Journal of Colloid and Interface Science 413 (2014) 43–53

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science www.elsevier.com/locate/jcis

Surface properties of CNTs and their interaction with silica Anastasia Sobolkina a, Viktor Mechtcherine a,⇑, Cornelia Bellmann b, Vyacheslav Khavrus c, Steffen Oswald c, Silke Hampel c, Albrecht Leonhardt c a

Institute of Construction Materials, Faculty of Civil Engineering, Technische Universität Dresden, D-01062 Dresden, Germany Leibniz Institute of Polymer Research Dresden, Hohe Strasse 6, D-01069 Dresden, Germany c Leibniz Institute for Solid State and Materials Research, Dresden Helmholtzstrasse 20, D-01069 Dresden, Germany b

a r t i c l e

i n f o

Article history: Received 11 May 2013 Accepted 16 September 2013 Available online 25 September 2013 Keywords: Carbon nanotubes Cement-based composites Wetting behaviour Electro-kinetic potential Silica adsorption Adsorption mechanism

a b s t r a c t In order to improve the embedding of carbon nanotubes (CNTs) in cement-based matrices, silica was deposited on the sidewall of CNTs by a sol–gel method. Knowledge of the conditions of CNTs’ surfaces is a key issue in understanding the corresponding interaction mechanisms. In this study various types of CNTs synthesized using acetonitrile, cyclohexane, and methane were investigated with regard to their physicochemical surface properties. Significant differences in surface polarity as well as in the wetting properties of the CNTs, depending on the precursors used, were revealed by combining electro-kinetic potential and contact angle measurements. The hydrophobicity of CNTs decreases by utilising the carbon sources in the following order: cyclohexane, methane, and finally acetonitrile. The XPS analysis, applied to estimate the chemical composition at the CNT surface, showed nitrogen atoms incorporated into the tube structure by using acetonitrile as a carbon source. It was found that the simultaneous presence of nitrogen- and/or oxygen-containing sites with different acid–base properties increased the surface polarity of the CNTs, imparting amphoteric characteristics to them and improving their wetting behaviour. Regarding the silica deposition, strong differences in adsorption capacity of the CNTs were observed. The mechanism of silica adsorption through interfacial bond formation was discussed. Ó 2013 Elsevier Inc. All rights reserved.

1. Introduction Theoretical and experimental investigations have demonstrated the excellent mechanical properties of carbon nanotubes (hereinafter CNTs) with Young´s modulus as high as 1 TPa and a tensile strength of approximately 100 GPa [1–4]. Moreover, CNTs exhibit a low density, a large aspect ratio, and a high chemical resistance. These properties potentially make CNTs an ideal reinforcement material for various types of matrices. To use the reinforcing effect of CNTs on the mechanical properties of composites, an adequate CNT–matrix interface is essential. In line with the theory of fibrereinforced composites, the mechanical interlocking and/or chemical bonding between matrix and CNTs assist load transfer across the interface. The chemical interaction between surfaces can be achieved by both covalent and non-covalent (acid–base, hydrogen, ionic, van der Waals) bonds. The first report on improving the interfacial bond between cement matrix and CNTs functionalised covalently with carboxylic groups was published by Li et al. [5]. In the ionic functionalities, such as hydroxyl (AOH) and carboxyl (ACOOH) groups grafted to the CNT surface, protons can be replaced by high-valent cations ⇑ Corresponding author. Fax: +49 351 463 37 268. E-mail address: [email protected] (V. Mechtcherine). 0021-9797/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcis.2013.09.033

namely Ca2+ ions present in abundance in the aqueous phase of a cement paste. Since Ca2+ ions exhibit a high binding capacity, they can form more than one bond with an ionic character. Hence, the further development of the hardened cement paste structure can lead to anchoring of CNTs into cement hydration products. However, to achieve a high surface occupancy with oxygen-containing functional groups coupled covalently to the graphene tube walls, the CNTs should be treated primarily with oxidising acids, such as HNO3, H2SO4 or a mixture of the two. The corrosiveness of acid treatment causes structural defects in the outer shells of CNTs to appear [6]. This could lead to a significant deterioration of the mechanical properties of the single- as well as multi-walled CNTs as the tensile load is transferred only through the outermost graphitic shell, a so-called ‘‘sword-in-sheath’’ failure [3,7]. The experimental studies of the influence of CNTs functionalized with carboxyl groups on the properties of cement-based composites are quite inconsistent. For instance, Li et al. [5] reported an increase in the compressive and flexural strength of the mortar by 19% and 25%, respectively, by the addition of chemically functionalised CNTs in a concentration of 0.5% by wt. (hereinafter by cement weight). In contrast, Musso et al. [8] recorded a significant reduction in the flexural and compressive strength of mortar after adding 0.5% by wt. carboxyl-functionalised multi-walled CNTs. Cwirzen et al. [9] investigated the effects of the method of surface

44

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

modification on the wettability of multi-walled CNTs and the mechanical properties of hardened cement paste modified with CNT dispersions. The greatest increase of nearly 50% in compressive strength of hardened cement paste was observed using CNTs in the amount of only 0.045% by wt. In addition to the covalent surface modification, these CNTs were treated with a water solution of polyacrylic acid during dispersion. However, the authors did not detect any improvement in flexural strength. The controversial results represented in the studies can be hardly interpreted correctly due to the lack of knowledge about the physicochemical surface properties of the CNTs utilised. As a non-covalent functionalization, the covering of CNTs with silica obtained in a sol–gel process looks promising [10]. It was expected that due to the reaction of the silica coating with calcium hydroxide present in the cement matrix, secondary hydration products of cement would be formed directly around the tubes (see Fig. 1). This would increase the embedding of the CNTs in the cement matrix and promote load transfer. It can be assumed that the embedding degree of CNTs into the cement matrix will be determined by the uniformity of the silica coating. To estimate the adhesive properties of CNTs toward colloidal silica, the tube surface needs to be characterised. The investigation of the wetting behaviour of CNTs as well as their surface chemistry is intended to provide an understanding of the chemical and physical phenomena taking place in the CNT–silica interfacial area. The aim of the present study was to investigate the physicochemical surface properties of various CNTs with respect to the precursors and methods used for their synthesis. Different approaches, including measurements of the wetting and the electro-kinetic potential of the CNTs and characterisation of their surface chemistry by means of X-ray photoelectron spectroscopy, were applied to reveal the surface properties of the CNTs used. A further topic included the non-covalent modification of the both pristine and surfactant-pretreated CNTs by coating with amorphous silica using a sol–gel process. The interrelation between adsorption of colloidal silica on the CNTs and their surface properties was demonstrated.

