1. Introduction
As the interest in environmental pollution increases, regulations of automotive emissions are being made stronger around the world [
1,
2]. Automotive emissions are divided into evaporation gas and exhaust gas. The exhaust gas is generated by the combustion of fuel and consists of harmful substances such as CO
x, NO
x, SO
x, and particulate matter (PM) [
3,
4,
5]. The evaporation of unburned fuel from the fuel tank generally forms evaporation gas [
6]. Its main component is hydrocarbon (HC) molecules, and there are many studies that indicate that HC is one of the critical materials causing smog, mist, and lung cancer [
7,
8].
Evaporation gas is normally generated even when the car is parked or being refueled; thus, it is difficult to prevent this using a separate actuator [
9]. The evaporation gas is systematically collected and concentrated by means of fuel-vapor-emission control systems (a carbon canister and hydrocarbon trap sheet composed of activated carbon pellets and an activated carbon sheet, respectively; ACP and ACS) [
10,
11,
12]. When the engine starts, the collected evaporation gas is transferred with air to the engine and used to enhance the fuel efficiency [
13,
14].
The diurnal regulation value for unburned HC emissions in cars was previously 500 mg/test (LEV-II), but recently it was further strengthened to 300 mg/test (LEV-III) [
2]. Especially, canister bleed emission limits have been set at 20 mg/test for passenger car [
2]. In the past, fuel-vapor-emission control systems were studied to enhance the pore characteristics of activated carbon or to increase the apparent density in order to adsorb larger amounts of evaporation gas [
15,
16,
17]. However, for the fuel-vapor-emission control systems that meet the new stringent regulations (LEV-III), it is necessary to develop activated carbon capable of adsorbing a larger amount of evaporation gas and of completely adsorbing low concentrations of evaporation gas at the same time [
2]. In order to solve this problem, a method for applying a material for the selective adsorption of low-concentration evaporation gas by an existing carbon canister, and a method for adding a separate device such as a ceramic honeycomb, have been discussed [
10].
Activated carbon (AC) occurs in shapes classified as granules [
18], pellets [
19], and fibers [
20,
21,
22]. Granular and pellet AC have a higher bulk density than fibrous activated carbon (activated carbon fiber, ACF), and so they can absorb a larger amount of harmful substances because of the larger input mass with the same volume [
16,
23]. Pellet AC usually exhibits lower pressure drop performance compared to granular AC [
24,
25], and so it is widely used in adsorption towers, canisters, and air cleaners. The ACF has a faster adsorption rate and better adsorption for low-concentration harmful substances than granular and pellet AC due to the excellent micropores on its surface [
22,
26].
In this study, the HC adsorption characteristics of AC (granular and pellet) and ACF were investigated to remove the evaporation gas of an automobile. The ACF was fabricated using various H2O activation times for pitch fibers to observe the relationship between the adsorption capacity of low-concentration evaporation gas and the pore structure of the ACF. The pore development mechanism of ACF was confirmed through the pore characteristics and crystal structure. The HC adsorption characteristics of AC and ACF were analyzed using various concentrations of n-butane according to the ASTM D5228 standard.
3. Results and Discussion
Isothermal adsorption–desorption curves are the most powerful method for analyzing the pore characteristics of the ACF.
Figure 2 exhibits the N
2/77 K isothermal adsorption–desorption curves of the ACF. The curves of most ACF variants (ACF-H-9-2 to ACF-H-9-5) were classified as Type I by the IUPAC classification and were found to include mainly micropores [
33]. Meanwhile, the curve of ACF-H-9-6 was classified as Type IV, and thus the volume ratio of mesopores to micropores was relatively high [
33]. The hysteresis patterns were hardly observed in the curves of all the ACF variants, and only a few hysteresis patterns were observed in ACF-H-9-6. Therefore, it is considered that the pores of all the ACF are wedge-shaped, and that the pores are well-developed on the surface of the ACF. The adsorption–desorption isotherm curves of the AC (BAX1500, BAXLBE, and WVA1100) for canisters were determined to be Type IV, and very large hysteresis was observed. It is recognized that the AC has a high mesopore volume fraction within the total pore volume, and pot-shaped pores are present inside the pore structure.
