3.1. Characterization of Sorbents
To identify the synthesized solid particles by their intrinsic crystallinity, XRD was performed. The obtained XRD patterns (
Figure 2a) revealed two main broad peaks in each profile of carbon nanocrystallite located at a 2θ of around 24° and 43° and one peak of carbide located at a 2θ of around 37°. The peak at a 2θ of 24° corresponded to the (002) plane of the turbostratic carbon [
23], while that at 43° was the (100)/(101) plane of carbon, reflecting the hexagonal ring structure of carbon (JCPDS card no.75-1621). The peak at a 2θ of 37° represented the (111) plane of tungsten carbide (WC
1-x), formed from the sputtering of the tungsten electrodes during the plasma discharge in the SPP [
25]. No other peaks were observed in the XRD patterns, inferring no contamination (impurity) in the carbon crystallite. Observing the peak at a 2θ of 24° for the Cn and OCn samples, its maximum was shifted towards a smaller 2θ value in Ocn than in Cn, which indicates that some oxygen was doped on the carbon framework, and also led to an expansion of the interlayer carbon sheet (d
002) space from 0.370 nm for Cn to 0.383 nm for Ocn (
Table 1). After nitriding, the Cn and Ocn samples at 800 °C to obtain NCn and NOCn, respectively, the shift in the maximum 2θ peak of 24° was still found. A change in the interlayer spacing value (d
002) was also found (
Table 1). The magnitude of d
002 was about 0.375 nm for NCn and 0.372 nm for NOCn. This inferred that the mechanism of nitridation of Cn and Ocn was not the same. However, the d
002 values of the particles synthesized here are higher than the 0.335 nm interlayer spacing of ideal graphite, reflecting a lower degree of graphitization on the carbon framework. Additionally, after nitriding, the peak at a 2θ of 37° was shifted slightly to a larger 2θ value and became broader. This was because the tungsten carbide might have been reduced by hydrogen gas [
26] produced from the decomposition of NH
3 at the high temperature (800 °C) used in the nitridation.
Representative Raman spectra of the Cn, Ocn, NCn and NOCn samples are shown in
Figure 2b. Only two bands were found in the spectra; the G band at around 1585 cm
−1 and the D band at around 1335 cm
−1. The G band represents the vibration of sp
2 orderly bonded carbons, while the D band corresponds to the vibration of sp
3 carbon atoms, indicating the defects on the graphene layer. The intensity ratio of the D and G bands (I
D/I
G) was used to evaluate the degree of defection or disorder on the carbon framework [
27,
28], with the results summarized in
Table 1. The Cn and Ocn samples showed a similar I
D/I
G ratio value of around 0.83 and 0.81, respectively. After nitriding at a high temperature to obtain NCn and NOCn, the I
D/I
G ratio was increased to 0.93 for NCn and 0.97 for NOCn. This indicated that nitrogen doping caused defects on the carbon framework [
29], and then the structure became less crystalline. A slight shift towards a higher Raman shift of the G band was found in the spectra of the NCn and NOCn compared to that in the Cn and Ocn, respectively. The inverse of I
D/I
G ratio was applied to calculate the size of the in-plane crystallite, L
a, via the Tuinstra-Koenig relationship [
30], shown in Equation (1):
where C is a constant (2.4 × 10
−10 nm
−3), and λ is the excitation laser wavelength (532.1 nm). From
Table 1, the L
a crystallite diameter of Cn was around 23.18 nm, while it was around 23.75 nm for OCn. Thus, an increase in the magnitude of L
a was found when doping oxygen on the carbon framework. After nitriding, the crystallite diameter became smaller, being 20.69 nm for NCn and 19.83 nm for NOCn, which might be due to the decomposition of the carbon during nitriding at a high temperature.