2. Materials and methods 2.1. Materials Three types of CNTs were chosen for this study. They differed with respect to the precursors, synthesis methods used, and tube dimensions; see Table 1 and Fig. 2. For the first type, a mixture of single-, double- and multi-walled CNTs (hereinafter ‘‘mixed

CNTs’’) was synthesized by chemical vapour deposition (CVD) from the methane utilised as a carbon source in an Ar/H2 atmosphere using a Fe/Mo–MgO catalyst with a Fe/Mo atomic ratio of 5.5:1. Before the CH4 was introduced, the catalyst was heated in a reactor tube up to 600 °C in an H2 atmosphere in order to reduce catalyst particles. After a certain period of CNT growth at 900 °C, the CH4 was replaced by pure Ar, and the reactor was cooled to room temperature. The fixed-bed CVD led to the bundle-like agglomeration of the nanotubes consisting of the mixture of single-, double-, and multi-walled CNTs, cf. Fig. 2a. To reduce the residual catalyst, the CNTs were subsequently treated by sonication in diluted hydrochloric acid. To synthesize two other types of CNTs, an aerosol-assisted CVD method was employed using either acetonitrile or cyclohexane as the carbon source and ferrocene as the catalyst. The pyrolysis temperature was set at 800 °C in both cases. The CNTs were deposited in an Ar/H2 environment. Using this method, carpets of oriented, multi-walled CNTs were formed primarily, cf. Fig. 2b and c. As the catalyst was mostly covered by carbon, CNTs produced by aerosol CVD did not undergo any purification procedure. The application of acetonitrile as a precursor led to the formation of multiwalled CNTs with a bamboo-like structure due to the incorporation of nitrogen into the CNT shells [11].

2.2. Characterisation of CNT surface To evaluate the polarity of the surface, the pristine CNTs were subjected to measurements of their electro-kinetic potential (zeta potential (f)) by means of laser Doppler electophoresis (LDE), performed using a Zetasizer 3000 (Malvern Instrument, UK). CNTs were dispersed in 103 mol/L KCl solution at a concentration of approximately 0.1 mg/mL. The measurement was carried out after sonication for 2 min by means of a cup-horn ultrasonic homogenizer with a cylindrical tip. The sonifier was operated at an amplitude of 140 lm. The zeta potential was calculated according to the Henry equation using the Smoluchowski approximation. In order to estimate an isoelectric point (IEP = pH|f=0), i.e., the pH value at which the net charge on the particles is zero, a plot of zeta potentials versus solution pH was determined. The pH of the CNT suspensions was adjusted using manual titration by adding either 0.1 M HCl or 0.1 M KOH. The position of the IEP was determined through fitting of 3rd order polynomial trend lines to the plotted data points. In addition to the surface potential, the wettability of the CNTs was investigated by optical contact angle measurement with DataPhysic equipment, model OCA35L. For this purpose, the pristine

Fig. 1. Schematic representation of the interaction between silica coating and cement matrix.

45

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53 Table 1 Selected properties of carbon nanotubes.

a b c

Labelling

Mixed CNTs

N-CNTs

MWCNTs

Method of synthesis Precursor Type of CNTs Diameter (nm) Length (lm) Specific surface areac (m2/g) Chemical purification

Fixed-bed CVD Methane (CH4) Single-, double-, multi-walled 1. . .15a ca. 10 354 +

Aerosol-assisted CVD Acetonitrile (CH3CN) Multi-walled 70 ± 20 50. . .200b 77 

Aerosol-assisted CVD Cyclohexane (C6H12) Multi-walled 90 ± 20 100. . .300b 29 

Single-wall CNTs form rope-like bundles with a diameter of up to approximately 20 nm, resulting from the agglomeration. The length of carpets of aligned CNTs measured by SEM. Specific surface area was determined using the Brunauer–Emmett–Teller method.

Fig. 2. SEM and high-magnification TEM images of CNTs produced using (a) methane (mixed CNTs), (b) acetonitrile (N-CNTs) and (c) cyclohexane (MWCNTs).

CNTs were dispersed in water and filtered through a membrane with a pore size of 0.4 lm using a vacuum pump. During filtration, the CNTs connected with each other and formed a sheet, also called buckypaper, through their random orientation (Fig. 3a). After drying the buckypaper surface was subjected to the wetting measurements using the sessile drop method (Fig. 3b). For this purpose, distilled water (pH  6) was used as test fluid. It should be noted that due to a certain roughness of the surface and a penetration of the water drop in the bulk, the evaluation of wetting behaviour in the CNTs was performed only qualitatively. The surface chemistry of the pristine and silica-coated CNTs was characterised using X-ray photoelectron spectroscopy (XPS). XPS determines the binding energies of both core and valence electrons by measurement of the kinetic energies of electrons photo-emitted from a solid when irradiated in vacuo with X-rays. Chemical state characterisation can be carried out by analysing peak shift characteristics for changing chemical surroundings. Since the core electron binding energies are characteristic of the elements present, XPS provides elemental analysis of the outermost 5–10 nm for all elements except hydrogen and helium [12]. The XPS experiments were carried out in an ultrahigh vacuum system equipped with a hemispherical electron analyser SPECS

PHOIBOS 100. Photoelectrons in all XPS measurements were excited with Mg Ka (1253.6 eV) radiation and analysed with constant pass energy of 15 eV. The X-ray source was running at a power of 300 W.

2.3. Coating CNTs with amorphous silica The sol–gel method is widely used in synthesising silica nanoparticles at room temperature [13]. The first report about coating of CNTs with silica via a sol–gel process was made by Seeger et al. [14]. The authors pretreated CNTs with polyethylenimine in order to create positive charges on the nanotubes and subsequently deposit negatively charged colloidal silica on the CNT surface. In this work, the adsorption of colloidal silica on the surface of the pristine CNTs (see Table 1) and CNTs modified using a surfactant was investigated. The present study specified the coating process as follows: CNTs were first dispersed in distilled water in a concentration of 0.5% by wt. either without or in the presence of surfactants. A non-ionic polyoxyetylene(23)laurylether (Brij 35) and a cationic cetyltrimethylammonium bromide (CTAB) were used as surfactants to reduce the surface tension of the CNTs and to promote their

Fig. 3. (a) Photo of a carbon nanotube buckypaper prepared for the contact angle measurement, (b) cross-section of a water drop wetted on the buckypaper surface; the initial drop volume was set at 5–8 ll.

