Imogolite is a natural mineral observed for the first time by Aomine et al. in 1962 [1]. It consists of a single-walled aluminosilicate nanotube with inner and outer diameters of 1 and 2 nm, respectively, and a length ranging from a few tens to several hundreds of nanometers. The tube walls are composed of a curved gibbsite (Al(OH)₃) layer on the outer surface and silicate tetrahedra linked to six aluminum octahedra inside the tube [2]. The empirical formula of imogolite is (HO)₃Al₂O₃SiOH. Imogolite was first prepared by Farmer et al. in 1977 by coprecipitation of aluminum and silicon monomer with millimolar initial concentrations at 95 °C [3]. Due to its hollow tubular structure, imogolite is nowadays considered as a competent substitution for carbon nanotubes. However imogolite differs from carbon nanotube by its hydrophilic surface which can be easily modified to obtain hybrid materials [4]. Thus, several applications of this material have been reported including catalysis [5], gas storage [6], adsorption [7] and heat exchange [8]. Several synthesis strategies have been developed recently in order to optimize imogolite synthesis by varying some parameters such as nature of reactants, temperature and growth phase duration, desalination process, recovering and drying. Nevertheless it is still difficult to produce a high yield of imogolite. Furthermore, some advances have been made about determination of the formation mechanism [9,10,11,12,13,14,15], but some problems remain to be overcome for industrial development of imogolite.

From these considerations, we have investigated the possibility to use the fluoride route for the synthesis of imogolite. Indeed, the fluoride route is extensively used for synthesis of aluminosilicate, e.g., zeolites and clays such as beidellites, montmorillonites and saponites [16,17]. The F⁻ fluoride ion acts as a mineralizing agent such as the hydroxide anion because of its similar charge and size. The fluoride ions used in small amounts, were proven to decrease crystallization times and increase the crystallinity of the solids. Moreover the substitution of hydroxyl groups by fluoride ions in the structure allows increasing the thermal stability of the material. In addition, the ¹⁹F nucleus, thanks to its ½ nuclear spin and 100% natural abundance, is an interesting probe for structural investigations by solid-state NMR. To the best of our knowledge, this route has never been explored for the imogolite synthesis.

The yield of the freeze-dried products was 64% for F-IMO against 44% for IMO. The XRD diffraction patterns of F-IMO and IMO are presented in Figure 1(a). Both diffractograms show four broad reflections at 4.2°2θ (2.07 nm), 9.1°2θ (0.97 nm), 14.3°2θ (0.61 nm) and 26.7°2θ (0.32 nm) characteristic of imogolite structure [2,13,21]. The most intense reflection at 4.2°2θ is attributed to the (100) plane of the overall structure, and is related to the diameter of the nanotubes (2.07 nm) [2]. The reflection at 9.1°2θ (0.97 nm) is assigned to the (001) plane and characterizes the fiber structure [9]. It is noteworthy that the sharp reflection at 18.6°2θ (0.47 nm) clearly observed for F-IMO and barely for IMO is not yet assigned.

Figure 1(b) shows similar FTIR spectra of the F-IMO and IMO. The bands characteristic of imogolite appear at 992, 953, 697, 562, 512 and 425 cm⁻¹. The band doublet observed at 992 and 953 cm⁻¹ corresponds to the Si-O stretching vibration, which are specific for tubular structure [22]. The bands at 697 and 425 cm⁻¹ are attributed to O-Al-O and Al-OH vibrations. The bands observed at 562 and 512 cm⁻¹ are related to O-Si-O vibrations. The FTIR results are consistent with the formation of imogolite in fluoride medium and confirm XRD results.

The morphology of F-IMO was observed by TEM. The micrographs in Figure 2 show that, like IMO, F-IMO forms well-distinguished individual fibers. The fibers are aligned in bundles. A spiderweb-like structure was also observed for both samples. The insufficient dilution of the sample prevents an accurate determination of the length and the diameter but by using Image J software the diameter was evaluated. No significant difference concerning the diameter value is observed between F-IMO and IMO. The average outer diameter is roughly 2.4 nm for both samples, which is slightly higher than the one estimated from XRD results.

