Nitrogen doped activated carbon derived from chitosan/hexamethylenetetramine: structural and CO2 adsorption properties

Chitosan and chitosan/hexamethylenetetramine (HMT) macrospheres were prepared respectively by dropping the solution of chitosan and chitosan/HMT (in wt/wt ratios of 1:1 and 1:3) in aq. NaOH solution. Here, HMT served as an additional nitrogen precursor for in situ N-doping with chitosan derived activated carbon (AC). The as-formed macrospheres were impregnated using ZnCl2, freeze-dried and carbonized at 500 °C under inert atmosphere to yield ACs of size ranging from 2.5 to 2.8 mm. All the samples were characterized using SEM, EDS, CHNS Analyzer, TGA, FT-IR, Raman, XRD and BET surface area analyzer. All the samples showed mesoporous characteristics. The surface area of the AC without HMT, with 1:1 HMT and 1:3 HMT were 391.502, 259.017 and 111.717 m2/g respectively. Similarly, the pore volumes of AC without HMT, with 1:1 HMT and 1:3 HMT were 0.138, 0.095 and 0.075 cc/g respectively. The chitosan/HMT derived AC possessed higher N-content and better thermal stability, however exhibited lower surface properties with increasing HMT content. The CO2 adsorption capacity of chitosan derived AC was 97.98 mg/g while that of AC with 1:1 HMT and 1:3 HMT were 72.95 mg/g, 55.11 mg/g respectively at 25 °C and 1 bar. Deterioration in adsorption capacity of chitosan/HMT derived AC may be attributed to the physical cross-linking of chitosan polymer chains induced by increase in pH with addition of HMT and also the intermolecular H-bonding interaction between HMT and chitosan, which in turn reduces the surface area of as-formed N-doped ACs progressively.


Introduction
One of the major problems faced by mankind in the twentyfirst century is global warming caused by an increase in concentration of CO 2 . According to the recent studies, accumulation of CO 2 in the atmosphere has been increased up to 413 ppm by 2021 [1]. Researches suggest a wide range of processes for the CO 2 removal from the atmosphere, such as absorption, membrane separation and adsorption [2]. Among them the widely adopted method is adsorption, where different porous materials are used as adsorbents, which include activated carbons [3][4][5], mesoporous silica [6,7], metal organic frameworks [8], covalent organic frameworks [9], zeolites [10], zeolitic imidazolate frameworks [11], etc. Activated carbon with tuned porosity is of significant interest in the adsorption of CO 2 among the proposed adsorbent materials. The adsorption capacity of activated carbons could be increased by doping nitrogen, as this approach improves the surface properties as well as the basicity of the carbon [12]. The improvement of basicity while providing electron donor moiety to the carbon network increases the adsorption of acidic CO 2 gas molecules. Basicity of activated carbon can be improved either by in situ nitrogen doping or by amine functionalization. Shivadas et al. reported the synthesis of activated carbon using sucrose as the carbon precursor and urea as the in situ nitrogen dopant. Degradation of urea during carbonization produced NH 3 , which resulted in the doping of nitrogen in the carbon matrix [13]. Similarly, Liu et al. also reported the use of urea as an in situ nitrogen dopant during the synthesis of activated carbon from furfural carbon precursors [14]. In another study, Wang et al. reported the synthesis of activated carbon using polyacrylonitrile (PAN), and the as-prepared activated carbon was functionalized using tetraethylene pentamine (TEPA) where the amine is adhered onto the support via a non-covalent attachment [15].
Chitosan is a potential precursor for activated carbon as it is the second most abundant polysaccharide next to cellulose [16] and second largest nitrogenous natural organic matter after protein. Chitosan is the de-acetylated derivative of chitin, which is obtained from the shells of crabs and shrimps [17] as well as cell walls of fungi. Also, it is a highly cost-effective material with an ease in processability, which makes it a potential carbon precursor. Several studies have been made previously on the preparation of activated carbon from chitosan. In 2013, Fan et al. reported the synthesis of chitosan derived activated carbon using K 2 CO 3 as the activating agent, which possessed a CO 2 adsorption capacity of 3.86 mmol/g [18]. Similarly Lee et al. in 2017 synthesized activated carbon from chitosan using KOH as the activating agent and further urea as the nitrogen dopant [19].
Few reports are available on activated carbon prepared from chitosan for the CO 2 capture studies [18,20,21]. But to the best of our knowledge, not much information was found on the enhancement of nitrogen functionalities on chitosan derived activated carbon. Nitrogen dopants such as HMT can act as an additional nitrogen source on activated carbon derived from biomass [22,23]. The incorporation of nitrogen into the carbon framework will modify the surface as well as CO 2 adsorption properties. Therefore, in the present study, chitosan was used as the carbon precursor and was mixed with HMT to prepare spherical N-doped activated carbon. As HMT possess four nitrogen atoms in its structure, nitrogen is expected to be doped into the carbon framework. The surface properties as well as the CO 2 adsorption capacity of both undoped and N-doped ACs were compared.

