capacity and speed of adsorption, than high surface area alone. Mesoporous Starbon^® (A300) was evaluated against commercially available activated carbon (Norit) and demonstrated a superior adsorption capacity of MB; 186 mg g⁻¹vs. 42 mg g⁻¹. The kinetic activity of Starbon^® was also determined with A800 showing the fastest rate of adsorption compared to S800 and Norit, suggesting that it is a more suitable material for water purification.

---

1 Introduction

Clean potable drinking water is one of the most precious and finite resources on the planet. Around the world it is becoming increasingly more common that industrial effluents must undergo costly treatment prior to discharge into water courses (lake, rivers and seas).¹

Chemicals such as dyes and pigments are used in various sectors of industry including textile manufacture, leather tanning, paper production, food technology and cosmetics manufacture, to name but a few.² These synthetic organic compounds consume vast quantities of water in their use and have a visible effect on water quality. The annual production of dyes and pigments is in excess of 7 × 10⁵ tonnes, of which an estimated 2–15% are lost in the effluent during the dyeing process.³ In some cases water colouration can be observed in dye concentrations of less than 1 ppm.⁴ Many of these dyes are toxic, even carcinogenic, persistent in the environment and non-biodegradable.⁵ The impact of releasing such waste streams into water courses can be dramatic and poses a significant hazard to both aquatic life and animals further up the food chain including humans.⁶ Many dyes are resistant to aerobic digestion, stable to light, heat, water and oxidation, making treatment a significant challenge.⁷ Biological, chemical, and physical methods are available for the removal of such dyes. Physical adsorption using activated carbons as an adsorption matrix is efficient and considerable attention has been focused on their use.⁸ These materials are effective for treating wastewater containing dyes that show difficulty in biodegradation e.g. azo dyes. There are a wide variety of biomass-derived activated carbons that have shown effective dye adsorption, however, these materials can be limited due to cost, availability, both in abundance and worldwide location and are frequently limited in practical use because of micropore sizes resulting in large volumes required to improve rates of extraction.⁹

Polysaccharides, including starch and alginic acid, are relatively inexpensive, non-toxic, biodegradable, possess polyfunctionality, have great potential for chemical modification and are found in nearly every geographical location on the planet.¹⁰ The extensive use of non-ionic polysaccharides (such as starch) as adsorbents, has been restricted by low surface area (<1 m² g⁻¹), low degree of mesoporosity and limited potential to hydrogen bond with dyes due to extensive intramolecular chain interactions. Anionic polysaccharides (such as alginic acid) have also previously shown limited success for use in adsorption, the main reason being electrostatic repulsion between the polymer surface and the dye.¹¹ The development of polysaccharide-derived mesoporous materials with large pore volumes and surface areas may open new doors to adsorbents. Previously we reported that through controlled pyrolysis it is possible to form tuneable, nano-structured, graphitizable and mesoporous carbons (Starbon^®) using no templating agents.¹² These Starbon^® materials have already demonstrated great promise as catalytic supports, catalysts, nanoparticle delivery systems and chromatographic stationary phases.¹³ The surface chemistry, functionality and porosity of these materials can be controlled through varying the temperature of preparation and selection of the polysaccharide precursor. Incorporation of polysaccharides which are abundant in regions of developing countries including starch, alginic acid, okra, chitin and chitosan would be of great benefit for potential water treatment application.

Herein, it is demonstrated that waste polysaccharides (starch and alginic acid), can be expanded and pyrolysed to produce effective mesoporous adsorbents, Starbon^®. By varying the adsorbent type and adsorbent preparation temperature a range of mesoporous materials are produced and these are utilised to clarify the importance of pore texture, surface functionality and dye molecular size on adsorption capacity.

2 Experimental

Chemicals

All materials were used as received without further purification. Methylene blue (high purity, biological stain) was purchased from Alfa Aesar. Acid blue 92 was purchased from Sigma-Aldrich (Fig. 1). Activated charcoal (Norit) was purchased from Fluka and was washed with deionised water at 80 °C, filtered and dried in a vacuum oven before use.

Fig. 1 Structures and dimensions of (A) methylene blue and (B) acid blue 92.

