Polyethyleneimine-impregnated activated carbon nanofiber composited graphene-derived rice husk char for efficient post-combustion CO₂ capture

2.1 Synthesis of graphene-derived rice husk char (GRHC)

GRHC is a graphene-based material that is made from rice husks through a chemical activation method. Rice husks were procured from a local paddy mill in Kedah, Malaysia. They were chemically treated with potassium hydroxide (KOH; ≥85% pellets) under pure air flow at 200°C to produce rice husk char (RHC). The RHC were ground to form a powder. GRHC were formed by a simple chemical activation method. Details on the synthesis method to convert rice husks into GRHC has been explained in our previous study [14].

2.2 Fabrication of ACNF composites

PAN (M _w 150,000 Da), N,N-dimethylformamide (DMF), and PEI were purchased from Sigma Aldrich and used as received. PAN and DMF were used to make the polymer dope solution and PEI was used for amine functionalization of the ACNF. Nitrogen gas (99.99%), CO₂ (99.99%), and purified air (99.99%) were acquired from Alpha Gas Solution Sdn Bhd. For the polymer dope preparation, 8 wt% (4 g) of PAN and 92 wt% (46 g) of DMF were used for the total weight of 50 g of solution. Prior to electrospinning, 1 wt% of GRHC relative to PAN was stirred in DMF for a minimum 6 h to form a homogeneous solution and then PAN was added. The dope solution was left stirring for 24 h at room temperature until homogeneity was achieved. The same procedure was repeated with varied GRHC loadings (2.5, 5, 10 g). The composition of all denoted samples are listed in Table 1 with ACNF acting as the control for this study.

The dope solution was then loaded into the electrospinning machine to fabricate the nanofiber. The optimum electrospinning parameters were carried out based on our previous work [15]. The syringe pump was set at 1.0 mL/h, voltage supply at 12 kV, and the needle–collector gap at 15 cm while the environment of the electrospinning chamber was controlled at 30°C temperature and 50% relative humidity. Pyrolysis in a horizontal furnace (model Carbolite 12/65/550) that included three stages, namely stabilization, carbonization, and activation, was carried out next. In the stabilization stage, the NFs were heated up to 275°C under O₂ flow with a heating rate of 2°C/min. Then, carbonization of the stabilized NFs under N₂ atmosphere was carried out with heating up to 600°C (heating rate of 5°C/min) followed by the activation of the carbonized NFs to 800°C (heating rate of 5°C/min), respectively. The gas flow rate and dwelling time used throughout the process were 0.2 L/min and 30 min, respectively.

2.3 Impregnation of PEI on GRHC/ACNF composites

Amine impregnation was adapted from the study by Khalil and coworkers with some modifications. First, 2 g of ACNFs was mixed with 2 g of PEI and 10 g of DI water. The mixture was magnetically stirred at room temperature for 3 h to make sure the PEI were fully attached on the surface of ACNF samples. Then, the mixture was put into a vacuum oven at 100°C for 5 h for complete drying. A similar procedure was also performed on GRHC/ACNFs. The dried PEI-impregnated samples were further investigated, characterized, and tested for CO₂ adsorption performance.

2.4 Characterization

Before the N₂ adsorption/desorption test, the VacPrep Degasser (VacPrep 061, Micromeritics Instrument Corporation) was used to degas the samples in a processor under vacuum (1 × 10⁻¹ kPa) at 150°C, minimum for 12 h. Then, the textural properties of the porosity of the samples were examined using a porosity analyzer (MicrotracBEL Belsorp-Max) with N₂ (99.9999% purity) at −196°C. The Brunauer–Emmett–Teller (BET) method has been used in determining the specific surface area (S _BET), total pore volume (TPV), and mean pore diameter of samples. Meanwhile, the t-plot and Barrett–Joyner–Halenda (BJH) methods were used to determine the micropore volume (V _micro) of samples. The mesopore volumes (V _meso) were evaluated by subtracting the V _micro from TPV. In order to study the crystallinity of the materials and their phase structure, X-ray diffractometer (XRD, Rigaku SmartLab) using Cu Kα (λ = 1.54184 Å) at a scanning rate of 1.5°/min was utilized. A Raman spectrophotometer (RMP-510S, Jasco USA) with 514 nm wavelength Ar-ion laser excitation and 20 mW laser power was applied to describe the degree of graphitization and structural transformation of the samples. The Raman spectrum peaks were recorded within the range of 1,000–3,000 cm⁻¹ laser excitation wavelength.

