Adsorption properties of advanced functional materials against gaseous formaldehyde

Journal Pre-proof Adsorption properties of advanced functional materials against gaseous formaldehyde Kumar Vikrant, Minkyu Cho, Azmatullah Khan, Ki-Hyun Kim, Wha-Seung Ahn, Eilhann E. Kwon PII:

S0013-9351(19)30469-4

DOI:

https://doi.org/10.1016/j.envres.2019.108672

Reference:

YENRS 108672

To appear in:

Environmental Research

Received Date: 9 April 2019 Revised Date:

13 August 2019

Accepted Date: 14 August 2019

Please cite this article as: Vikrant, K., Cho, M., Khan, A., Kim, K.-H., Ahn, W.-S., Kwon, E.E., Adsorption properties of advanced functional materials against gaseous formaldehyde, Environmental Research (2019), doi: https://doi.org/10.1016/j.envres.2019.108672. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc.

Adsorption Properties of Advanced Functional Materials against Gaseous Formaldehyde Kumar Vikranta#, Minkyu Choa#, Azmatullah Khanb, Ki-Hyun Kima*, Wha-Seung Ahnc, Eilhann E. Kwond* a

Department of Civil and Environmental Engineering, Hanyang University, 222 Wangsimni-Ro, Seoul 04763, Republic of Korea; bDepartment of Civil Engineering, Balochistan University of Information

Technology, Engineering and Management Sciences, Quetta, Pakistan; cDepartment of Chemistry and Chemical Engineering, Inha University, Incheon 402-751, Republic of Korea; dDepartment of Environment and Energy, Sejong University, Seoul 05005, Republic of Korea

Abstract Intense efforts have been made to eliminate toxic volatile organic compounds (VOCs) in indoor environments, especially formaldehyde (FA). In this study, the removal performances of gaseous FA using two metal-organic frameworks, MOF-5 and UiO-66-NH2, and two covalent-organic polymers, CBAP-1 (EDA) and CBAP-1 (DETA), along with activated carbon as a conventional reference material, were evaluated. To assess the removal capacity of FA under near-ambient conditions, a series of adsorption experiments were conducted at its concentrations/partial pressures of both low (0.1-0.5 ppm/0.01-0.05 Pa) and high ranges (5-25 ppm/0.5-2.5 Pa). At the high-pressure region (e.g., at 25 ppm FA), a maximum adsorption capacity of 69.7 mg g-1 was recorded by UiO-66-NH2. Moreover, UiO66-NH2 also displayed the best 10% breakthrough volume (BTV10) of 534 L g-1 (0.5 ppm FA) to 2,963 L g-1 (0.1 ppm FA). In contrast, at the high concentration test (at 5, 10, and 25 ppm FA), the maximum BTV10 values were observed as: 137 (UiO-66-NH2), 144 (CBAP-1 (DETA)), and 36.8 L g1

(CBAP-1 (EDA)), respectively. The Langmuir isotherm model was observed to be a better fit of the

1

adsorption data than the Freundlich model under most of the tested conditions. The superiority of UiO66-NH2 was attributed to the van der Waals interactions between the linkers (framework) and the hydrocarbon “tail” (FA) coupled with interactions between its open metal sites and the FA carbonyl groups. This study demonstrated the good potential of these advanced functional materials toward the practical removal of gaseous FA in indoor environments.

Keywords: formaldehyde; adsorption; air pollution controls (APCs); metal organic frameworks (MOFs); functionalized adsorbent

*Correspondence: [email protected] (K.-H. Kim); [email protected] (E.E. Kwon) #

These authors are considered as co-first authors as they contributed equally to this work.

2

Contents 1.

Introduction

2.

Materials and Methods

2.1. Chemicals and adsorbent synthesis 2.2. Sorbent characterization 2.3. Preparation of gaseous working standard (G-WS) and experimental outline 3. Results and Discussion 3.1. Characterization of analyzed adsorbents 3.2. Formaldehyde breakthrough curves and performance evaluation of the tested sorbents 3.3. Characterization of formaldehyde adsorption phenomenon 3.4. Isotherm analysis 4. Performance comparison and analysis 4.1. Adsorption performance 4.2. Reusability of adsorbent 5. Conclusions Acknowledgments References

3

1. Introduction Formaldehyde (FA) is considered one of the most hazardous volatile organic compounds (VOCs) due to its toxic propensity (Hadei et al., 2018; Zhang and Rana, 2018). The inhalation of even low amounts of concentrated FA results in vomiting, headaches, dizziness, and nausea (Delikhoon et al., 2018; Soni et al., 2018). FA is also well known as an allergen and carcinogen (Spencer, 2018; Xiao et al., 2017). FA is frequently found in articles of daily use (e.g., sealants, adhesives, electronic devices, furniture, and carpeting), which makes FA the most common indoor VOC (Dai et al., 2018; de Falco et al., 2018a). Nevertheless, the elimination of indoor FA is complicated in that it is widely utilized as a cheap precursor in the synthesis of numerous materials and complex organics (e.g., resins, polymers, paints, adhesives, explosives, and disinfectants) (Nomura and Jones, 2014). Thus, the removal of gaseous FA from confined interior spaces is more crucial today as people are spending more of their time indoors (Brilli et al., 2018). For the abatement of gaseous FA, numerous materials and associated techniques have been reported in the literature (Suresh and Bandosz, 2018). In particular, adsorption and catalytic degradation of FA have been acknowledged as two viable technical options (de Falco et al., 2018b; Fang et al., 2018). In the catalytic approach, FA is oxidized completely, resulting in the formation of H2O and CO2 (Kim et al., 2018). However, one of the major drawbacks of this method is its use of precious metals (e.g., palladium and platinum) coupled with the high energy input required to achieve complete or maximum oxidation of FA (Feng et al., 2018; Guan et al., 2018). Nevertheless, recent literature indicates that some advanced catalysts (e.g., PtNi(OH)x/γ-Al2O3 (Yang et al., 2017) and long-rod Pt/β-FeOOH (Chen et al., 2019)) have the capability to effectively remove gaseous FA under ambient temperatures. Such research endeavors look promising for prospective practical applications for the treatment of FA. However, the performances of these novel catalysts under harsh real-world conditions and their cost

4

economics remain to be explored (Chen et al., 2019; Yang et al., 2017). In addition, other technologies, such as plasma-based methods and photocatalysis, are used to break down FA, but their technical effectiveness and economic viability are inferior to the catalytic oxidation of FA (Assadi et al., 2018; Kim et al., 2018). It is commonly reported that adsorption is a promising technical option for the elimination of FA to improve indoor air quality, due to its simple unit operation and economic superiority (Suresh and Bandosz, 2018). At present, several sorbents, such as metal oxides and carbon-based materials, have been reported to be effective adsorptive agents toward gaseous FA (Liu et al., 2018; Suresh and Bandosz, 2018). However, most previous studies have focused on the removal of FA at higher concentrations (≥ 1 ppm). Because FA can pose a significant health threat even at ppb levels (e.g., 100 ppb), there is a great demand for developing removal methods that are capable of dealing with FA in lower concentration levels. Advanced novel adsorbents with high surface areas and desirable functionalities can enhance the sorptive removal of FA (sub-ppm levels) (Dutta et al., 2018). In this respect, metal-organic frameworks (MOFs) and covalent-organic polymers (COPs) are attractive agents for the sorptive removal of VOCs due to their unique porous structures and facile surface modifications; these technical merits greatly enhance the sorptive selectivity of FA (Bian et al., 2018; Vellingiri et al., 2016). The target-specific functionalization of sorbents results in better sorptive performance by tailoring their affinity and selectivity toward the target analyte (Wang et al., 2016). Considering the aforementioned technical requirements, amine-functionalized adsorbents have attracted great interest in FA sorption study due to the proclivity of the amino groups to form covalent bonding with FA (Song et al., 2017; Wu et al., 2017). For example, it was found that an aminefunctionalized polymeric amino silica displayed a sorption capacity of 129 mg g-1 toward 200-ppm 5

gaseous FA (Nomura and Jones, 2014). Also, an adsorption capacity of 27.4 mg g-1 was recorded by amine-functionalized graphene aerogels for gaseous FA at 16.28 ppm concentration (Wu et al., 2017). On this trend, an MOF (MIL-101) was used to attach amine functionalities to the Cr3+ open metallic sites for enhancing its capacity for FA (Wang et al., 2016). Interestingly, the enhancement of FA sorption capacity of 164.7 mg g-1 for the ethylenediamine functionalized variant (ED-MIL-101) relative to the pristine MIL-101 (100.2 mg g-1) indicates the great potential of the amine functionality. The sorption mechanism between FA molecules and ED-MIL-101 consisted of a proton exchange through the attack on the FA carbonyl group by the nitrogen lone pairs which resulted in the formation of an unstable hemiaminal intermediate (Nomura and Jones, 2014; Wang et al., 2016). The hemiaminal intermediate subsequently expelled water to form imine (Nomura and Jones, 2013). Several studies have revealed that the presence of moisture greatly interferes with the sorptive removal of VOCs (Thevenet et al., 2018). In a case of highly humid conditions (relative humidity ≥ 50%), the VOC breakthrough on activated carbon occurred much earlier as compared to dry conditions (Tao et al., 2004). As a result, from a practical viewpoint, it is essential to study adsorption processes in the presence of moisture. Specifically, it is desirable to conduct the sorptive removal of gaseous FA with sub-ppm concentration levels under humidified conditions. In this study, the sorption behaviors of several functionalized adsorbents, the metal-organic frameworks (MOFs) MOF-5 [M5] and UiO-66-NH2 [UN], and the covalent-organic polymers (COPs) ethylenediamine/diethylamine functionalized carbonyl-incorporated aromatic polymers [CBAP-1 (EDA) or CE and CBAP-1 (DETA) or CD], were examined in reference to activated carbon [AC] as a traditional reference sorbent to remove gaseous FA at low (0.1-0.5 ppm) and high (5-25 ppm) concentrations. As discussed earlier, amine-based adsorbents are of particular interest for the sorptive

6

removal of FA. In this respect, an amine functionalized MOF, UiO-66-NH2 was selected as a possible media to test the performance as FA sorbent. Moreover, di and tri amine functionality-based porous COPs (i.e., CBAP-1 (EDA) and CBAP-1 (DETA), respectively) have also been studied in this work due to the ease with which their structure can be pre-designed or functionalized. Note that all the amine functionalized adsorbents used in the present study are known to be water stable in nature (Vellingiri et al., 2019). In recent years, as COPs have drawn a good deal of attention in the fields of air quality management and carbon capture, they were also chosen as a prospective sorbent to treat FA in this study (Puthiaraj et al., 2017b; Ravi et al., 2017a). Also, MOF-5 has been reported to be used as a sorptive media for the effective FA capture under ambient conditions and its subsequent practical quantification as analytical sorbent for thermal desorption gas chromatography-mass spectrometry (Dutta et al., 2018; Gu et al., 2010; Kim et al., 2017). In this regard, MOF-5 was also included in the present study to provide a detailed assessment of its FA adsorption capabilities as compared to other potentially superior adsorbents. To impart a more practical understanding, we prepared gaseous FA standards under humidified conditions (relative humidity = 12% for the highest FA concentration tested, i.e., 25 ppm) to simulate real conditions. Isotherm analysis was then conducted to elucidate the underlying phenomena. Based on this research, we offer better insights into the guidance for assessing the actual performance of advanced functional materials toward the practical removal of gaseous FA under ambient conditions.