Characterisation

SEM micrographs were recorded using a JEOL JSM-6490LV. Samples were mounted on alumina plates and coated with a 7 nm layer of Au/Pd using a high resolution sputter SC-7640 coating device prior to analysis. TEM micrograph images were recorded using a Tecnai 12 BioTwin at 120 kV. Samples were suspended in ethanol and deposited onto carbon grids via solvent evaporation. UV-vis analysis was performed using a Jasco V-550 UV-vis spectrometer and IR analysis was performed using a Bruker Vertex 70 FT-IR fitted with a Specac Golden Gate ATR. The ¹³C-MAS NMR spectra were obtained using a Varian VNMRS spectrometer at 100.56 MHz for ¹³C. The spin rate of the MAS was set to 12 kHz spinning, with a wide spectral width and a rotor-synchronised echo. Spectral referencing was with respect to neat tetramethylsilane.

Starch-derived Starbon^® preparation¹²

1.6 kg corn starch was added to 8 L of deionised water and heated at 120 °C and 80 kPa for 45 min. The resulting gel was retrograded at 5 °C for 48 h. The gel was subject to 5 solvent exchanges with ethanol. The filtered material was then oven dried to yield expanded starch. This was doped with p-toluene sulfonic acid (5% w/w) and refluxed for 6 h. The resulting material was heated at 1 °C min⁻¹ in an inert atmosphere to the required temperature. Starbon^® samples used in adsorption studies were prepared at up to 300 °C (S300) and 800 °C (S800). (These materials are also available from Sigma-Aldrich at these preparation temperatures.)

Alginic acid-derived Starbon^® preparation^13b

40 g of alginic acid was added to 800 mL of deionised water and heated at 90 °C for 2 h. The gel was retrograded for 12 h at 5 °C. The gel was subjected to 5 solvent exchanges with ethanol and dried using supercritical CO₂ at 40 °C, 100 bar, 40 g min⁻¹ for 3 h. The resulting material was heated at 1 °C min⁻¹ in an inert atmosphere to the required temperature. Starbon^® samples used in adsorption studies were prepared at up to 300 °C (A300) and 800 °C (A800).

Dye adsorption studies

Standard solutions of the two dyes used were prepared in volumetric flasks using deionised water at the following concentrations: methylene blue (10 mg L⁻¹), acid blue 92 (40 mg L⁻¹). For the isotherm and kinetic adsorption studies a flask was filled with dye solution volumes ranging from 30 mL to 2.5 L and stirred for 5 min prior to adding the adsorbent and sealing the vessel. Samples were taken at 5, 15, 30, 60, 240 min and 24 h. For the eight adsorbents used, concentrations ranged from 3 mg L⁻¹ to 300 mg L⁻¹.

3 Results and discussion

Physical and chemical properties of the adsorbents

Four mesoporous adsorbents with different pore sizes, surface areas and pore volumes were synthesized using a method of expansion of the raw material, retrogradation and carbonisation. Among the products, S300 and S800 are the starch-based Starbon^® materials carbonised to 300 °C and 800 °C respectively; and A300 and A800 are the alginic acid-derived Starbon^® materials carbonised to 300 °C and 800 °C respectively. Norit, a commercially available microporous material, was used as a basis for comparison.

A quantitative analysis of the textural properties, determined using N₂ adsorption, of the adsorbents are compiled in Table 1. BET surface areas of Starbon^® increase from 332 to 535 m² g⁻¹ with increasing preparation temperature, S300 to S800, which correlates with an increase in micropore formation. The BET surface area of A300 is low at 280 m² g⁻¹, but exhibits a significantly smaller microporous volume and a greater average pore diameter as compared to S300 and S800 (Table 1)

Table 1 Porosimetry characteristics of Starbons^® and Norit

	S300	S800	A300	A800	Norit
a Calculated through the Dubinin-Radushkevich method.
S BET (m^2 g^−1)	332	535	280	265	798
S > 2 nm (m^2 g^−1)	153	133	214	247	102
Total pore volume (cm^3 g^−1)	0.82	0.75	1.41	1.08	0.57
Average pore diameter (nm) desorption	17.4	11.8	19.3	14.5	4.3
Microporous volume (cm^3 g^−1)^a	0.17	0.22	0.11	0.04	0.42
% Microporosity	20	29	8	4	79
E DR (kJ mol^−1)^a	16.8	24.9	13.9	23.4	19.2

