Tuning retention and selectivity in reversed-phase liquid chromatography by using functionalized multi-walled carbon nanotubes

2.1 Materials and reagents

Pristine MWCNTs (outer diameter 20–50 nm, inner diameter 5–10 nm, length 10–20 μm, specific surface area 60 m² g⁻¹, purity >95 wt%) were obtained from Cheap Tubes Inc. (Brattleboro, Vermont, USA). Silica microspheres (average particle size 5 μm, average pore diameter 100 Å, surface area 200 m² g⁻¹) were purchased from Agilent Technologies Italia S.p.A (Cernusco sul Naviglio, Italy). Ultrapure water (resistivity 18.2 MΩ cm⁻¹) was produced by a Milli-Q system (Millipore, Milan, Italy). Ultrapure HCl (37%, w/w) was from Merck Chemicals (Milan, Italy). H₃PO₄ (85%, w/w) and ethylbenzoate (⩾99%) were from Carlo Erba Reagents (Milan, Italy). Triton X-100, m-chloroperbenzoic acid, benzene, toluene, dichloromethane, HPLC grade acetonitrile (ACN), PB (M_n 5000, 80% cis + trans, 20% vinyl), 2,2′-(ethylenedioxy)diethylamine (98%), ethylbenzene (99%) and 2-chlorotoluene (99%) were supplied by Sigma–Aldrich (Milan, Italy). Other chemicals were obtained from Sigma–Aldrich. An Orion 420A pH meter (Thermo Electron, Rodano, Italy) was used.

2.2 Preparation of MWCNT-R-NH₂-PB-silica

Firstly, MWCNTs were chemically modified with amino-terminated alkyl chains and the derivatization was checked by TGA, as earlier reported (Merli et al., 2011; Speltini et al., 2010). MWCNT-R-NH₂-PB-silica was prepared as described in the following. A modification of a known procedure (Bukanova et al., 1976) was followed to prepare MWCNT-R-NH₂-PB: 5 g of 1,4-polybutadiene (M_n 5000, 80% cis + trans, 20% vinyl) was dissolved in 100 mL anhydrous benzene. To the solution, 10 g m-chloroperbenzoic acid was added and the mixture was refluxed for 72 h under nitrogen. The suspension was centrifuged to separate the benzoic acid formed; the solution was collected and washed with three aliquots of water (200 mL each). The organic phase was evaporated under vacuum to obtain a viscous yellowish liquid. The so obtained epoxide was dissolved in 150 mL toluene and 2 g MWCNTs-R-NH₂ were added. The mixture was refluxed for five days under nitrogen. The suspension was centrifuged, and the precipitate (3.2 g of MWCNT-R-NH₂-PB, containing the initial MWCNTs-R-NH₂) was washed four times with 40 mL toluene. The so obtained black mass was dried for 24 h at ambient temperature. To obtain MWCNT-R-NH₂-PB-silica, 250 mg of MWCNT-R-NH₂-PB was dispersed by sonication (30 min) in 85 mL dichloromethane, to which 250 μL PB and 40 μL Triton X-100 were added; after addition of 4.2 g silica, the suspension was magnetically stirred for 30 min, sonicated for 5 min, again stirred for 30 min, and finally the solvent was evaporated under vacuum. The obtained grey powder (MWCNTs-R-NH₂ content 3.6 wt%) was transferred in a glass vial and maintained in high vacuum (10⁻⁵ Torr) for 5 h to remove traces of solvent and oxygen. The vial was glass-sealed while kept under vacuum and then irradiated by a ⁶⁰Co source for a total dose of 50 kGy.

2.3 Preparation of MWCNT-PB-silica

MWCNTs were purified by HCl treatment as earlier reported (Merli et al., 2011; Speltini et al., 2010); the purified MWCNTs were grafted onto the silica particles by gamma radiation following the procedure previously described (Speltini et al., 2013b), to obtain MWCNT-PB-silica (CNT content 5 wt%).

2.4 Characterization of the synthesized materials

ATR FT-IR measurements were performed using a Perkin Elmer 1600 FT-IR equipped with a Specac Mk II Golden Gate system. A total of 256 scans were recorded over 400–4000 cm⁻¹ (resolution 2 cm⁻¹).

