Metal-organic framework and Tenax-TA as optimal sorbent mixture for concurrent GC-MS analysis of C1 to C5 carbonyl compounds

We report a multi adsorbent-based method using combinations of metal-organic frameworks (MOFs) and a commercial sorbent Tenax-TA for sampling and thermal desorption (TD) gas chromatography-mass spectrometry (GC-MS) quantification of mixtures of six (C1 to C5) aldehydes. The feasibility of this approach was demonstrated along with the optical analytical conditions for maximum recovery. Optimal TD conditions for adsorption and desorption of aldehydes using MOF-5 (Zn-based MOF)+ Tenax-TA were determined as −25 °C and 150 °C, respectively (purge volume: 100 ml). These conditions yielded good linearity (R2 = 0.997), precision, and high sensitivity. Analysis of the aldehyde mixtures yielded slightly smaller R2 values than the analysis of single species. Additionally, the performance of MOF-5+ Tenax-TA was compared with other combinations comprising of Cu-based MOF-199 and Zr-based MOF of UiO-66 topology. The results of the theoretical modelling analyses propose simultaneous interaction of the C=O group of aldehydes with open metal sites of the studied MOFs and van der Waals interaction of hydrocarbon “tail” of aldehydes with linkers of MOFs. The combined interactions significantly increased the enthalpy (eV/molecule) of formaldehyde adsorption on MOF. Our findings unravel a potential way to extend the application of GC-based detection toward concurrent analysis of organic molecules of variable sizes.

Carbonyl compounds (CCs), which consist of a carbon atom double bonded to an oxygen atom (C=O), are ubiquitous in the troposphere and play a significant role in secondary photochemical reactions 17 . Formaldehyde (FA) and acetaldehyde (AA) are classified as toxic air pollutants due to their adverse effects on public health and the environment 18 . Carbonyl compounds (except FA) are most commonly analyzed by trapping the compounds onto commercial sorbents and transferring them into a thermal desorption unit connected to a GC or GC-MS system 19 . FA, on the other hand, cannot be quantified by trapping on commonly available sorbent materials. Analysis of FA generally requires complex derivatization for stable sampling and accurate analysis for any kind of chromatographic system (e.g., GC or LC). Derivatization using the 2,4-dinitrophenylhydrazine (DNPH) cartridge method is frequently used to convert low molecular weight FA to a high molecular weight complex for analysis using HPLC-UV methods 20 . Likewise, mixtures of CCs containing FA cannot be quantified using a TD-GC/MS system due to the same sorptive limitations of commercial sorbents.
The reliability of MOF-5 (organic linker: 1,4-benzenedicarboxylates and metal cluster: Zn 4 O 13 ) as a sorptive media for the collection of FA (and the subsequent analysis by TD-GC/MS) has been demonstrated previously 21,22 . Nonetheless, the use of such media for simultaneous quantification of low and high molecular weight C1 to C5 aldehydes has not been investigated. In this work, we developed a multi-adsorbent TD-GC/MS method to analyze a mixture of environmentally important carbonyl species, i.e., formaldehyde (FA), acetaldehyde (AA), propanaldehyde (PA), butaraldehyde (BA), isovaleraldehyde (IA), and valeraldehyde (VA). A new sorption method was developed to absorb both light and heavier carbonyls simultaneously by combining MOF-5 and a well-known commercial sorbent, Tenax-TA. Mixtures of CCs were analyzed using sorbent tubes (ST) and cold traps (CT) packed with MOF and Tenax-TA. The MOFs for this study were chosen based on their high thermal, mechanical, and chemical stability: zinc (Zn)-based MOF-5, copper (Cu)-based MOF-199, and zirconium (Zr)-based UiO-66 (amine derivative). All three MOFs are stable (250 to 300 °C) at the operating temperatures in TD-GC/MS. MOF-199, which contains both copper (Cu 2+ ) ions and open metal sites, exhibits greater chemical stability than MOF-5, which contains zinc (Zn 2+ ) tetrahedrons. Likewise, although zirconium (Zr)-based MOFs (of UiO-66 topology) are thermally less stable than Zn-based MOF (MOF-5), they exhibit greater mechanical stability than MOF-5 and MOF-199. The reliability of the analytical system was evaluated by measuring basic quality (QA) assurance and quality control parameters. Furthermore, the outcomes of our experiments were discussed in light of theoretical interactions between MOFs and the main active site of aldehyde (C=O group) molecules based on density functional theory (DFT). As such, this study reveals for the first time the use of MOF-based multi-adsorbents in sorbent tubes (at room temperature) and thermal desorption-based analysis of a mixture of carbonyl compounds with varying molecular weights.