---

A800 exhibits the lowest surface area (265 m² g⁻¹), but the total pore volume and surface area are almost completely contributed by mesopores. The reduction in surface area and total pore volume between A300 and A800 can be attributed to the coalescence of micropores to form larger pores at higher temperatures. A reduction in average pore diameter was also observed between A300 and A800. This is contrary to expectation, but is likely to be due to the general shrinking of large pores at high temperatures.¹⁴ Norit exhibited the highest BET surface area of the materials, at 798 m² g⁻¹, characteristic of an almost completely microporous material such as this (Fig. 2). Thus, it is a suitable material for comparison with Starbon^® to determine the importance of mesoporosity in adsorbance of dye molecules.

Fig. 2 Porous distribution of S800, A800 and Norit determined by N₂ adsorption.

SEM images (Fig. 3) of all materials show the formation of an increasingly porous structure from the parent material to the finally carbonised product. The textural properties of starch-derived Starbons^® differ significantly to those derived from alginic acid where the structure is not granular in nature like the starch Starbon^®, but can be described as a fluffy, porous, web-like structure. TEM images also exemplify this feature (see SF1 in electronic supplementary information†).

Fig. 3 SEM images (A) starch, (B) expanded starch, (C) S300, (D) alginic acid, (E) expanded alginic acid, (F) A300.

The ¹³C-MAS NMR spectra of a range of Starbon^® materials demonstrate near identical changes in surface functionality on heating the materials (Fig. 4, original data available SF2 in electronic supplementary information†). This can be attributed to the inherent hydroxyl-rich nature of all polysaccharide materials. At 300 °C resonances between 50–100 ppm disappear, signifying a loss of the acetal and alcohol functionalities of the polysaccharides. The appearance of aliphatic carbons resulting from the increased thermal degradation of starch and alginic acid are represented by a broad resonance between 10–50 ppm. As the preparation temperature increases, aromatic rings such as substituted phenols, benzenes and furan form cross linked structures. At 450 °C the relative intensity of aromatic carbon resonances is significantly increased and the high field aliphatic resonance is decreased. This is in good agreement with FT-IR data for both starch and alginic acid Starbon^® samples (SF3 in electronic supplementary information†) and concisely shows the transformation from hydroxyl-rich polysaccharides to aliphatic/alkene groups and further into highly aromatic materials.

Fig. 4 Overlaid ¹³C MAS NMR spectra of starch/S300/S450 (red) and alginic acid/A300/A450 (grey).

This NMR data would indicate a move towards aromatic structures at higher temperatures, however, due to a combination of the nature of the bulk material and the probe tuning range it was not possible to record a spectra. XPS analysis was used to enable surface characterisation of high temperature materials (all scans available SF4 in electronic supplementary information†). Table 2 shows different elements and concentrations that were observed for the materials.

Table 2 Atomic content of elements in adsorbents by XPS

	% Atomic content
	C	O	S	Na	Ca
S300	85.7	14.3	0.1	—	—
S800	98.3	1.8	—	—	—
A300	81.9	16.1	0.4	1.0	0.7
A800	88.7	8.2	0.4	2.2	0.5

---

As expected from NMR and IR analysis, the oxygen content is lower for the high temperature materials, S800 and A800; this data also corresponded well with CHN analysis of the materials (ST1 in electronic supplementary information†). S300 shows the presence of SO₂, BE 168.0 eV, which is likely due to some residual p-toluene sulphonic acid catalyst used in the carbonisation process. The presence of elemental Na, BE 1072 ± 1.0 eV, CaSO₄, BE 347.5 ± 0.2 eV and CaO, BE 351.0 ± 0.5 eV in A300 and A800 is most likely a result of the alginic acid extraction process.¹⁵ Deconvolution of the C1s and O1s spectra of the adsorbents gives the individual component peaks (Fig. 5). Depending on binding energy (BE) each peak corresponds to different carbon and oxygen species (Table 3).¹⁶ S800 and A800 spectra clearly show the presence of aromaticity in good agreement with NMR observation. S300 and S800 exhibit a relatively large amount of surface C–OH groups compared to A300 and A800 which may affect the adsorbent properties.¹⁷

Fig. 5 XPS spectra of C1s (A) S300, (B) S800, (C) A300, (D) A800 and O1s (E) S300, (F) S800, (G) A300, (H) A800.

