Preparation, characterization, and performance evaluation of UiO-66 analogues as stationary phase in HPLC for the separation of substituted benzenes and polycyclic aromatic hydrocarbons

UiO-66 analogues are good candidates as stationary phase in HPLC because of their chemical/thermal stability, large surface area, and two cage structures. Here, two UiO-66 analogues, UiO-66-NH2 and UiO-67, were synthesized and used as stationary phase in HPLC to evaluate their performance in the separation of substituted benzenes (SBs) and polycyclic aromatic hydrocarbons (PAHs). The results showed that SBs could be well separated on UiO-66-NH2 column but not on UiO-67 column. Nonetheless, PAHs could be well separated on UiO-67 column. The separation mechanisms of SBs and PAHs on UiO-66 analogues may be involved in the pore size and functional group in the frameworks of UiO-66 analogues. Introduction of the–NH2 into UiO-66 significantly reduced its adsorption capacity for SB congeners, which resulted in less separation of SBs on UiO-66-NH2. As for the separation of PAHs on UiO-67 column, the π-π stacking effect was supposed to play a vital role.


Introduction
Metal-organic frameworks (MOFs) are a category of complexes constructed by organic ligands and inorganic nodes [1]. Most MOFs have infinite network structures and specific properties, such as large surface area, high adsorption affinity, accessible cages, tunable pore dimension, and excellent thermal stability [2,3]. Consequently, MOFs have aroused increasing research interests for their potential application in many fields [4][5][6][7]. Particularly, because of their pore dimension tunability and in-pore functionalization without topological change, MOFs endow great potential for application in analytical chemistry [8]. Recently, several MOF materials, including MOF-5, MIL-101(Cr), MIL-47, ZIF-8, and HKUST-1, have been tentatively applied a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 in chromatography [9][10][11][12][13]. Especially, due to the considerable difference of MOFs in topological structure, the effects of pore sizes in MOFs on chromatographic separation have been intensively studied [13,14]. In most cases, excellent separation was achieved based on a combination of molecular sieving and adsorption effects. However, due to their potential instability in most solvents, MOFs were only limitedly used in high performance liquid chromatography (HPLC) [15,16].
UiO-66 is Zr-cluster-based MOF consisting of cationic Zr 6 O 4 (OH) 4 nodes and organic linkers. Due to the combination of strong Zr-O bonds, the UiO-66 possesses high thermal stability and broad solvent resistance [15]. Because of these physicochemical properties, UiO-66 has been used in chromatographic separation process [14,17]. In our previous study, UiO-66 was slurry-packed in a chromatographic column and used as stationary phase to separate substituted benzenes (SBs) and polycyclic aromatic hydrocarbons (PAHs) in NP-and RP-HPLC [17]. However, the mechanism for the separation of SBs or PAHs by UiO-66 as well as the structure-performance relationship of UiO-66 analogues remains little understood. In the present study, two UiO-66 analogues, UiO-66-NH 2 and UiO-67, were used as stationary phase in HPLC to separate SBs and PAH, and the structure-performance relationship of UiO-66 analogues was assessed.
The amino-functionalized metal-organic framework UiO-66-NH 2 and the linker-lengthened metal-organic framework UiO-67 are isoreticular MOFs both derived from UiO-66. The UiO-66-NH 2 was obtained by replacing the UiO-66 organic linker H 2 BDC with amino-functionalized linker H 2 BDC-NH 2 , and the UiO-67 was obtained by replacing H 2 BDC with longer linker H 2 BPDC [18,19]. The presence of -NH 2 in UiO-66-NH 2 endows it with some special properties and further results in some potential application, such as photocatalysis in oxidation/reduction processes and adsorption of nitrogen-containing compounds [20][21][22][23][24][25][26]. While considering the high surface area, big pore volume and large pore size of UiO-67 due to its long linker, UiO-67 was mostly used in adsorption of small molecules, such as CO 2 , CH 4 , and H 2 [27][28][29]. Both UiO-66-NH 2 and UiO-67 have similar topological framework, which is also similar to that of UiO-66. For example, the UiO-66, UiO-66-NH 2 and UiO-67 have the same metal coordination, similar secondary building unit (SBU), and similar interlinking mode of SBUs. The three MOFs have a certain solvent stability property, which is essential for the potential application in HPLC separation. However, the UiO-66, UiO-67 and UiO-66-NH 2 has different pore size from each other, which enables an investigation of the effects of pore size on the performance of MOFs in chromatographic separation. In the present study, UiO-66-NH 2 was applied for NP-and RP-HPLC separation, whereas UiO-67 was used only for NP-HPLC separation due to its lower solvent stability. What's more, the structure-performance relationship of UiO-66 analogues in the same topology structures was demonstrated.