The morphology, structure, and elemental properties and mapping of the samples were conducted with a high-resolution transmission electron microscope (HR-TEM 120 kV; Hitachi HT7700), field-emission scanning electron microscope (FE-SEM), and electron dispersive X-ray (EDX) (Hitachi SU8020; Hitachi Co. Ltd Japan). Prior to characterization, all samples were placed on a copper grid with a carbon film (after ion bombardment) and were left to dry at an ambient atmosphere. Fourier-transform infrared (FTIR, Thermo Scientific/Nicolet iS10) was employed in determining the chemical functionalities in the samples by setting the scanning range of 4,000–1,000 cm⁻¹. For determining the thermal properties of the samples, such as the percentage of carbon yield and weight loss, thermal gravimetric analysis (TGA) (TA Instruments Q600 Simultaneous DSC/TG) was carried out in the range of 25–800°C with a heating rate of 10°C/min under N₂ flow.

2.5 CO₂ adsorption performance

CO₂ sorption was assessed through a volumetric gas adsorption system (Micromeritics TriStar II). Prior to adsorption, all moisture in the samples was firstly removed by degassing the samples for 12 h at 150°C. Each sample with a weight of ±0.1 g was filled into the adsorption column and constant flow of CO₂ (0.015 L/min) was used for the adsorption study. Consequently, samples with the best adsorption performance were subjected to the adsorption test at different temperatures (0, 25, and 50°C) to study their adsorption properties at low to moderate temperatures. As both PEI-impregnated and nonimpregnated GRHC/ACNFs exhibited better adsorption performance over pristine ACNFs, these two samples were used for further assessments including the kinetic and equilibrium characteristics as well as stability studies.

2.6 Kinetics and adsorption equilibrium studies

In order to examine the experimental data of CO₂ adsorption on the ACNF samples at different times, Lagergren’s pseudo first-order and pseudo second-order of kinetic models have been employed. According to pseudo first-order characteristics, the adsorption rate is proportional to the number of free adsorption sites on the surface of the adsorbent, and it was calculated based on Equation (1):

where q _e is the weight-specific adsorbed amount of the adsorbate at the end of equilibrium (mmol/g), q _t is the weight-specific adsorbed amount of the adsorbate at time t (s) during the adsorption process (mmol/g), and k ₁ is the rate constant of adsorption of the pseudo first-order model.

Thus, it can be assumed that CO₂ adsorption conforms to the pseudo first-order model if the representation of ln ( q e − q t ) versus time can be well fitted to a straight line. Commonly, the pseudo first-order model describes reversible interactions between adsorbates and adsorbent, and therefore indicates physisorption processes [16].

However, according to the pseudo second-order model, the adsorption rate is proportional to the square of vacant adsorption sites as shown by equation (2):

where k 2 (mmol/g) is the rate constant of adsorption for the pseudo second-order equation. Consequently, it can be assumed that if the representation of t/q _t versus time can be well fitted to a straight line, CO₂ adsorption highly obeys the pseudo second-order model. This model describes relatively strong and chemical interactions between the adsorbents and gas molecules. Thus, the process would be signified by chemical adsorption as the rate-controlling step [17]. Both k 1 and k 2 values were attained from the slopes of the corresponding fitted straight lines.

Similar to kinetic study, only GRHC/ACNFs (PEI-impregnated and nonimpregnated) were used for adsorption equilibrium study. The CO₂ adsorption isotherms show the relationship between the CO₂ adsorbed amount at equilibrium at different pressures (1 and 15 bar). The data obtained were adjusted to the Langmuir and Freundlich models to determine the stages and mechanism of the adsorption of CO₂.

2.7 Regeneration and stability

One of the crucial parameters that need to be considered for the practical application of any adsorbent in the adsorption process is the long-term cycling stability. Therefore, we carried out the desorption of the cyclic CO₂ after each cycle of adsorption to determine the stability of GRHC/ACNFs. In this study, the cycles were repeated five times at the same adsorption conditions in order to evaluate the regeneration of the GRHC/ACNFs in the early stage. This result will give a better understanding of the adsorbent stability for further use in the following cycles. The cycles were conducted under conditions similar to real-life situations as this temperature is where the CO₂ is stripped out from the flue gas pipe. The regeneration of both exhausted GRHC/ACNFs and PEI-impregnated GRHC/ACNFs was conducted by sweeping the adsorbents with 0.1 L/min upward pure N₂ at 100°C for 5 h. The regenerated adsorbents were reused in an experiment to adsorb CO₂ [18].