Representative TEM images of the different types of particles are shown in
Figure 3. The size of these particles was less than 50 nm, indicating they are nanoparticles. From the negative bright-field images of the Cn, OCn, NCn, and NOCn samples, the fringes were in a non-directive arrangement. This contributes to the 002 plane of carbon surrounded by an amorphous phase. However, the skeleton carbon in each particle was similar to each other. As can be seen from the SAED images of Cn (
Figure 3c), NCn (
Figure 3f), OCn (
Figure 3i) and NOCn (
Figure 3l), the structure of the particles was mainly amorphous with some polycrystalline. Compared to the SAED of the Cn and OCn samples, more blurred bright onion rings were observed in the SAED of the OCn sample, since some oxygen confounded with the carbon framework, while NCn and NOCn showed more blurred bright onion rings than C and OCn, respectively. During nitriding, some nitrogen atoms passed into the carbon framework, while some carbon atoms at the edge reacted with active species of NH
3 at high temperature. Thus, the crystallinity of the nitrided samples was decreased, and they became more amorphous.
Representative equilibrium N
2 adsorption/desorption isotherms at −196 °C for the Cn, OCn, NCn and NOCn samples are demonstrated in
Figure 4a. At the initial relative pressure, the first and steepest slope was found. The isotherm of NOCn and NCn exhibited a greater slope than the Cn and OCn samples, which indicated the development of ultramicropores with a diameter of less than 0.7 nm, during the nitriding at a high temperature. From the micropore size distribution via the MP-plot, as shown in
Figure 4b, it was evident that most pores in NCn and NOCn had a size of around 0.6 nm. Further increasing the relative pressure up to 0.8, N
2 uptake gradually rose. A higher amount of N
2 uptake was found when increasing the relative pressure from 0.8 to 1.0. Moreover, an adsorption/desorption hysteresis loop was observed in all isotherms (upper right corner of
Figure 4a), which is mainly caused by capillary condensation. Based on IUPAC classifications [
31], the isotherms of Cn and OCn expressed the integration of type II and type IV with an H3 hysteresis loop, meaning a non-porous sorbent with interparticle mesopores. The isotherms of NCn and NOCn were attributed to type I and type IV with an H3 hysteresis loop, meaning a porous sorbent with interparticle voids. It is noted that the isotherm of these samples changed from type II to type I after nitriding at 800 °C.
The specific surface area of the samples was calculated using the BET equation (S
BET) [
32], t-plot method (S
t) [
33,
34], and the results summarized in
Table 2. The S
BET of Cn and OCn were closed to the S
t obtained from t-plot method. The Cn sample had a higher specific surface area with a smaller average pore size than the OCn sample. However, the S
t of NCn and NOCn were higher than their specific surface area obtained from the BET equation, due to a presence of micropores. Differences in the type of working solution used in the SPP leads to a disparity of the synthesized particles. After nitriding, the OCn, the S
t of NOCn (679.8 m
2 g
−1) was about five-fold higher than that for OCn (137.5 m
2 g
−1), while the average pore diameter decreased 2.6-fold from 22.4 nm to 8.6 nm. For the nitridation of Cn to obtain NCn, the St was increased around 2.7-fold from 191.4 m
2 g
−1 to 521.3 m
2 g
−1, while the average pore diameter size was decreased approximately 2.1-fold from 19.7 nm to 9.3 nm. This reflects the development of micropores during the nitriding at 800 °C. The t-plot method [
33,
34] was then suggested to determine the micropore area and volume of NCn and NOCn. Note that since both Cn and OCn are non-porous, they could not express any information on micropores. The NOCn showed a two-fold higher micropore area and volume than NCn, and so NOCn should exhibit a higher CO
2 adsorption capacity.