Results and Discussion
Characterization of synthesized MOFs. The morphology, degree of crystallinity, thermal stability, and chemical functionalities of synthesized MOFs were determined through PXRD, SEM, FTIR, and TGA analyses. The obtained PXRD data of MOF-5 (see Supplementary Fig. S1) were consistent with the previously published results (See Gu et al. 21 ). The relative peak intensities of activated MOF-5 showed characteristic diffraction peaks at 2Ѳ values of 6.7° and 9.6°. Likewise, the PXRD patterns of UiO-66-NH 2 (diffraction peaks at 2Ѳ of 7.2° and 8.3°) (see Supplementary Fig. S2) and    [23][24][25] . The FE-SEM morphologies of the synthesized MOFs confirmed the formation of crystals with particle sizes ranging from 150 to 300 nm.
The two sharp FTIR bands of MOF-5 located in the region of 1650-1300 cm −1 were attributed to the attachment of the carboxylate ligand to the Zn 4 O cluster 26 . The FTIR bands of pristine and used MOF-5 were consistently demonstrating minimum disruption of metal-ligand binding by the adsorption and subsequent desorption of carbonyls ( Supplementary Fig. S4). The peaks centered at 3500 cm −1 indicated OH groups involved in H-bonding 27 . FTIR peaks in the region of 3100-2900 cm −1 indicated the aromatic and aliphatic ν(CH) stretching vibration of benzene rings and DMF. The relatively shorter bands located between 800-1200 cm −1 indicated Zn-O stretching. FTIR peaks of the MOF-199 and the amine derivative of UiO-66 were in good agreement with previously published results. MOF-199 showed two peaks at 1108 and 760 cm −1 , indicating C-O-Cu stretching vibrations 28 . The symmetric and asymmetric stretching vibrations of the carboxylate groups were observed at 1554, 1447, and 1367 cm −1 . UiO-66(NH 2 ) showed characteristic bands at 1257 and 1383 cm −1 , representing C-N stretching 29 . IR peaks at 765 and 660 cm −1 were attributed to N-H vibrations.
The TGA of MOF-5 was in excellent agreement with the previously reported results of Gu et al. 21 . The TGA results showed an initial weight loss of 12% from 100 to 300 °C, representing the loss of DMF molecules in the MOF pore space. This was followed by another weight loss of 35% from 300 to 500 °C due to the breakdown of the MOF structure. Thus, the activated MOF-5 was found to be thermally stable at the operation temperatures (<200 °C) used in this study. Similarly, UiO-66(NH 2 ) and MOF-199 showed good thermal stability (demonstrated by less than 15% loss of weight due to removal of solvent molecules) below 200 °C.
Quantification of a mixture of CCs (including FA) using multi-adsorbent-based TD-GC/MS. In this study, L-WS of FA (on its own) and a mixture of 6 CCs (including FA as a component of the mixture) ( Table 1) were analyzed using a multi-adsorbent (MOF and Tenax-TA)-based TD-GC/MS system ( Table 2). The use of MOF-5 for the analysis of FA was investigated previously by Gu et al. 21 and Kim et al. 22 . Gu et al. 21 developed an in-field sampling and pre-concentration technique for the TD-GC/MS determination of atmospheric FA without requiring chemical derivatization of the FA molecules 22 . Kim et al. 22 determined the optimal analytical conditions for maximum FA recovery. However, the application of MOFs to the analysis of a mixture of CCs has Scientific REPORTS | (2018) 8:5033 | DOI:10.1038/s41598-018-23391-6 not been reported previously. This could be because high molecular weight carbonyls like AA, PA, BA, IA, and VA can be easily quantified using various commercially available sorbents. However, in the presence of FA, the commercial sorbents become ineffective, requiring more advanced techniques. Combining commercial sorbents with MOFs (e.g., MOF-5), which has strong adsorptivity for FA, overcame the sorptive limitations of commercial sorbents at very low molecular weights (Refer to Fig. 1 and Supplementary Fig. S5 for the experimental scheme). Thus, the combined use of MOFs with commercial sorbent enabled simultaneous quantification of low and high molecular weight CCs in the TD-GC/MS system. All six CCs were detected simultaneously using the MOF-based multi-adsorbent technique. To the best of our knowledge, this is the first study that demonstrated the use of MOFs for the quantification of mixtures of CCs (including FA).

