TEPA impregnation of electrospun carbon nanofibers for enhanced low-level CO2 adsorption

The CO2 adsorption selectivity of plain activated carbon nanofibers (ANF) is generally low. For enhancement, nitrogen functionalities favorable for CO2 adsorption are usually tethered to the ANF. In the current study, we adopted chemical impregnation using 0.5 wt% tetraethylenepentamine (TEPA) solution as an impregnant. To enhance the impregnation of TEPA further, preliminary oxidation of the nanofibers with 70% HNO3 was conducted. The effects of HNO3 and TEPA treatments on the modified ANFs were investigated for physical (using N2 monosorb, thermogravimetric analyzer, scanning electron microscopy) and chemical (X-ray photoelectron spectrometer) changes. From the results, we found that although TEPA impregnation reduced the specific surface area and pore volume of the ANFs (from 673.7 and 15.61 to 278.8 m2/g and 0.284 cm3/g, respectively), whereas the HNO3 pre-oxidation increased the number of carboxylic groups on the ANF. Upon TEPA loading, pyridinic nitrogen was tethered and further enhanced by pre-oxidation. The surface treatment cumulatively increased the amine content from 5.81% to 13.31%. Consequently, the final adsorption capacity for low (0.3%) and pure CO2 levels were enhanced from 0.20 and 1.89 to 0.33 and 2.96 mmol/g, respectively. Hence, the two-step pre-oxidation and TEPA treatments were efficient for improved CO2 affinity.


Introduction
Besides the highest contributor to anthropogenic global warming, CO 2 could also be harmful at relatively low concentrations, especially in confined indoor spaces, depending on whether they are, stationary (such as offices, homes, and subway stations) or mobile environments (such as cars, airplane, and submarines) [1,2]. Therefore, the control of indoor CO 2 at 1,000 ppm or higher concentrations [the limit set by the Environmental Protection Agency (EPA)] require efficient technologies for safe indoor activities with a long duration [3].
In the last two decades, indoor air quality (IAQ) researchers have intensified efforts to improve the adsorption of elevated levels of indoor CO 2 . The use of amine-based or amine-functionalized adsorbents is popular among such research approaches [4]. Usually, these adsorbents are fabricated using grafting or the impregnation method [5]. Some researchers have confirmed that both impregnation and grafting are effective for increasing the CO 2 adsorption capacity of carbonbased sorbents [6][7][8]. Furthermore, Na Rao et al. [9] compared the two methods and their results showed that impregnation is superior to grafting with respect to amine-loading efficiency and eventual CO 2 adsorption capacity. Impregnation of amines is usually a wet process during which amines are physically adhered onto a support via non-covalent attachment to improve the affinity of the support for a target adsorptive [10]. For example, the effects of three types of organic amines (i.e., diethylenetriamine, triethylenetetramine, and tetraethylenepentamine (TEPA)) on CO 2 adsorption capacity has been reported [11]. The results showed that if TEPA was used as a chemical impregnant for a solid support, it could improve CO 2 adsorption of the support.
TEPA is a linear molecule bearing five amine groups per molecule: two primary amines (RNH 2 ) and three secondary amine (R 2 NH) groups. The mechanism by which primary amine and secondary amine react with CO 2 are expressed in Eqs. (1) and (2), respectively [5]: However, to achieve optimum wet impregnation, adequate knowledge of certain properties of the support such as the specific surface area, pore size, pore structure, and surface pH is essential. In general, a porous support with a relatively high specific surface area and large pore volume has a good potential for organic amine immobilization [11].
In this study, various activated nanofibers (ANFs) were prepared by either physical or chemical activation. Further, all the ANFs were impregnated with TEPA solution. To improve the TEPA loading, some ANFs were pre-oxidized with nitric acid solution prior to impregnation.