3.1 Activated carbon nanofibers composited GRHC (GRHC/ACNF) composites

The N₂ isotherm of all prepared samples are shown in Figure 1, and their microstructure and textural properties are recorded in Table 2. All prepared NF-based materials, including with GRHC shows no microporous structure, (V _micro = 0 cm³/g). After activation, the textural properties of the ACNFs shows the development of meso- and micropores and increment in the surface area. Type I(b) isotherms derived from the N₂ adsorption/desorption in Figure 1 explains the microporous attributes in all ACNF samples. [19,20]. Typical adsorption response of microporous materials can be seen from the stagnant plateau isotherm of all the ACNFs at p/p ₀ = 0.05–0.95; however, above 0.95 the adsorption curve continually increases. Additionally, as Qi et al. [19] suggested, H3-type of hysteresis loop was observed from the N₂ adsorption/desorption isotherms (p/p ₀ = 0.4–1.0), which was due to the capillary condensation in the mesoporous pores [21,22]. It is confirmed that all the activated samples were mainly microporous with a high proportion of V _micro in relation to the TPV as shown in Table 2. Along with it, the values of V _meso indicating the proportion of mesoporous structure are also tabulated.

Figure 1

N₂ adsorption/desorption isotherms of ACNFs synthesized at different loadings of GRHC after activation.

At an activation temperature of 800°C, it can be significantly observed that S _BET increases in all ACNF samples as shown in Table 2, which represents the formation of mesoporous/microporous structures. This could possibly be due to the catalytic effects of GRHC that takes place at higher temperatures [23]. For example, the S _BET of ACNF drastically improved from 17 m²/g to 139 m²/g after the activation. Prior to activation, it can be seen that the S _BET is very low in all samples, and it is considered that there is no significant catalytic ability at low temperatures. The S _BET considerably increases after activation indicating that GRHC has good catalytic ability for creating porosity at higher temperature. Interestingly, the highest S _BET of 597 m²/g was measured for GRHC/ACNF0.1 while further increasing GRHC loading diminished its textural properties. As shown in Table 2, ACNFs with the highest loading of GRHC show the lowest S _BET and V _micro. Importantly, the adsorbents with high S _BET and V _micro are preferable as these are important factors for the adsorption of gas molecules [24]. The V _meso increases and V _micro decreases with the increasing loading of GRHC. This increment of GRHC loading leading to poor textural properties of the ACNF composite could possibly be due to the effects of high conductivity of GRHC that could interfere with the jet ejection during electrospinning for producing nanofibers with larger diameter and formation of beads that results in pore blockage (see Supplementary Materials, Figure S1).

The XRD spectra of all samples in this study are shown in Figure 2. For GRHC, a sharp peak at 28.6° representing the (002) lattice plane and a broad peak at range 17°–31° representing hexagonal graphite were observed. All ACNFs shows two diffraction peaks ranging between 17°–31° and 41°–43°, representing hexagonal graphite and rhombohedral graphite, respectively [25], with variation only in their intensities. Peaks appear around 17°–31° corresponding to the (002) plane of the carbon skeleton [25]. This represents the formation of ladder structures within the ACNF and the composites after the cyclization of PAN activation. Moreover, XRD analysis reveals their crystallographic (100) and amorphous (110) planes [26]. Apparently, the presence of the low and broad peaks in the range of 41°–43° indicates that samples have amorphous structure and possess very low graphitization degree. This finding corresponds well to the one obtained by Raman spectroscopy. All amorphous ACNFs obtained in this study could possibly be due to the destruction of the in-plane aromatic lattices [27], leaving a large number of defect structures in the ACNFs [28,29]. Remarkably, there is no appearance of the GRHC peak in all ACNF composites and this is believed to be due to the very small amount of GRHC used in all composite samples, which is below 10% (relative to PAN wt) that contributed to very weak peak or insignificant peak in the amorphous structure of the carbon materials. In 2012, Guo et al. [30] found that the carbon materials with amorphous and defective structures were considered to have greater ability in modifying the textural characteristics of the ACNFs, which was also observed in this study (Figure 1 and Table 2).

Figure 2

XRD spectra of GRHC, pristine, and composite ACNFs with different loadings of GRHC.

Raman spectroscopy was used in this study to identify the structural features and degree of graphitization of the GRHC and pristine ACNFs, as well as their composites, as shown in Figure 3. All activated samples exhibited two distinguishing peaks, which are at 1,368 cm⁻¹ and 1,612 cm⁻¹ but with different intensities. These two peaks indicated the appearance of the D-band and G-band of graphitic materials in the samples, respectively. In conjunction, the D-band spectra of all samples stipulated the existence of defective and amorphous carbon structure whereas the presence of the G-band signified the orderly structure of the graphene with sp ² hybridization of the carbon atom. The intensity is relatively high at a low loading of GRHC (1 wt%), and further increasing the loading diminishes their intensity.

Figure 3

Raman spectra of GRHC, pristine, and composite ACNFs with different loadings of GRHC.

