Surface-modified spherical activated carbon materials for pre-combustion carbon dioxide capture

Surface modification of activated carbon beads via HNO3 oxidation and subsequent amination at elevated temperatures was investigated as a means to improve their performance for CO2 capture, and the effects of the resultant changes in porosity and surface chemistry on adsorption characteristics of the samples were studied. Characterisations conducted with elemental analysis, physical adsorption, X-ray photoelectron spectroscopy and scanning electron microscope demonstrate that both the porosity and surface chemistry of the carbon beads were tuned by the modification without any alteration of the integrity of the desirable spherical morphology. Adsorption evaluation with both thermogravimetric analysis and high pressure volumetric analysis under various conditions indicate that one of the modified samples had a high CO2 adsorption capacity (8.64 mmol g 1 at 20 bar and 30 C) with fast adsorption/desorption kinetics, superior durability and good selectivity over N2 and H2. Both the unique spherical form (diameter 1⁄4 1.2 0.2 mm) and the superior adsorption performance render the modified carbon beads a promising candidate for CO2 capture especially in pre-combustion capture using pressure swing adsorption.


Introduction
Various world energy outlooks suggest that in spite of the great efforts and heavy investments that have been made to improve energy efficiency and increase the share of renewable energy in primary energy supply, fossil fuels will continue to dominate the world primary energy supply for the coming decades. [1][2][3][4] Consequently, signicant contributions from carbon capture and storage (CCS) will be required to combat climate change to increase the opportunity for the continual use of fossil fuels. [5][6][7] However, the current state-of-the-art aqueous amine scrubbing technology for CO 2 capture is prohibitive for wide deployment due to its high operational costs and high energy penalty associated with absorbent regeneration. 8,9 On the other hand, solid adsorbents looping technology (SALT) offers a viable alternative that has many advantages over solution-based technologies, including substantially reduced energy consumption, greatly minimised health, safety and environmental hazards and greater structural variability of the adsorbents that can be selected.
Over recent years, a wide array of solid adsorbents have been used for CO 2 capture, those including zeolites, carbon materials, inorganic oxides, amine modied silicates and more recently, the more niche metal-organic frameworks (MOFs) and covalent organic frameworks (COFs) materials. [10][11][12] In previous investigations, a high adsorption capacity is the most pursued parameter among many others. 10,11 In practical carbon capture applications however, the so-called "working capacity" should be more important as it concerns both material and engineering issues. Other challenges facing real carbon capture applications, such as the large volume of ue gas at high ow rate, complicated chemical compositions of the ue gas and fast capture rate that is required, should also be emphasized. 13 Therefore, one needs to consider process costs, process efficiencies and engineering feasibilities for a real carbon capture process, and these criteria impose further requirements for applicable CO 2 adsorbents as illustrated in Fig. 1. For example, the factor of adsorption/desorption kinetics (factor 6 in Fig. 1) is located on the le-bottom of the triangle in Fig. 1 because adsorbents with fast kinetics can increase the efficiencies of a capturing process and lowering related costs by accelerating the adsorption/desorption cycles for a given carbon capture facility. Meanwhile, faster cycling requires a better designed and integrated adsorption/desorption subunit, giving rise to further challenges in process engineering and limitations in adsorbents selection. The comprehensive requirements for CO 2 adsorbents are also discussed in a recent review by Sayari et al. 14 A screening of the possible adsorbents for CO 2 capture indicates that carbon based adsorbents represent a good option to balance all the factors in Fig. 1 due to its promising properties such as low cost for preparation, chemical inertness, easyto-regeneration, etc. 10,11,14,15 As a result, the development of effective carbon-based adsorbent materials for CO 2 capture has received signicant attentions over recent years, 15 related works including: (i) preparation of carbons by templating methods 16,17 or direct carbonization of coal, 18 polymers 19,20 and biomass residuals; [21][22][23][24][25] (ii) chemical activation of carbon by KOH 21 and K 2 CO 3 ; 22,26 and (iii) modication of carbon by N doping, 17,20,27 S doping 28 and amine loading. 29,30 More recently, some carbon materials with exceptionally high CO 2 uptake were reported. For example, Srinivas et al. 31 prepared MOF-derived porous carbons with CO 2 adsorption capacities as high as 27.0 mmol g À1 at 27 C and 30 bar. By carbonation and activation of selected coal tar pitches, a CO 2 uptake of 20.5 mmol g À1 was obtained by Shao et al. at 25 C and 18 bar. 23 In addition to these results, Sjostrom et al. 32 has reached the conclusion that carbon can be the viable adsorbents for practical CO 2 capture especially if their CO 2 uptake can be improved, which emphasized again that carbon-based materials have great potential in CO 2 capture, particularly for pre-combustion capture where the CO 2 partial pressure in the gas stream is much higher than in postcombustion ue gases.