46

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

dispersion, respectively. The CNT-to-surfactant weight ratio was invariable and set at 1:1. As a silica precursor, tetraethyl orthosilicate (TEOS, >98%) from Merck was utilised. Ethanol (98%) was used as a co-solvent. The hydrolysis and condensation of TEOS occurred without addition of a catalyst under neutral pH conditions. The aqueous dispersion of CNTs, ethanol, and TEOS was mixed using a weight ratio of 20:6:1, respectively, in a vessel mounted onto the reflux cooler and subsequently heated to 70 °C with continuous stirring for 30 min in order to accelerate the sol–gel reaction. Thereafter, the reaction proceeded at room temperature for 10 h. To remove the residual colloidal solution, the CNTs were filtered by washing with a water–ethanol mixture and subsequently dried at 110 °C. The silica adsorption layer on the CNT surface was examined by means of high-resolution transmission electron microscopy (TEM, here: FEI Teacnai T20), operated at an acceleration voltage of 200 kV, and coupled with an EDX detector for elemental analysis. The silica-coated CNTs were placed on a copper grid for TEM investigation. 2.4. Theoretical approach of dissociation constant and its calculation The heteroatoms such as nitrogen and oxygen present on a nanotube surface are the nuclei of ionisable active centres that can be characterised by the general equilibrium process stated in the following equation: pK a

RAAH þ H2 O $ RAA þ H3 Oþ

ð1Þ

where pKa is the negative logarithm of the acid ionisation constant. The low pKa value of an ionisable site manifests itself in a high degree of dissociation in aqueous solution, with shift of the equilibrium of Eq. (1) to the right side. This indicates the high acidity of the ionisable surface group. The pH of the solution plays a key role in the position of equilibrium of the reaction (1). An increase or decrease in the solution pH relative to pKa is, according to the Henderson–Hasselbach equation (cf. Eq. (2)), accompanied by a shift of the equilibrium of the reaction (1) towards its right or left side, respectively. The pKa is equivalent to the pH of the solution at which it is exactly half-dissociated.

pH ¼ pK a þ logð½RAA =½RAAHÞ

ð2Þ

In order to estimate the dissociation constants of nitrogen functionalities on the N-CNT surface, a GALAS model developed by ACD/Labs [15] was employed. The algorithm of pKa prediction defined ionisable nitrogen-containing centres in the input CNT structure through removing protons. The pKa value of an ionogenic site was calculated, taking into account its actual surrounding including charge influences of any neighbouring ionised centres. 3. Results and discussion 3.1. Surface characterisation of the CNTs To observe wetting behaviour the CNTs synthesized from acetonitrile (N-CNTs), methane (mixed CNTs) and cyclohexane (MWCNTs), water contact angle measurements were carried out. Because of the surface roughness affecting water drop spreading, the advancing contact angle was shown for comparison only to illustrate the phenomena. The magnitude of the contact angles was quantified very roughly. The images of water droplets on the surface of the buckypapers produced from the N-CNTs, mixed CNTs, and MWCNTs are represented in Fig. 4. A complete wetting was observed for the N-CNT buckypaper (see Fig. 4a). The sessile water drop was drastically reduced in less than 0.3 s to achieve a contact angle of approximately 10° with the surface. The buckypa-

per made of the mixed CNTs had a moderately wettable surface. The contact angle of the water drop was diminished from 110° down to 35° within 10 s (cf. Fig. 4b). No visible penetration and/ or spreading of the water droplets were detected on the MWCNT buckypaper: The surface was non-wettable, i.e., hydrophobic. Fig. 5 shows the zeta potential of the CNTs measured as a function of pH. The IEP was estimated to be around 7.6, 6.6, and 4.0 for the mixed CNTs, N-CNTs, and MWCNTs, respectively. As is well known, the IEP is governed by the relative acidic strength (pKa) of the functional surface groups [16]. The presence of acidic or basic surface groups, such as carboxylic or amine groups, leads to a shift of the IEP towards lower or higher pH values, respectively, due to their dissociation or protonation. From this point of view, it should be expected that the IEP of hydrophobic surfaces without chemical functional groups is located at neutral pH. However, as seen from the electro-kinetic measurements (see Fig. 5), the IEP of the hydrophobic MWCNTs is shifted towards acidic pH while a shift in the IEP for the hydrophilic N-CNTs is not observed. In order to illustrate the surface chemical composition of the CNTs, XPS analysis was employed. The survey spectra and the atomic concentration of elements detected on the nanotube surface are represented in Fig. 6 and Table 2, respectively. The XPS analysis yielded the expected C1s bands, the distinct N1s peak attributable to 2.4 at.% of a nitrogen atom in the surface layer of the N-CNTs, and the O1s signals for all CNT types. The surface atomic concentrations of oxygen were 1.6 at.% for the N-CNTs, 1.0 at.% for the mixed CNTs and 1.4 at.% for the MWCNTs. Since the CNT syntheses occurred in a reducing atmosphere, the oxygen signal evidently originated from functional groups created on the external nanotube surface due to exposure to air in subsequent storage. The high surface area as well as the abundance of dangling bonds at the edges and defects of the outermost graphene layer of nanotubes could favour the chemisorption of molecules from the air onto the CNT surface [17–20,34]. The mild hydrochloric acid treatment, employed in purifiying the mixed CNTs, is considered to be non-oxidative [21] and could not, therefore, be the reason for the incorporation of oxygen into the mixed CNTs. With respect to metal contaminations utilised in the catalytic growth of the nanotubes, the XPS analysis revealed only a negligible amount of iron and a trace of molybdenum for the N-CNTs and the mixed CNTs, respectively. Still, it should be taken into account that the XPS reveals the chemical composition of a sample over a depth of only a few nanometres, from which the electrons characterised can be emitted with no energy loss. In this way the catalyst material encapsulated in a carbon layer and/or present in the bulk of CNT agglomerates might not be detected (cf. Fig. 11h). To define the bonding configuration of the elements found, high-resolution XPS spectra in the C1s, N1s and O1s regions were recorded. The overlapping transitions obtained by the XPS were numerically identified through modelling of the line-shapes fitted to the data. In the C1s region both Gaussian/Lorentzian and Gaussian functions were applied to reconstruct the peak shapes of the sp2 and sp3 hybridised C atoms. In line with previous works [18,20,22,23], the C1s spectra were resolved into six individual components (Fig. 7): the sp2-hybridized carbon peak at 284.8 ± 0.1 eV, the peak at 285.4 ± 0.1 eV originating from defects in the nanotube structure (e.g., sp3-hybridizied carbon atoms and sp2 carbon bonded to nitrogen), three weak peaks in the range 286–289 eV (assigned to carbon atoms bonded to oxygen atoms of various functional groups formed during exposure of the CNTs to the ambient air or N-doping), and the peak at 291 eV from p–p*-transitions in aromatic rings. As can be seen from the results of the fitting, the nitrogen doping of the N-CNTs produced more defective carbon nanotubes, which was accompanied by the increase in the area of the ‘‘defect peak’’ centred at 285.4 ± 0.1 eV, cf. Table 3. This can be traced back to the

47

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

Fig. 4. Images of water droplets on the surface of buckypapers produced from different precursors: (a) acetonitrile (N-CNTs), (b) methane (mixed CNTs) and (c) cyclohexane (MWCNTs).