2. Materials and Methods Adsorption experiments were conducted in two FA concentration ranges, namely, low (0.1-0.5 ppm) and high (5-25 ppm). A TD/GC-MS system was used for the low-concentration experiments as it is capable of detecting trace quantities of FA (0.25 ng from samples as small as 20 mL of air, at 7

precision of 1.79%) (Kim et al., 2017). This gas chromatography (GC)-based method was developed to measure FA at low concentration levels without the involvement of traditional derivatization techniques (Kim et al., 2017; Pal and Kim, 2008). For measurements of FA sorption in the high concentration range, an FA sensor was used.

2.1. Chemicals and adsorbent synthesis All the chemical reagents used in this study were commercially accessible and were used without any further processing. Also, the detailed list of chemical reagents used in this study along with the synthesis protocols applied for the functional materials have been discussed in detail in the supplementary information.

2.2. Sorbent characterization The MOFs and COPs were activated at 150°C for 2 h and subsequently analyzed using an HR-XRD diffractometer (Rigaku, Tokyo, Japan) to generate the powder X-ray diffraction (PXRD) patterns. The PXRD data were recorded at a scan speed of 4° min-1, a step size of 0.02°, and a 2 range of 5. The Brunauer-Emmett-Teller (BET)-based surface properties were computed by analyzing the N2 adsorption isotherms obtained at 77K by a BEL sorp Mini (BEL Corporation, Japan). Prior to the measurements, the samples were heated at 150 oC for 12 h under vacuum to remove moisture. The KBr pellet method was also used to identify the functional groups present on the adsorbents through Fourier transform infrared (FTIR; Thermo-Fisher FTIR analyzer Nicolet 5700, Japan) spectroscopy in the range of 400-4000 cm-1 (Zheng et al., 2018). Accordingly, the functional materials

8

were confirmed to be properly synthesized, and the surface morphology of the adsorbents was investigated via a field emission scanning electron microscope (FE-SEM, Hillsboro, OR, USA). Thermogravimetric analyses (TGA) were carried out using an SDT Q600, Auto-DSCQ20 system (Eden Prairie, Minnesota, USA). For the TGA analysis, the material samples were kept on alumina pans and heated at a rate of 10 °C min-1 from 25 to 800 ˚C under a N2 flow of 100 mL atm min-1.

2.3. Preparation of gaseous working standard (G-WS) and experimental outline The gaseous primary standard (G-PS) of FA was prepared by injecting 50 µL of aged formalin (FA: ~5% (w/w %) in H2O with 10%-15% methanol) into a 1-L polyester aluminum (PEA) bag (Top Trading Co., Korea) filled with N2 (99.99% pure) at 1 atm. The bag was then left to vaporize formalin for 24 h at 25˚C. The relative humidity (RH) and FA concentration in that G-PS were calculated mathematically to be ~100% and 16,694 ppm, respectively. Nonetheless, the actual RH level employed in the actual experiment dropped significantly due to dilution in the preparation of gaseous working standards (G-WS). The RH level went below 12% even for the highest FA concentration tested (i.e., 25 ppm) as explained in the subsequent paragraphs. The actual FA concentration of the G-PS standard was quantified with the aid of a well-established method, i.e., the combination of the 2,4-dinitrophenylhydrazine (DNPH) cartridge derivatization method and a high-performance liquid chromatogram with ultraviolet detector (HPLC-UV). The G-PS was pulled through the DNPH cartridge at 1 L min-1 for 5 min. After derivatization, the DNPH cartridge was eluted with 5 mL of acetonitrile, and 20 µL of the extract was analyzed by HPLC-UV (Shimadzu, Japan) (Table 1). The HPLC-UV system was calibrated using a Supelco DNPH-hydrazone

9

standard. A mixture of acetonitrile (70%) and distilled water (30%) was used as the mobile phase at a 1.5 mL min-1 flow rate for 16 min. After the determination of the actual G-PS FA concentration, the required G-WS were made by suitably mixing/diluting the G-PS with ultrapure N2. The prepared G-WS concentration of the FA was checked each time using the above methods before beginning the sorption experiments. The sorption experiments utilized quartz tubes, 4 mm (inner)/6 mm (outer) diameters and 89 mm in length (Top Trading Co., Republic of Korea). Each sorbent to be tested was packed into a tube and held in position using quartz wool. Before each adsorption experiment, the sorbent tube was conditioned for 3 h at 150°C (99.99% pure N2 as a purge gas at a flow rate of 0.2 L min-1) to remove the pre-adsorbed target species or any impurities if present. Note that the term, loaded volume refers to the total G-WS volume that has been passed through the sorbent tube at a particular flow rate over a given duration. The low-concentration FA samples (0.1-0.5 ppm) were analyzed by a gas chromatograph (Shimadzu GC-2010) fitted with a mass spectrometer (MS; Shimadzu GCMS-QP2010 Ultra) equipped with a thermal desorber (TD; Unity II, Markes International Ltd., UK) front end. For the determination of FA, Tenax-TA and MOF-5 were used to pack the cold trap in a 1:1 volumetric ratio separated by quartz wool (Kim et al., 2017). The detection limit of this approach was approximately 0.1 ng (Kim et al., 2017). Triplicate analyses of a chosen calibration point gave relative standard error (RSE) values in the range of 2-4%. The unit had a bed length of 7 cm (2.4 cm each of Tenax-TA and MOF-5) and a 2mm inner diameter. The FA loaded onto the sorbent tube was transported to the TD for separation of the target species by a CP-wax column (60-m length, 0.25-mm diameter, and 0.25-µm thickness). The final detection of FA from each adsorption experiment was done using the MS system. The

10

quantification of FA was done in the selected ion monitoring (SIM) mode at 29, 30, 43, 44, and 58 m/z (Table 1). Figure 1 provides an overview schematic of the experimental setup. The high-concentration FA samples (5-25 ppm) were analyzed by an FA sensor (KINSCO Tech., Korea). The sensor flow rate was set at 50 mL min-1 to match the AS/TD-GC/MS method (Table 1). The sensor had near-real-time sampling data-logging at every 3 or 4 s, which was transferred to a personal computer for analysis. Before each sample analysis, the calibration was performed by the connection of the sorbent tube and the G-WS bag (Figure 2). Similar to the AS/TD-GC.MS method, the G-WS was directly connected to the FA sensor to determine the inlet concentration before each sorption experiment. The operational details of this sensor system are provided in Table 1.

3. Results and Discussion

3.1. Characterization of analyzed adsorbents The structure, chemical functionality, and surface morphology of the synthesized materials were characterized using PXRD, FTIR, and SEM, respectively. Note that the detailed characterization results of the commercial activated carbon used in this study can be found elsewhere (Khan et al., 2019a). An analysis of Figure S1 indicated that the PXRD patterns of the investigated MOFs and COPs were in good agreement with the literature: MOF-5 (2θ = 6.7 and 9.7°) (Gu et al., 2010); UiO66-NH2 (2θ of 7.4 and 8.5°) (Garibay and Cohen, 2010; Vellingiri et al., 2017); CBAP-1 (EDA) (broad and diffused peaks in the vicinity of 2θ = 23 and 43°) (Ravi et al., 2017a); and CBAP-1 (DETA) (broad and diffused peaks in the vicinity of 2θ = 22 and 42°) (Puthiaraj et al., 2017a). As can be seen from Table S1, the BET surface areas of the tested sorbents were found to decrease in the following order: activated carbon (1,004 m2 g-1) > UiO-66-NH2 (963 m2 g-1) > CBAP-1 (EDA) 11

(674 m2 g-1) > CBAP-1 (DETA) (667 m2 g-1) > MOF-5 (424 m2 g-1). In case of pore volume, the observed order was altered slightly with a change in position (i.e., between CBAP-1 (DETA) and CBAP-1 (EDA)): activated carbon (0.71 cm3 g-1) > UiO-66-NH2 (0.58 cm3 g-1) > CBAP-1 (DETA) (0.32 cm3 g-1) > CBAP-1 (EDA) (0.23 cm3 g-1) > MOF-5 (0.22 cm3 g-1) (Table S1). Furthermore, the pore sizes of all the adsorbents were noted to be comparable as observed by the average pore diameter values: 2.07 nm (MOF-5), 1.56 nm (UiO-66-NH2), 1.69 nm (CBAP-1 (EDA)), 1.9 nm (CBAP-1 (DETA)), and 1.21 nm (activated carbon) (Table S1). The FTIR spectra showed the presence of abundant oxygen functionalities on the surface of the synthesized materials (Figure S2). The FTIR data of the MOF-5 exhibited characteristic bands at 1,399 and 1,571 cm-1, which were ascribed to the symmetric and asymmetric stretching vibrations of CO, respectively. The presence of sorbed water molecules was confirmed by the sharp peaks at 3,605 and 3,547 cm-1. Also, Zn-O stretching was observed through the relatively shorter bands present in between 700 and 900 cm-1. The FTIR spectra of UiO-66-NH2 confirmed distinct peaks at 1,258 and 1,338 cm-1, which corresponded to the stretching vibrations of C-N. Moreover, the N-H wagging of secondary amides, C=O, and N-H single bond vibrations of UiO-66-NH2 were observed at 765 and 1,568, 1,699, and 3,360 cm-1, respectively. The FTIR spectrum of CBAP-1 (EDA) showed characteristic peaks at 1,656 cm-1, which was assigned to the aromatic C=C stretching band. The symmetric and asymmetric stretching vibrations of –CH bonds were also observed at 2,871 and 2,909 cm-1, respectively. A small peak appeared at 1,048 cm-1, indicating the C-N stretching vibration. In addition, the N-H stretching was detected through a characteristic peak at 3,300 cm-1 (Ravi et al., 2017b). The FTIR spectrum of CBAP-1 (DETA) showed characteristic peaks at 1,052 and 1,655 cm-1, which were assigned to the C-N and C=C stretching bands, respectively. The C-H stretching bands

12

appeared at 2,889 and 2,950 cm-1, and the N-H stretching was detected at 3,300 cm-1 (Kim et al., 2012; Ravi et al., 2017b). The surface morphologies of the synthesized functional materials and activated carbon are presented in Figure S3. MOF-5 analysis revealed the formation of agglomerated cubic morphologies with particles ranging in size from 150 to 500 nm. The particle size of UiO-66-NH2 was in the range of 40 to 100 nm. The SEM images showed that CBAP-1 (EDA) and CBAP-1 (DETA) consisted of particles ranging from 2.0 to 4.5 µm in size. The structural morphologies and particle sizes of the synthesized MOFs, COPs, and activated carbon were in excellent agreement with previously reported results (Puthiaraj et al., 2017a; Vellingiri et al., 2017). The TGA profiles of the synthesized functional materials are plotted in Figure S4 and are in agreement with those reported previously to confirm their proper synthesis. Two major weight losses were observed for MOF-5. The removal of DMF from the MOF-5 pores was primarily responsible for the first weight loss event (8.9%) in the 100-350°C temperature range. The structural breakdown of the MOF-5 framework constituted the second weight loss event (30.8%) in the 400-500°C temperature range (Gao et al., 2010). For UiO-66-NH2, the complete removal of DMF took place at 400°C (15%↓), elucidating its high thermal stability. In principle, the great thermal stability of UiO-66-NH2 originates due to the presence of high polarity in its framework owing to the existence of amino groups. Essentially, the collapse of the porous channels is averted due to the containment of DMF molecules inside the pores by the amino groups (Žunkovič et al., 2015). In the 400-600°C temperature range, a minor and steady weight loss of 15% was observed to indicate the collapse of the UiO-66-NH2 framework (Dutta et al., 2018). A gradual weight loss of 4.2-6.4% was observed for CBAP-1 (EDA) and CBAP-1 (DETA) in the 200-350˚C temperature range which can be attributed to the loss of solvent and water molecules. At temperatures

13

greater than 350˚C, their gradual weigh loss indicating the structural breakdown was observed as 12.9 and 13.5%, respectively. The TGA profiles confirm that all the materials used in this study are satisfactorily stable at the temperature used for their thermal activation (150˚C) prior to the adsorption experiments.