Preparation of activated carbon
Chitosan solution (2% wt/vol) was prepared in 5% aqueous lactic acid and was dropped into 2.5% NaOH solution using a peristaltic pump at a flow rate of 3 mL/min (nozzle size: 5 mm) in order to form chitosan macrospheres. Similarly, chitosan solution was mixed with hexamethylenetetramine in different ratios (wt/wt) and dropped into NaOH solution to form chitosan/HMT macrospheres. The macrospheres were stirred for 1 h, filtered, washed with distilled water and impregnated with ZnCl 2 (1:1 ratio with respect to the weight of chitosan) followed by drying by freeze-drying method. Carbonization of the macrosphere beads were carried out in a tubular furnace at 500 °C for 90 min. under a nitrogen atmosphere. The carbonization temperature of chitosan was chosen from previous literatures, which showed a temperature between 300 °C and 600 °C yielded an improved surface characteristics for the chitosan derived activated carbon [24]. The resulting activated carbons were washed with 1 M HCl solution and hot distilled water to remove excess ZnCl 2 , until the negative result for Cl − ion with AgNO 3 . The samples before carbonization without HMT is named as CB-0, while the samples with 1:1 HMT and 1:3 HMT before carbonization are named as NCB-1 and NCB-3 respectively (Fig.  S1a). The activated carbons without HMT incorporation is denoted as C-0 while the activated carbons with 1:1 HMT and 1:3 HMT are denoted as NC-1 and NC-3 respectively (Fig. S1b).

Characterization
Iodine values of the activated carbons were determined following ASTM D4607-94 method. The surface morphology of the samples were studied using Scanning electron microscope (Vega3 Tescan). The N 2 adsorption-desorption isotherms were obtained at 77 K using Quantachrome Nova 2200e system. From the isotherm, the specific surface area was calculated using the Brunauer-Emmett-Teller (BET) equation, pore size distribution was obtained using Barrett-Joyner-Halenda (BJH) method and the total pore volume was calculated at 0.99 P/P 0 . X-ray diffraction studies for the samples were performed using PANalytical X'pert Pro. FT-IR studies were performed using ThermoNicolate Summit FTIR spectrometer. Raman spectra was collected using ReinshawinVia confocal Raman spectrometer. The elemental mapping was carried out by EDS method using EDAX Ametek and the elemental composition by CHN using Elementar Vario EL III. TGA Q500 V20 was used to evaluate the thermal properties of the activated carbon samples.