Although many carbon-based adsorbents currently under development have superior adsorption capacity, they are produced predominantly in low-density powder form which is unsuitable for practical applications and therefore needs to be further fabricated via pelletisation or granulation process with the aid of binders or additives, which can oen lead to signicant loss of their adsorption performance. [33][34][35][36][37][38] To address this issue, some researchers tried to prepare carbon monolith/ honeycomb adsorbents using purpose-designed molding processes, 20,39,40 however these methodologies oen involves the use of complicated processes, and the fabricated materials, mostly in their cylindrical and/or rod shapes, are not particularly suitable for use in SALT processes (e.g. circulating uidized bed) due to serious attrition problems that may arise from the undesirable material morphology.
In our previous work, a cost-effective and easy-to-scale-up hydrothermal technique has successfully been developed to produce spherical active carbon materials from a variety of polymeric precursors. The prepared porous carbon beads possess the most favourable properties for applications in continuous CO 2 capture using circulating uidised bed, such as their tuned porosity, high mechanical strength and desirable spherical forms with uniform diameters. 41 In this paper, different post-treatments, including HNO 3 oxidation and amination were investigated as an effective means to further improve the CO 2 adsorption performance of the carbon beads, the adsorption behaviour of the samples was rationalized based on extensive characterization tests, and their application potential in CO 2 capture was evaluated based on the analysis of adsorption capacity, selectivity, kinetics and reusability.

Preparation of the original carbon beads
The parent carbon beads were prepared from phenolic resins using a novel hydrothermal process. Typically, novolac-type phenolic resins were rst dissolved in a formulated solvent to form a solution which was later mixed with aqueous polyvinyl alcohol (PVA) and sealed in an autoclave. The mixture was then heated to 130 C at a rate of 5 C min À1 and held at this temperature for 1 h with rigorous stirring (400 rpm). The autoclave reactor was then cooled down to ambient temperature to allow the formed resin beads to be recovered. The resin beads was rst washed adequately with distilled water and dried at 110 C for 24 h before it was carbonised in N 2 at 800 C for 1 h to yield the carbon beads (sample AC, hereaer) for further treatments.

Surface modication of sample AC
The AC sample was further submitted to HNO 3 oxidation by mixing 10 g of AC and 250 mL of concentrated HNO 3 in a beaker at room temperature (25 C) for 1 h (sample oxAC), aer washing and drying, the obtained oxAC was aminated in a tube furnace under NH 3 at 300, 600 and 800 C to afford sample oxAC-300NH3, oxAC-600NH3 and oxAC-800NH3, respectively. For comparison, a sample thermally treated at 800 C in N 2 aer HNO 3 oxidation was also prepared (oxAC-800N2).