3.2. Formaldehyde breakthrough curves and performance evaluation of the tested sorbents For effective sorptive removal, it is imperative to seek a suitable sorbent with superior sorptive performance on the target pollutant. In this regard, the basic properties of the sorbents should be assessed with respect to the dynamic performance of the sorption process against VOCs (e.g., breakthrough capacity). The breakthrough behavior of FA on the tested sorbents is plotted in Figure 3 (for the low-pressure range, 0.01-0.05 Pa) and Figure 4 (for the high-pressure range, 0.5-2.5 Pa). Essentially, these breakthrough curves are useful to analyze the evolution of FA in terms of Cout/Cin vs. the volume of the loaded G-WS. Here, Cout and Cin refer to the FA concentration at the outlet and inlet of the sorbent tube, respectively; the Cout/Cin values were calculated using the following formula:

(

)

(

)

(1)

By definition, breakthrough volume (BTV) is the volume of the carrier gas that needs to be passed through a desorption tube at a fixed temperature to purge out a target analyte for a given mass of the adsorbent (Jo et al., 2017). The maximum strength of a sorbent (its potential to capture the target pollutant) can be represented as 100% BTV (Szulejko et al., 2019). However, the general properties of

14

sorbents can also be approximated at the initial stage of BTV (e.g., 5% or 10%), so from a practical point of view the 10% breakthrough volume (BTV10) offers a fair judgment of a sorbent’s performance. In this study, in the lower-pressure region, the BTV10 value was the smallest for MOF-5 (10.8 L g-1 for 0.5 ppm FA, to 112 L g-1 for 0.1 ppm FA), and was highest for UiO-66-NH2 (534 L g-1 for 0.5 ppm FA, to 2,963 L g-1 for 0.1 ppm FA) as presented in Table 2. Similar observations were made in the high-pressure region, wherein MOF-5 displayed the lowest BTV10 value of 9.78 L g-1 for 10 ppm FA, while UiO-66-NH2 had the highest value of 137 L g-1 for 5 ppm FA. The observation data in Table 2 indicate the great capability of the amine-functionalized MOF for FA adsorption. Also, the lowest BTV10 value of MOF-5 may be accounted for by its relatively small surface area (424 m2 g-1) and lack of target specific functionalities in its structure (Table S1). At both the lowest and highest end of the low-pressure region (0.1 and 0.5 ppm FA), the BTV10 values were measured consistently in the following order: UiO-66-NH2 > CBAP-1 (EDA) > CBAP-1 (DETA) > activated carbon > MOF-5. However, at the intermediate concentration of FA (0.25 ppm), activated carbon was observed to perform slightly better than CBAP-1 (DETA) in terms of the BTV10 value (Table 2). On the other hand, a clear and consistent trend could not be observed for the BTV10 values in the high-pressure region, and the performance of the sorbents varied dynamically across varied FA concentrations. In general, in the low-pressure region, the strengths of the sorbents were inversely proportional to the slopes of the breakthrough curves, such that MOF-5 < CBAP-1 (DETA) < CBAP-1 (EDA) < activated carbon < UiO-66-NH2 (Figure 3). In the high-pressure region, the patterns were found in the order of MOF-5 < activated carbon < CBAP-1 (DETA) < CBAP-1 (EDA) < UiO-66-NH2 (Figure 4). As such, UiO-66-NH2 exhibited the highest resistance levels of intraparticle mass transfer for FA that was accompanied by a delayed saturation of the surface active sites, e.g., good adsorptive performance (Chen et al., 2014; Kosuge et al., 2007). The percentage Cout/Cin was 15

observed to increase slowly for all the sorbents at the tested FA concentrations with passing sorption time, and the values were expected to reach the sorbent saturation at 100% BTV. It is difficult to obtain full-scale isotherm data when adsorption reactions take place at relatively low concentration levels of pollutants, especially for very strong sorbents such as UiO-66-NH2. In our lowconcentration experiments (below 0.5 ppm FA), there are considerable technical difficulties in operating the sorption testing system, as the standard loading on to the sorbent tube is directly interfaced with the TD-GC-MS unit. This implies that the system could malfunction or be contaminated if adsorption breakthrough takes place. Hence, the acquisition of isotherm data in our low-concentration experiments was often restricted to 50% BTV levels. The results also varied across the concentration levels of the standard gas tested, as shown in Figure 3. Thus, the experiments conducted at maximum concentration levels of FA in the low-pressure region (0.5 ppm) were capable of reaching near-saturation status. In contrast, the data obtained using the low concentration (0.1 ppm) were limited to reaching values considerably below saturation. As discussed, the isotherm data for all the tests conducted in low concentrations of FA (0.1-0.5 ppm) could not be evaluated in full scale. Consequently, the results obtained at the low and high concentration ranges could not be compared on a parallel basis at maximum capacity. Given this limitation, the breakthrough results (and the corresponding partition coefficient values) were compared at 10% BTV, as described in Section 3.3. These comparative data were thus still sufficient to evaluate the actual performance of the sorbents at the two concentration ranges. The observed peak area was transformed into a mass unit (ng) using a numerical integration following the trapezoidal rule: ∑

∑

(

)

(2)

16

where Δmn is the mass sorbed in the nth run, Cin is the inlet concentration, ΔVn is the volume pulled per run for the nth run, A is the analyte, and n is the run number. The variations in the adsorption capacity of the tested sorbents with the loaded G-WS volume at different FA concentrations is presented in Figures 5 and 6. Table 2 provides a comprehensive account of the adsorption capacity values at different breakthrough experiments (5%, 10%, 50%, and 100% BTV), providing a clear and better assessment of the sorbent performance. In the low-pressure region, the maximum sorption capacity was recorded by UiO-66-NH2 (1.72 mg g-1 at 0.5 ppm FA). Moreover, UiO-66-NH2 showed the best adsorption capacity at each of the breakthrough levels in the entire lowpressure region (0.29 mg g-1 at BTV5, 0.54 mg g-1 at BTV10, and 1.07 mg g-1 at BTV50, each at 0.5 ppm FA). MOF-5 displayed the minimum sorption capacity at each breakthrough level throughout the low- as well as high-pressure regions (Table 2). In the high-pressure region, the maximum adsorption capacity was recorded by UiO-66-NH2 (69.7 mg g-1 at 25 ppm FA). UiO-66-NH2 had the highest adsorption capacities at each breakthrough level throughout the high-pressure region (0.85 mg g-1 at BTV5, 2.87 mg g-1 at BTV10, and 11.9 mg g-1 at BTV50), each at 25 ppm FA). In the low-pressure region, the adsorption capacity values (at BTV50) were observed to vary in the following order: UiO66-NH2 > activated carbon > CBAP-1 (EDA) > CBAP-1 (DETA) > MOF-5. However, at the high end of the low-pressure region (0.5 ppm), CBAP-1 (EDA) performed better than the activated carbon. For the high-pressure region, the following order was observed for the maximum adsorption capacity values (at 25 ppm FA): UiO-66-NH2 > CBAP-1 (EDA) > CBAP-1 (DETA) > activated carbon > MOF-5. These observed trends indicate that COPs may not perform better than activated carbon at very low concentrations of FA, although they began to outperform carbon as the FA concentration increased (≥ 0.5 ppm).

17

The good performance of UiO-66-NH2 towards FA sorption may be a consequence of the van der Waals attraction between the FA molecules and the linkers (amino terephthalic acid) present in the structural matrix of UiO-66-NH2 (Nomura and Jones, 2014). Also, the carbonyl group present in FA might interact with the open metallic sites in the MOF structure (Dutta et al., 2018). Figure 7 (a) briefly summarizes the interactions taking place between the UiO-66-NH2 framework and FA molecules. The carbonyl bond has been known to be highly reactive owing to the difference in electronegativity between oxygen and carbon which gives rise to remarkable contribution to the dipolar resonance form. (Carbon and oxygen atoms bear positive and negative charges, respectively.) (Feeney et al., 1975). Interestingly, as a nucleophilic molecule approaches the carbonyl bond, the polarity increased further owing to the shift in the position of the π electrons towards oxygen (Feeney et al., 1975). In this respect, the presence of amine groups greatly helped enhance the sorption of FA molecules primarily through the formation of Schiff bases, i.e., imine derivatives (Nomura and Jones, 2013). In principle, the lone pair bearing nitrogen in the amine group attacks the carbon of the carbonyl functionality in FA (Nomura and Jones, 2014). This process results in an exchange of protons to ensue the formation of unstable hemiaminals which were subsequently dehydrated to form imines (Wang et al., 2016) (Figure 7 (a)). To observe the interactions between the FA molecules and UiO-66NH2 structure, the FTIR spectra were recorded before and after the sorption experiment (Figure 7 (b)). A magnified view of the spectra (in the region of probable imine formation) is presented in Figure 7 (c). In principle, the characteristic imine peaks can be observed at 1,580, 1,620, and 1,690 cm -1 of the FTIR spectra. The peaks at 1,580 and 1,620 cm-1 could not be recognized in the complex spectra owing to their relatively low intensity (Larkin, 2011). A magnified view highlights the beginning of the emergence of a new small peak at 1,630 cm-1, indicating the possible presence of imine after interactions with FA. Since the FA concentration used in the present study is relatively smaller (and

18

also in the presence of moisture) than what is reported in the literature, it was not possible to find strong evidence of the imine functionalities by the FTIR. Note that the presence of humidity in G-WS may retard the dehydration process (leading to the formation of imine) through stabilization of the hemiaminal intermediate. This will further explain the absence of a sharp imine peak (Nomura and Jones, 2014). Interestingly, similar evidence has also been provided by other authors who were not able to observe a distinct imine peak due to the presence of water in the system (Nomura and Jones, 2014; Vellingiri et al., 2019). Note that based on stoichiometry, the ideal interaction between primary amines and carbonyls should proceed in a one-to-one ratio. However, the efficiency of amine groups to capture FA (in case of gas-solid sorption systems) has been observed to be less than that predicted by the ideal stoichiometry for many functionalized sorbents such as amino activated carbon and polymeric amino silica sorbents (Drese et al., 2011; Nomura and Jones, 2013; Nomura and Jones, 2014; Tanada et al., 1999). This is primarily due to the fact that for solid sorbents, many amine functionalities may be buried deep into the porous structure to be inaccessible to the FA molecules (Nomura and Jones, 2014). As a consequence, the presence of amine functionalities in an adsorbent does not necessarily guarantee enhanced performance towards FA. Essentially, the porous sorbent network may also play a significant role in the overall sorption process. Interestingly, the apparent significance of some variables (e.g., porosity, pressure conditions, and amino groups) on the sorption of FA was recognized in this study. In the low concentration region, activated carbon performed slightly better than the COPs. However, the trend was reversed in the high concentration region such that the COPs outperform activated carbon. Essentially, at high concentration conditions, the partial pressure is high enough to push the FA molecules deep into the pores to effectively induce the favorable interactions with the amino groups. In contrast, under low FA concentrations, the conditions should not be

19

favorable to induce such favorable interactions with amine functionalities. The COPs (CBAP-1 (EDA) and CBAP-1 (DETA)) showed similar sorption capacities at very low FA concentrations, as FA molecules are packed inside the cages formed by favorable pore cavity sizes (host-guest interactions) (Vellingiri et al., 2017). The findings of consistently poor performance by MOF-5 against FA can be attributed to the lack of favorable surface sites to bind the target molecules. Also, the relatively better performance of activated carbon in the lower FA concentration range may be due to the easy accessibility of pores for FA molecules coupled with prevalent van der Waals interactions (Chen et al., 2014; Lee et al., 2013).