Scanning electron microscopy images of the powders were acquired using a Zeiss EVO^®-MA10-HR microscope. Particle size determination was performed on the raw powders, while morphology was studied on gold sputtered conductive samples. Particle size distribution curves were obtained by the Digital Micrograph^® 3.11.0 software. Specific surface area measurements on the samples were performed by a Sorptomatic 1990 instrument by Thermo Fisher Scientific. About 0.350 g of powders was charged in the glass sample holder and degassed at 100 °C for 60 h. Subsequently, samples were cooled down at −196 °C and 2 adsorption–desorption cycles followed by a last adsorption run were performed (B.E.T. method, analysing gas N₂, 50 points for run; blank done in He before the 1st adsorption run).

2.5 Liquid chromatography

The materials were slurry packed into 4.6 × 150 mm HPLC columns, as described elsewhere (Speltini et al., 2013b) and tested as stationary phases. A commercial Hypersil C18 column (4.6 × 150 mm, 5 μm) from Varian (Turin, Italy) was used for comparison. The chromatographic system consisted of a Shimadzu (Milan, Italy) LC-20AT solvent delivery module equipped with a DGU-20A3 degasser and interfaced with a SPD-20A UV detector. Injection volume was 20 μL; elution and detection details are reported in figure captions.

3.1 Characterization and properties of the obtained materials

The reaction between epoxidized PB and derivatized MWCNTs has been confirmed by IR spectroscopy. The ATR FT-IR spectrum, recorded on MWCNTs-R-NH₂ linked to the polymer, showed peaks attributable to the N—H bond (2842–2909 cm⁻¹) typical of amino and amide groups, a peak at 1683 cm⁻¹ characteristic of the amidic C⚌O bond, and two main signals at 1428 and 963 cm⁻¹ ascribed to PB (see Fig. S1).

The SEM images of the prepared materials (Fig. 1) show MWCNT bundles interlaying the silica particles, which act as supporting material and as spacers. As expected, no evident differences in the morphology were seen in the presence of unmodified or functionalized nanotubes. The particle size distributions of the (derivatized)MWCNT-based composites were determined by SEM. Results, reported in Table 1, indicate that the particle size was not significantly different from that of the starting silica (average size 5.4 μm).

Close

Figure 1

Scanning electron microscopy images of (a) MWCNT-PB-silica and of (b) MWCNT-R-NH₂-PB-silica microspheres: batch 1 (b), batch 2 (c) and batch 3 (d).

Export to PPT

From the BET results (Table 1), a decrease of the silica surface area (from ca. 25% to 40%) was observed after derivatization. These variations prove the effective functionalization of silica by MWCNTs through PB linking, with the MWCNT-PB network covering and partially filling the silica particles surface cavities. A sixfold difference in the pore volume between MWCNT-PB-silica and MWCNT-R-NH₂-PB-silica was also noticed, probably due to the distribution of the MWCNT functional groups in the silica pores.

The characterizations of the novel material showed a good reproducibility in the preparation procedure, as apparent from the results collected on three independent batches of MWCNT-R-NH₂-PB-silica.

3.2 Chromatographic tests

As above anticipated, the aim of this work was to highlight the possibility of selectivity/retention tuning in reversed-phase LC by using chemically modified CNTs. In particular, MWCNTs derivatized with amino-terminating alkyl chains with polar embedded groups were chosen in order to prepare a sort of mixed-mode material able to establish multiple interplays with the solutes, e.g. aspecific interaction, hydrophobic effect, π–π stacking, dipole–dipole, and electrostatic interaction (Speltini et al., 2012).

To begin with, the batch-to-batch reproducibility of the novel material MWCNT-R-NH₂-PB-silica was assessed by evaluating the capacity factors (k′) calculated on three columns packed with 3 batches of the material independently prepared starting from pristine silica (see Section 2.2). Relative standard deviations (RSDs) lower than 15% were obtained (see Table 2). Since the contribution of PB in terms of retention is negligible and the CNT pristine structure is preserved after gamma radiation (Merli et al., 2019; Speltini et al., 2013b), the k′ values experimentally obtained can be ascribed to the intermolecular interplays between solutes and MWCNTs. Good reproducibility in retention times was observed, with RSDs < 2% (n = 3).