Table 2 Atomic % concentration of elements detected on the surface of the CNTs. Sample

N-CNTs Mixed CNTs MWCNTs

Fig. 5. Zeta potential of the different types of the CNTs as well as silica measured as a function of pH.

incorporation of the nitrogen into the hexagonal graphitic structure and the breaking of its long-range order. This further indicates that the nitrogen atoms are partially bonded to sp2-coordinated carbon. The deconvolution of the C1s core-level spectra of the mixed CNTs and the MWCNTs did not reveal any significant differences in the chemical state of the carbon. Solely in the case of the MWCNTs, the area of the peak assigned to the p–p* electronic transition in aromatic rings is smaller compared to the N-CNTs and the mixed CNTs. Analysis of the N1s band of the N-CNTs (cf. Fig. 8a) clearly shows that nitrogen was present in the different types of species. In order

Total element content (at.%) C1s

O1s

N1s

95.9 98.6 98.6

1.6 1.0 1.4

2.4 0.4 –

C/O

C/N

59.9 98.6 70.4

40 246.5 –

to identify the bonding configurations of the nitrogen, the N1s spectrum was fitted using Gaussian/Lorentzian functions. From the curve fitting, the N1s signal can be deconvolved into five bands representing pyridinic nitrogen (NP) at a binding energy of 398.6 eV, pyrrolic nitrogen (NPYR) at 400.4 eV, quaternary nitrogen (NQ) at 401.5 eV, and nitrogen oxides (NOX) at 403.2 eV [24–27]. It should be noted that the nature of a band near 405 eV is still uncertain. According to Choi and Park [28] the peak appears only when a high photon energy of 1265 eV was used for excitation, leading to a higher photoelectron escape depth. Thus, the authors assigned this band to gaseous N2 intercalated between the tubes’ walls in the vicinity of an internal cavity of the CNTs and not originating from the tubes’ surfaces themselves. As mentioned in Section 2.3, the XPS measurements represented here were also performed using a magnesium Ka anode with a photon energy of 1253.6 eV, which might cause the origin of this band. On the other hand, in agreement with [17,29,30], the peak with a binding energy of 405.2 eV can be attributed to the chemisorbed ANOx species. The relative surface concentration of the nitrogen configurations calculated from the modelled peak area is summarized in Table 4. As indicated in the

Fig. 6. XPS survey scan spectra measured using magnesium Ka anode (1253.6 eV) for the N-CNTs, the mixed CNTs, and the MWCNTs.

48

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

Fig. 7. High-resolution XPS spectra of the C1s region recorded for the N-CNTs, the mixed CNTs and the MWCNTs; dotted lines depict the results of the XPS curve fitting; dot-dashed lines represent the envelopes obtained by fitting the experimental XPS curves (solid line).

literature [24,25,27,31], the pyridinic nitrogen is generally bonded to two sp2-hybridized C atoms and located at the edge-like defects present in the graphene sidewalls of CNTs; cf. Fig. 8b. In the case of the pyrrolic-like type, the structure of a five-membered ring containing one sp2-coordinated heteroatom of nitrogen is formed. The quaternary configuration implies a direct substitution of a carbon atom in the graphene layer by a nitrogen atom, where N is bonded to three sp2-coordinated C atoms [31,32]. The form of the nitrogen oxides is mainly represented by a pyridine-N-oxide [27]. A fitting of the N1s band for the mixed CNTs was not performed due to the low peak intensity relative to the background signal, which suggests a low nitrogen concentration of 0.4 at.% (cf. Table 2) at the nanotube surface. The deconvolution of the O1s spectra for the N-CNTs, the mixed CNTs and the MWCNTs (cf. Fig. 9 and Table 3) results in peaks corresponding to oxygen doubly bonded to carbon (i.e., C@O) at 531.6 ± 0.2 eV, oxygen singly bonded to a carbon atom (i.e., CAOH and/ or CAOAC) at 533.3 ± 0.15 eV, as well as chemisorbed oxygen and/or water with a peak barely above noise level at 535.5 eV [18,21,33]. Peak IV, with a binding energy of 529.9 eV in the spectra of the N-CNTs, can be attributed to the oxygen in the metal oxides. The oxygen-containing groups grafted to the graphene sidewalls of nanotubes can be represented by a variety of surface functionalities, such as carboxyl-, phenol-, anhydride-/lactone-, ether-, pyrone-, chromene- and carbonyl/quinine-types [34]. Since oxygen atoms in the majority of these functional groups have both single and double bonds with carbon atoms, this complicates their assignment to a specific group. As seen from the XPS results, various nitrogen- and/or oxygencontaining functionalities are present at the surface of the N-CNTs.

Table 3 Relative composition of C1s and O1s from XPS spectra. Sample

N-CNTs Mixed CNTs MWCNTs

‘‘defect’’ peak/sp2 C@C

C1s (peak area %) sp2 C@C

‘‘Defect’’ peak (sp3 CAC, sp2 C@N)

CAO, sp3 CAN

C@O

C(@O)O

p–p

36.8 54.8 60.7

31.5 16.7 16.3

5.7 4.6 4.2

6.0 4.0 4.3

4.0 3.2 3.9

16.1 16.7 10.6

0.86 0.31 0.27

O1s (peak area %) I

II

III

IV

10.1 11.7 11.8

25.1 36.1 70.3

40.9 52.2 17.8

23.8 – –

Table 4 Relative surface concentration of the nitrogen functionalities determined by fitting the N1s core level XPS spectra. Sample

N-CNTs *

Total nitrogen content* (at.%)

Content* of nitrogen functionalities (%)

N1s

NP 398.6 eV

NPYR 400.4 eV

NQ 401.5 eV

NOX 403.2 eV

N2/ANOX 405.2 eV

2.40 (0.07)

18.1 (4.3)

12.4 (0.2)

25.6 (2.3)

9.90 (2.1)

33.9 (4.4)

Mean and standard deviation of three measurements for different positions on the sample.

Fig. 8. (a) High-resolution XPS spectra of the N1s region recorded for the N-CNTs; dotted lines depict the results of the XPS curve fitting; dot-dashed lines represent the envelopes obtained by fitting the experimental XPS curves (solid line); (b) Nitrogen functionalities incorporated into a graphene layer of the N-CNTs.

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

Fig. 9. (a) High-resolution XPS spectra of the O1s region recorded for the CNTs; dotted lines depict the results of the XPS curve fitting; dot-dashed lines represent the envelopes obtained by fitting the experimental XPS curves (solid line).

Due to a localized lone pair of electrons, the pyridinic-like nitrogen (pKa  5.2 for the closely related pyridine [35]) located on the surface of the N-CNTs can act as an electron-donating site, which gains a positive charge via protonation in neutral and acidic aqueous media. The pyrrolic constituent also belongs to the basic nitrogen surface groups, but exhibits an extremely low basicity compared to the pyridinic form. Since the electron pair of the pyrrolic nitrogen is part of the aromatic p electron system, it is considered to be nonbonding. The full deprotonation of pyrrole to form a pyrryl anion with a pair of electrons localized on the nitrogen atom occurs only in the presence of a strong base and has a pKa value of 17.5 [36]. The quaternary nitrogen already exhibits a cationic structure and does not require protonation to gain a positive electrical charge. This configuration of nitrogen is defined as having a relatively more positively charged functionality as compared to the pyridinic one [27]. However, in line with the Lewis and Brønsted-Lowry theories, nitrogen in its quaternary form has no acidic properties, since it has no empty orbital to accept an electron-pair as well as no proton to donate. Similar to the quaternary nitrogen, a permanently positive electrical charge is also present on the nitrogen atom of the pyridine-N-oxide [27]. On the other hand, the oxygen attached to nitrogen is a bearer of a lone electron pair and can therefore act as a weak Lewis base (pKa  0.79 for conjugated acid of pyridine-N-oxide [37,38]). To screen such negatively charged sites, more acid is required, which is accompanied by a shift of the IEP towards lower pH values. Assuming that the nitrogen oxides, such as -NO2 (pKa  3.2 for protonated nitrobenzene radical-anion [39–41]), are still present on the surface of the N-CNTs, they could also have contributed to the increase in the tube polarity and have led to a shift of the IEP to lower pH values. The distinct nature of the nitrogen functionalities results in their different ionisation conditions. To predict the extent of deprotonation for the pyridinic, pyrrolic, and pyrridine-N-oxide sites at the tube surface versus pH, the GALAS modelling method was