3.3. Characterization of formaldehyde adsorption phenomenon A survey of the adsorption literature suggests that sorption capacity is the most commonly used parameter to assess sorption performance. However, sorption capacity is highly biased toward the initial loading conditions of the target pollutant and, hence, is of not much use when comparing the results of two different sorbents under varying conditions (Khan et al., 2019b; Vikrant and Kim, 2019; Vikrant et al., 2019b). In this regard, one of the key criteria in the selection of a suitable sorbent is its partition coefficient (PC) against the target pollutant, because this reveals the true performance of the sorbent free from biases involved in differences due to the initial loading concentrations (or partial pressures). This interaction can be effectively visualized and characterized as follows: (

)

(

)

(

)

(3)

Essentially, the PC values are computed by following a similar concept as observed for the partition of a target analyte between its gaseous and aqueous phases. In general, Henry’s law is applied for a given compound, when its distribution between the gaseous and liquid phases is equilibrated 20

(Vellingiri et al., 2017). However, since most gas-solid sorption systems do not necessarily obey Henry’s law, PC values are often computed to analyze the true removal capacities of adsorbents or the strength of interactions between gaseous pollutants/solid adsorbents (Szulejko et al., 2019). The added advantage of computing the PC is that one can easily assess and see the real performance of sorbents (sorption affinity and sorbent heterogeneity), which is not possible through adsorption capacity values as they are heavily biased on the initial concentration of the target pollutant (Szulejko and Kim, 2019; Vikrant et al., 2019a; Vikrant et al., 2019b). For most practical operations, quantifying the PC at low BTV levels, such as BTV10, is considered an important figure of merit for the sorption operation (Khan et al., 2019a). Figures 8 and 9 elucidate the variation of PC with the loaded G-WS volume. Table 2 reveals that UiO-66-NH2 showcased the highest PC at BTV10 (10.1 mol kg-1 Pa-1) at the lowest tested concentration of 0.1 ppm FA. Moreover, in the low-pressure region, the following order was observed for PC values at BTV10: UiO-66-NH2 > CBAP-1 (EDA) > activated carbon > CBAP-1 (DETA) > MOF-5. Likewise, in the high-pressure region, UiO-66-NH2 was observed to have the highest PC value at BTV10 (0.53 mol kg-1 Pa-1) at 5 ppm FA, confirming the superiority of this amine-functionalized MOF toward FA sorption. However, unlike in the low-pressure region, a clear order could not be observed for the PC values, probably due to the involvement of complex dynamics and variations in the sorption affinity under high partial pressure conditions. For example, UiO-66-NH2 had the highest PC value at 5 ppm FA. In contrast, CBAP-1 (DETA) outperformed the other sorbents in the high-pressure region (PC value of 0.7 mol kg-1 Pa-1 at 10 ppm FA, and 0.18 mol kg-1 Pa-1 at 25 ppm FA). As such, the results confirmed that the COPs performed best as the FA partial pressure increased (Table 2).

21

3.4. Isotherm analysis A sorption isotherm aims to mathematically model the mass of the target pollutant sorbed on a solid adsorbent surface at a constant temperature at constant loading pressure of the target component in the gaseous phase. In general, the physical explanation of the nature of an isotherm profile can be derived from the surface morphology of the sorbent (e.g., the specific surface area and porosity) (Ayawei et al., 2017; Chen, 2015; Vikrant et al., 2019b). As such, information on sorption isotherms can provide better insights into the nature and properties of sorbate-sorbent interactions (Vellingiri et al., 2017). Many nonlinear and complex models are available in the literature; however, for the purpose of simplifying the mathematics and for a better and easier understanding, linearized forms of the Langmuir and Freundlich isotherm models are commonly employed (Foo and Hameed, 2010). In the Langmuir theory, the sorption of molecules from the gas phase takes place in the form of a monolayer on distinct homogeneous surface active sites on the micro-porous sorbents (Foo and Hameed, 2010). The Langmuir theory helps in analyzing the partitioning of targets between the solid and

gaseous

phases

through

the

application

of

dynamic

equilibrium

(balancing

of

adsorption/desorption rates) (Zhang et al., 2018). This model approximates Henry’s law at lowpressure conditions by assuming that sorption takes place in proportion to the number of uncovered sites on the sorbent surface (Ayawei et al., 2017; Chen, 2015). The Langmuir model can be described as follows: (4) where C is the sorption capacity (mg g-1), Cm is the sorption capacity at monolayer coverage (mg g-1), p is the pressure of the adsorbate (Pa), and KL is the Langmuir constant (Pa-1). In order to determine the constants KL and Cm, the Langmuir model can be expressed in the linearized form as follows: 22

(

)

(5)

The linearized Langmuir isotherm model was realized by plotting 1/C against 1/P. The Freundlich sorption model is an exponential equation that depicts a multilayered and reversible adsorption phenomenon (Foo and Hameed, 2010):

(6) where KF is the Freundlich constant and n is a measure of sorption intensity that helps in describing the heterogeneous sorbent surface (Foo and Hameed, 2010). The Freundlich parameters can be easily calculated by using the linearized form of the Freundlich model: (7) The linearized Freundlich isotherm model was obtained by plotting ln C against ln P. Figures 8 and 9 show the Langmuir and Freundlich models for all the tested sorbents, respectively. The high- and low-pressure regions have been suitably highlighted in the graphs to facilitate a better comparison and understanding of the prevalent sorption mechanisms and sorbent-sorbate interactions in these regions. Table 3 showcases the R2 values for the Langmuir and Freundlich isotherm model fits. Table 4 elucidates the isotherm parameters obtained by the model fit. The results summarized in Table 3 indicate that the Langmuir isotherm should be a better fit for both the low- and high-pressure regions. However, UiO-66-NH2 and activated carbon demonstrated more affinity toward the Freundlich isotherm in the high-pressure region by a small margin, indicating the presence of complementary effects. At the low-pressure conditions, the Langmuir assumption of monolayer sorption was adequately satisfied by these sorbents to make the better fit. However, in the

23

high-pressure region the Freundlich model was the better fit while the Langmuir assumption was no longer valid. The small margin of difference between these two model fits indicates that the assumptions of both the Langmuir and Freundlich models were somewhat meaningful in explaining the behavior of these sorbents. Given the enhanced adaptability of our experimental data to the Langmuir isotherm, only the Langmuir parameters have been organized in Table 4. Figure 9 shows the general trend of the Freundlich isotherms. From the Langmuir model, the highest absolute Cm values were recorded by UiO-66-NH2 in both the low (1.43 mg g-1) and high (49.5 mg g-1) pressure regions. In the low-pressure region, the largest absolute KL value of 389 Pa-1 was observed for UiO-66-NH2. In contrast, CBAP-1 (EDA) exhibited the highest absolute KL value of 52.2 Pa-1 in the high-pressure region. Also, the largest absolute Henry’s constant (indicating the extent of formaldehyde partitioning between the gaseous and solid phases) values in the low- and high-pressure regions were recorded by UiO-66-NH2: 18.5 and 7.58 mol kg-1 Pa-1, respectively. The reduced value of Henry’s law constant in the high-pressure region further indicates a decrease in the affinity between formaldehyde molecules and the sorbent surface which is accompanied by the lack of availability of active sorption sites. As such, these results confirm the superior uptake capabilities of the aminefunctionalized MOF toward FA. The variations in the behavior of isotherms in the low- and high-pressure regions can be explained by using the Langmuir two-surface theory: ( )

( )

( )

(8)

Where C(p) is the net sorption capacity (mg g-1) at a given exit pressure of the sorbent tube (p in Pa). C1 C2 represent the sorption capacity for site-1 (low-pressure region) and site-2 (high-pressure region), respectively (Foo and Hameed, 2010). Interestingly, Table 4 reveals that the absolute Henry’s constant 24

value decreases with an increase in the pressure condition for most adsorbents. This can be explained by the fact that the partitioning behavior of the sorbent is inversely proportional to (1+KLp). Also, at high-pressure conditions, more repulsive forces are experienced by the adsorbate molecules among themselves, resulting in a reduction of the Henry’s constant.

4. Performance comparison and discussions on sorbent reusability 4.1. Adsorption performance The results of some available data from recent investigations of FA adsorption are summarized in Table 5. As explained earlier, the performance of different sorbents has often been assessed in terms of the maximum adsorption capacity (Table 5). However, because the sorption capacity is a highly biased figure of merit, the partition coefficient is a more reliable metric for the real performance of the sorbent (Khan et al., 2019a; Na et al., 2019; Vikrant et al., 2019b). Interestingly, in our study, MOF-5 was observed to have an enhanced adsorption capacity of 1.66 mg g-1 compared to 0.11 mg g-1 as reported in earlier research (Gu et al., 2010); both studies were performed at a similar FA concentration of 25 ppm under ambient conditions (Table 5). The observed difference in MOF-5 performance towards FA can be primarily ascribed to the variations in the operating conditions during the experimental study. The present study was conducted at a comparatively low relative humidity of 12% (~3.5 times less than the previous study). As such, the difference between two data sets is suspected to signify the potent role of water molecules in occupying the sorption sites with an increase in the moisture content. In addition, it is possible to infer that a fraction of FA molecules might interact with water molecules, if present in the G-WS, to form methylene hydrate, which may be bound together to form paraformaldehyde (Trincado et al., 2017). 25

The generation of paraformaldehyde may cause a negative impact on the adsorption affinity due to the relative inactivation of the (FA) carbonyl group towards the surface active adsorption sites of the adsorbent. Also, such a conversion may lead to a deficiency of FA molecules in G-WS, rendering the overall sorption process ineffective (as the target FA molecules will escape into the effluent in the form of paraformaldehyde). Our experiments revealed the superior performance of UiO-66-NH2 for FA removal as compared to any other sorbents reported in the literature (Table 5). UiO-66-NH2 exhibited a very high sorption capacity of 69.7 mg g-1 with a large maximum partition coefficient value (14.4 mol kg-1 Pa-1) at a relative humidity of 12%. It should be noted that, an amine-functionalized silica exhibited an abnormally high sorption capacity of 1,200 mg g-1 with a very small partition coefficient value of 0.005 mol kg-1 Pa-1 (Srisuda and Virote, 2008) (Table 5). Interestingly, those authors used an initial FA concentration of 80,000 ppm under bone-dry conditions to obtain such a high sorption capacity value. However, the very low partition coefficient value (0.005 mol kg-1 Pa-1) suggested a significantly poorer performance of the sorbent. This observation further elucidates the concentration bias associated with the usage of adsorption capacity data for practical analysis purposes (Srisuda and Virote, 2008). On similar lines, ethylenediamine (ED) appended MIL-101 showcased a very high maximum adsorption capacity value of 164.9 mg g-1 under bone-dry conditions (Wang et al., 2016) (Table 5). However, its low partition coefficient value of 0.37 mol kg-1 Pa-1 indicates the effect of very high loading conditions of FA (e.g., initial concertation of FA at 150 ppm) (Wang et al., 2016) (Table 5). In another report made recently, a chitosan grafted β-cyclodextrin (CGC) polymeric material showcased a relatively inferior adsorptive performance with a maximum adsorption capacity and partition coefficient values of 15.5 mg g-1 and 0.14 mol kg-1 Pa-1, respectively at the initial FA concentration of 37.5 ppm (Yang et al., 2019) (Table 5). This relatively low performance can be

26

attributed to the predominance of weak chemical interactions (e.g., hydrogen bonding) between the sorbent surface and FA molecules along with a very high moisture content (40% relative humidity) in the G-WS (Yang et al., 2019) (Table 5). Nevertheless, the actual practical merit of CGC may lie in the fact that this polymeric material is completely biodegradable with enhanced regenerability (e.g., up to 4 adsorption-desorption cycles) without noticeable loss in the performance (Yang et al., 2019). Also, the potential utility of UiO-66-NH2 for practical applications is further supported by its superiority over conventional sorbents (e.g., activated carbon) (Table 5). Essentially, at 25 ppm FA, activated carbon in our study displayed a maximum adsorption capacity of 2.18 mg g-1 and a maximum partition coefficient of 0.06 mol kg-1 Pa-1 at BTV10. Also, in a previous study at 2.18 ppm FA at a relative humidity of 30% (Ma et al., 2011), activated carbon was observed to have a maximum sorption capacity of 0.08 mg g-1 and a partition coefficient of 0.138 mol kg-1 Pa-1.