The reversed-phase behaviour of MWCNT-R-NH₂-PB-silica has been investigated at first by testing a series of alkylbenzenes (Table 3). As it can be seen, the elution order is the same as for MWCNT-PB-silica, with increased retention in going from benzene to n-butylbenzene. These findings (mean values of three independent measurements) prove that the material owns reversed-phase characteristics (as expectable). However, remarkably lower k′ values were observed on the mixed-mode sorbent under the same elution conditions (see also Fig. S2). This indicates that MWCNT-R-NH₂-PB-silica is less hydrophobic than MWCNT-PB-silica, well highlighting the role of CNT chemical functionalization in addressing the retention capability.

In order to better evidence the behaviour of the novel MWCNT-R-NH₂-based material in comparison with unmodified MWCNTs and the conventional mono-modal C18 phase, three classes of analytes were tested. Specifically, barbiturates, steroid hormones and alkaloids, compounds of different molecular structure and polarity, were expected to exhibit different affinity for each of the three stationary phases. All the separations presented in the following have been obtained on each sorbent under the best elution conditions.

Barbiturates are usually separated on C18 columns using mixtures of water and ACN as the eluent phases (Cole, 2003), as also described in detail for routine analysis (http://www.chem.agilent.com/Library/applications/59682635.pdf). The four drugs tested resulted baseline-separated on the two MWCNTs-based columns, showing symmetric peaks (see Figs. 2a and 3a). The main difference between the two separations is the composition of the mobile phase. Indeed, in the case of MWCNT-R-NH₂-PB-silica only a little percentage of ACN (<10%) was necessary to achieve complete separation, requiring the same run time. This suggests that, despite the lower interaction with the sorbent, MWCNTs functionalization significantly increased the selectivity towards these amphiphilic molecules consisting of 2,4,6(1H,3H,5H)-pyrimidinetrione nucleus bearing aliphatic/aromatic substituents. The separation of these drugs on the conventional C18 column (Fig. 4a) was as well satisfactory in terms of resolution and run rimes, but higher percentage of organic eluent was necessary.

Close

Figure 2

Chromatograms obtained on the MWCNT-PB-silica column (4.6 × 150 mm) for a mixture of (a) barbiturates (20–30 mg L⁻¹): barbital (1), allobarbital (2) and its impurity (2′), phenobarbital (3), pentothal (4); (b) alkaloids (30–50 mg L⁻¹): ephedrine (1), colchicine (2), hyoscyamine (3), vincamine (4), strychnine (5), quinine (6); (c) sex hormones (30 mg L⁻¹): methandrostenolone (1), testosterone (2), medroxyprogesterone (3), progesterone (4), hydroxyprogesterone caproate (5), testosterone propionate (6), stanozolol (7); (d) testosterone esters (30 mg L⁻¹): acetate (1), propionate (2), isobutyrate (3), isocaproate (4), enanthate (5) cypionate (6), decanoate (7), undecanoate (8). Conditions: (a) ACN–water 0.8 mL min⁻¹, 18% ACN for 8 min, linear gradient to 27% ACN in 10 min, 214 nm; (b) ACN–25 mM H₃PO₄ 1 mL min⁻¹, 25% ACN for 3 min, linear gradient to 80% ACN in 30 min, 210 nm; (c) ACN–25 mM H₃PO₄ 1 mL min⁻¹, 30% ACN for 3 min, linear gradient to 60% ACN in 30 min, 230 nm; (d) ACN–25 mM H₃PO₄ 1.2 mL min⁻¹, 50% ACN for 9 min, linear gradient to 80% ACN in 30 min, 230 nm.

Export to PPT

Close

Figure 3

Chromatograms obtained on the MWCNT-R-NH₂-PB-silica column (4.6 × 150 mm) for a mixture of (a) barbiturates (20–30 mg L⁻¹); (b) alkaloids (30 mg L⁻¹); (c) sex hormones (30–70 mg L⁻¹); (d) testosterone esters (30 mg L⁻¹). For peak identification see Fig. 2 caption. Conditions: (a) ACN–water (8:92) 1 mL min⁻¹, 214 nm; (b) ACN–25 mM H₃PO₄ (28:72) 0.8 mL min⁻¹, 210 nm; (c) ACN–25 mM H₃PO₄ (37:63) 1.2 mL min⁻¹, 230 nm; (d) ACN–25 mM H₃PO₄ 1.2 mL min⁻¹, 45% ACN for 8 min, linear gradient to 60% ACN in 20 min, 230 nm.