Instrumentations
The powder X-ray diffraction (PXRD) patterns of UiO-66 analogues were obtained on a BRU-KER D8-Focus Bragg-Brentano X-ray Powder Diffractometer equipped with a Cu sealed tube (λ = 1.54178 nm) at room temperature. Thermogravimetric analysis (TGA) was performed on a TGA-50 (SHIMADZU) thermogravimetric analyzer with a heating rate of 2˚CÁmin -1 under N 2 atmosphere. The N 2 adsorption isotherms were measured on a Micromeritics ASAP2020 surface area and pore analyzer. The scanning electron microscopy (SEM) images were recorded on a Shimadzu SS-550 scanning electron microscope at 15.0 kV.
The MOFs-packed columns were prepared using a HY-HPLC-M chromatographic column packing equipment (Singapore Hydratech Industries Pie Ltd., Beijing, China). HPLC separations were performed on a LC-20AT chromatograph equipped with a tunable UV-Vis detector, an auto-sampler, a binary pump, a column oven, and a LC-solution chromatography data system (Shimadzu, Kyoto, Japan).

Synthesis and activation of UiO-66 analogues
UiO-66 was synthesized and activated following to the methods described in our previous study [17]. Briefly, 0.932 g of ZrCl 4 , 1.32 g of terephthalic acid, 0.665 mL of concentrated hydrochloric acid, and 24 mL of DMF were mixed in a Teflon-lined bomb. After ultrasonic treating for 5 min, the bomb was sealed and then heated at 120˚C for 16 h. The white solid product was obtained after cooling down and centrifuging at 8000 rpm for 10 min. The product was washed three times sequentially with DMF and acetone, and then dried at 60˚C.
UiO-66-NH 2 was synthesized in the same way as that of UiO-66 but replacing H 2 BDC with H 2 BDC-NH 2 , according to the method described by Zhang et al. [30]. The product UiO-66-NH 2 was activated using solvent exchange method (washed with DMF and acetone in turn). UiO-67 was synthesized by adding 0.12 g of ZrCl 4 , 0.125 g of H 2 PDC, and 1.83 g of H 2 BDC into 20 mL of DMF in a glass vial. The vial was sealed and then heated at 100˚C for two days. The resultant white powder was collected and washed with DMF and acetone in sequence for three times, and then dried at 80˚C.

Preparation of packed columns for HPLC analysis
The UiO-66 packed column for RP-and NP-HPLC separation was prepared following the method as described in our previous study [17]. For the preparation of UiO-66-NH 2 packed column, 1.20 g of UiO-66-NH 2 was dispersed in 50 mL n-hexane under ultrasonication for 5 min. The resultant suspension was then downward packed into a stainless steel column (5 cm long, 4.6 mm i.d.) under 5000 psi for 15 min. The suspension was kept intact by the stainless steel frit with a pore size of 2 μm. The packed column used for NP-HPLC was conditioned with n-hexane at a flow of 1.0 mLÁmin −1 for 1 h. The packed column used for RP-HPLC was conditioned with methanol at a flow of 0.5 mLÁmin −1 for 1 h. The UiO-67 packed column was prepared as follows: the well-dispersed suspension of 1.30 g of UiO-67 in 50 mL n-hexane was downward packed into a stainless steel column (5 cm long, 4.6 mm i.d.) under 5000 psi for 15 min. The UiO-67 packed column was assembled in a similar way to that of the preparation of UiO-66-NH 2 packed column except that1.3 g of UiO-67 rather than 1.2 g of UiO-66-NH 2 was used.

HPLC separation experiments
The prepared packed column was connected to the HPLC system to start the HPLC separation experiments under the indicated chromatographic conditions. The RP-HPLC separations using UiO-66-NH 2 as the stationary phase were performed using MeOH/H 2 O as mobile phase at a flow rate of 0.5 mLÁmin −1 , with UV detection at 254 nm and column temperature at 30˚C. The NP-HPLC separation experiments were conducted using UiO-66-NH 2 and UiO-67 packed columns, with n-hexane/DCM as mobile phase at a flow rate 1.0 mLÁmin −1 , UV detection at 210 nm. Data acquisition and processing were carried out using a LC-solution chromatography data system.