4.2. Reusability of adsorbent As already discussed earlier (refer to section 3.2), one of the prominent pathways involved in the capture of FA molecules by amine functionalized sorbents (e.g., UiO-66-NH2) is the formation of stable imines, i.e., an overall chemisorptive process. According to an early study on an aminosilica sorbent, the spent material (after FA adsorption at 10% relative humidity which is comparable to 12% used in the present study) could not be regenerated even after heating it at very high temperatures (≥ 130°C) (Nomura and Jones, 2014). In contrast, some studies have reported the complete regeneration of materials spent for FA adsorption with the formation of Schiff bases (e.g., imine) (Vellingiri et al., 2019; Yang et al., 2019). Some amine-based adsorbents were regenerated completely after removing FA from an aqueous solution (Vellingiri et al., 2019). Also, as mentioned earlier, CGC was easily 27

regenerated after removing FA under a high relative humidity condition (40%) by simply heating the spent adsorbent at 60°C for 5 h (Yang et al., 2019). The common feature determining the regenerability of sorbents (in association with the formation of a Schiff base during FA adsorption) is tightly controlled by the lack or excess of water in the system. In the presence of excess moisture, the formed hemiaminal intermediate is stabilized without being dehydrated to yield the final imine product (although it is thermodynamically more feasible) (Nomura and Jones, 2014). Overtime, the stabilized hemiaminal species may revert back to the original amine form by liberating FA or convert to the imine form, i.e., formation of a Schiff base (Nomura and Jones, 2013) (Figure 7 (a)). However, the complete regeneration of amine functionalities through the liberation of adsorbed FA molecules can be favored by supplying a suitable external agent (e.g., heat or a specific regeneration procedure) to the spent material just after the FA adsorption process (conducted in the presence of excess water) (Nomura and Jones, 2014; Vellingiri et al., 2019; Yang et al., 2019). In contrast, the complete regeneration of amine functionalities may not be possible under low humidity conditions because the formed hemiaminal species would readily dehydrate to be converted into its final form (highly stable Schiff base) (Nomura and Jones, 2013; Nomura and Jones, 2014) (Figure 7 (a)). In case of largely non-regenerable amine functionalized adsorbents, the spent materials can be subjected to high-intensity ultraviolet irradiation so as to photolyze the formed Schiff base into other useful chemicals (Szulejko and Kim, 2015; Uchiyama et al., 2007). Additionally, if one desires to reuse/recycle the spent MOF materials (e.g., UiO-66-NH2), then one should consider the various factors of the MOF such as the chemical/thermal stability, structural features, and the nature of solvent utilized during the synthesis process (Kumar et al., 2019). The spent MOFs can also be suitably pyrolyzed, calcined, or carbonized under specific conditions to yield nanoparticles (e.g., nanotubes and 28

quantum dots) (Li et al., 2018; Maya et al., 2017; Yap et al., 2017). Such a process may be employed as a win-win strategy if one considers the inherently high costs involved in the synthesis of fresh nanoparticles (Kumar et al., 2019; Wakefield, 2008; Yap et al., 2017). Interestingly, the spent MOF particles can be regenerated via a stepwise treatment approach which involves the breaking down of parent MOF structure (dissolution in strong acidic solutions (e.g., nitric or hydrochloric acid)) followed by its recrystallization (subsequent treatment with an alkaline solution) (Han and Lah, 2015; Kumar et al., 2019). The abovementioned processes have been demonstrated to be effective for the reuse of spent MOFs on the lab-scale. Nonetheless, further investigations are imperative to evaluate the overall cost economics and feasibility of such reuse/regeneration strategies. Also, it is noteworthy that the practicality of non-regenerable or partially regenerable sorbents can be justified if they possess sufficiently high adsorption capacity towards the target pollutant (Nomura and Jones, 2014). The overall one-time usage of such materials may also be considered as a realistic option if it takes a sufficiently long time to reach saturation of target VOCs that are present under real-world conditions (e.g., 4-100 ppb range of FA in typical indoor environments) (Duncan et al., 2018).

5. Conclusions At present, relatively little is known about the sorptive removal of FA under real-world conditions (e.g., low FA concentration and typical pressure and ambient temperature/humidity conditions). This work analyzes the FA removal performances of several novel sorbents: the metal-organic frameworks MOF-5 and UiO-66-NH2, and the covalent-organic polymers CBAP-1 (EDA) and CBAP-1 (DETA). Activated carbon was the reference sorbent against FA under varying ambient conditions: two different FA concentration ranges, low (0.1-0.5 ppm) and high (5-25 ppm), and different partial pressures (0.01-0.05 Pa and 0.5-2.5 Pa). UiO-66-NH2 exhibited the maximum adsorption capacity of 29

69.7 mg g-1 at 25 ppm FA. In the low-pressure region, UiO-66-NH2 also exhibited the best breakthrough volume (BTV10: 534 L g-1 for 0.5 ppm FA, to 2,963 L g-1 for 0.1 ppm FA). In contrast, in the high-pressure region, the maximum BTV10 values varied considerably with the concentration range, such as 137 L g-1 of UiO-66-NH2 at 5 ppm FA; 144 L g-1 of CBAP-1 (DETA) at 10 ppm FA; and 36.8 L g-1 of CBAP-1 (EDA) at 25 ppm FA. A comparative analysis of our results with the available literature data indicated that UiO-66-NH2 is the most effective sorbent for the sorptive removal of FA under practical conditions. It has a very high maximum partition coefficient (14.4 mol kg-1 Pa-1) and BT performance compared to the other nanomaterials as well as conventional sorbents like activated carbon. The majority of our test conditions favored the Langmuir isotherm model over the Freundlich model. The superior performance of UiO66-NH2 was ascribed to the van der Waals interactions between the FA’s hydrocarbon “tail” and the linkers present in the MOF framework, along with the synergistic interactions between the C=O and the open metallic centers of the MOF. This investigation supports the possibility for the prospective application of advanced functional materials toward the sorptive removal of FA under ambient pressure/temperature and real-word conditions, especially at low FA concentrations and high humidity.

Acknowledgments The authors acknowledge the support made by the R&D Center for Green Patrol Technologies through the R&D for Global Top Environmental Technologies funded by the Ministry of Environment (MOE 2018001850001) as well as a grant from the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (Grant No: 2016R1E1A1A01940995). KHK also acknowledges support from the Korea Ministry of Environment (2015001950001) as part of “The Chemical Accident Prevention Technology Development Project” and the support of “Cooperative

30

Research Program for Agriculture Science and Technology Development (Grant No: PJ012521032018)” Rural Development Administration, Republic of Korea.

References Assadi, A. A., et al., 2018. Pilot scale degradation of mono and multi volatile organic compounds by surface discharge plasma/TiO2 reactor: Investigation of competition and synergism. Journal of Hazardous Materials. 357, 305-313. Ayawei, N., et al., 2017. Modelling and Interpretation of Adsorption Isotherms. Journal of Chemistry. 2017, 11. Bian, Y., et al., 2018. Metal–organic framework-based nanofiber filters for effective indoor air quality control. Journal of Materials Chemistry A. 6, 15807-15814. Brilli, F., et al., 2018. Plants for Sustainable Improvement of Indoor Air Quality. Trends in Plant Science. 23, 507-512. Chen, D., et al., 2014. Adsorption–desorption behavior of gaseous formaldehyde on different porous Al2O3 materials. Colloids and Surfaces A: Physicochemical and Engineering Aspects. 441, 433440. Chen, J., et al., 2019. Pt supported on long-rod β-FeOOH as an efficient catalyst for HCHO oxidation at ambient temperature. Catalysis Science & Technology. 9, 3287-3294. Chen, X., 2015. Modeling of Experimental Adsorption Isotherm Data. Information. 6. Dai, X., et al., 2018. Modeling and controlling indoor formaldehyde concentrations in apartments: Onsite investigation in all climate zones of China. Building and Environment. 127, 98-106. de Falco, G., et al., 2018a. A New Generation of Surface Active Carbon Textiles As Reactive Adsorbents of Indoor Formaldehyde. ACS Applied Materials & Interfaces. 10, 8066-8076. de Falco, G., et al., 2018b. Role of sulfur and nitrogen surface groups in adsorption of formaldehyde on nanoporous carbons. Carbon. 138, 283-291. Delikhoon, M., et al., 2018. Characteristics and health effects of formaldehyde and acetaldehyde in an urban area in Iran. Environmental Pollution. 242, 938-951. Drese, J. H., et al., 2011. Aminosilica Materials as Adsorbents for the Selective Removal of Aldehydes and Ketones from Simulated Bio-Oil. ChemSusChem. 4, 379-385. Duncan, S. M., et al., 2018. Oxygenated VOCs, aqueous chemistry, and potential impacts on residential indoor air composition. Indoor Air. 28, 198-212. Dutta, T., et al., 2018. Metal-organic framework and Tenax-TA as optimal sorbent mixture for concurrent GC-MS analysis of C1 to C5 carbonyl compounds. Scientific Reports. 8, 5033. Fang, R., et al., 2018. Efficient MnOx supported on coconut shell activated carbon for catalytic oxidation of indoor formaldehyde at room temperature. Chemical Engineering Journal. 334, 2050-2057. Feeney, R. E., et al., Carbonyl-Amine Reactions in Protein Chemistry. In: C. B. Anfinsen, et al., Eds.), Advances in Protein Chemistry. Academic Press, 1975, pp. 135-203. Feng, X., et al., 2018. Synergistic effects and mechanism of a non-thermal plasma catalysis system in volatile organic compound removal: a review. Catalysis Science & Technology. 8, 936-954. Foo, K. Y., Hameed, B. H., 2010. Insights into the modeling of adsorption isotherm systems. Chemical Engineering Journal. 156, 2-10. Gao, S., et al., 2010. Palladium nanoparticles supported on MOF-5: A highly active catalyst for a ligandand copper-free Sonogashira coupling reaction. Applied Catalysis A: General. 388, 196-201. 31

