Graphene oxide (GO) is a single-layered nanosheet with epoxy, hydroxyl and carboxyl groups on the basal plane and edges of graphitic backbone [1-3], offering high surface area and abundant interaction sites with analytes. It has received tremendous attention [4,5] due to its novel physicochemical properties. Like other nanomaterials used in separation science [6], GO has been reported as a separation material in capillary electrochromatography [7-11] and capillary gas chromatography (GC) [12]. Qu and coworkers reported the use of GO as GC stationary phase, proving its potential for this purpose, although severe peak tailing was observed for alcohols [12]. In fact, there is great difficulty in preparing a GO capillary column with satisfactory efficiency since GO nanosheets tend to aggregate in dichloromethane used in GC column preparation. To address the problem, 3-aminopropyldiethoxymethylsilane (3-AMDS) [11,12] was used as a coupling agent for covalently bonding with GO in two steps, first, coating 3-AMDS onto the capillary and then coating GC nanosheets onto it.

Herein, we report a one-step column coating approach for fabrication of GO capillary column for GC separations. This proposed coating approach combines the two steps into one and makes the coating process more feasible and efficient and achieves an improved separation performance, showing advantages over the reported column coating method [12]. In this work, the GO capillary column was evaluated in terms of column efficiency, McReynolds constants, separation performance and reproducibility. 2. Experimental 2.1. Chemicals and instruments

All the analytes in this work were of analytical grade. GO was prepared following the modified Hummers method [13-15]. Untreated fused-silica capillary tubing (0.25 mm i.d.) was purchased from Yongnian Ruifeng Chromatogram Apparatus Company (Hebei, China). A commercial HP-INNOWAX capillary column purchased from Agilent Technologies (5 m long0.25 mm i.d.) was also used for comparison.

All the GC separations were carried out on a GC 7890A gas chromatograph with a flame ionization detector (FID) (Agilent Technologies, USA) under the following GC conditions: nitrogen of high purity (99.999%) as carrier gas at a flow rate of 1 mL/min, injection port at 250°C and FID at 300°C. Temperature programs for the separations of different samples are individually provided in their figure captions. 2.2. Capillary column fabrication

The capillary column coated with GO was fabricated as follows. First, a bare capillary column (5 m0.25 mm) was rinsed with dichloromethane for 10 min, heated at 100°C for 10 min, washed with 1.0 mol/L sodium hydroxide for 1.5 h and then with water until the eluate was neutral. After this, the column was heated at 120°C for 2.0 h. Then, a GO dispersion was prepared by dispersing 200mL GO aqueous solution (1 mg/mL) in a mixture of 100mL3-AMDS and 500mL ethanol under ultrasonication for 10 min. Next, the capillary was filled with the GO dispersion and maintained at room temperature for 1.0 h with both ends sealed. Afterwards, it was conditioned from 40°C to 170°Cat1°C/min and held at the high-end temperature for 5 h under nitrogen flow. Finally, the capillary column coated with GO nanosheets was obtained and used for the following GC separations. 3. Results and discussion 3.1. McReynolds constants and column efficiency

Column efficiency of the GO column was determined with naphthalene at 120°C. The determined height equivalent to a theoretical plate (HETP) was 0.74 mm, corresponding to the column efficiency of 1350 plates/m. McReynolds constants are an empirical measure of the polarity of a GC stationary phase for the characterization of the possible interactions of a stationary phase with analytes. They can be experimentally measured by determining the differences of five probes, namely benzene (X'), 1-butanol (Y'), 2-pentanone (Z'), 1-nitropropane (U') and pyridine (S'), in the retention indices on the given stationary phase and squalane. The sum and average of the retention index differences (ΔI) for the five probes correspond to a measure of the overall polarity and average polarity of a GC stationary phase, respectively. Table 1 shows the resulting McReynolds constants (ΔI)and their sum and average values of the GO capillary column, suggesting its moderately polar nature. As shown, it exhibits comparable Y'value (H-bond donor and acceptor) to that of the conventional Wax phase and lower S'(H-bond acceptor) and Z'(dipole-dipole interaction) values than the latter, suggesting that H-bonding and dipole-dipole interactions may play a major role in the retention of the GO stationary phase for analytes of different types.

Table 1

Table 1 McReynolds constants of GO and HP-INNOWAX capillary columns.

Table 1 McReynolds constants of GO and HP-INNOWAX capillary columns.

Separation performance of the GO capillary column was evaluated by GC separations of analytes of different functionalities, including n-alkanes, aromatic hydrocarbons, alcohols, amines and a mixture of 11 analytes. The corresponding GC separation chromatograms are shown in Figs. 1-3, respectively.