Characterization
Textural properties of the samples were measured by N 2 physisorption on an Micromeritics ASAP 2420 instrument at À196 C and CO 2 at 0 C. Prior to any measurements, all samples were rst degassed at 120 C overnight. The apparent surface area (S BET ) was calculated according to the method suggested by Parra et al. 42 The cumulative pore volumes (V total ) were calculated from the amount of nitrogen adsorbed at P/P 0 of ca. 0.99, and the average pore diameter (D avg ) was calculated by 4V total / SA BET . The micropore volume (V micro ) and surface area (SA micro ) were determined by the t-plot method. Narrow micropore volumes (W 0 ) of the samples were calculated by the Dubinin-Radushkevich (DR) method 43 from CO 2 isotherms obtained at 0 C. In theory, the DR method allows the calculation of average width of the narrow micropores (L 0 ) based on the empirical relationship between L 0 and the characteristic adsorption potential (E 0 ) in the DR equation. 44 However, it was reported that this relationship is inuenced by the surface chemistry of samples. 45 Therefore, L 0 was not calculated here due to the signicant differences in the surface properties of the samples.
X-Ray photoelectron spectroscopy (XPS) was carried out using a Kratos AXIS ULTRA instrument with an Al Ka source (energy 1486.6 eV). The high resolution scans were charge corrected to the main C1s peak (284.8 eV). Morphology of the samples was examined with a QUANTA 600 scanning electron microscope (SEM) instrument. An elemental analyser (Thermo EA1112 series) was used to determine the carbon, hydrogen, and nitrogen contents of the samples.

Evaluation of the CO 2 adsorption performance
The CO 2 capture behaviors of the samples were measured by a TA Q500 thermal gravimetric analyser (TGA) under ambient pressure. In brief, 20-30 mg of the samples were loaded in a platinum pan, heated to 115 C and held at this temperature in owing N 2 (100 mL min À1 ) for 10 min to remove any physisorbed moisture and/or CO 2 , and the temperature was then decreased (10 C min À1 ) to and equilibrated at 30 C. To start the adsorption tests, the gas ow was switched to 100 mL min À1 of pure CO 2 or 15 vol% CO 2 in N 2 to perform CO 2 adsorption tests and the temperature was held at 30 C for 30 min in order for the adsorption to reach equilibrium. The samples were then heated to 115 C at a ramp rate of 0.5 C min À1 . The use of such a slow ramp allows the adsorption to change with temperature in a quasi-equilibrium state and the mass change recorded can be used to assess the inuence of temperature on CO 2 adsorption capacity. Reversibility of CO 2 adsorption/desorption was previously checked to make sure the as-obtained data is valid.
High pressure CO 2 adsorption tests were carried out with a High Pressure Volumetric Analyser (Particulate Systems HPVA-100). Prior to any tests, the sample was rst degassed in vacuum at 120 C overnight to remove any physisorbed moisture and/or CO 2 . Table 1 presents the results from elemental analysis (EA) of the samples. The parent AC showed a high carbon content of 93.6 wt%. This result is expectable as the sample was derived from phenolic resin precursors containing little mineral substances. Aer HNO 3 treatment (sample oxAC), the contents of oxygen and nitrogen increased by an order of magnitude due to the introduction of surface nitrogen and oxygen-bearing functionalities, such as hydroxyl, carboxyl and quinine groups. 46 Signicant proportions of these functionalities were eliminated during the subsequent thermal treatment in N 2 as indicated by the decreased oxygen and nitrogen contents in the sample oxAC-800N2. Similarly, both the nitrogen and oxygen contents diminished when oxAC was treated by NH 3 at 300 and 600 C, which is a typical behaviour for the pyrolysis of carbon precursors. 26 Interestingly, although the oxAC sample showed the highest bulk nitrogen content, the highest surface nitrogen content was observed for the sample oxAC-300NH3 as measured by XPS. This result conrms the occurrence of surface reactions between NH 3 and oxygenated groups to introduce nitrogen functionalities onto the surface of oxAC-300NH3. Despite the decrease in surface nitrogen concentration observed for oxAC-600NH3, further increase of the amination temperature to 800 C led to the increase of nitrogen content in the resulted oxAC-800NH3, indicating a different mechanism of nitrogen incorporation process at different temperatures that might lead to the formation of different nitrogen-bearing functionalities. It is believed that at high temperatures such as 800 C, most of the oxygenated groups have decomposed, and therefore the surface amination was mainly dominated by the anchoring of free radicals such as NH 2 and NH from the decomposition of NH 3 , and the formed nitrogen functionalities had a higher opportunity to further cleavage the N-H bond, leading to the formation of pyridine-like structures. This assumption is also in line with decreased H/N mass ratio with the increasing of amination temperature (Table 1).