The surface elemental analyses (
Table 3) revealed that the oxygen level at the surface of the sample increased from 4.18 at % for Cn to 7.21 at % for OCn, with a 1.77-fold higher oxygen/carbon (O/C) ratio in OCn than that in Cn. This supported that the oxygen-doped samples were successfully synthesized in the one-step SPP at room temperature and atmospheric pressure. It is noted that the N-content was not observed in the Cn sample or the OCn sample. After nitriding at 800 °C to obtain the NCn and NOCn samples, the nitrogen/carbon (N/C) ratio increased to 0.013 and 0.024 for NCn and NOCn, respectively. The nitrogen content in NOCn was around 1.78-fold higher than in NCn, indicating a different way to react with the active species derived from NH
3 during nitriding. The NH
3 that routed to the nitriding process at high temperature was decomposed to active species that then react with the carbon on the surface of Cn to obtain the NCn with various N-bonds, while they replace the oxygen on the surface of OCn [
35], and form a bond with the carbon framework to achieve NOCn, as evidenced in the XPS results (
Table 4). Even though the N/C ratio after nitriding increased, the O/C ratio became smaller from 0.044 for Cn to 0.017 for NCn and from 0.078 for OCn to 0.025 for NOCn. Moreover, the bulk elemental analyses (
Table 3) showed that the NOCn and NCn samples had a lower carbon level than the OCn and Cn samples, respectively. The decreased O/C ratio and carbon level were perhaps, due to the decomposition of oxygen and carbon at the elevated temperature (800 °C) during nitridation. The change in the carbon level might drive the development of the microporosity, as evidenced in the pore structure studies (
Table 2).
Representative XPS spectra of C 1s, O 1s, and N 1s are shown in
Figure 5 and summarized in
Table 4. For the C 1s region, the major bands represent the C–C sp
2 bonding at a BE of 284.5 eV and C–C sp
3 ordering at a BE of 285.4 eV. This implied that the carbon framework of the samples was composed of ordered and disordered carbon. The long-tail peak in the C 1s region refers to the heteroatom bonds of C–O at 286.5 ± 0.1 eV, C=O at 288.1 ± 0.2 eV and O–C=O at 289.9 ± 0.2 eV. The C–N band at 285.8 eV was found in the spectra of NCn and NOCn, supporting the successful synthesis of NCn and NOCn via nitridation. For the O 1s region, the XPS spectra revealed quinone at 530.2 eV, O=C–OH at 531.4 ± 0.2 eV, C=O at 532.3 ± 0.1 eV, C–OH at 533.3 ± 0.1 eV, C–O at 534.0 ± 0.2 eV and O–H at 535.5 eV. The presence of the O–H bond was due to the moisture adsorbed on the sample. Only the spectra of the nitriding samples expressed the N 1s region, which represented various N-forms on the surface, such as pyridinic at 384 ± 0.1 eV, pyrrolic-N at 400.2 ± 0.1 eV, graphitic-N at 401.3 ± 0.2 eV and pyridinic N-oxide at 403.5 ± 0.2 eV. Most of N-contribution belonged to nitrogen bonding on edge, at around 90.4% for NCn and 84.2% for NOCn, whereas, the nitrogen chemical bond in bulk (graphitic-N) existed at about 9.6% and 15.8%, respectively. Doping nitrogen at the edge was done by replacing the surface oxygen-functional group with the active species of NH
3 (NH and NH
2 radicals) [
35,
36]. There is some difficulty in doping nitrogen atoms on the bulk, due to the requirement to break the strong bonds of the carbon matrix [
29]. The major N-contribution in the N 1s region of the NCn and NOCn samples was the pyridinic form, which is attributed to its adsorption performance. Moreover, the W4f band of elemental tungsten at a BE of around 33 eV was not observed in the XPS spectra of all samples, since the amount of contaminated tungsten was quite low.
3.2. Performance of CO2 Adsorption
The influence of the type of adsorbent and temperature on the CO2 adsorption capacity was investigated by volumetric analysis and is reported in terms of mmol CO2 adsorbed g−1 adsorbent (dry weight).