Table 1 shows the textural properties of the ACF and AC. The specific surface area and total pore volume of the ACF increased from 840 to 2630 m
2/g and from 0.33 to 1.34 cm
3/g, respectively, with increased activation time. From ACF-H-9-2 to ACF-H-9-5, most of the porous structure was found to consist of micropores. As the activation time increased, the mesopore volume continuously increased. ACF-H-9-6 was observed to consist of about 37% of mesopores (of the total pore volume) and to have the highest specific surface area of all the ACF variants.
The AC for canisters has completely different pore structures compared to the ACF. The AC mesopores accounted for more than 57% of the total pore volume. The specific surface areas of the AC were in the order of BAX 1500 > WVA 1100 > BAX LBE. The pore ratio that ACF-H-9-6 exhibited was similar to that of the AC for canisters. ACF-H-9-6 had higher specific surface area and micropore volume than BAX 1500, but BAX 1500 had a higher mesopore volume than that of ACF-H-9-6.
Figure 3a exhibits the micropore size distribution curves of the ACF by an NLDFT method. The micropore size diameter was determined by gas sorption and estimated using nonlocalized density functional theory (NLDFT) and the grand canonical Monte Carlo method (GCMC) using the BELSORP evaluation software from the computer simulation.
Figure 3a shows the typical pore size distribution of porous carbonaceous materials. The pore size distribution curves indicate that as the activation time increases to 40 min (ACF-H-9-4), the pore diameter increases and the width of the curve gradually increases. The pore size distribution curves of ACF-H-9-5 were observed to possess a narrow (~1 nm) and a broad (>2 nm) curve. ACF-H-9-6 had a broad curve, and the pore diameters increased from micropores to sub-mesopores. Therefore, it was confirmed that the mesopores were formed by the alteration of the micropores to mesopores by the widening of the pore diameters by further oxidation or by the collapse of the micropore walls. It was also confirmed that new micropore development continued in the variants up to ACF-H-9-6, as observed by the increase in micropore volume.
Figure 3b shows the mesopore size distribution of the ACF calculated using the BJH equation. The mesopore volume of the ACF was found to increase with increasing activation time. ACF-H-9-6 had the most massive mesopore volume among the ACF variants. From ACF-H-9-2 to ACF-H-9-5, mesopores of less than 10 nm in pore diameter were mainly developed, whereas ACF-H-9-6 had mesopores of pore diameter <50 nm. Because hysteresis of the ACF is hardly observed in
Figure 2 (which means that the pore shape is probably a wedge-shape, as shown in
Figure 2), it can be inferred that the location micropores of the ACF gradually shift to deep inside as the activation time increases, while maintaining a wedge-shaped pore structure (
Figure 4a). On the other hand, it is assumed that the AC has a pore structure of a unique jar shape, resulting in a high mesopore volume ratio and a large hysteresis curve (
Figure 4b).
Physical activation is the process of oxidizing crystallites to form various pores. Therefore, the changes of crystallite and pore structure with increasing activation time are very closely related. XRD is an advantageous method for observing the crystallite structure of ACFs.
In
Figure 5, the XRD curve of the pitch fiber (as-received) exhibits a clear 002 peak and a well-developed domain characteristic of the inherent crystallite structure of the pitch. The XRD curves of the ACF exhibit the typical appearance of an isotropic carbon material. The 002 peak diffraction angle of the graphite crystallite is 26.56°, but the 002 peaks of the ACF samples are located at about 23°, indicating that the crystallite structures are considerably different from the structure of graphite. Besides this, the widely spread 10
l peak indicates that each atomic layer is disordered and imperfectly laminated.
The XRD curves in
Figure 3 were determined using the Bragg and Scherrer equations, and the interplanar spacings (d
002 and d
10l) and the crystallite sizes (L
c and L
a) were measured. The structural parameters are shown in
Table 2 and
Figure 5.