Optimum analytical conditions for recovery of mixtures of carbonyls.
Factors influencing the analysis of individual CCs (FA, AA, PA, BA, IA, VA) in a mixture were examined using a system with MOF-5 and Tenax-TA-based sorbent tube (ST) sampling, a pre-concentration step, and a TD-GC/MS (Table 3). We collected a mixture of FA, AA, PA, BA, VA, and IA in a ST packed with MOF-5 and Tenax-TA and analyzed the concentration of each compound in a TD-GC/MS system. The quartz trap (of the thermal desorption system) packed with MOF-5 and Tenax-TA was used for subsequent adsorption and desorption of CCs before analysis in the GC/ MS system. The optimal analytical conditions for the recovery of carbonyls were determined by comparing the recovery of the CCs at different CT and ST adsorption and desorption temperatures and at different ST pre-purge conditions. The results of the analysis of CCs are summarized in Table 4.    Optimal TD conditions for the analysis of FA (in mixture of CCs) were: (i) an adsorption temperature of CT of −25 °C, (ii) desorption temperatures of CT and ST of 100 °C, and (iii) a pre-purge volume of 100 mL. The results indicated that the combined effects of several key factors (e.g., adsorption temperature, desorption temperature, and pre-purge volume) were critical to determine the analytical conditions for a mixture of CCs. For instance, the RF values of FA increased with decreasing adsorption temperature when the ST and CT were desorbed at 150 °C (ST pre-purge volume = 100 ml). The recovery of PA decreased with increasing adsorption temperature. This trend was reversed for BA, IA, and VA when a desorption temperature of 150 °C was used. The RF values of FA were slightly greater at ST and CT desorption temperatures of 150 °C, with one exception. The R 2 values for the FA calibrations in a CC mixture were smaller than the R 2 values of FA alone, which clearly indicates that the recovery of FA decreased in the presence of high MW compounds. The adsorption stability and solvent effect were assessed using different pre-purge volumes of L-WS of CCs. The FA peak was clearly distinguishable from the methanol (solvent), similar to the previous findings of Kim et al. 22 . No effect of pre-purge volume on CC recovery was found in this study. It is worth noting that previous FA analysis study using MOF-5 by Kim et al. 22 suggested a possible breakthrough of FA before the sampling volume reached 250 ml.
The basic calibration and QA data were obtained by analyzing the L-WS of the mixture of CCs using the optimal TD conditions used for the analysis of FA (in mixture of CCs), as mentioned above ( Table 5). The method detection limit (MDL) for the CC analysis was determined using a three-fold dilution of the first calibration point (i.e., 4.8 ng/µL) with methanol to reach a final concentration of 1.6 ng/µL (injection volume = 1 µL). The MDL was calculated as the product of the standard deviation of seven replicates and the Student's t-value at the 99% confidence level (for example: 6 df, t = 3.14). The MDL of FA in the CC mixture with high MW compounds (MDL ≥ 1 ng in this study) was greater than the analysis of FA alone (MDL (FA) = 0.1 ng 22 ). The relative standard error (RSE) values (calculated by triplicate analyses of the third (23.8 ng) calibration point) ranged from 2 to 4% (except for AA).