The four types of particles synthesized here (Cn, OCn, NCn, and NOCn) were evaluated for their CO
2 adsorption capacity at 25 °C (
Figure 6a). The capacities of each sample increased with pressure, due to the thermodynamic driving force [
37]. The adsorption phenomenon of gaseous CO
2 molecules on the solid adsorbent could be explained as when the gaseous molecules come close to the adsorbent surface, they induced attractive interactions between them [
38,
39]. Increasing the equilibrium pressure then led to an increased amount of gas molecules covered on the surface, resulting in higher adsorption capacity. The maximal capacity of each sample was found at 1 bar and was ranked (lowest to highest) as: OCn (0.20 mmol g
−1) < Cn (0.34 mmol g
−1) < NCn (1.28 mmol g
−1) < NOCn (1.63 mmol g
−1). Based on the textural properties of the sorbent (
Table 2), those with a higher specific surface area and smaller average pore diameter size expressed a higher adsorption capacity. Moreover, NOCn had a higher micropore area and volume than NCn, promoting a higher degree of CO
2 adsorption [
40,
41]. In addition, the presence of the nitrogen functionality, especially the formation of pyridine-like species, on the surface of the carbon (
Table 4) enhanced the CO
2 adsorption capacity, and so the nitrided samples expressed a higher CO
2 adsorption capacity than those without nitriding by about 8.15- fold for NOCn and 3.76-fold for NCn, compared to OCn and Cn, respectively. The higher adsorption capacity was effectively achieved by the improved interaction between the acidic CO
2 molecules and the active nitrogen basic sites on the adsorbent surface [
42]. In conclusion, the CO
2 adsorption capacity depended not only on the textural properties of the adsorbent, and particularly the micropore area, but also on the nitrogen functionality on the surface of the carbonaceous skeleton of the adsorbent. Due to the origin of Cn and the higher adsorption capacity of NOCn, they were selected for further study.
The effect of the temperature on the CO
2 adsorption capacity of Cn and NOCn at various pressures is shown in
Figure 6b (data provided in
Table S1). It can be seen clearly that, at the same equilibrium pressure, the CO
2 adsorption was decreased from 1.63 mmol g
−1 to 1.37, 1.18 and 0.99 mmol g
−1 for NOCn, and from 0.34 mmol g
−1 to 0.32, 0.28 and 0.24 mmol g
−1 for Cn, as the temperature increased from 25 °C to 35, 45 and 55 °C, respectively. The effect of pressure on the CO
2 adsorption capacity differed from that of the temperature. The pressure played an important role in the thermodynamic driving force to push the adsorption forward, and so the CO
2 adsorption capacity increased with increasing pressure (
Figure 6a). For the temperature, a decrease in CO
2 adsorption capacity was observed with increasing temperatures (
Figure 6b). Increasing the temperature caused the CO
2 molecules to diffuse faster. Therefore, fewer molecules were able to interact with the active site of the adsorbent. Moreover, at higher temperatures, the surface CO
2 molecules were desorbed into the surrounding gas once there was adequate energy to overcome the gas-solid interaction [
43]. Therefore, increasing the temperature attenuated the CO
2 uptake, in agreement with the exothermic nature of the process. The CO
2 adsorption capacity of NOCn was greater than that of Cn at the same pressure and temperature. This reflects that these adsorbents have different active sites for CO
2 binding on their surface. It also infers that CO
2 has stronger interactions with the active sites on NOCn than on Cn, and so more CO
2 molecules covered the surface of NOCn, leading to a higher CO
2 adsorption capacity.
3.3. Thermodynamic Studies
To gain insight into the adsorption phenomenon and to determine the thermodynamic parameters, the affinity of adsorbate-adsorbent interaction at the equilibrium state was investigated when there was no coverage of adsorbate on the surface of the fresh adsorbent. In principle, at a very low pressure (P → 0), the adsorbate-adsorbent forces are the most dominant, and so we can apply Henry’s law (Henry’s law region). Henry’s constant is directly related to the adsorbate-adsorbent interaction and represents affinity. To obtain Henry’s constant, we used the Virial equation [
44], as shown in Equation (2):
where K
H is the Henry’s constant (mmol g
−1 bar
−1), P is the pressure (bar), q is the amount adsorbed (mmol g
−1), and A
1 and A
2 are the virial coefficients. The virial plot between the natural logarithm function of the ratio of P to q (P/q) and q is then done. After linearization, it gives the straight line approaching the axis (q → 0), the -ln K
H is then obtained from the intercept, whereas, the virial coefficients with higher order (A
2, A
3, …) could be neglected. The K
H at different temperatures of CO
2 adsorption on Cn and NOCn are shown in
Table 5. Over the whole temperature range, higher Henry’s constants for adsorption were found on NOCn, which means that NOCn exhibited a higher affinity for CO
2 adsorption than Cn. It is noticed that the value of K
H became smaller with increasing temperatures, inferring that CO
2 adsorption on the adsorbent was less favorable at higher temperatures, and so the CO
2 capacity was decreased in accord with the temperature effect on the capacity, as discussed above and in
Section 3.2.