Preparation of PAN-based carbon nanofibers
First, two types of PAN solutions were prepared in weight ratios: (i) PAN/DMF solution (PAN:DMF = 1:9) for physical activation and (ii) KOH/PAN/DMF solution (PAN:DMF:KOH = 1:9:0.01-0.05) for chemical activation. Each polymer solution was electrospun at an inject velocity of 1.5 mL/h under 18 kV. The electrospun nanofiber was stabilized in an oven at 200 °C for 4 h before cooling to room temperature. Then, cooling it to room temperature, it was transferred into a tubular quartz reactor and heated at a ramping rate of 5 °C/min to 800 °C under N 2 flow that was fed at 200 cm. Samples prepared with weight ratios (i) were activated for 15, 30, 60 and 90 min, and denoted as 15-ANF, 30-ANF, 60-ANF, and 90-ANF, respectively, whereas those prepared with weight ratios (ii) were activated for 15 min, and similarly denoted as 0.01-ANF, 0.03-ANF, and 0.05-ANF. (1) Furthermore, 0.5 wt% TEPA solution in absolute ethanol was prepared as an impregnating solution. Then, 100 mg of each sample was added to the solution, and stirred for 6 h at room temperature. Afterwards, the samples were evaporated to dryness in an oven. However, some of the ANFs were pre-oxidized with 70% HNO 3 solution at 80 °C for 1 h before TEPA impregnation. The HNO 3 pretreatment was preferably carried out below 100 °C for no more than 2 h [12]. Aside from ensuring a complete surface treatment of the fibers, the depth of the ravines increased and the fiber matrix remained uncorroded under those conditions.

Characterization of nanofiber adsorbent
The specific surface area (S BET ) and porosity of all the ANFs were measured by N 2 adsorption at 77 K using a surface analyzer (Belsorp mini II, BEL, Tokyo, Japan) at relative pressures (p/p 0 ) of 0-1. To pre-clean the samples prior to measurement, outgassing under N 2 flow for 3 h at 100 °C was carried out. The equation based on BET theory (Brunauer-Emmett-Teller) was used to calculate the S BET , whereas the mesopore volume (V meso ) and micropore volume (V micro ) were obtained using Barrett-Joyner-Halenda (BJH) and MP-plot [5]. The total pore volume (V total ) was analyzed at a relative pressure of 0.99. The thermal stability of the adsorbents was investigated with the aid of a thermogravimetric analyzer (TGA N-1500, Scinco, USA). Each sample was heated from room temperature to 400 °C at 10 °C/min heating rate under N 2 atmosphere. The surface morphology and feature of the samples were observed by field emission scanning electron microscopy (FE-SEM, LEO SUPRA 55, Germany) with the magnification of 50,000×. It was coupled with an energy-dispersive X-ray spectroscopy (EDS) for simultaneous elemental distribution analysis. In details, the changes in the surface chemical speciation, especially of the nitrogen functional groups, were investigated using X-ray photoelectron spectroscopy (XPS, K-Alpha, Thermo Scientific, USA).

Measurement of CO 2 adsorption capacity
The adsorption capacity of the adsorbents for pure CO 2 was obtained using a monosorb instrument. The measurement was carried out under pressures ranging from vacuum to ambient (i.e., p/p 0 = 0-1) at 273 K. Furthermore, the selective adsorption capacity for 0.3% CO 2 (99.999%, in binary mixture with 99.99% N 2 ) was examined using a lab scale set-up as shown in Fig. 1. The 0.3% (3000 ppm) CO 2 represents the mean concentration of the gas commonly found in indoor spaces [13]. A CO 2 detector (SenseAir, Sweden), equipped with a non-dispersive infrared (NDIR) sensor was used to monitor the CO 2 concentration in the exiting gas stream. The same set-up was used for adsorption regeneration exercise carried out at 100 °C. Table 1 shows the textural characteristics of all of the ANF samples. We observed that the S BET and V total of ANFs were enhanced by either activation time or the quantity of the activation reagent. However, after impregnation with TEPA, the S BET and V total were reduced proportionately, an observation that is consistent with those of other studies [11,14]. In particular, micropores of the chemically activated ANFs undetectable because the percolation of the TEPA solution blocked the tiny pores. On the contrary, the microporosity of the ANFs derived by physical activation increased after impregnation, probably due to the thermal shrinking of boundary mesopores. Since the kinetic diameter of CO 2 is 0.33 nm, a factor that is significant to its physisorption, the proportion of microporosity is essential to the adsorption of the gas. Therefore, we inferred that ANFs prepared by physical activation are more suitable for surface modification by