Garibay, S. J., Cohen, S. M., 2010. Isoreticular synthesis and modification of frameworks with the UiO-66 topology. Chemical Communications. 46, 7700-7702. Gu, Z.-Y., et al., 2010. MOF-5 Metal−Organic Framework as Sorbent for In-Field Sampling and Preconcentration in Combination with Thermal Desorption GC/MS for Determination of Atmospheric Formaldehyde. Analytical Chemistry. 82, 1365-1370. Guan, S., et al., 2018. A review of the preparation and applications of MnO2 composites in formaldehyde oxidation. Journal of Industrial and Engineering Chemistry. Hadei, M., et al., 2018. Indoor and outdoor concentrations of BTEX and formaldehyde in Tehran, Iran: effects of building characteristics and health risk assessment. Environmental Science and Pollution Research. Han, S., Lah, M. S., 2015. Simple and Efficient Regeneration of MOF-5 and HKUST-1 via Acid–Base Treatment. Crystal Growth & Design. 15, 5568-5572. Jo, S.-H., et al., 2017. The combined effects of sampling parameters on the sorbent tube sampling of phthalates in air. Scientific Reports. 7, 45677. Katz, M. J., et al., 2013. A facile synthesis of UiO-66, UiO-67 and their derivatives. Chemical Communications. 49, 9449-9451. Khan, A., et al., 2019a. A comparison of figure of merit (FOM) for various materials in adsorptive removal of benzene under ambient temperature and pressure. Environmental Research. 168, 96-108. Khan, A., et al., 2019b. The potential of biochar as sorptive media for removal of hazardous benzene in air. Chemical Engineering Journal. 361, 1576-1585. Kim, M., et al., 2018. Oxidation of gaseous formaldehyde with ozone over MnOx/TiO2 catalysts at room temperature (25°C). Powder Technology. 325, 368-372. Kim, S.-N., et al., 2012. Post-synthesis functionalization of MIL-101 using diethylenetriamine: a study on adsorption and catalysis. CrystEngComm. 14, 4142-4147. Kim, Y.-H., et al., 2017. Metal-organic frameworks as superior media for thermal desorption-gas chromatography application: A critical assessment of MOF-5 for the quantitation of airborne formaldehyde. Microchemical Journal. 132, 219-226. Kosuge, K., et al., 2007. Effect of Pore Structure in Mesoporous Silicas on VOC Dynamic Adsorption/Desorption Performance. Langmuir. 23, 3095-3102. Kumar, P., et al., 2019. Regeneration, degradation, and toxicity effect of MOFs: Opportunities and challenges. Environmental Research. 176, 108488. Larkin, P., 2011. Infrared and Raman Spectroscopy. Elsevier. Lee, K. J., et al., 2013. Toward an effective adsorbent for polar pollutants: Formaldehyde adsorption by activated carbon. Journal of Hazardous Materials. 260, 82-88. Li, Y., et al., 2018. Metal–organic framework derived Co,N-bidoped carbons as superior electrode catalysts for quantum dot sensitized solar cells. Journal of Materials Chemistry A. 6, 2129-2138. Liu, F., et al., 2018. Tungsten doped manganese dioxide for efficient removal of gaseous formaldehyde at ambient temperatures. Materials & Design. 149, 165-172. Ma, C., et al., 2011. Removal of low-concentration formaldehyde in air by adsorption on activated carbon modified by hexamethylene diamine. Carbon. 49, 2873-2875. Maya, F., et al., 2017. Magnetic solid-phase extraction using metal-organic frameworks (MOFs) and their derived carbons. TrAC Trends in Analytical Chemistry. 90, 142-152. Momma, K., Izumi, F., 2008. VESTA: a three-dimensional visualization system for electronic and structural analysis. Journal of Applied Crystallography. 41, 653-658. Na, C.-J., et al., 2019. High-performance materials for effective sorptive removal of formaldehyde in air. Journal of Hazardous Materials. 366, 452-465. Nomura, A., Jones, C. W., 2013. Amine-Functionalized Porous Silicas as Adsorbents for Aldehyde Abatement. ACS Applied Materials & Interfaces. 5, 5569-5577. 32

Nomura, A., Jones, C. W., 2014. Enhanced formaldehyde-vapor adsorption capacity of polymeric amineincorporated aminosilicas. Chemistry. 20, 6381-90. Pal, R., Kim, K.-H., 2008. Gas chromatographic approach for the determination of carbonyl compounds in ambient air. Microchemical Journal. 90, 147-158. PubChem, PubChem Database. Formaldehyde, CID=712 (accessed on Aug. 12, 2019). National Center for Biotechnology Information (https://pubchem.ncbi.nlm.nih.gov/compound/Formaldehyde), Bethesda, Maryland, USA, 2019. Puthiaraj, P., et al., 2017a. Dual-functionalized porous organic polymer as reusable catalyst for one-pot cascade CC bond-forming reactions. Molecular Catalysis. 441, 1-9. Puthiaraj, P., et al., 2017b. Microporous amine-functionalized aromatic polymers and their carbonized products for CO2 adsorption. Chemical Engineering Journal. 319, 65-74. Ravi, S., et al., 2017a. Cyclic carbonate synthesis from CO2 and epoxides over diamine-functionalized porous organic frameworks. Journal of CO2 Utilization. 21, 450-458. Ravi, S., et al., 2017b. Aminoethanethiol-Grafted Porous Organic Polymer for Hg2+ Removal in Aqueous Solution. Industrial & Engineering Chemistry Research. 56, 10174-10182. Song, Y., et al., 2017. Magnetic mesoporous polymelamine-formaldehyde resin as an adsorbent for endocrine disrupting chemicals. Microchimica Acta. 185, 19. Soni, V., et al., Effects of VOCs on Human Health. In: N. Sharma, et al., Eds.), Air Pollution and Control. Springer Singapore, Singapore, 2018, pp. 119-142. Spencer, P. S., 2018. Formaldehyde, DNA damage, ALS and related neurodegenerative diseases. Journal of the Neurological Sciences. 391, 141-142. Srisuda, S., Virote, B., 2008. Adsorption of formaldehyde vapor by amine-functionalized mesoporous silica materials. J Environ Sci (China). 20, 379-84. Suresh, S., Bandosz, T. J., 2018. Removal of formaldehyde on carbon -based materials: A review of the recent approaches and findings. Carbon. 137, 207-221. Szulejko, J. E., Kim, K.-H., 2015. Derivatization techniques for determination of carbonyls in air. TrAC Trends in Analytical Chemistry. 64, 29-41. Szulejko, J. E., Kim, K.-H., 2019. Is the maximum adsorption capacity obtained at high VOC pressures (>1000 Pa) really meaningful in real-world applications for the sorptive removal of VOCs under ambient conditions (<1 Pa)? Separation and Purification Technology. 228, 115729. Szulejko, J. E., et al., 2019. Seeking the most powerful and practical real-world sorbents for gaseous benzene as a representative volatile organic compound based on performance metrics. Separation and Purification Technology. 212, 980-985. Tanada, S., et al., 1999. Removal of Formaldehyde by Activated Carbons Containing Amino Groups. J Colloid Interface Sci. 214, 106-108. Tao, W.-H., et al., 2004. Effect of Moisture on the Adsorption of Volatile Organic Compounds by Zeolite 13×. Journal of Environmental Engineering. 130, 1210-1216. Thevenet, F., et al., 2018. VOC uptakes on gypsum boards: Sorption performances and impact on indoor air quality. Building and Environment. 137, 138-146. Trincado, M., et al., 2017. Homogeneously catalysed conversion of aqueous formaldehyde to H2 and carbonate. Nature Communications. 8, 14990. Uchiyama, S., et al., 2007. Acid-catalyzed isomerization and decomposition of ketone-2,4dinitrophenylhydrazones. Analytica Chimica Acta. 605, 198-204. Vellingiri, K., et al., 2019. Amine-Functionalized Metal–Organic Frameworks and Covalent Organic Polymers as Potential Sorbents for Removal of Formaldehyde in Aqueous Phase: Experimental Versus Theoretical Study. ACS Applied Materials & Interfaces. 11, 1426-1439. Vellingiri, K., et al., 2017. Metal-organic frameworks for the adsorption of gaseous toluene under ambient temperature and pressure. Chemical Engineering Journal. 307, 1116-1126. 33

Vellingiri, K., et al., 2016. Coordination polymers: Challenges and future scenarios for capture and degradation of volatile organic compounds. Nano Research. 9, 3181-3208. Vikrant, K., Kim, K.-H., 2019. Nanomaterials for the adsorptive treatment of Hg(II) ions from water. Chemical Engineering Journal. 358, 264-282. Vikrant, K., et al., 2019a. Nanomaterials for the abatement of cadmium (II) ions from water/wastewater. Nano Research. 12, 1489-1507. Vikrant, K., et al., 2019b. Evidence for superiority of conventional adsorbents in the sorptive removal of gaseous benzene under real-world conditions: Test of activated carbon against novel metalorganic frameworks. Journal of Cleaner Production. 235, 1090-1102. Wakefield, G., 2008. Nanomaterials: costs and opportunities. Nano Today. 3, 48. Wang, Z., et al., 2016. Diamine-appended metal–organic frameworks: enhanced formaldehyde-vapor adsorption capacity, superior recyclability and water resistibility. Dalton Transactions. 45, 11306-11311. Wu, L., et al., 2017. CNT-enhanced amino-functionalized graphene aerogel adsorbent for highly efficient removal of formaldehyde. New Journal of Chemistry. 41, 2527-2533. Xiao, R., et al., 2017. An in-situ thermally regenerated air purifier for indoor formaldehyde removal. Indoor Air. 28, 266-275. Yang, T., et al., 2017. Efficient formaldehyde oxidation over nickel hydroxide promoted Pt/γ-Al2O3 with a low Pt content. Applied Catalysis B: Environmental. 200, 543-551. Yang, Z., et al., 2019. Enhanced Formaldehyde Removal from Air Using Fully Biodegradable Chitosan Grafted β-Cyclodextrin Adsorbent with Weak Chemical Interaction. Polymers. 11. Yap, M. H., et al., 2017. Synthesis and applications of MOF-derived porous nanostructures. Green Energy & Environment. 2, 218-245. Zhang, H.-Y., et al., 2018. A Computational Study of the Adsorptive Removal of H2S by MOF-199. Journal of Inorganic and Organometallic Polymers and Materials. 28, 694-701. Zhang, L., Rana, I., CHAPTER 11 Formaldehyde Toxicity in Children. Formaldehyde: Exposure, Toxicity and Health Effects. The Royal Society of Chemistry, 2018, pp. 240-264. Zheng, K., et al., 2018. Highly luminescent Ln-MOFs based on 1,3-adamantanediacetic acid as bifunctional sensor. Sensors and Actuators B: Chemical. 257, 705-713. Žunkovič, E., et al., 2015. Structural study of Ni- or Mg-based complexes incorporated within UiO-66NH2 framework and their impact on hydrogen sorption properties. Journal of Solid State Chemistry. 225, 209-215.