In
Figure 6a, the L
c (crystallite height) and L
a (crystallite size) of the ACF increases with increasing activation time. Generally, the XRD data of carbonaceous materials provides statistical data about the number of crystallite aggregates. It is known that amorphous and small crystallites are preferentially oxidized, compared to larger crystallites, during the activation process; that is, as the activation of a pitch fiber proceeds, amorphous parts and small crystallite are easily oxidized, which may appear to increase the relative size of the entire crystallite. Therefore, it is considered that L
c and L
a are increased by the oxidation of amorphous areas or small crystallites as the activation time increases. Moreover, a steady increase of L
c and L
a during long activation times seems to result in the sustained oxidation of amorphous regions or small crystallite. The low increase of L
a in ACF-H-9-6 is considered to be highly correlated with the increase of the mesopore volume (
Table 2). The increase in the micropore volume is the result of the oxidation of amorphous or small crystallites, leading to an increase in the L
c and L
a. By contrast, the increase in the mesopore volume is formed by the oxidation of the micropore walls (the edge of the large crystallite), resulting in a decrease in L
c and L
a. Therefore, ACF-H-9-6 is considered to result in a tiny increase in L
a because this sample shows an increase in the micropore volume and a significant increase in the mesopore volume.
In
Figure 6b, d
002 decreases, and d
10l does not change significantly with increasing activation time. In the graphite crystal structure, the 002 planes are composed of strong hybridized sp
2 bonds, and the vertical π bond on 002 planes displays weak interlayer bonding. The decrease of d
002 is considered to be the result of the continuous oxidation of amorphous parts and small crystallites.
The butane working capacity (BWC) is a useful analytical method for evaluating the canister performance of ACF and AC.
Table 3 lists the BWC, butane activity (BA), and butane retentivity (BR) of the ACF and AC measured according to ASTM D5228. The BA and BR indicate the butane adsorption capacity and the residual butane rate, respectively, after the desorption of the butane from the ACF. The BWC is defined as the difference between the butane adsorbed at saturation and that retained per unit volume of the ACF after a specified purge.
The BA and BR of the ACF increased with increasing activation time and were observed to range between 15.78–57.33% and 4.19–11.47%, respectively. The butane activity of the ACF was increased by the increase of the micropore and mesopore volume. It is generally known that AC with a high mesopore ratio can desorb the adsorbate more easily in the purging process. For example, as shown in
Table 3, the BR of the AC for the canister was found to be lower with the increase in the mesopore volume ratio. However, the mesopore volume ratio of the ACF increased from ACF-H-9-3 to ACF-H-9-6, but the BR decreased. As shown in
Figure 2, micropore development is more probable during pore development in the ACF. The micropores of the ACF were then transformed to mesopores, and new micropores developed deep inside the mesopores. Therefore, the micropores are uniformly formed on the surface of the ACF (up to ACF-H-9-5) and most of the pores are micropores. The ACF exhibited higher BA and BR values than those of AC with a similar specific surface area. The ACF is known to have faster adsorption characteristics than AC because of the formation of pores on its surface. This means that the ACF has a higher heat of adsorption than that of AC, resulting in desorption difficulty. Therefore, it is considered that ACF has better adsorption characteristics even if the specific surface area is similar to that of AC. On the other hand, because the AC has a higher mesopore volume ratio than the ACF, desorption can be easily performed during the purging process.
The BA of the AC was investigated by exploring its correlation with textural properties. In
Figure 7, the BA of the ACF and AC appears to be linearly dependent on the specific surface area and microporosity. On the other hand, the BA was not found to be linearly related to the mesoporosity. These results are consistent with the results in
Table 3 and suggest that the pore size may play an essential role in determining the BA of AC.
Figure 8 exhibits the result of plotting the pore volume according to pore diameter in 0.5 nm units using the NLDFT method and then plotting the coefficient of determination with BA. It is considered that the BA of the ACF is determined by pores with diameters from 2.0 to 3.5 nm. Primarily, the BA was strongly dependent on pores with diameters from 2.0 to 2.5 nm. This result is consistent with those in previous studies [
15] showing that mesopores (2.0–2.5 nm) are essential for providing the butane adsorption capacity.