Physical properties of the ANFs
Physically activated ANF alkaline solution impregnation, which would improve the chemisorption of the CO 2 molecules. The N 2 adsorption isotherms of a part of physically activated ANFs are compared in Fig. 2. All the isotherms followed a typical behaviour of Type I between 0 and 0.8 of the relative pressure, indicating single-layer reversible adsorption. Type I adsorption frequently occurs on microporous solids. The initial adsorption capacity is relatively large with the change of relative pressure, and then the adsorption capacity will not change greatly with the increase of relative pressure [15]. However, as P/Po approached 1, the adsorption capacity increased infinitely. This was caused by multiple molecular layers (physisorption) of the sorptive formed at various thicknesses on the solid surface, leading to the formation of Type II adsorption isotherm. Therefore, it can be seen that the ANFs prepared has both inner micropores and outer mesopores well developed.
To further enhance TEPA loading on the ANFs at the same impregnation concentration, the ANFs were oxidized with 70% HNO 3 solution for 1 h prior to TEPA impregnation. The HNO 3 oxidation formed crevasses and pits by eroding the graphene layer primarily from the basal planes of carbonaceous materials [16]. After HNO 3 oxidation, the surface became irregular and there were more oxygen-based active groups such as carboxyl, carbonyl and ether groups tethered onto the ANFs [17][18][19]. The presence of these active groups were advantageous for the attachment of TEPA onto the adsorbent surface via a displacement reaction [20,21].
The optimization of HNO 3 pre-oxidation was conducted with 60-ANF because the textural quality (S BET and V micro /V total ) of the 60-ANF-TEPA sample was found best in the TEPA-impregnation step. After pre-oxidation, the V total of 60-ANF decreased significantly from 15.61 to 0.463 cm 3 /g. This was a result of mesopore shrinkage, resulting in a large increase in the proportion of micropore volume (V micro /V total ) from 1.63 to 49.4%. The V micro /V total has been reported to be the driving force for selective adsorption of CO 2 [22]. Despite the significant improvement in this property, a slight decrease in the S BET of the 60-ANF-HNO 3 sample was observed, which was attributed to the collapse of some mesopores. From the TEPA impregnation results of the adsorbent (before and after oxidation), the pre-treated 60-ANF-HNO 3 -TEPA sample exhibited superior specific surface area and pore volume. However, the impregnation of TEPA depreciated the microporosity of the adsorbent (from 60-ANF-HNO 3 to 60-ANF-HNO 3 -TEPA), i.e., a much superior microporosity was developed compared to that of 60-ANF. Finally, the pre-oxidized TEPA-doped ANFs showed more superior properties than those that were chemically activated.
The surface morphology of fibers is shown in the FE-SEM micrographs of ANFs (Fig. 3), produced at 50,000× magnification with 10 kV accelerating voltage. It can be seen that various surface modified ANFs exhibited uniform size distribution of fiber thickness. However, there were obvious etching on the surface of the TEPAmodified ANFs (Fig. 3b, d), while those of 60-ANF-TEPA sample (Fig. 3b) was deformed. Generally, the structural effects of HNO 3 pre-oxidation on the ANFs correspond with the differences in the S BET and V total of 60-ANF-HNO 3 , which were higher than those of 60-ANF-TEPA and 60-ANF-HNO 3 -TEPA (Table 1).
To understand the thermal stability of modified ANFs, we performed a thermogravimetric analysis on the samples. The thermograms obtained are shown in Fig. 4. All of the samples showed different rates of weight loss as the volatile organics and moisture content were eluted below 100 °C. TEPA-treated samples evinced a less steep slope than those that were untreated, indicating more refractiveness of the contents introduced by TEPA treatment. Between 100 and 400 °C, the 60-ANF sample showed the most impressive thermal stability, as a steady plateau was observed. Such steadiness could be attributed to the homogeneity of the fibers, devoid of foreign dopants in the matrix as can be seen in Fig. 3a. However, the thermogram of pre-oxidized fibers exhibited steeper depreciation because the oxidized fiber surfaces were more easily decomposed under an increasing temperature [20]. The strong acidic functionalities such as carboxylic, anhydrides and lactones tend to be decomposed at low temperatures, while the weak acidic functionalities such as carbonyl, phenol and quinone are dissipated at high temperature [23,24]. The steady weight loss of TEPA impregnated samples (60-ANF-TEPA and 60-ANF-HNO 3 -TEPA) began at about 180 °C. From the difference in their thermograms, we inferred that chemical modification was most pronounced on the surface of the preoxidized 60-ANF-HNO 3 -TEPA sample.