34

Tables and Figures Table 1. Optimal conditions for FA analysis using TD-GC/MS and sensor systems a. High-performance Liquid Chromatography (HPLC) Settings HPLC/UV (Spectrasystem UV 2000, ThermoFisher Scientific, USA) Column: C18 (5 µm, 2.1 x 250 mm, Acclaim 120 C18) (1) Analysis setting (2) Detector setting Acetonitrile:H2O 70:30 Detector: -1 Flow rate: 1.5 mL min Injection: Time:

360 nm UV

20 µL 5 min

b. Gas Chromatography (GC) Settings GC (Shimadzu GC2010, Japan) and MS (Shimadzu GCMS-QP2010, Japan) Column: CP Wax (length: 60 m, diameter: 0.25 mm, and film thickness: 0.25 µm) (1) Oven setting (2) Detector setting Oven temp.: 40°C (5 min) Ionization mode: EI (70 eV) -1 Oven rate: 24°C min Ion source temp: 230°C Max oven temp.: 220°C (0 min) Interface temp: 230°C 29, 30, 43, 44 and 58 Total time: 10 min SIM mode m/z Thermal Desorber (UNITY2, Markes International, Ltd., UK) Cold trap: TENAX + MOF 5 (1:1) (TENAX: 5 mg + MOF 5: 5 mg) Split ratio: 1:10 Trap low: -25°C -1 Split flow: 10 mL min Trap high: 150°C Trap hold time:

5 min

Transfer line temperature:

150°C

c. FA Sensor setting (KINSCO Tech., South Korea) (1) Instrument setting (2) Sensor specification Flow rate:

50 mL min-1

Nominal Range: Maximum Overload: Resolution: Temperature Range:

35

0 -500 ppm 1,000 ppm 0.5 ppm -40°C - 50℃

Table 2. Performance of the tested adsorbents toward FA Order

Adsorbent

Formaldehyde concentration (ppm) 0.1

0.25

0.5

5

10

25

[A] 10% breakthrough volume (BTV10, L g-1) 1

MOF-5 [M5]

112

52.1

10.8

10

9.78

9.10

2

UiO-66-NH2 [UN]

2,963

613

534

137

31.4

30.20

3

CBAP-1 (EDA) [CE]

1,014

523

502

54.8

45.1

36.80

4

CBAP-1 (DETA) [CD]

630

159

146

49.5

40.2

35.60

5

Activated Carbon [AC]

372

170

57.9

31.1

26.9

21.40

[B] Adsorption capacity (mg g-1) At 5% breakthrough 1

MOF-5 [M5]

0.00

0.00

0.00

0.11

0.20

0.31

2

UiO-66-NH2 [UN]

0.11

0.14

0.29

0.51

0.71

0.85

3

CBAP-1 (EDA) [CE]

0.04

0.11

0.27

0.38

0.68

0.73

4

CBAP-1 (DETA) [CD]

0.03

0.08

0.10

0.14

0.27

0.71

5

Activated Carbon [AC]

0.02

0.10

0.12

0.17

0.31

0.42

At 10% breakthrough 1

MOF-5 [M5]

0.00

0.07

0.14

0.23

0.42

0.62

2

UiO-66-NH2 [UN]

0.39

0.41

0.54

0.84

1.43

2.87

3

CBAP-1 (EDA) [CE]

0.11

0.24

0.31

0.64

1.01

1.45

4

CBAP-1 (DETA) [CD]

0.06

0.11

0.24

0.41

0.98

1.13

5

Activated Carbon [AC]

0.07

0.21

0.29

0.44

0.89

1.02

At 50% breakthrough 1

MOF-5 [M5]

0.06

0.14

0.15

0.21

0.42

1.02

2

UiO-66-NH2 [UN]

1.22

1.60

1.98

8.36

10.87

11.90

3

CBAP-1 (EDA) [CE]

0.29

0.47

0.72

4.29

6.53

11.40

4

CBAP-1 (DETA) [CD]

0.21

0.23

0.30

1.32

3.52

3.91

5

Activated Carbon [AC]

0.55

0.59

0.63

0.95

1.02

1.05

-

0.44

0.98

1.66

At 100% breakthrough 1

MOF-5 [M5]

-

-

2

UiO-66-NH2 [UN]

-

-

1.72

8.72

9.84

69.70

3

CBAP-1 (EDA) [CE]

-

-

1.00

-

8.35

27.80

4

CBAP-1 (DETA) [CD]

-

-

0.70

-

4.92

9.74

5

Activated Carbon [AC]

-

-

0.90

-

1.83

2.18

[C] Partition coefficient (PC, mol kg-1 Pa-1) at BTV10

36

1

MOF-5 [M5]

0.05

0.08

0.01

0.13

0.00

0.06

2

UiO-66-NH2 [UN]

10.1

3.15

2.66

0.53

0.11

0.13

3

CBAP-1 (EDA) [CE]

3.23

1.15

2.62

0.21

0.46

0.16

4

CBAP-1 (DETA) [CD]

0.76

0.14

0.11

0.19

0.70

0.18

5

Activated Carbon [AC]

1.29

2.79

0.2

0.2

0.18

0.06

37

Table 3. Comparison of linearity between Langmuir and Freundlich for the tested sorbents Sorbent

Isotherm

CBAP-1 (EDA) [CE] CBAP-1 (DETA) [CD] MOF-5 [M5] UiO-66-NH2 [UN] Activated Carbon [AC]

Langmuir Freundlich Langmuir Freundlich Langmuir Freundlich Langmuir Freundlich Langmuir Freundlich

R2 values for the pressure regions Low (0.01 – 0.05 Pa) High (0.5 – 2.5 Pa) 0.9936 0.9845 0.9743 0.9607 0.7085 0.8852 0.9084 0.6313 0.9810 0.9230

0.9442 0.9051 0.8164 0.7832 0.7439 0.9194 0.9081 0.9451

Table 4. Langmuir isotherm parameters for the low and high pressure conditions

Sorbent

CBAP-1 (EDA) [CE] CBAP-1 (DETA) [CE] MOF-5 [M5] UiO-66-NH2 [UN] Activated Carbon [AC]

Pressure region

Cm = (1/intercept) (mg g-1)

KL = (1/intercept)/ingredient (Pa-1)

Henry’s constant (KH) = (Cm/Molecular weight)*KL (mol kg-1 Pa-1)

Low High Low High Low Low High Low High

0.57 22.1 0.45 -0.43 0.08 1.43 -49.5 -0.33 -2.65

181 2.07 169 -52.2 0.38 389 -4.59 -126 -29.0

3.47 1.53 2.51 0.75 0.001 18.5 7.58 1.38 2.56

38

Table 5. Performance comparison of different adsorbents toward gaseous FA Order

Adsorbent

Initial Concentration (ppm)

Temperature (K)

Relative Humidity (%)

Max. Adsorption capacity (mg g-1)

Partition coefficient (mol kg-1 Pa-1) (Assuming Henry’s law)

Reference

1

NH2-functionalized silica

80,000

-

-

1,200

0.005

(Srisuda and Virote, 2008)

2

MOF-5

23.2

298

45

0.11

1.34 x 10

-4

(Gu et al., 2010)

3

polymeric amino silica

200

-

10

129

0.21

(Nomura and Jones, 2014)

4

CNT-enhanced amino-functional graphene aerogel

16.28

-

-

27.4

0.55

(Wu et al., 2017)

5

Activated carbon

2.18

298

30

0.08

0.14

(Ma et al., 2011)

6

Chitosan grafted β-cyclodextrin (CGC)

37.5

293

40

15.5

0.14

(Yang et al., 2019)

7

Ethylenediamine (ED) appended MIL-101

150

-

-

164.9

0.37

(Wang et al., 2016)

8

MOF-5

25

298

12

1.66

0.13

This Study

9

UiO-66-NH2

25

298

12

69.7

14.4

This Study

10

CBAP-1 (EDA)

25

298

12

27.8

13.2

This Study

11

CBAP-1 (DETA)

25

298

12

9.74

1.7

This Study

12

Activated carbon

25

298

12

2.18

3.52

This Study

39

Figure 1. Schematic of AS/TD-GC/MS system for the analysis of lower FA concentrations (0.1 – 0.5 ppm)

40

Figure 2. Schematic of the sensor system for the analysis of higher FA concentrations (5 – 25 ppm)

41

(a) Cout/Cin ratio (%)_0.1 ppm 100

Cout/Cin (%)

75

CE

CD

M5

UN

AC 50

25

0 0

5

10

15 20 Loaded volume (L)

25

30

35

(b) Cout/Cin ratio (%)_0.25 ppm 100

Cout/Cin (%)

75

50

25

CE

CD

UN

AC

M5

0 0

5

10 Loaded volume (L)

15

20

(c) Cout/Cin ratio (%)_0.5 ppm 100

Cout/Cin (%)

75

50

25

CE

CD

UN

AC

M5

0 0

2

4

6 8 Loaded volume (L)

10

12

Figure 3. Breakthrough performances of the tested adsorbents toward low FA concentrations (0.1 – 0.5 ppm) (ambient temperature and 2 mg adsorbents)

42

(a) Cout/Cin ratio (%)_5 ppm 100

Cout/Cin (%)

75

50

25

CE

CD

M5

UN

AC 0 0

2

4

6 8 Loaded volume (L)

10

12

14

(b) Cout/Cin ratio (%)_10 ppm 100

Cout/Cin (%)

75

50

25

CE

CD

M5

UN

AC

0 0

1

2 3 Loaded volume (L)

4

5

(c) Cout/Cin ratio (%)_25 ppm

Cout/Cin ratio (%)

100

75

50 CE

CD

UN

AC

M5

25

0 0

0

3 Log (Loaded volume (L))

30

Figure 4. Breakthrough performances of the tested adsorbents toward high FA concentrations (5 – 25 ppm) (ambient temperature and 2 mg adsorbents) 43

(a) Adsorption capacity (mg g-1)_0.1 ppm Adsorption capacity (mg g-1)

1.80 1.60 1.40 1.20 1.00 0.80

CE

CD

UN

AC

M5

0.60 0.40 0.20 0.00 0

5

10

15 20 Loaded volume (L)

25

30

35

(b) Adsorption capacity (mg g-1)_0.25 ppm Adsorption capacity (mg g-1)

1.20 1.00 0.80

CE M5 AC

0.60

CD UN

0.40 0.20 0.00 0

2

4

6 8 10 Loaded volume (L)

12

14

16

Adsorption capacity (mg g-1)

(c) Adsorption capacity (mg g-1)_0.5 ppm 2.00 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00

CE

CD

M5

UN

AC

0

2

4

6 8 Loaded volume (L)

10

12

Figure 5. Adsorption capacities of the tested adsorbents toward low FA concentrations (0.1-0.5 ppm) (ambient temperature and 2 mg adsorbents) 44

Adsorption capacity (mg g-1)

(a) Adsorption capacity (mg g-1)_5 ppm 20 18 16 14 12 10 8 6 4 2 0

CE

CD

M5

UN

AC

0

3

6 Loaded volume (L)

9

12

(b) Adsorption capacity (mg g-1)_10 ppm Adsorption capacity (mg g-1)

12.00 10.00 8.00

CE

CD

M5

UN

AC 6.00 4.00 2.00 0.00 0

1

2 3 Loaded volume (L)

4

5

Log (Adsorption capacity (mg g-1))

(c) Adsorption capacity (mg g-1)_25 ppm 100

10

1

CE

CD

M5

UN

AC 0 0

0

4 Log (Loaded volume (L))

40

Figure 6. Adsorption capacities of the tested adsorbents toward high FA concentrations (5-25 ppm) (ambient temperature and 2 mg adsorbents) 45

(a)

110

110

(b)

100 %Transmittance

%Transmittance

100 90 80 70 Before

60

After

50

(c)

90 80 70

Beginning of the emergence of Imine

60

-1

Before 50

40 380

1380 2380 Wavenumber (cm-1)

40 1490

3380

peak (1,630 cm )

After 1540

1590 1640 -1 Wavenumber (cm )

Figure 7. Interactions between the FA molecules and the UiO-66-NH2 framework. Panel (a) Schematic representation of the major interactions involved. Panel (b) FTIR spectra of UiO-66-NH2 before and after FA adsorption. Panel (c) Magnified view of the spectra to show the emergence of Imine peak at 1,630 cm-1. Note that the UiO-66-NH2 framework was drawn using ‘Visualization for Electronic and Structural Analysis (VESTA)’ software (Momma and Izumi, 2008). Also, the FA ball-and-stick models were obtained from PubChem database (PubChem, 2019). The zirconium, carbon, and oxygen atoms in the UiO-66-NH2 structure are represented by green, brown, and red atoms, respectively. In case of FA molecules, red, grey, and white colors represent oxygen, carbon, and hydrogen atoms, respectively.