Thermodynamic analysis
Several parameters calculated by a van't Hoff model, Gibbs free energy change (ΔG), enthalpy change (ΔH) and entropy change (ΔS), were used to evaluate the thermodynamic properties of the HPLC separation [31].

Effects of mobile phase on the separation.
Mobile phases composed of MeOH and H 2 O at different ratios were used to determine the performance of UiO-66-NH 2 in RP-HPLC separation of benzene and toluene, or ethylbenzene (EB), styrene and xylenes (Fig 2). It was found that the increase of water proportion in the mobile phases from 0% to 10% resulted in a less retention time of SBs on UiO-66-NH 2 packed column. However, further increase of water resulted in the increase of retention time. The decreased retention time of SBs on UiO-66-NH 2 column with increasing H 2 O from 0% to 10% in mobile phases is most likely due to the high affinity of the open-metal site in UiO-66-NH 2 for lone pair of electrons in water molecule [32]. Water molecules in the mobile phase could occupy the open metal sites in UiO-66-NH 2 , reducing the interactions between SB analytes and UiO-66-NH 2 and resulting in a short elution time. However, with the proportion of H 2 O increased from 10% to 50% in mobile phase, the neutral SB molecules had a poor solubility in the strong polar solvent based on the similar dissolve mutually theory and consequently resulted in a longer retention time on UiO-66-NH 2 column.
Different ratios of the mixtures of at n-hexane and DCM were also used as mobile phases to investigate the performance of UiO-66-NH 2 in NP-HPLC separation of SBs. When n-hexane alone was used as a mobile phase, the SB congeners were eluted in the way of long retention time, poor resolution and broad peak. Adding of DCM even in a little amount to n-hexane would greatly improve the resolution and dramatically decrease the retention time. This phenomenon might be explained by the competitive adsorption theory [33,34]. When n-hexane/ DCM (97:3) was used as mobile phase, baseline separations of EB, styrene and o-xylene, as well as benzene and toluene were achieved (Fig 3). However, further increase of DCM content in n-hexane/DCM mixtures would remarkably decrease resolution and shorten retention time. Particularly, 10% or more of DCM in n-hexane/DCM would lead to the co-elution of analytes.

RP-HPLC separation of SBs on UiO-66-NH 2 .
Compared to the performance of SBs on UiO-66 column [17], SBs on UiO-66-NH 2 column could not be efficiently separated irrespective of the different retention times between SBs. The retention times of three xylene Separation of hydrocarbons on UiO-66 analogues packed columns in HPLC isomers on UiO-66-NH 2 column followed the order p-xylene < m-xylene < o-xylene, which was the same to the order on UiO-66 column [17]. However, the retention times of the three xylene isomers on UiO-66-NH 2 column were shorter than the corresponding retention times of each xylene isomer on UiO-66 column under the same MeOH/H 2 O. Since xylene isomers are neutral compounds, their adsorptions on UiO-66-NH 2 and UiO-66 might be attributed to the action of van der Waals force [25]. Because of the partial blocking of pores by-NH 2 groups, the surface area of UiO-66-NH 2 (1070 m 2 Ág -1 ) was smaller than that of UiO-66 (1200 m 2 Ág -1 ), which led to weaker van der Waals force between xylenes and UiO-66-NH 2 and resulted in faster elution (i.e., shorter retention times) of xylenes on UiO-66-NH 2 column than on UiO-66 column.

NP-HPLC separation of xylene isomers on UiO-66-NH 2 .
On UiO-66-NH 2 packed column, o-xylene could be separated from p-and m-xylene, while the separation of m-and pxylene was failed. Similar to UiO-66, UiO-66-NH 2 has a preferential affinity toward o-xylene. This might be explained in two aspects [17]. Firstly, O-xylene which has the largest molecular size among SBs had the strongest Van der Waals interaction with the UiO-66-NH 2 framework [14], and consequently retained longer on the UiO-66-NH 2 packed column than the other SBs did. Secondly, the two methyl groups in o-xylene had a strong interaction with carboxylate groups in UiO-66-NH 2 , which might also result in a longer retention of o-xylene on UiO-66-NH 2 packed column [35].
Unlike the performance of o-xylene on UiO-66 column, the baseline separation of m-xylene and p-xylene failed on UiO-66-NH 2 column [17]. This might be due to the lower porosity as the pores in UiO-66-NH 2 could be partially blocked by-NH 2 . As a result, the surface area decreased and the interaction between xylenes and UiO-66-NH 2 reduced . In addition, only one methyl group was available in m-and p-xylene and the methyl group could similarly interact with the carboxylate groups in UiO-66-NH 2 , which made m-and p-xylene be co-eluted from UiO-66-NH 2 [35].