To understand the nature of the adsorption phenomenon and type of adsorption, the other thermodynamic parameters were investigated via (i) van ’t Hoff equation for the enthalpy change of adsorption (
.) [
45], as shown in Equation (3); (ii) the fundamental Gibb’s free energy equations, for Gibb’s free energy (
), as shown in Equation (4), and the entropy change of adsorption (
), as shown in Equation (5):
The assumption is that the enthalpy and entropy changes are essentially constant over the small range of studied temperatures. The difference in each adsorption temperature interval should not exceed 10 K. These thermodynamic parameters can be used as a crucial key to characterizing the adsorption process.
The van ’t Hoff plot (natural logarithm function of K
H as a function of the reciprocal temperature) was a straight line (
Figure 7), where the gradient and intercept of the line were
/R and
/R, respectively. The thermodynamic parameters were then determined and are shown in
Table 5. At 25 °C, Cn and NOCn had negative
values, which means the adsorption of CO
2 on both sorbents was spontaneous at this temperature. The
values for NOCn at different temperatures were more negative than those for Cn, confirming that the CO
2 adsorption on NOCn is more thermodynamically feasible. Nitridation at high temperature caused micropore development and doped nitrogen atoms on the surface, which then enhanced the CO
2 adsorption capacity. When increasing the temperature, the
from NOCn and Cn became less negative, which means that the adsorption had a lower degree of spontaneity at higher temperatures. At lower temperatures, the CO
2 molecules diffused to the more energetically favorable active sites on the surface of adsorbent to form a surface layer of CO
2 molecules. Consequently, a higher adsorption capacity was obtained at a lower temperature, as discussed above. At a higher temperature, CO
2 diffuses faster, and CO
2 molecules with a weaker interaction with the less energetically favorable active sites on the sorbent surface can then desorb and diffuse back to the gaseous bulk phase.
The sign of
indicates the heat inflow/outflow to the adsorption system, where a negative sign meant that the nature of the adsorption process on Cn and NOCn was exothermic. The total energy released during the formation of a bond between CO
2 and the active sites on the sorbent was greater than the total energy used in breaking the adsorbed CO
2 bond. The absolute value of
from NOCn was greater than that from Cn, meaning that a larger amount of heat energy is released to the surrounding after adsorbing CO
2 molecules on the NOCn surface. This also reflected the stronger nitrogen-active site on NOCn. Moreover, the absolute magnitude of
revealed the type of adsorption, where
< 40 kJ mol
−1 and > 80 kJ mol
−1 for physisorption and strong chemisorption, respectively, [
46]. Thus, the major mechanism of CO
2 adsorption on Cn and NOCn was physisorption, where the active site on the adsorbent surface attracts CO
2 through weak Van der Waals forces.
For
, the sign corresponded to the degree of randomness of the adsorption process. It was clearly seen (
Table 5) that lower randomness of the system was observed during the CO
2 adsorption onto NOCn than onto the Cn surface. The gaseous CO
2 in the bulk phase moved randomly, while the CO
2 adsorbed on the surface could not move freely, due to their interaction force. The stronger interaction force between CO
2 and the nitrogen-active site on the NOCn caused a more ordered stage with less randomness.