49

applied. The determined protonation states as well as calculated pKa values are represented in Fig. 10. According to the species distribution diagram, the protonated nitrogen-containing functionalities (protonation state (PS) 1) dominate in a region of low pH. The pyridine-N-oxide centres are completely deprotonated beyond pH 3 and the protonation state 2 (PS2), with a net charge of +1, reaches its maximum under this condition. If the pH is adjusted to 4.5, the protonation probability of the pyridinic site with pKa  4.5 becomes 50%. It means that one half of the pyridinic sites present at the tube surface is deprotonated. The further increase in the pH leads to the prevalence of the protonation state 3 (PS3), at which the tube surface gains a net charge of zero, over the PS2. The next deprotonation process with pKa  15.6 occurs in a high pH region and leads to the formation of anionic sites from the pyrrolic nitrogen. As can be seen from the modelling results, the disagreement between the calculated pKa and the pKa values reported in the literature [35–38] are insignificant and lie within the standard deviations. The predicted pH region from 6.5 to 11, at which 99–100% of the nitrogen functionalities have no net charge, includes the experimental IEP values of 6.6 (determined in this work) and 9.2 ± 0.4 (reported by Maldonado et al. for the nitrogen-doped CNTs in the previous study [42]). The value of the IEP is defined by the average ratio of acid to base surface functionalities and their pKa’s. Obviously, the high IEP value reported by Maldonado for the N-doped CNTs originated from an abundance of basic surface species, whereas the IEP at about 6.6 for the Ndoped CNTs in this work indicated significant contribution of acidic centres. Based on this, it can be concluded that nitrogen doping can impart amphoteric characteristics to the surface of carbon nanotubes. The IEP at neutral pH testifies in this case as to the neutrality of the CNT surface, i.e. equal contribution of positively and negatively charged sites. Independent from the pH of an aqueous solution, such active surface sites are able to interact with water and thereby facilitate the wetting of the N-CNTs. The pronounced differences in the polarity as well as in the wettability of the mixed CNTs and the MWCNTs yield puzzling results. As mentioned above, the XPS data did not reveal any significant peculiarities in the surface contamination resulting in NAO or CAO bonds for these CNT types. The strongly hydrophobic and nonpolar MWCNTs synthesised from cyclohexane contrast with the moderately hydrophilic and basic mixed CNTs produced from methane. The basic feature causing the shift in the IEP of the mixed CNTs towards higher pH values may be explained by various factors, such as the basicity of the graphene layer of the CNTs and the presence of oxygen-containing basic groups, for example, chromene-type, pyrone-type, on the tube surface. Leon et al. demonstrated that the p-electron-rich basal plane surface of graphitized carbon can adsorb solvated protons through an electron donor– acceptor mechanism [43]. In the case of nanotubes, the sp2-hybridized orbitals of carbon atoms are deformed due to the curvature of graphene walls [44]. The asymmetry in the overlap expressed in both a closer proximity of the p-orbital lobes situated inside the tube cavity and their mutual spacing outside the carbon framework results in a shift of p-electron density towards the convex outer tube surface; thus the permanent dipole moment oriented approximately perpendicular to the tube surface is induced [45]. The electron-rich outer surface of the nanotubes now has a greater ability to attract protons or other cations in an aqueous solution and to hold them in the fixed (Stern) layer due to cation-p interactions [46]. The p-donating properties of the curved graphene layers evidently become more pronounced in nanotubes with a very small diameter, for instance, single- and double-walled CNTs, due to their large curvature and a greater resulting electronic charge displacement. A significant deformation of the p-electron system might increase the basicity of the graphene walls and facilitate their interaction with water. As a consequence, depending on the nanotube

50

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

Fig. 10. Calculated distribution of protonation states for the ionisable nitrogen species present on the N-CNTs surface as a function of pH.

diameter, a degree of hydrophilicity may be imparted to the CNT surface. Considering the above discussion, it can be assumed that the greater overall hydrophilicity and the basic properties observed for the mixed CNTs would have stemmed from the presence of the very small diameter tube (cf. Table 1), which exhibits a more pronounced p-electron-donating ability and could therefore affect the interaction of the nanotube bulk material with water. Another possible reason for the basicity of the mixed CNTs is the presence of the surface oxides with electron-donating properties. As known from previous studies [19,34,43,47], the oxygen chemisorbed on the carbon surface can exhibit both acidic and basic features. Basic properties have been observed when a carbon surface after undergoing a high-temperature heat treatment at ca. 900–1000 °C in vacuo or under an inert gas was exposed to air and aqueous acid, or water, after cooling to room temperature [18,48,49]. It has been believed that the pyrone-, chromene-types [34,43] as well as quinone-type [47] structures stabilised by resonance are formed, which increases the carbon alkalinity. The certain content of polar functional groups grafted to the tube surface could also contribute to the wetting improvement observed.

Regarding the hydrophobic MWCNTs, hydrophobic surfaces (e.g., Teflon, diamond) are, as known, non-polar and therefore have no surface charge. They could otherwise interact with water molecules. Until recently, it has been commonly assumed that the isoelectric point of hydrophobic materials is between pH 6 und 7. However, as reported in the literature [50–52], the interaction of hydrophobic surfaces with aqueous solution is more complex with some electrokinetic phenomena occurring in the interface zone. The low IEP of ca. 4.0 and the negative surface potential at neutral and alkaline pH, also detected for the MWCNTs (see Fig. 5), are caused by physical adsorption of anionic species such as hydroxide ions at an aqueous solution/hydrophobic interface. The driving force for the migration of the hydroxide ions from the aqueous phase is preferential orientation of the water in the vicinity of the hydrophobic surface. This ordering of the water molecules generates electrical potential gradients. The high hydrophobicity of the MWCNTs might originate from a deposition of carbon species, forming during cyclohexane pyrolysis, on the surface of the nanotubes. As shown in [53,54], the gaseous products of cyclohexane decomposition are chiefly represented by aliphatic hydrocarbons, such as ethylene, 1,3-buta-