Table 3 Composition of the components in adsorbents from XPS spectra

Peak	BE (eV)	Chemical State	%
			S300	S800	A300	A800
I	284.6 ± 0.2	C–C	53	32	66	37
II	285.1 ± 0.2	C–C (Ar)	—	28	—	45
III	285.8 ± 0.2	C–OH	30	18	11	—
IV	289.0 ± 1.0	O=C–O	17	22	23	18
I	531.3 ± 0.2	O=C	—	—	60	—
II	532.0 ± 0.2	O–C	34	—	40	100
III	533.4 ± 0.2	O–C=O	66	100	—	—

---

Dye adsorption related to material properties

To understand the adsorption on the material surface and to establish the relationship between pore textual properties and the adsorption ability, the adsorption of methylene blue (MB) and acid blue 92 (AB) were studied on the Starbon^® materials; for comparison Norit was also used as an adsorbent.

Adsorption isotherms were generated for adsorption on the different materials and fitted to the Langmuir model. Table 4 compiles the adsorption capacities of MB and AB on the various materials, this data enabled an estimation of the dye surface area coverage, using the size of each dye molecule as indicated by hyperchem (Fig. 1).

Table 4 Adsorption capacities for dye molecules and surface area of dye coverage

	Methylene blue		Acid blue 92
	Adsorbent capacity (mg g^−1)	Surface area (m^2 g^−1)	Adsorbent capacity (mg g^−1)	Surface area (m^2 g^−1)
S300	36	72	27	41
S800	52	104	39	59
A300	186	373	82	124
A800	97	195	108	164
Norit	42	83	49	74

---

The order of adsorption capacity for MB, a cationic dye and the smaller of the two dyes studied, was: A300 ≫ A800 > S800 > Norit > S300. A300 was a highly effective adsorber of MB, adsorption capacity was 186 mg g⁻¹ which is more than four times higher than for Norit at 42 mg g⁻¹ (Table 4). The calculated surface area of dye coverage was higher than the BET surface area for N₂ adsorption; this could be due to high carbonyl oxygen functionality of the adsorbent surface (Table 3) resulting in a high affinity between the dye molecule and the adsorbent, as represented below:¹⁷

C=O^δ− + MB^δ+ → C=O ⋯ MB (1)

It can also be attributed to large mesopores, average diameter of 19 nm, allowing for effective diffusion of dye into the pore structure (Table 1). The adsorbed MB is able to occupy all the porous volume, giving the appearance of a larger surface area of dye coverage.¹⁸

The adsorption capacities of the materials for AB are in the order of A800 > A300 > Norit > S800 > S300. A800 is the most efficient adsorbent for this dye, exhibiting twice the adsorption capacity of Norit (108 mg g⁻¹ compared to 49 mg g⁻¹), this is due to the high mesoporosity of the material and high aromaticity leading to strong π–π interactions with the dye molecule. A300 adsorption of AB is significantly lower than MB adsorption, caused by electrostatic repulsion between the anionic dye and anionic polymer surface.^11,17

The starch-based materials S300 and S800 exhibit significantly lower adsorption capacities than A300 and A800 and similar capacities compared to Norit for both dyes. As S300 and S800 exhibit pore volumes and diameters closer to the alginic acid-derived materials than Norit it can be assumed that this is not the factor limiting dye uptake, rather it could be due to the high concentration of hydroxyl groups on the surface of the starch-based adsorbents. In this case molecules are likely to be attracted to the surface due to (a) dipole–dipole interactions between the hydrogen on the adsorbent surface and the electropositive groups on the dye and (b) Yoshida hydrogen bonding between the OH groups and the aromatic rings and this correlates well with observations for similar systems (Fig. 6).^11,19 However, this highly functionalised surface can also give rise to hydrogen bonding with the water present in the system, thus, competing for adsorption sites and vastly reducing uptake of the dye.²⁰ The higher degree of graphitisation of the alginic acid derived materials and Norit prevents this competition occurring.