Chemical properties of the ANFs
Through high resolution XPS analysis of the samples, the major peaks of the scan spectra determined to be due to the C 1s , O 1s , and N 1s photoelectrons. Table 2 lists the surface elemental compositions and atomic ratios of optimized 60-ANF samples. As expected, the HNO 3 treated sample (60-ANF-HNO 3 ) showed higher O 1s / C 1s ratios (14.75%) and N 1s /C 1s ratios (7.95%) than the unoxidized precursor (60-ANF). With HNO 3 treatment, the amount of O 1s and N 1s increased significantly, while  the proportion of C 1s decreased due to surface eroding of the graphene layers, attributed to wet oxidation [25]. With the same impregnant concentration, the oxidized 60-ANF-HNO 3 -TEPA sample showed a higher N 1s /C 1s ratios (17.19%) than that of the 60-ANF-TEPA (12.81%) sample, indicating that the HNO 3 oxidation treatment improved the loading of TEPA.
To better understand the chemical speciation of C and N on the selected samples, deconvolution of the XPS spectra was performed in the C 1s and N 1s regions. Figure 5 shows the best curve fit for the high resolution XPS C 1s spectra of all of the samples. Table 3 [26]. With either HNO 3 or TEPA treatment, the appearance of -OH and C=O groups on the fiber surface resulted in a decrease in the relative proportion of C (sp 2 ) and C (sp 3 ). Wet oxidation often incorporates high amount of oxygen, as carboxylic and phenolic functionalities on the carbon surface [20]. As expected, upon HNO 3 oxidation (60-ANF-HNO 3 ), the amount of -COOH groups increased significantly. However, the content of -COOH groups in the TEPA impregnated 60-ANF-TEPA and 60-ANF-HNO 3 -TEPA samples decreased by 0.4% and 2.87% respectively, compared to that of 60-ANF and 60-ANF-HNO 3 . Of the two possible types of acidic reactive groups present on the surface (phenolic -OH and carboxylic -COOH), only the carboxyl groups could undergo a reaction with TEPA because both phenolic and aliphatic hydroxyl groups are essentially inert toward primary and secondary amino groups [27]. This result indicated the successful substitution reaction of the acidic oxygen functionalities by the N-bearing groups from TEPA.
The nature and amount of the N species tethered by the amination were investigated. Figure 6 shows the best curve fit for the high resolution XPS N 1s spectra of all samples, while Table 4 lists the percentage of nitrogen-containing functional groups. The primary surface nitrogen groups found on the ANFs  were pyridine-type (BE = 398.1 eV), pyridone and pyrrole (BE = 400.3 eV), quaternary-N (BE = 401.5 eV), oxidized-N (BE = 402.8 eV) [28], and chemisorption of NO 2 at 405 eV [14]. Pyridinic and pyridine/pyrrole N were the two dominant N-groups found on the 60-ANF. Both accounted for about 74% of the total N, while the other groups were present in minor to trace quantities. However, upon the impregnation of TEPA (60-ANF-TEPA), there was a significant increase in the pyridinic and quaternary N-groups, possibly derived from the total conversion of shake-up satellites and adsorbed NO 2 . Such observation which implied that the linear TEPA resulted in the increase of pyridine-like structures or six-member or five-member rings [14]. As expected, HNO 3 oxidation evinced lowering of basic N-groups (pyridines), while the least basic groups (shake-up satellites and adsorbed NO 2 ) were re-introduced to the ANFs. Furthermore, final TEPA amination of the pre-oxidized samples (60-ANF-HNO 3 -TEPA) ensured that the quantity of the desired pyridinic N was optimized, while the unwanted shakeup satellites and adsorbed NO 2 groups were significantly lowered. We concluded that although TEPA is a viable amination agent, it also incorporates refractory quaternary-N while expunging unwanted N groups. In addition, HNO 3 was confirmed as an excellent pre-oxidant for TEPA amination of ANFs, as it improved the tethering of preferred pyridine groups which are favorable for selective adsorption of CO 2 over N 2 [29].
To complement the elemental distribution analysis carried out with the XPS, an EDS analysis was performed on the optimized sample (60-ANF-HNO 3 -TEPA). As shown in the Fig. 7, the EDS spectrum identified only a trace amount of N on the sample surface. Also, finite but  uniform distribution of carbon and oxygen was observed. This was because XPS analysis is more sensitive than EDS, with a minimum detection concentration of 0.1%, although at the analysis depth of 10 -20 nm only, while EDS analysis occurs bulk measurement that reaches micron level [30,31]. Since the impregnated N-groups were populated on the surface of the adsorbent, the EDS analysis could only detect a lesser percentage compared to those of the XPS. Still, both measurements give comparatively similar results when compared in proportion. Table 5 shows the adsorption capacities for 0.3% and 100% CO 2 of all the test samples. We found that with either an increase in the activation time or the quantity of the activation reagent, the CO 2 adsorption capacity increased for both levels. This result is fundamentally true because gas adsorption is positively correlated with both the specific surface area and microporosity [32]. Although the S BET of chemically activated ANFs was smaller than that of physically activated ANFs, the adsorption capacities of the former for pure CO 2 were still significantly improved. In regards to the 0.3% CO 2 test feed (Table 5), physically activated ANFs exhibited significantly higher adsorption capacities than their counterparts. We ascribed this result to the larger specific surface area induced by physical activation, which, in turn, favored TEPA impregnation and eventually CO 2 capture. In addition, the pre-oxidized 60-ANF-HNO 3 -TEPA sample had higher amino loading (Table 2 and Fig. 4), resulting in the adsorption of 0.3% CO 2 , which was about quadruple that of not pre-oxidized 60-ANF-TEPA. Table 6 shows a comparison list of the CO 2 adsorption capacities of various adsorbents from open literatures.  For all the sorbents, the higher S BET and V micro /V total were, the higher the adsorption capacity of pure CO 2 was. In our study, an increase in amino group content on the surface of the modified samples also led to a relatively high CO 2 adsorption capacity. However, as most published works have focused on pure and flue gas CO 2 levels, our work is unique in that we attempted to improve the CO 2 adsorption selectivity from indoor low levels.