CO 2 capture studies
CO 2 capture studies were performed on the samples using U-tube experiments. Figure 1 depicts the adsorption setup to determine the CO 2 adsorption capacity of carbon samples. Briefly, 1 g of activated carbon sample taken in a U-shaped glass column (diameter = 1 mm, tapered column length = 7 cm) were preheated at 110 °C for 1 h and cooled in a desiccator for complete removal of moisture and weight of the sample in grams was noted prior to the adsorption process. High pure CO 2 gas (99.99%) at a flow rate of 100 mL/min was purged through the sample at 25 °C and change in weight (in grams) was monitored for every 5 min. until the adsorption profile reached saturation [25]. The CO 2 capture capacity of the activated carbon in mg/g was calculated using the following equation [26]: The desorption of CO 2 was performed on the above CO 2 saturated samples by heating at 150 °C. The regeneration studies were performed on the samples by continuing the adsorption-desorption process up to five cycles.
The CO 2 adsorption isotherm measurements were also performed using high-pressure Quantachrome iSorb HP1 instrument equipped with water circulation to maintain a temperature of 25 °C. The adsorption experiment was conducted under a pressure variation from 0 to 3 bar. Before the measurements the samples were degassed at 180 °C for 4 h in order for the complete removal of moisture and other gases. The higher magnification of carbonized samples showed that the incorporation of HMT caused a decline in the porous structure, which was observed to be the maximum in case of NC-3, followed by NC-1 and C-0 ( Fig. 2j-l). The decline in the porosity may be ascribed to the hydrogen bonding interaction between nitrogen atoms of HMT and -OH groups of chitosan (Fig. 3). The hydrogen bonding interaction brings the polymer chains closer together, resulting the hindrance for the evolution of volatile gases during the thermal activation process and thereby progressively reducing the porous structure.

Elemental analysis
EDS spot scans and mapping were used to evaluate the elemental composition survey of the activated carbon samples. The elemental mapping of the samples are presented in Fig. 4a-c and the spot analysis spectrum are shown in Fig. 4d-f. The elemental analysis by both the methods illustrate the distribution of C, N and O in all the samples that qualitatively confirmed the doping of nitrogen in the increasing order as the composition of HMT was increased. The percentages of C, H and N in the activated carbon samples were quantitatively obtained using CHN analysis (Table S1). The elemental analysis of C-0, NC-1 and NC-3 obtained from EDS spot scan as well as CHN analysis confirmed an improvement in the percentage of nitrogen as the concentration of HMT was increased. Thus upon carbonization, nitrogen from HMT was effectively doped on to the carbon framework [23]. Therefore, HMT acts as a better nitrogen dopant as compared to other nitrogen dopants such as urea and ammonia reported by Liu et al. [14] and Song et al. [27] respectively. The reported percentages of nitrogen obtained for N-doped activated carbon using urea as the nitrogen dopant was 7.5%, and ammonia as the dopant was 9.72% while that of HMT (present study) is 13.06%.