51

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

diene, and their free radicals at 725 and 810 °C, and by polycyclic aromatic hydrocarbons, such as anthracene, ethylpyrene, and pyrene, at 900 °C. It is assumed that some decomposition products could be adsorbed as non-volatile substances on the tube walls, which would have increased chemical and morphological heterogeneity of the CNT surface. Moreover, the adsorption of the pyrolytic carbon species might partially screen the p-electron cloud of the graphene tube walls and thus adversely affect their interaction with water. This assumption is to be verified in further investigations by the authors. Based on these observations, it can be concluded that both the carbon sources and synthesis conditions play a crucial role in establishing the surface properties of CNTs. 3.2. Coating of CNTs with amorphous silica Fig. 11 shows the TEM images of the CNTs after silica sol–gel coating. In contrast to the as-grown MWCNTs, a significant change

in surface morphology can be seen for the pristine N-CNTs; cf. Fig. 11a and b. The image Fig. 11a shows a primary uniform adsorption of silica on the N-CNT surface, but partly in the form of non-adherent fragments, likely caused by shrinkage and cracking of a silica gel film during final drying [55]. Adsorption of the colloidal silica on the hydrophobic surface of the MWCNTs was not observed. Silica was found in the form of agglomerates in the bulk of the nanotubes (cf. Fig. 11b). The modification of the MWCNTs through their dispersion in the Brij solution obviously promoted the silica adsorption, which consequently led to a uniform covering of the tube surface with precipitate (cf. Fig. 11d). Moreover, it seems to be that the application of Brij 35 contributed to the formation of a coherent continuous film on the CNTs by a reduction in the surface tension of the water and by the consequent shrinkage due to drying of the gel (cf. Fig. 11c). As a result, a uniform silica layer with a thickness of approximately 10– 40 nm could be achieved on the surface of the N-CNTs as well as the MWCNTs. In contrast to the Brij 35, the addition of CTAB as a surface modifier adversely affects the precipitation of silica, present mainly as individual clusters in the bulk of the CNTs and occasionally as agglomerations adsorbed on the tube surface; cf. Fig. 11e, f. Generally, the coating of the surface of the small-diameter mixed CNTs was impeded by the CNT entanglement and the formation of a dense network during intensive sonication, even if the surfactant was used. It was impossible to discover details of CNT coating with silica by TEM studies due to deposition of silica outside the CNT network (cf. Fig. 11g). The XPS data represented in Table 5 yielded a low surface carbon concentration along with a high content of oxygen as well as silicon for the silica-coated N-CNTs and the MWCNTs. Taking into consideration the TEM-results, this can be accounted for by a high degree of coverage with silica. The stoichiometric ratio of Si to O corresponds nearly to SiO2. The XPS analysis of the silica-coated mixed CNTs revealed a low surface concentration of oxygen and silicon as well as a high carbon content, which is in line with the TEM observation. 3.3. Mechanism of silica adsorption on the CNTs Numerous studies confirm the existence of silanol groups („SiAOH) on the surface of amorphous silica along with „SiAOASi„ bridges [56,57]. Silanol groups are formed on the silica surface in the course of its synthesis during the condensation– polymerization of TEOS and are differentiated by their multiplicity of sites as well as by the type of their association. The silanol sites as well as the degree of surface coverage with them are responsible for the silica surface charge which is induced by the protonation– deprotonation of the silanol groups with the establishment of equilibrium between „SiO, „SiOH and „SiOH2+. As known and once again demonstrated in the present work (cf. Fig. 5), the silica surface becomes negatively charged in neutral and basic aqueous solutions. The deprotonation of the silanol sites creating the „SiO form at the silica surface interface is intensified by a gradual increase in pH of the solution. The acidity of the silanol groups is strongly dependent on the hydrogen bonding among the silanols.

Table 5 Atomic % concentration of elements detected on the surface of silica-coated CNTs. Sample

N-CNTs* MWCNTs* Mixed CNTs* Fig. 11. TEM images of the CNTs after sol–gel dip coating with silica.

*

Total element content (at.%)

Si/O

C1s

O1s

N1s

Si1s

15.4 10.4 85.5

52.1 48.1 9.5

– – 0.1

32.5 41.5 4.9

Dispersed in Brij solution before the sol–gel dip coating.

0.62 0.86 0.52

52

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

The pKa value of silanols increases from a range of 7 to 8, which corresponds to the silanol sites’ participating in intramolecular hydrogen bonding, up to 8.9 to 10 attributed to isolated silanols [58]. Moreover, the various studies reported the existence of „SiOH groups with a very low acidity constant of only ca. 4.5 [59,60]. Based on this, it can be inferred that the silanol groups on the silica surface are partially deprotonated already under neutral pH conditions. Here too, another consequence is that the silanols bound through the hydrogen bonds, such as vicinal silanols, would be ionised in the first place. The acidity of the ionisable, nitrogen-containing species incorporated into the graphite surface of the N-CNTs has already been discussed above; cf. Fig. 10. As can be seen, almost all nitrogen configurations, with the exception of the pyrrolic form, are completely deprotonated in an aqueous solution with a pH level of 6–7, at which the coating of the nanotubes with colloidal silica was performed. 2–3% of the pyridine-like centres are still available in their non-ionised state in this pH region. Due to the presence of unshared electron pairs on the N and O atoms of the deprotonated pyridinic and oxidized nitrogen, respectively, these sites could donate electrons and attract hydrogen atoms of the acidic silanol groups through Brønsted acid–base interactions. (cf. Fig. 12a). Conversely, the NAH moiety of the non-ionised pyridinic sites on the N-CNTs can share its hydrogen atom with the electronegative oxygen of the deprotonated silanol groups and contributes, thereby, to the adsorption of silica from the solution. The graphitic nitrogen is the bearer of a permanent charge and could thereby act as a cationic centre on the outermost shells of the tubes. An electrostatic interaction could be a driving force to attract the negatively charged ionised silanol groups to such centres. However, the Coulomb interaction is limited by the effective distance separating the cationic quaternary nitrogen from the anionic oxygen atom of the ionised silanols in the aqueous medium and could be completely screened due to the high dielectric constant of water (e  80). The pyrrolic nitrogen cannot be converted to the pyrryl anion under neutral pH conditions [36]. Thus, the undissociable hydrogen of the pyrrolic NAH sites in the N-CNTs might be attached to the oxygen of silanol groups via either hydrogen bonds [61] or an acid– base type of bonding. Based on this discussion, it can be concluded that the adsorption of colloidal silica on the polar surface of the NCNTs is promoted by various types of chemical non-covalent bonds provided through acid–base interactions, electrostatic attraction and hydrogen bridging.