As mentioned in previous literature, the formation of the graphene material can be confirmed from the appearance of the 2D-band spectrum around 2,700 cm⁻¹ of the Raman shift [31]. Broad and weak 2D-bands around 2,700–2,800 cm⁻¹ can be observed in all ACNF samples in Figure 3. The very low intensity of the 2D-band in ACNF composites is believed to be due to too small amount of GRHC composited into the structures. It is worth noting that all ACNF-based samples exhibited defect structures and highly amorphous nature as indicated by the intensity of the D-band being higher than the G-band, concurring with the XRD result analysis. In contrast, the G-band is higher than the D-band in the GRHC sample proving its higher crystallinity as a graphene-based material.

Pristine GRHC exhibits the smallest intensity ratio R(I _D/I _G) of 0.96 for D-band to G-band among all materials under investigation due to the ordered carbon structures as well as the high degree of graphitization [32,33] whereas ACNF exhibits R(I _D/I _G) of 1.35 indicating a highly disordered structure. Incorporating GRHC into the ACNF shows the composite structure becoming more ordered as implied by the decrease in R(I _D/I _G) with increasing GRHC, with exception of GRHC/ACNF0.01. It shows that R(I _D/I _G) of GRHC/ACNF0.01 is higher than that of pristine ACNF and this is corresponding well with the BET and XRD results discussed earlier, indicating this sample possesses the most disordered structures. As more GRHC was incorporated into ACNFs, the composites became more ordered and this could be explained by the addition of the highly ordered structure of pristine GRHC that could affect the structure of the composites.

The graphitic structure of the GRHC/ACNF composites is expected to diminish with the increment of GRHC loading. Regardless of its lowest R(I _D/I _G) in comparison with other ACNF composites, the carbon material with this value is believed to possess disordered and defective structures. This is validated by Gayathri et al. [31], who found that the graphene synthesized in their research still possessed defect structure even at a significantly low value of R(I _D/I _G) of ∼0.2–0.4. Remarkably, in this study, their defective structures are more preferable for better gas adsorption due to porosity enhancement caused by the defects. This means the sample with higher R(I _D/I _G), such as GRHC/ACNF0.01, will show better performance toward CO₂ adsorption.

Figure 4 shows the morphological structure of the ACNF at different magnifications and loadings and summarized in Table 3. The development of the porous structure after the activation of the NFs was proved with the appearance of pores on the ACNF structure as shown in Figure 4(e). Among all samples, GRHC/ACNF composited with 1% of GRHC exhibited the smallest fiber diameter of 238 ± 79.97 nm as shown in Table 3. Bead formation was observed in ACNF composites likely due to conductive properties provided by the GRHC and the viscosity of the solution [34]. This would affect the final diameters of the fibers due to the electrostatic repulsion within the jet spray from the tip of the needle to the collector surface throughout the electrospinning process. Huang et al. [35] reported that high conductivity led to the production of electrospun fibers with smaller diameter due to the increment of the surface tension of the solution, which affects the jet ejection during electrospinning. Thus, it can be hypothesized that increasing GHRC loading would result in a smaller diameter. However, interestingly in this study, only GRHC/ACNF0.01 resulted in the smallest diameter whereas increasing GHRC loading further increases the fiber diameter (albeit smaller than ACNF). The results show correlation with the S _BET obtained (Table 2) where smaller diameter provides higher S _BET and vice versa. From the results obtained, only GRHC/ACNF0.01 has been used for further impregnation study and pristine ACNF has been used for comparison.

Figure 4

FE-SEM images of pristine ACNFs at different magnifications: (a) 5,000×; (c) 10,000×; (e) 100,000×; and with different loadings of GRHC at 10,000× (b) GRHC/ACNF0.01; (d) GRHC/ACNF0.05; (f) GRHC/ACNF0.1.

Moreover, another major point to be addressed in this study is the particle size of the GRHC/ACNF0.01 in powdered form. From the results of this current study (see Supplementary Materials, Figure S2), the average particle size of GRHC/ACNF obtained ranged from 22.2 to 35.4 µm. Compared with other sorbents, this range of particle size is a good indicator of the suitability of these samples as good sorbents. It is believed that the size of the particles can greatly affect the adsorption properties of the sorbents. This was supported by a study conducted by Li et al. [36] as they found that particle size of 52.5 µm or smaller resulted in higher absorption of CO₂ as compared with a particle size of 77.5 and 100.0 µm. Even though this sorbent has been ground into powdered form, however, under high magnification image, the thread-like structure of the nanofibers remains as shown in Figure 4.