Thermal analysis
The profiles of thermogravimetric analysis (TGA) as well as derivative thermogravimetry (DTG) carried out at 20 °C/ min. under N 2 atmosphere for the samples before and after carbonization are shown in Figs. 5 and 6 respectively. The TGA plot depicted the weight losses as a function of temperature, while the DTG plot portrayed derivative of weight loss as a function of temperature. From Fig. 5a and b, it is observed that the removal of moisture happened for CB-0 at 120 °C, which is not prominent for the samples, NCB-1 and NCB-3. This may be due to an increase in the intermolecular H-bonding between chitosan and HMT, which restricts the later samples to coordinate water molecules with functional groups of chitosan (Fig. 2). CB-0 exhibited the 1st decomposition stage from 175 to 330 °C with a The decomposition for NCB-1 started around 190 °C and ended at 410 °C with a maximum decomposition temperature at 275 °C. The initial shoulder at 220 °C for NCB-1 observed in Fig. 5b may be due to the removal of functional groups and side chains of chitosan followed by the decomposition of HMT. The corresponding decomposition for NCB-3 started at 175 °C and ended at 415 °C with a maximum at 290 °C as a single decomposition event, which may be due to the simultaneous decomposition of functional groups, side chains of chitosan and HMT. The enhancement in the thermal decomposition temperatures for NCB-1 and NCB-3 as compared to CB-0 may be due to the existence of H-bonding interaction between chitosan and HMT, due to which a higher thermal energy is required to overcome the bonding interaction and to degrade the structure. The second stage of decomposition for CB-0 started at 330 °C and ended at 570 °C with a maximum at 410 °C, while the corresponding decomposition temperature for NCB-1 and NCB-3 was observed between 430 and 700 °C with a maxima at 560 °C, which may be attributed to the degradation of polymer backbone structure [28]. The slow degradation profile for NCB-1 and NCB-3 as compared to CB-0 indicates the polymer chains are held by HMT due to stronger physico-chemical interaction. NCB-3 showed the highest thermal stability as compared to CB-0 and NCB-1 both in terms of thermal degradation profile and char yield. Figure 6a and b represented the TGA and DTG graphs of activated carbons respectively. The TGA and DTG plot reveals that the decomposition for activated carbon samples occurred around 500 °C, which was caused by the elimination of labile functional groups -CO and -OH forming a stable carbon-carbon bond in the framework [29]. The thermal stability of N-doped carbon was found to be better than the undoped activated carbon. While considering the residues around 800 °C NC-1, and NC-3 possessed about 61%, whereas C-0 had 57% of residues. A similar result was reported by Hai Long Peng et al.
where the doping of activated carbon with pentaethylenehexamine (PEHA) resulted in an improved thermal stability of the carbon [30], which is attributed to the confining effect of nanopores. Figure S2a shows the FT-IR spectra of samples before carbonization. The bands obtained around 3600-3400 cm −1 for the three samples correspond to the O-H stretching vibration [31]. The band at 1593 cm −1 for the samples corresponds to the primary amine N-H bending while the bands around 1050 cm −1 attributed to the C-N stretching vibration of amine groups [32]. The sharp peak at 1380 cm −1 could be assigned to the -CH 3 symmetrical deformation mode. Even though there arises a H-bonding due to the HMT addition in case of NCB-1 and NCB-3, it is not prominently observed due to the broadened FT-IR peaks. Fig. S2b represents the FT-IR spectra of samples after carbonization. The bands around 3600-3400 cm −1 correspond to the O-H stretching vibrations, which is similar to the samples before carbonization. The bands around 1740 cm −1 was attributed to the C=O stretching vibrations. The appearance of a band at 1452 cm −1 was attributed to the C-H stretching vibration [33].

XRD analysis
The diffraction patterns of C-0, NC-1, and NC-3 are depicted in Fig. S3. No significant difference in the diffraction patterns was observed among the three samples. All the samples possessed patterns of a non-ordered material, i.e. an amorphous structure with a very low degree of graphitization [34]. The broad and weak peaks obtained at 2ϴ = 25° was ascribed to the (002) plane of polycyclic aromatic C-sheets of amorphous carbon in irregular orientation and reflection of graphene, whereas the peak at 2ϴ = 42° represented the (100) plane of 2D in-plane diffraction of graphene sheets [31]. A similar result was observed by Limin et al. in the case of N-doped activated carbon prepared from urea modified coconut shell [35]. The percentage crystallinity from the diffraction patterns were calculated to be 52.3%, 34.7% and 31.1% respectively for C-0, NC-1 and NC-3, which indicated that the amorphous nature of the carbons increased with the increase in N-doping. The decrease in the crystallinity can be attributed to the fact that during nitrogen doping, elemental nitrogen enters into the carbon rings by replacing carbon atom, thereby destroying its graphitic lamellar structure [36]. Figure 7 represents the Raman spectra of C-0, NC-1 and NC-3. All the samples showed two peaks at 1350 cm −1 and 1580 cm −1 corresponding to the D and G bands of graphitic carbon respectively. D band corresponds to the A 1 g mode similar to the in-plane breathing vibration, which is Raman inactive in large crystals that becomes active due to the presence of disorder parts like grain edges. The G band is attributed to the E 2 g in-plane vibrational mode of sp 2 carbon framework in the planar hexagonal crystal lattice [37]. The degree of disorder in the graphitic structure is determined by the intensity ratio, I D /I G . In the case of C-0, NC-1 and NC-3 the I D /I G values were calculated from the peak heights and found to be 1.  [38]. Further, in a study performed by Spesatto et al. it was seen that the I D /I G value of N-doped activated carbon was higher that of the undoped activated carbon [22,31]. The increase in the I D /I G value was consistent with the XRD data and this reveals doping of nitrogen in case of NC-1 and NC-3.