NP-HPLC separation of EB and styrene on UiO-66-NH 2 .
EB and styrene could be separated on UiO-66-NH 2 packed column. And the retention time of styrene was shorter than EB. The retention order of EB and styrene on UiO-66-NH 2 is opposite to that on UiO-66 [17] (Fig 3). The preference of UiO-66 for styrene over EB was observed on other terephthalic linked MOFs, such as MIL-47 and MIL-53 [36]. The less preference to styrene than EB of UiO-66-NH 2 might be ascribed to its geometry configuration [37]. Styrene had a smaller molecule size (0.96×0.70×0.34 nm) than EB (0.95×0.67×0.53 nm), which enabled styrene to be easy to escape from the trapping of the small pores of UiO-66-NH 2 . Therefore, it was expected that the retention time of styrene on UiO-66-NH 2 column was shorter than that of EB.

Evaluation of separation performance on UiO-66-NH 2 .
The reproducibility and selectivity of UiO-66-NH 2 packed column for the NP-HPLC separation of SBs were evaluated using two blends: benzene and toluene, and EB, styrene and o-xylene (Fig 3, S1 Fig, S1 and S2 Tables). The relative standard deviations (RSD) of retention time (t R ), peak area, peak height, and half peak width (W 1/2 ) for five replicates ranged from 0.06 to 0.08%, 0.15 to 0.98%, 0.20 to 0.88%, and 0.17 to 0.40%, respectively. Therefore, it was concluded that the UiO-66-NH 2 packed column gave good reproducibility for NP-HPLC separation of SBs. Also, it was found that there was a linear increase of the chromatographic peak area and peak height with the increase of mass of analytes (S2 Fig). However, the retention time and selectivity were independent of the mass of analytes (S2 Table). Nonetheless, the column efficiency generally decreased with the increase of analyte mass (S3 Fig).

NP-HPLC separation of SBs on UiO-67.
To evaluate the separation performance of UiO-67 as stationary phase, one-component injected analysis of SBs using different ratios of n-hexane/DCM (100:0, 95:5, 90:10, 85:15, 80:20, 70:30, 40:60, v/v) as mobile phases were investigated (S3 Table, S4 Fig). It was found that SB analytes almost coeluted within a short time (approximately 0.6 to 2.2 min) and displayed the expected insensitivity of retention time to mobile phase composition. On the contrary, the separations of SBs on UiO-66 packed column were successful in terms of the reverse shape selectivity, size selectivity and stacking effect as results of MOFs cages and van der Waals force [17]. The kinetic diameter of SBs (5.85-7.4 Å) is comparable to the sizes of UiO-66 cavities (8 to 11 Å) [19]. Therefore, SB molecules could enter the cavities of UiO-66 and interacted with cavity walls, resulting in an effective separation of SBs. However, the sizes of UiO-67 cavities (12 to 16 Å) are much bigger than the kinetic diameters of SBs [38]. SB molecules could pass through the cavities of UiO-67 through diffusion, enabling the equilibration and desorption of SBs on UiO-67 column to be easy and fast [13]. Therefore, it is proposed that the cavity sizes of MOFs made great contribution to the chromatographic separation of SBs.