Comparison of MOF-5 with other MOFs. The method of analyzing the CC mixture using MOF-5 combined with Tenax-TA as the ST and CT sorbent for TD-GC/MS was compared with other MOFs with good thermal and mechanical stability including: (i) a Cu-based MOF (MOF-199) and (ii) a Zr-based MOF (of UiO-66 topology). The MOFs were carefully selected based on the outcomes of previous studies. MOF-199 has been used previously for competitive adsorption experiments 14,30 . Also, MOF-199 is a rare example of MOFs that has been developed at an industrial scale 31 . The amine derivate of UiO-66 (i.e., UiO-66(NH 2 )) was chosen based on its previous use for FA sensing 29 . As in the previous analysis using an MOF-5 based system, the analyses using Cuand Zr-based MOFs were carried out with L-WSs of both FA alone and a mixture of carbonyls including FA. The Different geometry of metal sites and pores of studied MOFs provides different contribution from two above-described types of bonds to the energetics of adsorption (see Fig. 2). In case of MOF-5, the distance between oxygen atom of aldehydes and Zn-ion is in the order of 2.5 Å. This distance corresponds to the leading role of metal-oxygen coordination bonds in the interaction of the aldehydes with MOF-5. In contrast, in the cases of MOF-199 and especially UiO-66-NH 2 , the distances between open metal site and oxygen "head" of the aldehydes are visibly larger. Because coordination bonds are much more robust than vdW bonds, the magnitude of the enthalpy in adsorption of MOF-5 is significantly larger than those of the other two MOFs (see Supplementary  Tables S1, S2 and S3).
On the other hand, the interaction between the hydrocarbon "tails" of aldehydes and the linkers of MOF-199 (or UiO-66-NH 2 ) provides visible distortion of the lattice of these MOFs. To characterize these distortions, we calculated the ratio of the volumes of the supercells of pristine MOFs and MOFs after adsorption. In the case of MOF-5, the change of the calculated volume of MOF in result of the adsorption is within 2%, in contrast to much larger values for other MOFs (6% for MOF-199, and 8% for UiO-66-NH 2 ). This result is in agreement with the previous calculations by Wu et al. 32 in which the rigidity of MOF-5 and flexibility of UiO-66-NH 2 (at low mechanical loading) were demonstrated. The magnitude of deviation of the volume after adsorption from the value for pristine systems increases with the increasing length of aldehydes i.e. with the increased number of vdW interactions between hydrocarbon "tail" and linkers. The observed distortion of the MOF increases the total energy of the system which makes adsorption less energetically favorable. Results of the calculations demonstrate that despite formation of additional vdW bonds between larger aldehydes and MOF-199 and UiO-66-NH 2 , the  magnitude of the enthalpy of adsorption change insignificantly (see Supplementary Tables S2 and S3) in contrast to MOF-5 (see Supplementary Table S1). The next step of our study was to assess how the enthalpy of adsorption and volume of MOF changed due to the adsorption of the second species of the same or another kind with an already adsorbed molecule of the first type. In case of adsorption of multiple species with larger sizes (PA, BA, VA, and IA) onto MOF-5, the values of the adsorption enthalpy and the local distortion of MOF were almost the same as for the cases of single species adsorption. However, in the presence of the paired carbonyl (eg., between FA and AA), a significant increase in the enthalpy of adsorption was observed for both species. This effect can be interpreted as the result of a mismatch in the local distortion of MOF caused by adsorption of small species which interacted only with the Zn-core of the MOF-5. As the enthalpy of adsorption remained largely negative, the total quantity of adsorbed aldehyde should be (almost) the same. Therefore, the results of this analysis suggest that larger molecules should be preferentially adsorbed over smaller molecules when a mixture of aldehyde is present. In case of adsorption of aldehydes on MOF-199 and UiO-66-NH 2 , the results of theoretical modeling also demonstrated the influence of already adsorbed species on further adsorption of other species. In case of MOF-199, the adsorption of small aldehydes provided rather large distortion of lattice (Table S2). Correspondingly, the adsorption of larger aldehyde molecules became energetically more favorable (than MOF-5) due to lattice stabilization by molecule-linker interactions. In case of UiO-66-NH 2 , the picture was more complicated despite some similarities in the adsorption patterns with MOF-5 and MOF-199 (Refer to Supplementary Table S3). This is due to greater flexibility of this  Table 5. Calibration and quality assurance data for the analysis of carbonyl compounds (using MOF-5 + Tenax-TA as the sorbent media) by TD-GC/MS analysis. a MDL was calculated as the product of the standard deviation of seven replicate measurements using a 1 µL injection volume of mixture of CCs (1.6 ng μL −1 ) (prepared by the 3-fold dilution of the 1st calibration point of L-WS, i.e., 4.8 ng μL −1 , with methanol) multiplied by the Student's t-value at the 99.9% confidence level (6 df, t = 3.14). b Calculated by assuming a total sample volume of 1000 mL (25 °C MOF (at low mechanical loading) and also due probably to additional contribution from integrations of -NH 2 groups of ligands with adsorbed molecules.