As shown by the TEM-images (cf. Fig. 11b, d, and f), the uniform coating of the nonpolar MWCNT surface with silica was achieved only after dispersion of the tubes in the aqueous solution of nonionic surfactant Brij 35. The amphiphilic surfactant molecules adsorbing on the CNTs by means of their self-association [62,63] evidently facilitated the subsequent precipitation of colloidal silica on the tube surface due to hydrogen bond formation (cf. Fig. 12b). In contrast to the Brij 35, the application of the cationic surfactant CTAB surprisingly seems to act disadvantageously. Instead of forming a uniform coating on the MWCNT surface, the silica is present as individual clusters by addition of CTAB. It is supposed that the negatively charged silica sites attract positively charged CTAB molecules, resulting in the formation of neutral agglomerates. Moreover, in agreement with Bellmann et al. [64], the application of oversaturated CTAB solution could change the surface charge of silica even to positive (the used CTAB concentration of 13.7 mM was much higher than its micelle concentration of 1.01. . .0.98 mM [65]). On the other hand, the lipophilic tails of CTAB molecules are adsorbed on the hydrophobic surface of the MWCNTs with their hydrophilic cationic heads exposed outside. Therefore, the positively charged CTAB/MWCNT agglomerates and the neutral or slightly positively charged CTAB/silica agglomerates are present in the solution. Obviously, an efficient attraction of such agglomerates is not favourable.

4. Conclusions The efficient use of the superior mechanical properties of CNTs in cement-based composites involves providing for adequate interaction between nanotubes and cement matrix. In order to improve the embedding of carbon nanotubes (CNTs) in a cement matrix, silica was deposited on the sidewall of CNTs by the sol–gel method. To observe the wetting of carbon nanotubes and the adhesion behaviour of silica thereon, surface characterisation of the CNTs synthesized from the different precursors was performed. It was found that the surface properties of the CNTs were considerably affected by their precursors as well as by the synthesis method used. The utilisation of acetonitrile as carbon source led to the incorporation of nitrogen atoms in graphene layers on the nanotubes. The active nitrogen- and/or oxygen-containing sites present at the N-CNTs increased the surface polarity of the tubes, giving them amphoteric characteristics and improving their affinity to water.

Fig. 12. Mechanism of silica adsorption on the surface of (a) N-CNTs and (b) hydrophobic MWCNTs pre-treated with solution of non-ionic surfactant Brij 35: non-covalent bonds are depicted by dotted lines; bond lengths are displayed arbitrarily for illustration only.

A. Sobolkina et al. / Journal of Colloid and Interface Science 413 (2014) 43–53

The moderate wettability and basic feature of the mixed CNTs obtained by fixed-bed CVD using methane as carbon source could be explained by two factors: (i) p-donating effect pronounced for the single- and double-walled nanotubes of very small diameter and/ or (ii) the presence of the oxygen-containing functional groups of a basic nature at the tube surface. It is proposed that the strong hydrophobic character of the MWCNTs produced from cyclohexane was raised due to the adsorption of some pyrolytic residue on the tube surface from the gaseous phase. The sol–gel coating of the CNTs with silica showed that nitrogen- and/or oxygen containing sites of the N-CNT surface contributed to the adsorption of silica by creation of non-covalent (acid– base, ionic, hydrogen) interactions with silanol groups. The uniform deposition of silica on the nonpolar and hydrophobic MWCNTs, synthesized from cyclohexane, was achieved only after pretreatment of the nanotubes with nonionic surfactant Brij 35. Acknowledgments The authors are grateful to their colleagues from the Leibniz Institute for Solid State and Materials Research namely Manfred Ritschel for the synthesis of CNTs as well as Diana Maier for the support in performing TEM investigations. Furthermore, the authors thank Steffi Kaschube from the Leibniz Institute for Solid State and Materials Research for assistance in the XPS measurements and data analysis, Anja Caspari from the Leibniz Institute of Polymer Research for performing electro-kinetic measurements, and Christoph Schroefl from the Institute of Construction Materials at the TU Dresden for fruitful discussions. References [1] C. Li, T.W. Chou, Compos. Sci. Technol. 63 (2003) 1517–1524. [2] T. Belytschko, S.P. Xiao, G.C. Schatz, R. Ruoff, Phys. Rev. B 65 (2002) 235430-1– 235430-6. [3] M.F. Yu, O. Lourie, M.J. Dyer, K. Moloni, T.F. Kelly, R.S. Ruoff, Science 287 (2000) 637–640. [4] B. Peng, M. Lacascio, P. Zapol, S. Li, S.L. Mielke, G.C. Schatz, H.D. Espinosa, Nat. Nanotechnol. 3 (2008) 626–631. [5] G.Y. Li, P.M. Wang, X. Zhao, Carbon 43 (2005) 1239–1245. [6] C. Kahattha, P. Woointranont, T. Chodjarusawad, W. Pecharapa, J. Microsc. Soc. Thailand 24 (2) (2010) 133–135. [7] J. Cumings, A. Zettl, Science 289 (2002) 602–604. [8] S. Musso, J.-M. Tulliani, G. Ferro, A. Taglaferro, Compos. Sci. Technol. 69 (2009) 1985–1990. [9] A. Cwirzen, K. Habermehl-Cwirzen, V. Pentalla, Adv. Cem. Res. 20 (2) (2008) 65–73. [10] P. Stynoski, P. Mondal, C. Marsh, Novel properties to improve CNT utility in cement, in: M.S. Konsta-Gdoutos (Ed.), NICOM4. Proceedings of the 4th International Symposium on the Nanotechnology in Construction, 2012 May 20–22, Agios Nikolaos, Grete, Greece. [11] V.O. Khavrus, A. Leonhardt, S. Hampel, C. Täschner, C. Müller, W. Gruner, S. Oswald, P.E. Strizhak, B. Büchner, Carbon 45 (2007) 2889–2896. [12] K.D. Bartle, D.L. Perry, S. Wallace, Fuel Process. Technol. 15 (1987) 351–361. [13] H. Giesche, J. Eur. Ceram. Soc. 14 (1994) 189–204. [14] T. Seeger, Ph. Redlich, N. Grobert, M. Terrones, D.R.M. Walton, H.W. Kroto, M. Rühle, Chem. Phys. Lett. 339 (2001) 41–46. [15] . [16] H.J. Jacobasch, F. Simon, C. Werner, C. Bellmann, Tech. Mess. 63 (1996) 439– 446. [17] V.O. Khavrus, H. Vinzelberg, J. Schumann, A. Leonhardt, S. Oswald, B. Büchner, Physica E 43 (2011) 1199–1207. [18] S. Biniak, G. Szyman´ski, J. Siedlewski, A. S´wiatkowski, Carbon 35 (12) (1997) 1799–1810. [19] H.P. Boehm, Carbon 40 (2002) 145–149. [20] H. Ago, T. Kugler, F. Cacialli, W.R. Salaneck, M.S.P. Shaffer, A.H. Windle, R.H. Friend, J. Phys. Chem. B 103 (1999) 8121–8126. [21] V. Datsyuk, M. Kalyva, K. Papagelis, J. Parthenios, D. Tasis, A. Siokou, I. Kallitsis, C. Galiotis, Carbon 46 (2008) 833–840. [22] B. Khare, P. Wilhite, B. Tran, E. Teixeira, K. Fresquez, D. Mvondo, C. Bauschlicher Jr., J. Phys. Chem. B 109 (2005) 23466–23472.