In terms of mechanical strength, this proposed GRHC/ACNF is believed to possess good mechanical strength, including high tensile strength, which could be due to the highly oriented fibers during the electrospinning process that affect the mechanical strength of the nanofibers. Due to limitations of time limitation and availability of instruments, the mechanical test on the GRHC/ACNF could not be conducted. However, superiority of the mechanical properties possessed by GRHC/ACNF can be seen from similar materials such carbon nanofibers studied in several previous work reviewed by Yadav et al. [37]. For example, Pu et al. [38] have successfully fabricated electrospun carbon nanofibers with excellent mechanical properties, such as high tensile strength up to 20.7 MPa. According to their studies, the stretching effect of the jet during electrospinning is advantageous to the arrangement of carbon nanofibers along the fiber axis, which significantly enhances the tensile strength and modulus of the fibers [38].

3.2 PEI-impregnated ACNFs and GRHC/ACNFs

Figure 5 shows the N₂ adsorption/desorption isotherms of PEI-impregnated ACNFs and GRHC/ACNF0.01. Both impregnated samples exhibit similar type of isotherm, which is type II indicative of nonporous or macroporous structures [39]. The isotherm type changes from type I(b) prior to activation into type II after activation. This type of isotherm explained the formation of a monolayer at relatively low pressure, and then multilayer formation at higher relative pressure [40]. In contrast with the type I isotherm, the long plateau is absent in type II isotherm as the adsorption uptake continuously increases with the relative pressure without significant halt and it even almost reaches unity, indicating the multilayer adsorption.

Figure 5

N₂ adsorption/desorption isotherms of PEI-impregnated ACNFs and PEI-impregnated GRHC/ACNFs.

Table 4 summarizes the corresponding textural properties of both impregnated samples. As compared with nonimpregnated samples, both PEI-impregnated samples have shown significant reduction in S _BET and TPV, with the disappearance of V _micro in the structure. This reduction of S _BET confirms that PEI was successfully impregnated onto the porous structures of the ACNF samples, and all of its porosity structure was filled with PEI. Rouzitalab and coworkers have reported that adsorbent materials that are high in TPV were easily accessible by the N-functional groups, which results in blocking the pores of the ACNFs, and were reflected by the decrement of the S _BET. In this case, the physisorption that mainly depends on S _BET, which predominantly lay in its pores, would be mostly excluded and chemisorption takes place.

The thermal stability of the impregnated ACNFs was evaluated using TGA analysis in the temperature range of 25–700°C (Figure 6). All impregnated ACNFs exhibit thermal decomposition at a higher onset temperature of ∼400°C. This result suggests PEI-impregnated ACNFs have better thermal stability as compared with nonimpregnated samples. From the thermogram, the weight percentage of the ACNFs improved from 24.2% to 36.2% as compared to PEI-ACNF. Meanwhile for PEI-impregnated GRHC/ACNFs, the final weight obtained is much higher than the nonimpregnated one, which is up to 68.2%. This is could possibly be due to the low volatility and good stability properties of PEI [41]. This result is expected to contribute to GRHC/ACNF stability and cyclability during the real-life flue gas separation in postcombustion CO₂ capture.

Figure 6

TGA thermogram of PEI-impregnated and nonimpregnated ACNFs and GRHC/ACNFs.

Figure 7 exhibits the FTIR spectra of the nonimpregnated and PEI-impregnated ACNFs and GRHC/ACNFs, respectively. Compared with nonimpregnated ACNFs. A broad band at 2,980 cm⁻¹ was observed, which indicates the existence of –CH₂ symmetric stretching modes of the PEI chain in the PEI-impregnated samples. The band at 1,557 cm⁻¹ also can be observed in all impregnated samples associated with the asymmetric bending of the primary amines (–NH₂). The appearance of this peak further supports successful impregnation of PEI [42,43].

Figure 7

FTIR spectrum of PEI-impregnated and nonimpregnated ACNFs and GRHC/ACNFs.

The SEM, TEM images, and EDX mapping of PEI-impregnated GRHC/ACNFs is presented in Figure 8(a)–(d). As shown in Figure 8(a), it can be seen that the morphology of PEI-GRHC/ACNF was agglomerated with appearance of large beads, which is believed happened during the impregnation process. The presence of black spots within the fibers in Figure 8(b) indicates the GRHC presence onto the structures. Furthermore, the presence of a small amount of silica (0.02 ± 0.02 wt%) in the nanowires was also observed [44] in Figure 8(c). This also can be seen with the appearance of silica in the resultant GRHC/ACNFs through EDX mapping (Supplementary Materials, Figure S3), which indicating the high content of silica in untreated (raw) rice husks (prior conversion into GRHC). Through EDX analysis, Table 5 confirms the existence of a high concentration of N atoms on the impregnated samples (35.21 ± 6.81 wt%). The other major elemental properties such as carbon (C) and oxygen (O) were also detected at 51.45 ± 0.74 wt% and 13.32 ± 1.71 wt%, respectively. As expected, C covers the highest percentage among other elements as ACNFs are carbon-based materials.