N 2 adsorption desorption isotherm
The N 2 adsorption desorption isotherms represented in Fig. 8a shows characteristics of mesoporous materials having a typical Type 4 isotherm with a hysteresis loop caused by the capillary condensation [39]. The N 2 adsorption capacity decreased as the ratio of HMT impregnation was increased. The surface areas of NC-1 and NC-3 were lesser than that of C-0 as shown in Table S2. Peng et al. reported a comparable results with PEHA loaded activated carbon [30]. Considering the BJH pore size distribution in Fig. 8b, the distribution around 2 nm showed the presence of micropores and the distribution between 2 and 50 nm showed the presence of mesopores. The mean pore diameters 2.211 nm, 2.105 nm and 2.125 nm for C-0, NC-1, and NC-3 respectively calculated from the BJH pore size distribution indicates that all the three samples possessed mesoporous characteristics [40].
The iodine values for the three samples obtained using ASTM D4607-94 method [41] were 1262 mg/g, 1196 mg/g and 1080 mg/g respectively, which also indicated that N-doping with HMT results in the decrease of porosity. The total pore volume estimated from the amount of N 2 adsorbed at P/P 0 = 0.99 for C-0, NC-1 and NC-3 were 0.138 cc/g, 0.095 cc/g and 0.075 cc/g respectively. Among the three samples C-0 possessed the maximum surface area and pore volume. Therefore, N-doped activated carbon prepared using HMT as an additional nitrogen precursor leads to deterioration in the structural characteristics, which corroborate with XRD results discussed earlier [42]. The surface characteristics, nitrogen content as well as the CO 2 adsorption capacity of these samples were compared with other N-doped activated carbons reported in previous literatures and is listed in Table 1. The activated carbon prepared from chitosan using simple synthesis method in this study exhibited better CO 2 adsorption capacity as compared to other N-doped activated carbons. In the case of CO 2 sorbents, which showed higher surface area and adsorption capacity in the previous literatures, the method of preparations were complex and further the raw materials used in the processes were costlier.