The local environment of Si, Al and F in F-IMO was probed by solid-state NMR spectroscopy. The ²⁹Si CP-MAS NMR spectra of F-IMO and IMO samples are displayed in Figure 3(a). CP-MAS usually do not allow a quantitative approach but is used to improve the signal to noise ratio and save spectrometer time. Nevertheless, the ¹H decoupled ²⁹Si MAS NMR spectrum recorded for the IMO sample lead to the same intensity ratio for all resonances. Therefore, in this particular case, it is possible to compare both samples. For both sample a main resonance at −79 ppm is detected and assigned to silanol groups connected to 6 Al through oxygen atoms, as expected for imogolite structure [23]. The width at half height, in the case of the IMO sample is 73 Hz instead of 60 Hz for the F-IMO sample. The broadening of the line is often related to a distribution of bond lengths or angles suggesting that imogolite prepared in fluoride medium seems to be better organized on a local point of view. Furthermore, additional weak and very broad resonances are detected on both ²⁹Si CP-MAS NMR spectra (Figure 3 inset). Decomposition of the lines reveals that these broad components appear between −83 ppm and −95 ppm and account for 10% and 32% of the total signal in F-IMO and IMO samples, respectively. Those components are assigned to less ordered silicon environments indicating the presence of poorly crystallized species called protoimogolite or/and allophone [21,22], but could also correspond to structural defects present in the imogolite walls as recently demonstrated by Yucelen et al. [24]. At the stage of the study it is not possible to discriminate between both hypotheses but whatever the assignment of these resonances the presence of fluorine in the synthesis medium appears to decrease the proportion of unwanted species. ²⁷Al MAS NMR spectra of both samples are shown in Figure 3(b). Two resonances are observed at 4 ppm and 60 ppm corresponding to the octahedral and tetrahedral aluminum, respectively. This tetrahedral aluminum could correspond to either transient phase formed in variable amount during imogolite synthesis or to aluminum sites at the end of the nanotubes as suggested by Nair et al. [14]. It is interesting to note that, the proportion of tetrahedral aluminum is negligible (2% of the total signal) when the synthesis was performed in fluoride medium instead of 6% without fluorine. Both ²⁷Al MAS and ²⁹Si CP-MAS NMR suggest a lower amount of these poorly crystalline species or defect sites when the fluoride route is used. As reported for other minerals, the synthesis in the presence of fluorine allows increasing the crystallization rate [19].

Elemental analysis of F-IMO was performed for the following elements: Al, Si, H and F. The results are listed in Table 1. It should be noted that a small amount of fluorine was incorporated in F-IMO (less than 1%). Based on these results and on ²⁷Al MAS and ²⁹Si CP-MAS solid state NMR data, the following formula can be given for F-IMO: (HO)₃O₃Al₂(Si_0.9Al_0.1)(F_0.08OH_0.92). It is noteworthy that there is slightly more Al in the F-IMO than in classical imogolite. However uncertainty of element analysis is about 5%.

¹⁹F MAS NMR spectra of F-IMO displayed in Figure 4 shows a broad resonance at about −131 ppm confirming the presence of fluorine. The very low signal to noise ratio of the spectrum suggests that only a weak amount of fluorine has been incorporated in the structure, in agreement with chemical analysis (<1 wt.%). The observed chemical shift value (−131 ppm) is close to the one observed for beidellite, a dioctahedral phyllosilicate prepared in fluoride medium [19]. In that case, a ¹⁹F signal at −132 ppm was detected and assigned to fluorine atom linked to Al-Al-□ in octahedral sheet (where □ represents a vacancy). A similar atom arrangement is also probably present in F-IMO. The broadness of the line suggests a distribution of environments for the fluorine.

In conclusion, aluminosilicate nanotubes were successfully obtained by a new synthesis method using fluoride medium with a yield as high as 64% (against 44% in absence of fluorine). According to solid state NMR, a lower amount of poorly crystalline species or defects sites are observed when the fluoride route is used. Studies are in progress to fine-tune the different parameters governing the synthesis and to explore the mechanism of formation taking advantage of the presence of fluorine.