Figure 8

(a) SEM image at magnification of 5,000×; (b and c) TEM images at different magnifications; and (d) EDX mapping of PEI-impregnated GRHC/ACNFs.

3.3 Adsorption performance and characteristics toward CO₂

GRHC/ACNFs showing better CO₂ adsorption than the pristine ACNFs (Figure 9) at pressure 1–15 bar indicate that the presence of graphene improves its affinity toward CO₂ as it contributes to the enhancement of the textural properties of the adsorbents [45]. These results are in agreement with the BET results where higher S _BET and V _micro [46] provide more surface for gas to be adsorbed. Compared with other previous works, both ACNFs possess higher CO₂ adsorption compared with other carbon-based materials as summarized in Table 6, even though with lower S _BET values. It is believed that the high adsorption value is not totally S _BET dependent, yet it also could be contributed by the available open micropores on the adsorbents. For example, most of the CNF-based adsorbents in Table 6 showed higher S _BET as compared with the samples in this study, however, with lower V _micro values and limited CO₂ adsorption capacities. This means V _micro is a very important characteristic for adsorbents to adsorb higher gas. From this, it can be seen that the resultant ACNFs can be good candidates for gas adsorbents, especially in flue gases, in order to reduce the CO₂ emission into the atmosphere, consequently combating the global warming problem.

Figure 9

CO₂ adsorption of ACNFs and GRHC/ACNFs at different adsorption pressures: (a) 1 bar; (b) 5 bar; (c) 10 bar; (d) 15 bar at 25°C.

Various sorbent materials have been widely used in current adsorption technology and further assessment on their performance needs to be clarified. Table 7 shows the comparison of the sorbents’ properties and their adsorption capacities to further assess the potential of currently synthesized GRHC/ACNFs compared with other available sorbent types, such as conventional activated carbon, zeolites, carbon nanotubes (CNTs), MOFs, graphene, and so on. Among the available adsorbents, carbon-based adsorbents have shown promising adsorption performance as excellent gas adsorbents due to their large surface area and availability (can be produced from wide range of raw materials). In the past few decades, activated carbon is the commonly used adsorbent in various applications. The wide utilization of activated carbon in adsorption industries is advantageous due to their wide distribution of pore structures and surface chemistry for gas adsorption. Most importantly, activated carbon is relatively low in cost, has high S _BET, widely available, has an easily modifiable pore structure, , susceptible to surface functionalization, and can be regenerated [53,54].

Recently, activated carbon has been obtained in different structures and forms, including powder, granular, tubular, and fibers. It was found that activated carbon in fiber form has shown significant structure improvement, especially in terms of surface area, micropores, active site availability, and gas adsorption capacity as compared with the traditional granular form [55]. The presence of fibril structure in ACNFs has ensured much higher adsorption capacity, which is due to the direct diffusion of gas adsorbates into the available micropores inside and on the surface of the adsorbents. This means there is no resistance for adsorbates to reach the adsorption sites as there are no macropore or mesopore network. Moreover, this fiber form can be prepared from a wide range of precursors such as coal, polymers, petroleum pitch, rayon, and viscose.

As listed in Table 7, the adsorbed volume of CO₂ by the ACF and GRHC/ACNF are the highest, which are 6.4 and 6.58 mmol/g, respectively compared with other adsorbents. Even though, the CO₂ adsorbed volume is comparable for both samples, the S _BET value is significantly different, which is 1240 m²/g for ACF and 597 m²/g for GRHC/ACNF. From this finding, it can be concluded that even at low S _BET value, the adsorption performance of the as-proposed adsorbent, GRHC/ACNF, is better than that of the other mentioned adsorbents. It is believed that tuning into higher S _BET value can potentially turn this GRHC/ACNF into an excellent CO₂ adsorbent. Besides S _BET value, another important factor that affects the adsorption performance is the compatibility of the adsorbents for real-life CO₂ capture applications. As mentioned in the TGA results above, GRHC/ACNF with good thermal stability (up to 700°C) can withstand the temperature during postcombustion CO₂ capture ranging from 40°C to 80°C, which makes this proposed GRHC/ACNF as one of the excellent CO₂ adsorbents.

The CO₂ adsorption performance of PEI-impregnated GRHC/ACNFs is presented in Table 8. The CO₂ adsorption capacity was higher at 107 cm³/g for PEI-ACNFs (55% enhancement) and 191 cm³/g for PEI-GRHC/ACNFs (34% enhancement) as compared with their nonimpregnated samples, which were 69 cm³/g and 142 cm³/g for ACNFs and GRHC/ACNFs, respectively. This has proved the involvement of chemical interaction between the basic N-functionalities in PEI and acidic CO₂ molecules. From this finding, it is believed that the high adsorption capacity in PEI-impregnated samples could be contributed by the involvement of both physical and chemical interactions between the adsorbent and adsorbates.