Fig. 6 Methylene blue–polysaccharide interactions: (a) dipole–dipole hydrogen bonding interactions, (b) Yoshida H-bonding.

The microporous material Norit, demonstrated lower adsorption than A300 and A800, with approximately only 10% coverage of the total surface area. It has been previously reported that for adsorbents with small pore diameters pore blockage may occur due to the aggregation of bulky molecules, such as dyes, in the pore orifice.¹⁷ Therefore the full surface area of the adsorbent cannot be utilised, reducing the effectiveness of adsorption.

Overall these results demonstrate that the effective adsorption of dyes of different sizes is dependent on both the pore structure and the surface functionality of the adsorbent. Hence, highly mesoporous materials with low hydroxyl and high aromatic functionality, A300 and A800, far outperform those with less mesoporosity and high surface OH.

Information relating to the kinetic adsorption activity of Starbon^® was vital for applications including, water treatment. Fig. 7 plots the adsorption of MB onto three adsorbents versus time. A800 showed the fastest rate of adsorption, with the majority of the dye being adsorbed in the first 15 min. This result gives further evidence of the importance of the large pores present in the adsorbent as these allow for free transportation of the dye within the pore structure, leading to fast uptake of the dye. Norit is the slowest adsorber of MB likely due to the small diameters of the pores leading to inhibited diffusion of the dye. A similar correlation was observed for AB adsorption (SF. 5 in supplementary material†).

Fig. 7 Adsorption of methylene blue on various adsorbents versus contact time.

4 Conclusion

Highly efficient adsorption of dyes from waste water has been achieved using polysaccharide-derived mesoporous adsorbents and a clear relationship has been established between the importance of mesoporosity, surface functionality and efficient dye adsorption. Our results show that a large surface area is not vital for high adsorption capacities, as is commonly thought, but instead large pore volumes and pore diameters are principally important for fast and effective dye uptake. In particular, A300 and A800 adsorbents which exhibit extremely high mesoporosity of above 80%, large average pore diameters, between 14–19 nm, and pore volumes above 1 cm³ g⁻¹, but with less than 300 m² g⁻¹ surface area, far outperform commercially available activated carbon adsorbents with significantly higher surface areas but only 20% mesoporosity, in both adsorption capacity and speed of adsorbate uptake. For example, A300 shows an adsorption capacity of 186 mg g⁻¹ for MB, more than four times higher than Norit.

Overall, it has been shown that by using our method of preparation polysaccharides can be effectively utilised as adsorbers. These promising results for dye wastewater treatment suggest that the materials have the potential to be effective for the treatment of aqueous waste containing other organic molecules. Due to the worldwide availability of polysaccharides this could be a cheap and accessible solution for a water treatment alternative to activated carbon.

Acknowledgements

The authors would like to acknowledge Glaxo Smith Kline and the EPSRC for their financial support. Dr Duncan Macquarrie and Dr Simon Breeden (University of York) are thanked for their input into discussions. Meg Stark is gratefully acknowledged for her assistance in electron microscopy. Dr D. J. Morgan (EPSRC XPS service, University of Cardiff, UK) and Dr D. C. Apperley (EPSRC Solid State NMR service, University of Durham, UK) are thanked for their assistance.