NP-HPLC separation of PAHs on UiO-67.
In order to further explore the roles of pore size of UiO-66 analogues in chromatographic separation, NP-HPLC separation of large molecule PAHs on UiO-67 column were studied. Several PAHs with different numbers of rings, including naphthalene (two rings), fluorene, acenaphthene, anthracene and phenanthrene (three rings), pyrene and chrysene (four rings) were used. Also, different ratios of nhexane/DCM were used as mobile phases. Based on the one-component injected experiments, two PAH mixtures: (1) naphthalene, anthracene and chrysene; and (2) naphthalene, phenanthrene and chrysene were used for NP-HPLC separation on the UiO-67 column.
The ratio of n-hexane/DCM showed significant effect on NP-HPLC separation of PAHs on UiO-67 column (Fig 4). When n-hexane alone was used as mobile phase, the PAHs in both mixtures could generally be separated with long elution time needed and a poor peak pattern observed. Addition of 5% DCM to n-hexane (n-hexane/DCM = 95:5) remarkably improved the separation and baseline separations (R > 1.5) of PAHs in both mixtures, and the elution time of PAHs became shorter compared to that using n-hexane alone as mobile phase. However, further increase of DCM in n-hexane/DCM mixtures would lead to inefficient separation of PAHs, and even resulted in co-elution.
As shown in Fig 4, the retention order of PAHs on UiO-67 was as follows: PAHs with four rings > PAHs with three rings > PAHs with two rings. The π-π interactions between PAHs and aromatic framework walls of UiO-67 might be strong and dominantly important for the separation of PAHs. The more rings PAHs have, the stronger interactions between PAHs and the framework walls, which resulted in a longer retention time. It worth noting that inverse retention orders of PAHs on UiO-66 column (5 cm long, 4.6 mm i.d.) have been reported: PAHs with four rings < PAHs with three rings < PAHs with two rings [17]. This discrepancy might be ascribed to the size-exclusion effect. When UiO-66 was used as stationary phase to separate chrysene, fluorene and naphthalene, the PAHs with larger molecule sizes could be more likely to be excluded from the pores of UiO-66 and hence be eluted in shorter time.

Thermodynamics of HPLC separation on UiO-66 analogues
Thermodynamics of HPLC separation of analytes on UiO-66-NH 2 and UiO-67 were investigated under different temperatures. The mixture of EB, styrene and o-xylene and the mixture of benzene and toluene were separated on UiO-66-NH 2 column with temperatures ranged from 20˚C to 60˚C (S5 Fig), while the mixture of naphthalene, anthracene and chrysene and the mixture of naphthalene, phenanthrene and chrysene were separated on UiO-67 column with temperatures ranged from 20˚C to 40˚C (S6 Fig). As the column temperature increased, the retention time of analytes on either UiO-66-NH 2 or UiO-67 decreased, indicating the involvement of the exothermic HPLC separation process. It is interesting that the selectivity of analytes on UiO-66-NH 2 column slightly decreased as the temperature increased (S4 Table). However, the column temperature had no significant effect on the selectivity of analytes on UiO-67 column (S5 Table).
The van 0 t Hoff equation was used to explain the relationship between retention factor k' and column temperature [31]. A good linearity between lnk' and 1/T suggested that there was no change in the mechanism for the NP-HPLC separation of SBs on UiO-66-NH 2 at the temperature range of 20 to 60˚C (Fig 5). The thermodynamic parameters obtained from the van't Hoff equation for the transfer of analytes from mobile phase to the UiO-66-NH 2 stationary phase were summarized and presented in Table 1. The negative values of ΔG indicated that the analytes transfer processes on UiO-66-NH 2 were thermodynamically spontaneous. The smaller negative values of ΔG indicate the more favorable transfer of the analytes from the mobile phase to the stationary phase and thus the longer retention time on the stationary phase. In addition, comparative analysis of ΔH and ΔS on different analytes showed that the values of ΔH of different analytes were comparable but the values of ΔS varied considerably. Thus, the NP-HPLC separations of these analytes on UiO-66-NH 2 stationary phase were controlled only by positive ΔS.

Conclusions
The performances of UiO-66-NH 2 and UiO-67 packed columns for HPLC separation of SBs and PAHs were evaluated in the present study. Parameters influencing HPLC separation, such as mobile phase composition, analyte mass, and column temperature were determined. Although not all analytes could be efficiently separated on UiO-66-NH 2 column especially for RP-HPLC separation, the efficient separation of EB, styrene and o-xylene is applicable, suggesting a potential application of UiO-66-NH 2 as stationary phase in NP-HPLC. To some extent, UiO-67 can be effectively used as stationary phase in NP-HPLC separation of molecules with relatively large diameter. Additionally, the functional groups and pore size of MOFs play important roles in the separation of organics on the HPLC instrument coupling with MOFs packed column.  Separation of hydrocarbons on UiO-66 analogues packed columns in HPLC S1