In terms of adsorption in more practical dynamic conditions in multicomponent mixtures such as flue gases, this newly produced both impregnated and nonimpregnated GRHC/ACNFs are believed to have good gas separation properties. This statement was made by considering and comparing the N₂ adsorption obtained from BET results and the CO₂ adsorption capacity. From BET results, the PEI-impregnated sample shows very low N₂ adsorption capacity of 2.67 cm³/g whereas the nonimpregnated sample possesses higher adsorption capacity of 33.7 cm³/g. The low N₂ adsorption on PEI-GRHC/ACNFs indicates the blockage of the available porous structure (very low S _BET as previously discussed) by PEI for N₂ molecules to be adsorbed onto the structure. As N₂ is an inert gas, it does not interact with N-functionalities on the surface of PEI-GRHC/ACNFs. In contrast, the nonimpregnated sample exhibits higher N₂ adsorption due to the numerous available porous structures, which contributes to physisorption. According to both N₂ and CO₂ adsorption capacity, it is believed that the resultant GRHC/ACNFs more selective toward CO₂. From this finding, GRHC/ACNFs can potentially act as a good adsorbent for flue gas separation in postcombustion CO₂ capture.

As previously discussed, the GRHC/ACNFs possessed the highest S _BET and TPV prior to PEI impregnation but the lowest after impregnation. It can be attributed to the presence of amine in the GRHC/ACNFs. This phenomenon explains the high CO₂ adsorption capacity in PEI-GRHC/ACNFs even though they possessed smallest values of S _BET and TPV as compared with GRHC/ACNFs, which is highly believed to be influential in both physisorption and chemisorption. In conclusion, it can be said that high CO₂ adsorption capacity in PEI-GRHC/ACNFs is attributed to a synergic effect of amino groups, thus multiplying the CO₂ uptakes through chemisorption [71]. With low value of S _BET (presence of very low V _meso or macropores), it can be said that PEI does not totally block the surface of the GRHC/ACNFs leaving spaces for physisorption to take place. Both GRHC/ACNFs and PEI- GRHC/ACNFs show the highest CO₂ uptakes as compared with other samples as shown in Table 8. For further adsorption studies, both samples have been used and were denoted as PEI-GRHC/ACNFs for PEI-impregnated samples and GRHC/ACNFs for nonimpregnated ones.

Figure 10 shows the CO₂ adsorption isotherm of the GRHC/ACNFs and PEI-GRHC/ACNFs at atmospheric pressure and temperatures of 0°C, 25°C, and 50°C. Figure 10 reveals that the desorption of both impregnated and nonimpregnated samples at different temperatures shows no hysteresis. This means that both samples have very good adsorption/desorption capacity. It was also observed that CO₂ adsorption capacity reduces with increasing temperature for both samples. The adsorbed amount of CO₂ in PEI-GRHC/ACNFs and GRHC/ACNFs decreased from 522 to 142 cm³/g and 293 to 69 cm³/g, respectively. At higher temperature, the energized gas molecules were rapidly occupied on the GRHC/ACNFs surface, therefore, resulting in fewer CO₂ molecules captured [72]. Moreover, low equilibrium of CO₂ at higher temperature was also due to the exothermic nature of the adsorption [73].

Figure 10

CO₂ adsorption/desorption on PEI-impregnated and nonimpregnated GRHC/ACNFs at different temperatures (a) 0°C; (b) 25°C; and (c) 50°C under atmospheric pressure.

Table 9 summarizes the kinetic studies of the PEI-GRHC/ACNFs and GRHC/ACNFs in order to determine the most fitted kinetics adsorption for the adsorption of CO₂. Based on the R ² values, both GRHC/ACNFs samples are best fitted to the pseudo first-order kinetic model at low adsorption pressure of 1 bar by displaying greater R ² value of 0.891 and 0.899. Meanwhile, the significantly low values of R ² found in both samples in pseudo second-order kinetic model. This means that the CO₂ adsorption process of GRHC/ACNFs involved a physisorption phenomenon at low pressure as per our expectation, which means the physisorption is more related to the adsorption toward the surface with higher S _BET, TPV, and V _micro by weak dipole van der Waals interactions. These results are also in agreement with the results of S _BET, N₂, and CO₂ adsorption previously discussed. Remarkably, even though the nonimpregnated GRHC/ACNFs show the results as expected, however, the PEI-impregnated GRHC/ACNFs show result beyond our expectations as they also obeyed the pseudo first-order kinetic model. This is believed to be because of the small amount of amino groups presence in the impregnated samples (due to the low ratio of PEI:GRHC/ACNF:DI water; 1:1:5 used) that led to weaker interaction between PEI and CO₂ molecules resulting to minimum effects on the chemical interaction while involving physical interaction simultaneously [74]. In low pressure, physical interaction exhibits more dominant effect on the adsorption process. This is corresponding with previous study conducted by Liu et al. as they found that PEI-impregnated porous polymer-based adsorbents suggest both physical adsorption and chemical adsorption toward CO₂ [75].