Comparison of results using MOF and commercial sorbent and effect of MOF storage time.
The results of analyzing CC mixtures using a MOF-5 + Tenax-TA-based TD-GC/MS system were compared with the results using the system with Tenax-TA only ( Fig. 3 and Supplementary Table S4). Accordingly, detection of C2 to C5 aldehydes was achieved using both Tenax-TA (on its own) and the MOF-Tenax combination. However, FA (a C1 molecule) alone could only be detected with the multi-adsorbent-based system. FA was not detected using Tenax-TA on its own. Simultaneous detection of all aldehyde species with the multi-adsorbent system is likely a consequence of the partial adsorption of heavier molecules onto the commercial sorbent (Tenax-TA) lessening the competition (between FA and larger aldehyde species) for the available MOF pores (MOF-5). The effect of storage time of MOF was also investigated. The results using old and freshly prepared MOFs were comparable (Refer to Supplementary Table S4), although the use of fresh MOF resulted in slightly greater RF values (with larger RF values for larger molecules) than when older MOFs were used.

Conclusions
In this work, the feasibility of a MOF-based multi-adsorbent technique was demonstrated successfully for the first time for the simultaneous analysis of aldehyde molecules containing 1 to 5 C atoms by TD-GC/MS. The multi-adsorbent-based approach offered higher sensitivity and better reproducibility than that built with a commercial sorbent without the requirement of complex derivatization of small aldehyde molecules. The performance of Zn-based MOF-5 was demonstrated to be superior both theoretically and experimentally to that of Cu-based MOF-199 and Zr-based MOF (of UiO-66 topology). Furthermore, the performance of MOF-5 system was not affected considerably by the storage of MOF. The result of the theoretical modeling analyses revealed that the structural and chemical characteristics of the MOF component of the binary sorbent material should have a direct effect on the efficacy of TD-GC/MS quantification of carbonyl mixtures. Nevertheless, there are still some difficulties associated with co-adsorption on the MOF, which may affect the uptake of lighter compounds. Further work should thus be directed to help resolve these issues, perhaps by establishing more reliable method based on the analysis of ambient air samples and by continuing efforts to develop more materials with advanced functionalities.

Materials and Methods
Chemicals and synthesis of the adsorbents. Reagents were commercially available products that were used without any further purification. Zinc nitrate hexahydrate (Zn(NO 3 ) 2 ·6H 2 O) (98%), terephthalic acid (1,4-benzenedicarboxylate, BDC) (98%), amino terephthalic acid (BDC-NH 2 ), and trimethylamine (TEA) (≥99%) were procured from Sigma-Aldrich. Chloroform (>99.5%) was procured from Daejung, Korea. Dimethylformamide (DMF, 99.0%) and ethanol (99.5%) were obtained from Samchun Chemicals, Korea. Tenax-TA (35/60 mesh) was obtained from Markes International, UK. MOF-5 was synthesized as follows: 25 mL of DMF were placed in a 100-mL beaker. While the DMF was stirred continuously at 1,000 rpm using a magnetic stir bar on a digital hotplate stirrer (DAIHAN Scientific, Korea), 0.3 g of terephthalic acid was added to prepare the linker solution. Similarly, 1.4 g of Zn(NO 3 ) 2 ·6H 2 O was added to 25 ml of DMF to make the metal solution. The linker solution was mixed with the metal solution and stirred constantly. Then, 2 mL of TEA was added drop-wise to the reaction mixture as a catalyst (in situ synthesis method) to reduce the reaction time for MOF-5 synthesis. The solution was then covered with aluminum foil and left for 2 hours at 25 °C. The precipitate in the solution was filtered using a glass microfiber filter (diameter 47 mm, Whatman TM , UK) under vacuum applied using a mini diaphragm vacuum pump (N86KT.18, KNF, UK) (pressure difference = 2 bar and filtering time = 5 min). The residue was washed two times with DMF to remove excess BDC. The DMF was exchanged with 40 ml chloroform and allowed to soak for 12 hours before it was exchanged with fresh chloroform. This process was repeated twice. The chloroform was filtered without letting the MOF dry, and the product was dried in a convection oven (CO-150, Hanyang Scientific Equipment Co., Ltd, Korea) for 8 hours at 90 °C.