3.4. The Selectivity of CO2 Adsorption
A potential sorbent should express a selective adsorption ability. Herein, the CO
2 and N
2 adsorption isotherms on Cn and NOCn at 25 °C (
Figure 8) were then applied to study the selective CO
2 adsorption via the affinity of adsorbate-adsorbent interactions in terms of Henry’s constant (as shown in Equation (2)). From
Figure 8, it can be seen that Cn and NOCn expressed an N
2 adsorption capacity of around 0.02 mmol g
−1 and 0.21 mmol g
−1, respectively, and CO
2 adsorption capacity of about 0.34 mmol g
−1 and 1.63 mmol g
−1, respectively. The Henry constant for the N
2 adsorption (
) and CO
2 adsorption (
) on Cn and NOCn are summarized in
Table 6, where the selective CO
2 adsorption on Cn and NOCn was then exhibited as the ratio of
, and shown in
Table 6.
was about 0.11 for Cn and 0.42 for NOCn, while
was around 1.86 for Cn and 12.26 for NOCn, respectively. It was noted that the magnitude of
for both sorbents was higher than that of
. This reflected that both sorbents had a greater affinity to CO
2 than N
2. Higher values of
and
for NOCn inferred that NOCn had more favorable adsorption. After nitriding to obtain NOCn, the affinity was increased about 6.6-fold or CO
2 adsorption and 3.8-fold for N
2 adsorption. This enhanced the CO
2 adsorption affinity could be because not only micropores were developed during nitridation, but there was also nitrogen doping onto the carbon structure. The ratio of
was about 16.29 for Cn and 29.22 for NOCn, respectively. This revealed that the sorbents had a selective CO
2 adsorption ability and the CO
2 adsorption was more favorable on NOCn than Cn. This is because CO
2 was attached to the surface of sorbent with stronger induced dipole interactions. Even though CO
2 and N
2 are nonpolar molecules, the polarizability and quadrupole moment are 26.5 × 10
−25 cm
3 and 4.3 × 10
−26 esu·cm
2 for CO
2 and 17.6 × 10
−25 cm
3 and 1.52 × 10
−26 esu·cm
2 for N
2, respectively [
47]. Thus, the carbon surface could form stronger interactions at the electron-rich region, particularly the π-system and/or lone pair electron of N atom in the case of NOCn. Therefore, it is more attractive to the position of C
δ+ of CO
2.
3.7. Adsorption Mechanism
To explain the mechanism of CO
2 adsorption on NOCn, the principles of classical chemistry were applied. Nitrogen atoms in the carbon matrix have a greater affinity for CO
2, since the lone pair electron on the nitrogen atom, especially the pyridinic-N, acts as Lewis base, while the C atom on CO
2 is an electrophile in nature [
20], allowing them to form a Lewis acid-base interaction through N donating an electron to C
δ+ on the CO
2 molecule. Since the adsorption of CO
2 on NOCn was mainly physisorption through weak forces, there was no change in the electronic properties of the CO
2-adsorbent complex and no significant change in the molecular orbital (MO) level [
48]. From the perspective of quantum chemistry, electron donation and electron backdonation, based on the highest occupied MO (HOMO) of the sorbent’s surface interact attractively with the lowest unoccupied MO (LUMO) of the sorbate molecule. The HOMO and LUMO of the CO
2 molecule and the interaction between the nitrogen atom on the carbon structure and the C atom on CO
2 are shown in
Figure 10. For electron donation, the electron pair in the non-bonding MO (1π
g), which is located on the oxygen atom of the CO
2 molecule, was donated to the LUMO on the C site of NOCn. The electrons in the HOMO of the nitrogen atom of NOCn were then back donated to the LUMO of CO
2 (2π
u), which in this MO is mainly a 2p
xC [
49]. In the case of Cn, the pristine carbon without a nitrogen atom had a higher adsorption barrier, compared to NOCn. Therefore, the larger energy gap between the HOMO and LUMO of Cn and the CO
2 adsorbate was exhibited, leading to a lower adsorption capacity as described above. This also leads to less favorable adsorption, which is in good agreement with the thermodynamic studies. The Cn had a lower Henry’s constant than NOCn, meaning that Cn had a lower CO
2 adsorption affinity. When doping nitrogen, the energy gap of the electron transfer between CO
2 and the adsorption site was reduced, which induced the local density of state below the Fermi level [
21,
50].