53

[23] H. Estrade-Szwarckopf, Carbon 42 (2004) 1713–1721. [24] S. van Dommele, N. Romero-Izquirdo, R. Brydson, K.P. de Jong, J.H. Bitter, Tuning nitrogen functionalities in catalytically grown nitrogen-containing carbon nanotubes, Carbon 46 (2008) 138–148. [25] P. Ayala, R. Arenal, M. Rümmeli, A. Rubio, T. Pichler, Carbon 48 (2010) 575– 586. [26] X.Y. Tao, X.B. Zhang, F.Y. Sun, J.P. Cheng, F. Liu, Z.Q. Luo, Diamond Relat. Mater. 16 (2007) 425–430. [27] J.R. Pels, F. Kapteijn, J.A. Moulijn, Q. Zhu, K.M. Thomas, Carbon 33 (11) (1995) 1641–1653. [28] H.C. Choi, J. Park, J. Phys. Chem. 109 (2005) 4333–4340. [29] D.N. Hendrickson, J.M. Hollander, W.L. Jolly, Inorg. Chem. 8 (12) (1969) 2642– 2647. [30] J.L. Hueso, J.P. Espino9 s, A. Caballero, J. Cotrino, A.R. González-Elipe, Carbon 45 (2007) 86–89. [31] R. Lv, Q. Li, A.R. Botello-Méndez, T. Hayashi, B. Wang, A. Berkdemir, Q. Hao, A.L. Elías, H.R. Gutiérrez, Y.A. Kim, H. Muramatsu, J. Zhu, M. Endo, H. Terrones, J.C. Charlier, M. Pan, M. Terrones, Science Reports 2(586) (2012) doi: 10.1038/ srep00586. [32] H. Liu, Y. Zhang, R. Li, X. Sun, S. Désilets, H. Abou-Rachid, M. Jaidann, L.S. Lussier, Carbon 48 (2010) 1498–1507. [33] T.I.T. Okpalugo, P. Papakonstantinou, H. Murphy, J. McLaughlin, N.M.D. Brown, Carbon 43 (2005) 153–161. [34] H.P. Boehm, Carbon 32 (5) (1994) 759–769. [35] P.M. Dewick, Essentials of Organic Chemistry: for Students of Pharmacy, Medicinal Chemistry and Biological Chemistry, John Wiley & Sons, Chichester, 2006. [36] J.A. Joule, K. Mills, Heterocyclic Chemistry at a Glance, second ed., John Wiley & Sons, Chichester, 2013. [37] C. Klofutar, Š. Paljk, D. Kremser, Spectrochim. Acta Part A 29 (1973) 139–145. [38] R.E. March, C.J. Stadey, E.G. Lewars, Rapid Commun. Mass Spectrom. 19 (2005) 984–1004. [39] Z.V. Todres, Ion-Radical Organic Chemistry: Principles and Applications, second ed., CRC/Taylor&Francis, Boca Raton, 2009. [40] G.A. Olah, T.E. Kiovsky, J. Am. Chem. Soc. 90 (23) (1968) 6461–6464. [41] G.A. Olah, D.A. Klumpp, Superelectrophiles and Their Chemistry, John Wiley & Sons Ltd., Hoboken, 2008. [42] S. Maldonado, S. Morin, K.J. Stevenson, Analyst 131 (2006) 262–267. [43] C.A. Leon y Leon, J.M. Solar, V. Calemma, L.R. Radovic, Carbon 30 (5) (1992) 797–811. [44] R.C. Haddon, J. Phys. Chem. A 105 (2001) 4164–4165. [45] T. Dumitrica˘, C.M. Landis, B.I. Yakobson, Chem. Phys. Lett. 360 (2002) 182–188. [46] M. Keiluweit, M. Kleber, Environ. Sci. Technol. 43 (2009) 3421–3429. [47] A. Contescu, M. Vass, C. Contescu, K. Putyera, J.A. Schwarz, Carbon 36 (3) (1998) 247–258. [48] C.H. Tessmer, R.D. Vidic, L.J. Uranowski, Environ. Sci. Technol. 31 (1997) 1872– 1878. [49] A. Bismarck, C. Wuertz, J. Springer, Carbon 37 (1999) 1019–1027. [50] R. Zimmermann, N. Rein, C. Werner, Phys. Chem. Chem. Phys. 11 (2009) 4360– 4364. [51] R. Zangi, J.B.F.N. Engberts, J. Am. Chem. Soc. 127 (2005) 2272–2276. [52] P. Koelsch, P. Viswanath, H. Motschmann, V.L. Shapovalov, G. Brezesinski, H. Möhwald, D. Horinek, R.R. Netz, K. Giewekemeyer, T. Salditt, H. Schollmeyer, V. Klitzing, R. Diallant, J. Guenoun, Colloids Surf. A: Physicochem. Eng. Aspects 303 (2007) 101–136. [53] Y. Kawabuchi, M. Kishino, S. Kawano, D.D. Whitehurst, I. Mochida, Langmuir 12 (1996) 4281–4285. [54] F. Billaud, P. Chaverot, M. Berthelin, E. Freund, Ind. Eng. Chem. Res. 27 (1988) 759–764. [55] R.K. Iler, The Chemistry of Silica. Solubility, Polymerization, Colloid and Surface Properties, and Biochemistry, John Wiley & Sons, Chichester, 1979. [56] L.T. Zhuravlev, Colloids Surf. A 173 (2000) 1–38. [57] H.B. Bergna, W.O. Roberts (Eds.), Colloidal Silica: Fundamentals and Applications, CRC/Taylor&Francis, Boca Raton, 2006. [58] D.A. Loy, Sol–gel processing of hybrid organic–inorganic materials based on polysilsesquioxanes, in: G. Kickelbick (Ed.), Hybrid Materials: Syntheses, Characterization and Applications, WILEY-VCH, Weinheim, 2007, pp. 225–253. [59] S. Ong, X. Zhao, K.B. Eisenthal, Chem. Phys. Lett. 191 (3–4) (1992) 327–335. [60] C. Chaiyasut, Y. Takatsu, S. Kitagawa, T. Tsuda, Electrophoresis 22 (2001) 1267–1272. [61] Y. Matsumoto, K. Honma, J. Chem. Phys. 130 (2009) 054311-1–054311-10. [62] L. Vaisman, H.D. Wagner, G. Marom, Adv. Colloid Interface Sci. 128–130 (2006) 37–46. [63] A. Sobolkina, V. Mechtcherine, V. Khavrus, D. Maier, M. Mende, M. Ritschel, A. Leonhardt, Cement Concr. Compos. 34 (2012) 1104–1113. [64] C. Bellmann, A. Synytska, A. Caspari, A. Drechsler, K. Grundke, J. Colloid Interface Sci. 309 (2007) 225–230. [65] M.A. Malik, Z. Khan, Colloids Surf., B 64 (2008) 42–48.