CO 2 adsorption studies
The CO 2 adsorption capacity obtained from the U-tube experiments for C-0, NC-1 and NC-3 were 97.98 mg/g, 72.95 mg/g and 55.11 mg/g respectively ( Fig. 9) with an average uncertainty of 2%. TGA and DTG studies were performed on the CO 2 adsorbed carbon samples (Fig. S4) to evaluate the optimal regeneration temperature. The initial weight loss between 30 °C and 150 °C with a maximum weight loss at about 60 °C indicated the removal of adsorbed CO 2 from the carbon samples (Fig. S4b). The amounts of residue after the initial weight loss was found to be 78%, 81% and 86% for C-0, NC-1 and NC-3 respectively (Fig.  S4a), which also confirms that C-0 possess the maximum CO 2 adsorption capacity. The TGA plot also revealed that the complete desorption of CO 2 from the samples occurred around 150 °C. Therefore, regeneration studies were performed on the samples by heating at 150 °C (Fig. S5) and all the samples showed an excellent regeneration capacity. CO 2 adsorption isotherms for all the AC samples were also studied under a pressure variation from 0 to 3 bar at 25 °C. The shape of isotherms of all the samples is monotonically concave and can be classified as Type 1 according to the IUPAC nomenclature [49]. The CO 2 adsorption isotherms (Fig. 10) obtained for the AC samples convey that C-0 has the highest adsorption capacity as compared to NC-1 and NC-3. The adsorption capacities for C-0, NC-1 and NC-3 at 1 bar were 96.40 mg/g, 72.30 mg/g, and 54.90 mg/g respectively. These results corroborate well with the results obtained from U-tube experiments. As pressure is increased, the adsorption capacity of the samples further increased. The adsorption capacities at ~ 3 bar for C-0, NC-1 and NC-3 were 135.00 mg/g, 113.66 mg/g and 86.95 mg/g respectively. The increased adsorption capacity is due to the fact that an increase in partial pressure of CO 2 results an increase in the diffusion CO 2 molecules in to the pores of the activated carbon. Hence, it indicates that at lower pressure, adsorption happens only in the surface of macropores while at higher pressures adsorption starts to occur in the meso as well as the micropores of the AC samples [50]. Generally, it is perceived that the CO 2 adsorption capacity of materials with larger surface area and pore volume will be higher than those with smaller surface area and pore volume. In this study also the adsorption capacity positively correlate with the surface area and pore volume of the AC samples. Among the three samples C-0, which has the maximum surface area and pore volume, possessed the highest CO 2 adsorption capacity. However, in the case of NC-1 and NC-3, the surface area as well as pore volume decreased significantly and as a result the adsorption capacity also got reduced. In order to study further about the CO 2 adsorption process, the experimental data were fitted to Langmuir and Freundlich isotherm models following non-linear fitting method [51,52]. The correlation coefficient (R 2 ) values obtained from the curve fitting method are also presented in Fig. 10. Among the two models Langmuir model suited the most for NC-1 and NC-3 samples, whereas C-0 followed Freundlich model as shown by their higher correlation coefficient (R 2 ) values. The models reveal that the CO 2 molecules are adsorbed on C-0 through physisorption method and in case of nitrogen doped activated carbons, the adsorption follows primarily chemisorption method [22,42]. As discussed in the previous section, the nitrogen doped activated carbons are being characterized with enhanced basicity as compared to undoped activated carbon. Hence in cases of NC-1 and NC-3 samples, the Lewis acid-base interaction between nitrogen moieties of carbon and CO 2 molecules induces chemisorption due to an enhanced polarization of electron density from the former to the later. However, the adsorption of CO 2 on undoped activated carbon occurred mostly due to physisorption method [52].
In general, N-doped activated carbons exhibit an enhanced CO 2 adsorption as compared to their undoped counterparts as reported in previous studies [53,54]. The nitrogen functionalities due to its basic nature react with Lewis acidic CO 2 and increase the chemisorption to a greater extent. However, in the present study, the decline in the adsorption capacity of N-doped activated carbon was due to the deterioration of their structural characteristics (i.e. surface area and pore volume) as obtained from structural studies discussed in previous sections. The plausible reason may be due to (i) the hydrogen bonding interaction between HMT and chitosan polymer chain, which reduces the free volume between the polymer networks as discussed in previous paragraph (Fig. 3) and (ii) increase in pH of chitosan precursor solution with the addition of HMT. The addition of HMT to chitosan solution marked a significant increase in the pH. The pH of chitosan solution was measured to be 2.56, while that of chitosan solution mixed with 1:1 HMT and 1:3 HMT were 3.28 and 4.19 respectively. According to the course grind (CG) model reported by Hongcheng Xu et al. the chitosan polymer chains self-aggregate with gradual increase in pH of solution [55]. As depicted in Fig. 11, the self-assembly of chitosan polymer chains starts to form a physically cross-linked network as the pH of the solution is increased from 2.56 to 3.28 and then to 4.19 with respect to the increase in HMT concentration. The resulting physical cross-linking of polymer chains resulted in the deterioration in the surface characteristics such as surface area and pore volume. As a result, the adsorption capacity of N-doped activated carbon prepared from chitosan/HMT precursor system reduced as compared to that of the undoped activated carbon obtained from chitosan alone.

Conclusion
N-doped porous activated carbon was prepared by using chitosan as the carbon precursor and HMT as an additional nitrogen source. A comparative study on the structural as well as the CO 2 adsorption properties was performed on the undoped and N-doped activated carbons. The nitrogen content for chitosan/HMT derived activated carbons were higher than that of the undoped activated carbon as observed from CHN as well as EDS analysis. The incorporation of HMT with chitosan solution for the preparation of N-doped activated carbon resulted in the deterioration in properties such as surface area, porosity and pore volume of the activated carbon. The N-doped activated carbon exhibited a reduced CO 2 adsorption capacity as compared to the undoped carbon and the extent of reduction increased as the ratio of HMT to the chitosan was increased. The decline in the surface characteristics as well as the adsorption capacity were attributed to (i) the H-bonding interaction between chitosan and HMT resulting in the reduction in the free volume of the polymer network; and (ii) the physical cross-linking of chitosan polymer network effected by the increase of pH due to incorporation of HMT.