Regeneration capacity of the ANFs
The regeneration of our representative samples for each treatment were examined by a simple temperature programmed desorption (TPD) using the TGA. The desorption temperature was set at 100 °C under N 2 atmosphere. It has been confirmed that the desorption of CO 2 form CO 2 -TEPA interaction is likely occurred in two stages: (i) slow transport via the decomposition of primary ammonium-carbamate species, followed by (ii) rapid desorption of primary and secondary ammonium-carbamate species at 100 °C. At 100 °C, the ammonium-carbamate species was eliminated, which liberated most of the adsorbed CO 2 [38]. Figure 8 shows the 5-cycle regeneration adsorptions of CO 2 (3000 ppm). Quantitatively, the adsorption capacity of sample 60-ANF, 60-ANF-TEPA and 60-ANF-HNO 3 -TEPA decreased mildly by 2.62%, 11.49% and 3.58% respectively, after 5 cycles. The regeneration performance of the TEPA impregnated samples was relatively weak, which may be caused by the thermal-induced elution of some TEPA molecules from the surface.

Conclusions
Nanofibers were prepared by electrospinning before physical or chemical activation to obtain activated nanofibers (ANFs) with varying structural and chemical characteristics. All of the ANFs were impregnated with a TEPA solution of varying amounts. By examination, we found that the ANFs that were physically activated were more suitable for TEPA impregnation because of the enhanced specific surface area and porosity. After TEPA impregnation, the specific surface area and total pore volume of the adsorbents decreased significantly. After the HNO 3 pre-oxidation treatment, the amine loading improved the incorporation of useful, basic pyridinic nitrogen, whereas the least basic N functionalities were effectively removed or converted. Therefore, the adsorption capacity of the optimized 60-ANF-HNO 3 -TEPA for 0.3% and 100% CO 2 increased from 0.12 and 1.89 mmol/g (by 60-ANF-TEPA) to 0.33 and 2.96 mmol/g, respectively. A 5-cycle regeneration test showed a relatively stable adsorption of thermally refreshed adsorbents for proficient reusability. Consequently, HNO 3 pre-oxidation of TEPA-doped ANF is efficient for enhancing low-level CO 2 adsorption from indoor spaces.