Elemental analysis.
3.1.2 Textural properties. Fig. 2 and Table 2 summarise the textural properties of the samples measured by the physical adsorption of N 2 at À196 C and CO 2 at 0 C. According to the IUPAC classication, all but the oxAC sample exhibited a type-I isotherm, suggesting the materials are mainly microporous. 47 Table 1 Weight based composition measured by EA and XPS (dry base) The isotherm of oxAC is characterized by an opening knee and a considerable increase of N 2 uptake at higher partial pressure region, indicating its higher fraction of mesopores. As shown in Table 2, the parent carbon beads (sample AC) have a moderate surface area (SA BET ) of 972 m 2 g À1 with a total pore volume (V total ) of 0.45 cm 3 g À1 . Aer HNO 3 oxidation, the SA BET and V total of oxAC decreased with slightly increased average pore diameter (D avg ). Upon removal of the generated oxygen-containing groups by thermal treatment in N 2 , a signicant improvement in porosity can be observed for oxAC-800N2, which has a surface area comparable to that of the parent sample AC. It has been frequently reported that HNO 3 oxidation can reduce the porosity of carbon materials, 47,48 due to either the blockage of micropores by the formed oxygencontaining functionalities or partial destruction of the carbon framework by damaging the basal planes. 49 Although pore blockage cannot be ruled out based on the present data, the high microporosity of the oxAC-800N2 convinced us that extra micropores, which could be recovered by the removal of oxygencontaining functionalities during thermal treatment, were formed via the penetration of the carbon framework by the oxidant. As shown in Table 2, both oxAC and oxAC-800N2 exhibit W 0 values similar to that of the parent sample AC of ca. 0.24 cm 3 g À1 , suggesting that the inuence of HNO 3 treatment on the narrow micropores is small. This can be expected, given the difficulty for the oxidant to enter these narrow micropores as a result of the high surface tension of the aqueous solution and the hydrophobic nature of the carbon surface.
A substantial increase in microporosity was observed for the 300 C aminated sample (oxAC-300NH3). It is interesting to note that while the total micropore volume of the oxAC300NH3 increased by over 20% as compared with oxAC (from 0.28 to 0.35 cm 3 g À1 ), the total pore volume of this sample remained relatively unchanged (0.41 versus 0.39 cm 3 g À1 ). This implies that the mesopores in the sample oxAC diminished due to pore narrowing during the amination, which might be trigged by the reactions of NH 3 with the oxygenate groups. [50][51][52] Similar assumption can be applied to explain the increase of W 0 for this sample because carbon activation by NH 3 becomes less effective under temperatures as low as 300 C. With increasing the amination temperature from 300 to 800 C, further activation/ surface modication of the carbon beads can be achieved, which imposed a range of changes on the porosity of the samples such as formation of new narrow-micropores, widening/blocking of the existing wide-and narrowmicropores, etc. 18,27 These synergetic effects led to a gradual increase in S micro /V micro and a decrease in W 0 . Consequently, both the SA BET and V total increased for the oxAC-600NH3 and oxAC-800NH3 samples. It seems that the increase was most likely induced by the enhancement of wide-microporosity as indicated by the similar increasing of V total and V micro . Similar results, namely the increased wide-microporosity during enhanced activation, were also observed by An et al. during their attempts to carbonise activated carbon bre-phenolic resin composites. 39 3.1.3 Morphology. Morphology of the samples observed by SEM are presented in Fig. 3. In general, surface modication imposed little inuence on the morphological integrity of the desired spherical form, and all the samples showed similar morphology (details not shown). It has been reported this kind of spherical materials possess great potential and engineering exibilities in large scale process as their shape and texture exhibit the ability to minimize material attrition in circulating uidized bed reactors. 53 Interconnected 3D porous network structures are also evident inside the carbon beads with irregular openings and cracks in mm-scale on both the external and internal surfaces of the samples (Fig. 3a and b). These features serve as channels for uids to pass through and thus facilitate  the more effective diffusion of CO 2 molecules within the interior porous networks present in the carbon beads, which is of vital importance for the adsorption efficiency and kinetics of the materials. 22 A closer check of the carbon beads revealed a relatively smooth outer surface (Fig. 3c), indicating a better capability for these samples to resist attrition. In contrast, as showed in Fig. 3d, the surface of cavities inside the carbon beads are sponge-like frameworks comprising of relatively uniform microspheres with a diameter of ca. 100 nm.