Interestingly, at higher adsorption pressure of 15 bar, both samples can be ascribed by pseudo first-order and pseudo second-order kinetic models as the R ² values obtained were high in both models. The R ² values of pseudo first order of PEI-GRHC/ACNFs and GRHC/ACNFs are 0.7978 and 0.9154, respectively. Meanwhile, the R ² values of pseudo second order of PEI-impregnated GRHC/ACNFs and nonimpregnated GRHC/ACNFs are 0.996 and 0.986, respectively. These findings indicate that the adsorption properties of both GRHC/ACNFs involve both chemical and physical reactions between the free active sites and open pores available on the GRHC/ACNFs and the CO₂ molecules. Moreover, this also confirms the main idea of PEI impregnation is to improve the CO₂ adsorption by introducing the basic N-functional groups onto the GRHC/ACNFs for better binding with acidic CO₂ molecules through chemisorption without cancelling the physisorption properties. To conclude, both kinetic models used in this study are suitable for the present data obtained for the adsorption of CO₂ at 15 bar and 25°C; however, it fits the best with pseudo second order and followed by pseudo first order [76].

Table 10 summarizes the linearized plot of the experimental data correlated to the Langmuir, Freundlich, and BET models to determine the homogeneity and heterogeneity energy distribution on the surface of the adsorbents. The R ² values show that the resultant GRHC/ACNFs best fitted in Langmuir model and followed by BET. This implies that the adsorption of CO₂ involves monolayer phenomenon with homogeneous distribution on adsorbent active sites as previously discussed by Kim et al. [24]. In Langmuir isotherm and BET model, the dynamic equilibrium can be obtained by balancing the relative rates of adsorption and desorption based on the GRHC/ACNFs surface coverage [77].

It is believed that the adsorption process on the resultant GRHC/ACNFs does not only involve the formation of monolayer but also the formation of multilayers simultaneously [78]. This result was also supported by N₂ adsorption/desorption mentioned previously, in which the adsorption is monolayer at lower pressure but forming multilayers at higher pressures. This multilayer formation is best described by the Freundlich model. This model explains the adsorption process on surface adsorption sites which are energetically homogeneous.

Apart from a high CO₂ uptakes and fast adsorption/desorption kinetics, a good adsorbent should be stable and regenerated under mild conditions, in order to avoid high costs of the energy penalty. In this study, the cyclic behavior and stability of both impregnated and nonimpregnated GRHC/ACNFs were investigated via five consecutive CO₂ adsorption/desorption cycles for 14 h at 50°C (Figure 11). The selection of this adsorption temperature is to replicate the real condition of postcombustion CO₂ capture that is commonly used for adsorbent materials. As shown, there is a slight loss in adsorption capacity after each cycle. After five cycles, the stability of the impregnated GRHC/ACNFs shows almost similar drop percentage of 23% (from 32.2 to 24.6 cm³/g) as compared with nonimpregnated GRHC/ACNFs of 22% (15.1–11.8 cm³/g). This is could possibly be due to imperfect regeneration of the adsorbents due to the high affinity of PEI toward CO₂ and the current desorption method is insufficient to evacuate all CO₂ from the adsorbents. Moreover, PEI-modified ACNFs are dominated by chemisorption, and hence require higher energy (i.e. temperature) to evacuate CO₂. It has been found that this dropping percentage is considered acceptable for amine-impregnated samples after five cycles, as the capacity of the carbon capture under the same conditions on tetraethylenepentamine (TEPA) impregnated electrospun carbon nanofibers by a study conducted by Wang et al. [79] reported the drop is more than 11%. Rather than its higher drop percentage as compared with previously reported studies, the GRHC/ACNFs possessed comparable or superior adsorption capacity over other samples which is up to 27.1 cm³/g at 50°C. Even though the GRHC/ACNFs obtained in this study do not show similar adsorption capacity compared with the first cycle, however it become more stable from the fourth cycle to fifth cycle, which shows that the percentage drop is very low. It has been found that the resultant GRHC/ACNFs have good recyclability as well as mild regeneration conditions, which is believed to have high potential for practical applications in energy-efficient carbon capture processes.

Figure 11

Cyclic ability of the CO₂ adsorption on PEI-impregnated and nonimpregnated GRHC/ACNFs at 50°C and 1 bar.