Export to PPT

Close

Figure 4

Chromatograms obtained on the C18 column (4.6 × 150 mm) for a mixture of (a) barbiturates (20–50 mg L⁻¹); (b) alkaloids (30–50 mg L⁻¹); (c) sex hormones (30 mg L⁻¹); (d) testosterone esters (30 mg L⁻¹). For peak identification see Fig. 2 caption. Conditions: (a) ACN–water (35:65) 1 mL min⁻¹, 214 nm; (b) ACN–10 mM NaHCO₃ (90:10) 0.5 mL min⁻¹, 210 nm; (c) ACN–25 mM H₃PO₄ 1 mL min⁻¹, 40% ACN for 20 min, linear gradient to 80% ACN in 35 min, 230 nm; (d) ACN–25 mM H₃PO₄ 1 mL min⁻¹, 70% ACN for 10 min, linear gradient to 90% ACN in 30 min, 230 nm.

Export to PPT

Different chromatographic profiles and elution conditions were observed between the two MWCNTs-containing columns also for the separation of six alkaloids, chosen for bearing to different classes and having a different molecular structure. As apparent from Fig. 2b, the analytes were separated in ca. 30 min on MWCNT-PB-silica but with a marked band broadening and peak tailing for compounds 5 and 6. A gradient elution up to high percentages of ACN (80% v/v) was required to shorten analysis time, thus indicating a stronger interaction between nanotubes and the latest eluting compounds. On the contrary, despite partial overlapping for some analytes, MWCNT-R-NH₂-PB-silica allowed easy desorption in less than 10 min under isocratic elution (Fig. 3b), using far lower amounts of ACN. Results were better than those obtained on the C18 sorbent. Indeed, on the commercial column compounds 3, 4, and 5 showed large bandwidth and poor resolution, and peak tailing was observed for the two most retained analytes (Fig. 4b) working with ACN-10 mM HCO₃⁻ as the eluent (Qiu et al., 2008). This can be ascribed to the different properties of the C18 stationary phase compared to the MWCNT sorbents, as also evidenced by the different elution order observed on the commercial column.

The results so far presented highlight the different adsorption affinity of the two classes of different polarity analytes (barbiturates and alkaloids) for MWCNTs and for C18, and show the advantages of using functionalized MWCNTs.

The influence of chemical derivatization in tuning retention and selectivity is further corroborated by the results obtained in the separation of steroids, compounds having a cyclopentanoperhydrophenanthrene lipophilic core. In particular, various sex hormones (including natural and synthetic androgens, extrogens and progestin) were tested. As it can be seen comparing Figs. 2c and 3c, apart from the different elution order the best chromatographic profile was gained on MWCNT-PB-silica. This can be explained considering that these lipophilic molecules better interact with the unmodified nanotubes with respect to the more polar MWCNTs-R-NH₂. Indeed, the higher affinity of steroids for MWCNTs (indicated by the need of using an organic-rich eluent in the chromatographic elution) accounts for the improved resolution and peak shape.

Testosterone esters were well resolved on MWCNT-PB-silica (Fig. 2d); on MWCNT-R-NH₂-PB-silica, these compounds were better separated (Fig. 3d) than the steroids mixture (Fig. 3c). This can be ascribed to the interplays between derivatized CNTs and esters substituents on the steroid nucleus. As expected for these kind of molecules (Mesmer and Satzger, 1997), all the steroid compounds resulted well separated on the hydrophobic C18 column (Fig. 4c and d), although with organic-rich eluents and longer run times compared to MWCNT-PB-silica.

In agreement with previous work about the role of CNT functionalization (Hussain et al., 2009), the derivatization with polar functionality, as in our case, tunes the MWCNTs sorption properties improving selectivity for polar compounds, while the unmodified MWCNTs are better suited for the retention of apolar and aromatic compounds.