46

1690

20

4.0

(a) MOF-5

(b) UIO-66-NH2

3.5 3.0

y = 0.0044x - 0.0202 y = 0.0018x + 0.7004 R² = 0.9084 R² = 0.7439

2.5 1/cap.

y = 0.2052x - 12.482 R² = 0.7085

10

2.0 1.5

5

High pres. region Low pres. region Linear (High pres. region) Linear (Low pres. region)

1.0

Preesure region

0.5 0.0

0 0

50

100

0

150

100 200 300 400 500 600 700 800 900 1000 1/P

1/P

8

12

(c) CBAP-1 (EDA)

(d) CBAP-1 (DETA)

10

6

y = 0.0096 x + 1.7405 R² = 0.9936

2

y = 0.0218 x + 0.0452 R² = 0.9442

50

High pres. region

1/cap.

Low pres. region 2

Linear (High pres. region) Linear (Low pres. region)

0 0

100 150 200 250 300 350 400 1/P 20

50

100

150

(e) Activated carbon

15 y = 0.0242x - 3.0564 R² = 0.981

10

y = 0.013x - 0.3771 R² = 0.9081

5

High pres. region Low pres. region Linear (High pres. region) Linear (Low pres. region)

0 0

200

y = 0.0133x + 2.245 R² = 0.9743

6 4

High pres. region Low pres. region Linear (High pres. region) Linear (Low pres. region)

0 0

y = 0.0442x - 2.3055 R² = 0.8164

8

4

1/cap.

1/cap.

1/cap.

15

400

600

800

1/P

Figure 8. Linearized Langmuir isotherm model fit for the tested adsorbents

47

1000

200 1/P

250

300

350

400

-1.0

1.0

(a) MOF-5

y = 1.0433x + 5.6795 R² = 0.6998

ln(Cap.)

0.5 y = 1.2932x - 1.0351 R² = 0.8852

-2.0

-2.5

y = 0.581x + 3.114 R² = 0.925

0.0

-0.5

Preesure region

-3.0

High pres. region Low pres. region Linear (High pres. region) Linear (Low pres. region)

-1.0 -1.5

-1

-0.5

0

-7

-6 ln(P)

ln(P) 0.0

0.0

(c) CBAP-1 (EDA) y = 0.9785x + 3.7077 R² = 0.9403

y = 1.7498x + 7.2874 R² = 0.815

-1.0 ln(Cap.)

-0.5

y = 0.5497x + 1.605 R² = 0.9897

-1.0

High pres. region Low pres. region Linear (High pres. region) Linear (Low pres. region)

-1.5

-5

(d) CBAP-1 (DETA)

-0.5 y = 0.5977x + 1.5765 R² = 0.9459

-1.5 High pres. region Low pres. region Linear (High pres. region) Linear (Low pres. region)

-2.0 -2.5 -3.0

-2.0 -6

-5 ln(P)

-6

-4

1.0

-5 ln(P)

(e) Activated carbon

0.5 y = 1.1508x + 5.31 R² = 0.9256

0.0 -0.5 ln(Cap.)

ln(Cap.)

lnCap.)

-1.5

(b) UIO-66-NH2

y = 1.4556x + 6.8881 R² = 0.992

-1.0 -1.5 -2.0 -2.5

High pres. region Low pres. region Linear (High pres. region) Linear (Low pres. region)

-3.0 -3.5 -4.0 -7

-6 ln(P)

Figure 9. Linearized Freundlich isotherm model fit for the tested adsorbents

48

-5

-4

Supplementary Information

Chemicals and adsorbent synthesis

Zinc nitrate hexahydrate (Zn(NO3)2•6H2O, 98%), 1,4-benzenedicarboxylate (terephthalic acid, BDC, 98%), trimethylamine (TEA, 99%), zirconyl chloride octahydrate (ZrOCl2•8H2O, 98%), terephthaloyl chloride (TC, 99%), 1,3,5-triphenylbenzene (97%), dichloromethane (DCM, 99.8%), anhydrous aluminum chloride (AlCl3, 98%), and amino terephthalic acid (BDC-NH2, 99%) were purchased from Merck. Ethanol (99.5%), methanol (99%), ethylenediamine (EDA, 99%), diethylenetriamine (DETA, 99%), sodium borohydride (NaBH4, 98%), and dimethylformamide (DMF, 99.0%) were purchased from Samchun Chemicals (Korea), and chloroform (99.5%) from Daejung (Korea). Charcoal-activated carbon (212 mesh) was obtained from Duksan Pure Chemicals (Korea). For the synthesis of MOF-5, the protocol given in (Dutta et al., 2018) was used after slight modifications. The metal precursor solution was made by dissolving 1.4 g of Zn(NO3)2•6H2O into 25 mL DMF (poured into a 100-mL beaker) that was homogenized at 1,000 rpm (hotplate stirrer; DAIHAN Scientific, Republic of Korea) via a magnetic bar. The linker solution was prepared similarly by dissolving 0.3 g of BDC in 25 mL DMF. The metal precursor solution was then mixed with the linker solution and stirred. The TEA (2 mL) was added drop wise into the reaction mixture to act as a catalyst for the in situ synthesis of MOF-5. The obtained solution was left to stand for 120 min at room temperature after being covered with standard aluminum foil. The precipitate formed in the solution was filtered out (WhatmanTM glass microfiber filter, 47-mm diameter, UK) under 5 min using a minidiaphragm vacuum pump (N86KT.18; KNF Neuberger Ltd., Seoul, South Korea) operating at a pressure difference of 0.01 bar. The surplus BDC present in the residue was removed by washing twice with DMF. Chloroform (40 mL) was used to replace the DMF with a 12-h standing time before replenishment with fresh chloroform. The chloroform was removed via filtration after 48 h and the obtained product was dried at 90°C for 8 h in an oven (CO-150; Hanyang Scientific Equipment Co., Ltd., Republic of Korea). UiO-66-NH2 was synthesized by following the procedures reported in (Katz et al., 2013). Essentially, the linker solution was prepared by dissolving 134 mg BDC-NH2 in 10 mL DMF. The

49

metal solution was made by dissolving 125 mg ZrOCl2•8H2O in DMF (5 mL), followed by the addition of hydrochloric acid (1 mL). The obtained solution was ultrasonicated for 20 min. Both solutions were then mixed and ultrasonicated for 30 min. The resultant solution was heated for 12 h at 80°C to obtain pale yellow precipitates, and the obtained solids were filtered and washed with 30 mL DMF and ethanol, twice each. The product was soaked in 10 mL ethanol for 24 h (repeated three times) and then heated for 12 h at 90°C. The synthesis of the COPs was done by following the protocol in (Puthiaraj et al., 2017b). Briefly, 3.05 g of TC, 3.06 g of 1,3,5-triphenylbenzene, and 180 mL DCM were mixed and purged with N2 for 15 min. Subsequently, 2 g of anhydrous AlCl3 was added to the reaction mixture, followed by refluxing for 12 h to produce dark brown powders. The powdered product was filtered and washed three times (each 30 mL) with DCM and methanol. Finally, the product was dried under vacuum at 130°C, producing a dark brown solid of CBAP-1. The freshly prepared 1 g of CBAP-1 was solubilized in 40 mL methanol, followed by addition of 2 mL EDA (or DETA). The mixture was refluxed at 80°C for 15 h under vigorous stirring and finally cooled to room temperature. The resulting Schiff-base intermediate was reduced with excess NaBH4 before being vigorously stirred for 10 h at room temperature. Then, the product was filtered and washed with methanol and water several times. Finally, the resultant products of CBAP-1 (EDA) and CBAP-1 (DETA) were dried in an oven at 130°C for 12 h.

50

BET surface area

Single point surface area (P/Po)

Pore volume

Adsorption average pore diameter (4V/A)

m2 g-1

m2 g-1

cm3 g-1

nm

424 963

535 749

0.22 0.58

2.07 1.56

Adsorbents

MOF-5 UiO-66-NH2

Table S1. Surface properties of the analyzed adsorbents.

51

CBAP-1 (EDA) CBAP-1 (DETA) Activated Carbon

674 667 1,004

558 541 831

52

0.23 0.32 0.71

1.69 1.90 1.21

Figure S1. PXRD patterns of (a) MOF-5, (b) UiO-66-NH2, (c) CBAP-1 (EDA), and (d) CBAP-1 (DETA).

53

Figure S2. FTIR spectra of (a) MOF-5, (b) UiO-66-NH2, (c) CBAP-1 (EDA), and (d) CBAP-1 (DETA).

54

(b)

(a)

(d) (c)

(e)

Figure S3. SEM images of (a) MOF-5, (b) UiO-66-NH2, (c) CBAP-1 (EDA), (d) CBAP-1 (DETA), and (e) activated carbon.

55

Figure S4. TGA plots of (a) MOF-5, (b) UiO-66-NH2, (c) CBAP-1 (EDA), and (d) CBAP-1 (DETA).

56

(a) PC (mol kg-1 Pa-1)_0.1 ppm

Log (PC (mol kg-1 Pa-1))

100.00

10.00

1.00

CE

CD

UN

AC

M5

0.10 0

5

10

15 20 Loaded volume (L)

25

30

35

(b) PC (mol kg-1 Pa-1)_0.25 ppm

Log (PC (mol kg-1 Pa-1))

100.00

CE

CD

UN

AC

M5

10.00

1.00

0.10 0

2

4

6 8 10 Loaded volume (mL)

12

14

16

(c) PC (mol kg-1 Pa-1)_0.5 ppm

Log (PC (mol kg-1 Pa-1))

10.00

1.00

CE M5 AC

0.10

CD UN

0.01 0

2

4 6 8 Loaded volume (mL)

10

12

Figure S5. Partition coefficients of the tested adsorbents toward low FA concentrations (0.1-0.5 ppm) (ambient temperature and 2 mg adsorbents) 57

(a) PC (mol kg-1 Pa-1)_5 ppm

Log (PC (mol kg-1 Pa-1))

1.00

0.10

CE

CD

UN

AC

M5

0.01 0

0 Log (Loaded volume (L))

4

(b) PC (mol kg-1 Pa-1)_10 ppm

Log (PC (mol kg-1 Pa-1))

10.00

1.00

0.10

0.01

CE

CD

UN

AC

M5

0.00 0

0 Log (Loaded volume (L))

4

(c) PC (mol kg-1 Pa-1)_25 ppm

Log (PC (mol kg-1 Pa-1))

10 CE M5 AC

1

CD UN

0

0 0

0

4 Log (Loaded volume (mL))

40

Figure S6. Partition coefficients of the tested adsorbents toward high FA concentrations (5-25 ppm) (ambient temperature and 2 mg adsorbents) 58

Highlights

• • • •

The adsorptive removal of gaseous formaldehyde is investigated using novel materials. Two MOFs and two COPs were evaluated for the adsorptive removal of formaldehyde. Sorption experiments were conducted in two partial pressure regions (0.01-0.05 and 0.5-2.5 Pa). UiO-66-NH2 displayed the best performance amongst all the tested adsorbents.