In order to meet the enhanced regulation of evaporation gas, it was necessary to develop an adsorbent with excellent adsorption capacity for HC at low concentrations. Because ASTM D5228 is used to evaluate the adsorption capacity of butane (100%) by mass change, it was impossible to conduct an experiment measuring butane at low concentration because the change in mass would be minimal.
Figure 9 exhibits the change in concentration of the AC and ACF at various concentrations of butane using a Q-mass for the ASTM D5228 measurement.
As the butane concentration decreased, the slope of the curve near the breakthrough time of the AC decreased. On the other hand, the decrease in butane concentration did not significantly affect the curve slope of the ACF near the breakthrough time. In general, as the gas concentration decreases, the adsorption rate of the adsorbent decreases because the adsorption and the desorption reaction occur simultaneously.
The ACF has a faster adsorption rate than AC because micropores develop on the surface due to its inherent pore structure characteristics. Therefore, the adsorption rate of AC decreased as the concentration of butane decreased, suggesting that the slope of the curve decreased. On the other hand, because the adsorption rate of the ACF is faster than that of AC, it is considered that a change of the curved ACF slope does not appear, even if the concentration of butane decreases.
Figure 9a exhibits the change in concentration caused by the adsorption of 100% butane by the AC and ACF. In
Table 3, the butane activity of the ACF is higher than that of AC. However, in
Figure 9a, it can be seen that the breakthrough times of AC are longer than those of ACF. The apparent density of AC is about 1.9–2.3 times higher than that of ACF, so the mass of AC is higher, given the same volume. The butane adsorption capacity of the ACF was found to be proportional to the pore characteristics. In particular, ACF-H-9-6 exhibited the highest adsorption performance in the 100% butane-adsorption test because it had the highest specific surface area and highest total pore volume among the ACF variants.
Figure 9b shows the change in concentration caused by the adsorption of 10% butane by AC and ACF. In
Figure 5b, the AC exhibits better butane adsorption characteristics than the ACF (
Figure 5a), but the breakthrough times between the AC and ACF are very close. The breakthrough times of the ACF variants were almost the same, except for ACF-H-9-2. This result is attributed to the decrease in the adsorption rate of the ACF and AC due to the lower butane concentration.
Figure 9c shows the change in concentration caused by the adsorption of 1.0% butane by the AC and ACF. The slope of the curve near the breakthrough time of the AC was further reduced, while the curve of the ACF maintained a high slope. Moreover, ACF-H-9-4 and ACF-H-9-5 show breakthrough times similar to those of the AC. These results indicate that the adsorption rate of the AC is significantly lower than that of the ACF at very low butane concentrations. Among the ACF variants, although ACF-H-9-4 had a lower specific surface area and total pore volume than those of ACF-H-9-5 and ACF-H-9-6, it exhibited the best adsorption capacity for 1.0% butane. As shown in
Figure 4, it can be recognized that ACF-H-9-4 may have micropores which formed mostly on the surface. In conclusion, ACF-H-9-4 exhibited excellent adsorption characteristics for 1.0% butane due to its well-developed micropore structure.
Figure 9d exhibits the change in concentration caused by the adsorption of 0.1% butane by AC and ACF. At this butane concentration, the ACF exhibited better adsorption characteristics than the AC. Moreover, the shorter the activation time of the ACF, the better the adsorption capacity observed. In
Figure 2, the isothermal curves of the ACF prepared in this work showed Type 1 curves, and the hysteresis of the samples was negligible. This could mean that the pore shapes of the samples might be wedge-shaped or cylindrical slit-shaped. Moreover, as the activation time increases, the diameter of the pore inlet located on the surface of the ACF increases gradually. Therefore, as the activation time increases, the positions of pores with diameters of 2.0–2.5 nm, which affect the butane adsorption, are expected to gradually shift to the inside of the pore structure, meaning that the adsorption rate decreases.