MOF-199 was synthesized by dissolving 10 g of Cu (NO 3 ) 2 •2.5H 2 O and 5 g of H 3 BTC in 250 ml of solvent containing a 1:1:1 ratio of DMF, ethanol, and water. Later, the mixture was transferred to a tightly capped glass vessel and heated at 85 °C for 20 h in a forced convection oven (CO-150, Hanyang Scientific Equipment Co., Ltd, Korea). The resulting blue crystals were washed with 30 mL of DMF two times to remove the unbounded ligand materials. The resulting product was then immersed in DMF (30 mL) for 24 h to remove the undissolved impurities. The product was filtered using a glass fiber filter (diameter 47 mm, Whatman TM , UK) under vacuum using a mini diaphragm vacuum pump (N86KT, KNF, UK) (pressure difference was 2 bar, and filtering time was 10 minutes). Finally, the filtered product was kept in an oven at 170 °C for 24 h.
The synthesis of UiO-66-(NH 2 ) was based on the protocol reported by Katz et al. 15 . Briefly, the metal solution was prepared by dissolving 125 mg of ZrCl 4 •8H 2 O in 5 mL of DMF to which 1 mL of concentrated HCl was added. The resulting mixture was then subjected to ultra-sonication for about 20 min. The organic linker solution was prepared simultaneously by dissolving 134 mg of amino terephthalic acid (BDC-NH 2 ) in 10 mL of DMF. The above metal and organic linker solutions were mixed and ultrasonicated for an additional 30 min to complete the reaction. The resulting solution produced a pale yellow solid that precipitated after heating at 80 °C for 12 h. These solids were filtered and then thoroughly washed two times with DMF (30 ml) and two times with ethanol (30 ml). The resulting product was allowed to soak in ethanol (10 mL) three times in three days and heated at 90 °C for 12 h. The final solid product was stored in a closed vial at room temperature under atmospheric conditions. Finally, an empty quartz sorbent tube (ST) (length: 89 mm, OD: 6 mm, and ID: 4 mm, Top Trading Co., Korea) was packed with MOF and Tenax-TA (at a ratio of 1:1 by mass). The ST was first packed with Tenax-TA and then with MOF (refer to Supplementary Fig. S5 for details.) The two adsorbents were separated and held in place using quartz wool. The total mass of the sorbent was 60 mg. The ends of the quartz tube were plugged with quartz wool. The ST was thermally conditioned at 150 °C for 20 hours (purge gas = 99.999% nitrogen and purge flow rate = 200 mL min −1 ) to make sure no impurities remained adsorbed.

Characterization of MOFs.
The powder X-ray diffraction (PXRD) patterns of the MOFs activated at 150 °C for 2 hours were recorded with an HR-XRD diffractometer (Rigaku, Japan). Data were collected with an angular 2Ɵ range of 5-60°, a step size of 0.02°, and a scan speed of 4° min −1 . The thermal stability of MOF-5 was determined by monitoring the loss of mass of MOF-5 as a function of temperature in a thermogravimetric analysis (TGA) experiment. TGA experiments were performed using a SDTQ600 Auto-DSCQ20 system (Eden Prairie, MN, USA). Samples were placed in alumina pans heated from 25 to 800 °C at a ramping rate of 10 °C min −1 while supplying ultrapure N 2 at 100 ml min −1 . Additionally, the morphological characteristics of the MOFs were studied using a NOVA nanoSEM-450 field emission scanning electron microscope (FE-SEM, Hillsboro, OR, USA) to obtain images of MOFs. The metal-ligand binding of pristine and used MOF was examined using Fourier transform infrared spectroscopy (FTIR) analysis. This analysis confirmed that the MOFs had been prepared correctly.