Fig. 1 describes the GC separations of n-alkanes (a), aromatic hydrocarbons (b) and alcohols (c) on the GO column, respectively, showing the good separation performance of GO stationary phase for analytes from non-polar to polar. Especially, the GO stationary phase exhibited good peak shapes for alcohols that are known to be prone to peak tailing in GC analysis due to their strong H-bonding tendencies with a polar stationary phase. This suggests moderate interactions with alcohols as well as good column inertness. Based on this finding, we proceeded to further investigate the separation performance of GO capillary column for other H-bonding analytes, such as amines.

Fig. 1

Download: JPG larger image

Fig. 1. GC separations of n-alkanes (a) aromatic hydrocarbons (b) and alcohols (c) on GO capillary column. Peaks for (a): (1) n-nonane, (2) n-decane, (3) n-undecane, (4)ndodecane, (5) n-tridecane; (6) n-tetradecane, (7) n-pentadecane and (8) n-hexadecane; for (b): (1) naphthalene, (2) biphenyl, (3) fluorene, (4) phenanthrene and (5) fluoranthene; for (c): (1) 1-propanol, (2) 1-butanol, (3) 1-pentanol, (4) 1-hexanol, (5) 1-heptanol, (6) 1-octanol, (7) 1-nonanol, (8) 1-decanol, (9) 1-undecanol and (10) 1-dodecanol. Temperature programs: (a) 40°C (1 min) to 130°Cat10°C/min, (b) 40°C (1 min) to 160°Cat10°C/min and (c) 40°C (1 min) to 140°Cat10°C/min.

Fig. 2 shows the GC separations of amines on GO (a) and conventional Wax (b) capillary columns, respectively. Amines are proto n-acceptor and liable to peak tailing in GC analysis. As can be observed, the GO column exhibited good peak shapes, especially for di- n-hexylamine (peak 2) and dodecylamine (peak 3). Dodecylamine had a shortened peak height on the commercial column, suggesting the possible existence of active sites on the indicated column. The above results demonstrated the good separation performance of the GO stationary phase for H-bonding analytes, such as alcohols and amines.

Fig. 2

	Download: JPG larger image
Fig. 2. GC separations of amines on GO (a) and HP-INNOWAX (b) capillary columns. Peaks: (1) dimethylformamide, (2) di- n-hexylamine, (3) dodecylamine, (4) di-noctylamine. Temperature programs: (a) 40°C (1 min) to 150°Cat10°C/min and (b) 40°C (1 min) to 200°Cat10°C/min.

Fig. 3 presents the GC separations of a mixture of 11 analytes on GO (a) and conventional Wax (b) capillary columns. As shown, the GO stationary phase achieved baseline resolution for all the analytes and exhibited different retention behaviors for some of the analytes from the conventional Wax phase. Specifically, the GO stationary phase completely resolved chlorobenzene (peak 2) and n-dodecane (peak 4), which were partially separated on the conventional phase (R=0.66). Of particular note, the nonpolar analytes, such as n-dodecane (peak 4) and methyl nonanoate (peak 7) are retained longer on the GO column, which eluted earlier on the Wax column and resulted in the switched elution order of bromobenzene/n-dodecane (peaks 3/4) and 1,2-dichlorobenzene/benzonitrile (peaks 5/6), demonstrating their different separation performance and retention behaviors. Additionally, separation reproducibility of GO column in retention times of analytes was examined by separation of the mixture of 11 analytes. The obtained relative standard deviations (RSD%) were in the range of 0.03%-0.24% (n = 5), indicating the good reproducibility of GO column for the GC separations.

Fig. 3

	Download: JPG larger image
Fig. 3. GC separations of a mixture of 11 analytes on GO (a) and HP-INNOWAX (b) capillary columns. Peaks: (1) toluene, (2) chlorobenzene, (3) bromobenzene, (4) n-dodecane, (5) 1,2-dichlorobenzene, (6) benzonitrile, (7) methyl nonanoate, (8) naphthalene, (9) aniline, (10) 3-toluidine and (11) biphenyl. Temperature program: 30°C (1 min) to 140°Cat10°C/min.

This work describes the separation performance of GO nanosheets as a capillary GC stationary phase. The GO column fabricated by a new single-step coating approach exhibited different resolving ability from the conventional Wax column and achieved good peak shapes for H-bonding analytes, such as alcohols and amines that are prone to peak-tailing in GC analysis. These differences may originate from their different molecular interactions. The GO stationary phase exhibits moderate H-bonding and dipole-dipole interactions with H-bonding analytes and offers strongerp-pinteractions with aromatic analytes than the latter.

The comprehensive result of these molecular interactions distinguishes the GO stationary phase from the conventional stationary phase in term of retention behaviors and resolving ability.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (Nos. 21075010, 21174019) and the 111 Project B07012 in China.