References

1. G. Crini, Bioresour. Technol., 2006, 97, 1061–1085.
2. A. Bhatnagar and A. K. Jain, J. Colloid Interface Sci., 2005, 281, 49–55.
3. (a) J. W. Lee, S. P. Choi, R. Thiruvenkatachari, W. G. Shim and H. Moon, Dyes Pigm., 2006, 69, 196–203; (b) I. A. W. Tan, B. H. Hameed and A. L. Ahmad, Chem. Eng. J., 2007, 127, 111–119; (c) T. Robinson, G. McMullan, R. Marchant and P. Nigam, Bioresour. Technol., 2001, 77, 247–255; (d) I. J. Muhammad, A. N. Muhammad and J. Hazard, Mater., 2007, 139, 57–66.
4. (a) I. M. Banat, P. Nigam, D. Singh and R. Marchant, Bioresour. Technol., 1996, 58, 217–227; (b) Y. Al-Degs, M. A. M. Khraisheh, S. J. Allen and M. N. Ahmad, Water Res., 2000, 34, 927.
5. M. Dogan, Y. Ozdemir and M. Alkan, Dyes Pigm., 2007, 75, 701–713.
6. (a) K. R. Ramakrishna and T. Viraraghavan, Water Sci. Technol., 1997, 36, 189–196; (b) C. O'Neill, F. R. Hawkes, D. L. Hawkes, N. D. Lourenco, H. M. Pinheiro and W. Delere, J. Chem. Technol. Biotechnol., 1999, 74, 1009–1018; (c) G. E. Walsh and L. H. Bahner, Environ. Pollut., Ser. A., 1980, 21, 169; (d) K. G. Bhattacharyya, A. Sharma and J. Environ, Manag., 2004, 71, 217.
7. (a) V. Golob and A. Ojstrsek, Dyes Pigm., 2005, 64, 57–61; (b) Q. Sun and L. Yang, Water Res., 2003, 37, 1535–1544.
8. (a) K. K. H. Choy, G. McKay and J. F. Porter, Resour., Conserv. Recycl., 1999, 27, 57–71; (b) B. H. Hameed, A. L. Ahmad and K. N. A. Latiff, Dyes Pigm., 2007, 75, 143–149; (c) J. Avom, K. J. Mbadcam, C. Noubactep and P. Germain, Carbon, 1997, 35, 365–369.
9. B. K. Singh and N. S. Rawat, J. Chem. Technol. Biotechnol., 1994, 61, 307–317.
10. (a) V. Budarin, J. H. Clark, F. E. I. Deswarte, J. J. E. Hardy, A. J. Hunt and F. M. Kerton, Chem. Commun., 2005, 2903–2905; (b) R. J. White, V. L. Budarin and J. H. Clark, ChemSusChem, 2008, 1, 408–411.
11. R. S. Blackburn, Environ. Sci. Technol., 2004, 38, 4905–4909.
12. V. Budarin, J. H. Clark, J. J. E. Hardy, R. Luque, K. Milkowski, S. J. Tavener and A. J. Wilson, Angew. Chem., Int. Ed., 2006, 45, 3782–3786.
13. (a) V. L. Budarin, J. H. Clark, R. Luque and D. J. Macquarrie, Chem. Commun., 2007, 634–636; (b) J. H.Clark, V. L. Budarin, T. Dugmore, R. Luque, D. J. Macquarrie and V. Strelko, Catal. Commun., 2008, 8, 1709–1714; (c) V. Budarin, J. H. Clark, R. Luque, D. J. Macquarrie, K. Milkowski, R. J. White, P. C. T. International patent, WO 2007104798 A2 20070920, 2007; (d) R. J. White, C. Antonio, V. L. Budarin, E. Bergström, J. Thomas-Oates and J. H. Clark, Adv. Funct. Mater., 2010, 20, 1834–1841.
14. L. Tellez, J. Rubio, M. A. Valenzuela, F. Rubio and J. L. Oteo, Mater. Charact., 2009, 60, 506–512.
15. A. Haug, U. S. Patent, 3,396,158, 1968.
16. M. Jung, K. Ahn, Y. Lee, K. Kim, J. Rhee, J. T. Park and K. Paeng, Microchem. J., 2001, 70, 123–131.
17. M. Valix, W. H. Cheng and G. McKay, Langmuir, 2006, 22, 4574–4582.
18. X. Zhuang, Y. Wan, C. Feng, Y. Shen and D. Zhao, Chem. Mater., 2009, 21, 706–716.
19. (a) Z. Yoshida, E. Osawa and R. Oda, J. Phys. Chem., 1964, 68, 2895–2898; (b) L. Abramian and H. El-Rassy, Chem. Eng. J., 2009, 150, 403–410.
20. (a) C. Moreno-Castilla, Carbon, 2004, 42, 83–94; (b) C. Moreno-Castilla, J. Rivera-Utrilla, M. V. Lopez-Ramon and F. Carrasco-Marin, Carbon, 1994, 33, 845–851.

---

Footnote

† Electronic Supplementary Information (ESI) available. See DOI: 10.1039/c2ra21367b/

---