Preparation of liquid working standards (L-WS). Reagent grade chemicals (RGC) were purchased
from Sigma-Aldrich (USA), and the FA solution and mixture of aldehydes were prepared. The liquid primary standard (LPS) was diluted with methanol for the preparation of working standards (L-WS) at five different concentrations (Supplementary Tables S5 and S6). The structure of MOF-5 can be altered by intrusion of water. However, the very low concentration of water impurities (<35 ng μL −1 ) in the L-WS made the extent of this degradation negligible.
Instrumental (TD-GC/MS) system. All carbonyl samples in this study were analyzed using a GC (Shimadzu GC-2010, Japan) equipped with an MS (Shimadzu GCMS-QP2010 Ultra, Japan) and a TD (Unity II, Markes International Ltd, UK) front end. The cold trap (CT) in the TD unit was packed with MOF-5 and Tenax-TA at a ratio of 1:1 by volume. These components were separated by quartz wool (QW), and the unit had an inner diameter of 2 mm and a total (sorbent + QW) bed length was 7 cm (2.4 cm for both MOF-5 and Tenax-TA) (refer to Supplementary Fig. S5). The analytes (i.e., individual or mixtures of aldehydes) loaded onto the ST were subjected to thermal desorption to transfer them for separation using a CP-Wax column (diameter: 0.25 mm, length: 60 m, and thickness: 0.25 µm). Figure 1 depicts the method of replacement of the cold (quartz) trap in the TD unit interfaced with the GC/MS system along with those for sorbent tube (ST) sampling and injection of L-WS into the ST. This study comprised of a total of seven experimental stages (E1 to E7) as the basic experimental scheme of each stage is outlined in Table 2. Final detection of all aldehydes was performed using the MS detection system. FA was quantified in selected ion monitoring (SIM) mode at 29, 30, and 44 m/z, and the other aldehydes were quantified in the total ion chromatogram (TIC) mode at 35 to 600 m/z (Table 3).
Density functional theory (DFT) calculations. The interaction mechanism between the MOF and the guest analytes was studied by the DFT calculations. For this purpose, the pseudo-potential code SIESTA model was applied 33 . All the calculations were carried out using the generalized gradient approximation (GGA-PBE) with spin-polarization 34 and the +vdw correction 35 required for the description of weak interactions. A full optimization of atomic positions was performed. In the course of the optimization, the ion cores were described by norm-conserving non-relativistic pseudo-potentials 36 with cut-off radii of 2.52, 2.37, 2.08, 1.15, 1.14, 1.45, and 1.25 au for Zr, Zn, Cu, O, C, N, and H, respectively. Moreover, the wave functions were expanded with a double-ζ plus polarization basis of localized orbitals for all species (excluding hydrogen), while a double-ζ basis for hydrogen. Optimization of the force and total energy was done with an accuracy of 0.04 eV/Å and 1 meV, respectively. These calculations were performed with an energy mesh cut-off of 300 Ry and a k-point mesh of 4 × 4 × 4 in the Mokhorst-Pack scheme. Electronic structure of all molecules was calculated for a single molecule located in the empty cubic box with sides of 20 Å. This procedure was considered to reflect realistic atomic structures of MOF-5 and MOF-199 based on the Cambridge Database (CCDC) and of UiO-66-NH 2 taken from Valenzano et al. 37 . The DFT-based 0 K adsorption enthalpy (ΔH ab ) was calculated by where E host+guest is the total energy of the MOF and the guest analyte, E host is the total energy of the pristine MOF, and E guest is the energy of the guest molecule in the gas phase.
Data availability. The datasets generated from this study are included in this published article and in Supplementary information.