Surface chemistry.
It has been recognized that the CO 2 affinity of N functionalities incorporated into graphene structures were governed by the mobility and delocalization of the lone pair electrons on the N atoms, which is determined by adjacent chemical environment. 54 Therefore, XPS was carried out over the oxidized and aminated samples to investigate their surface chemistries. The quantied surface compositions are listed in Table 1. For clearer comparison with the EA results, the initially obtained atomic compositions were converted to weight-percentage concentrations. As can be seen, increases in nitrogen and oxygen contents were evident as a result of HNO 3 oxidation, being consistent with the EA results. Substantially higher surface N/C mass ratios measured by XPS were found for all aminated samples, while the values obtained for oxAC from EA and XPS were similar. On the other hand, it is interesting to note that all the O/C mass ratios measured by XPS were considerably lower than those measured by EA, being in contrast to the trend observed for the N/C mass ratios. This clearly indicates again that the oxidation by liquid phase HNO 3 was so aggressive that not only the surface, but also the carbon framework was altered by involving nitrogen and oxygen atoms. The hypothesis can be evidenced by the considerably higher O/C mass ratios measured by EA than that by XPS, which is indicative of the preferential occurrence of oxygenated functionalities within the carbon framework in addition to the outmost surface. As mentioned in Section 3.1.2, due to the hydrophobic nature of the AC surface, it is reasonable to assume that during oxidation, the distribution of aqueous HNO 3 over the carbon surface is unlikely to be uniform but preferably in the form of micro droplets. This might lead to the localised penetration of the corrosive HNO 3 into the carbon framework, creating interior "channels" with abundance of oxygenate groups. By comparing the textural parameters of oxAC and oxAC-800N2 (Table 2), it can be concluded that these "channels" can be regarded as blocked micropores where oxygen present in various functionalities that can hardly be detected by XPS. In contrast, there is little limitation on the accessibility of the NH 3 molecule to enter the micropores, and thus amination occurs more evenly and preferably on the surface, leading to pore widening and surface enrichment of nitrogen as revealed by the higher N/C ratio measure by XPS than that of EA. Fig. 4 presents the N1s spectra and the peak tting results of the N-containing samples, for the sample oxAC, peaks at 399.4, 401.0, and 406.0 eV can be attributed to pyridinic, quaternary and nitro-nitrogen according to the NIST XPS Database. Aer amination, a new peak was detected at 400.0 eV, which is attributable to pyrrolic and/or pyridone nitrogen. Although these two functionalities cannot be effectively differentiated by XPS, it was reported that pyridone nitrogen is more stable than pyrrolic nitrogen at elevated temperatures, 55 therefore pyridone has a higher opportunity to survive on the aminated samples. Among the identied groups, pyridinic nitrogen has lone pair of electrons, which may enhance the alkaline character of the surface, and thus would facilitate the adsorption of acidic CO 2 . As for the pyridone nitrogen, Fan et al. has reported that the p-p conjugation between the -OH and the aromatic rings endows the ortho-N with a higher electron density, which can lead to higher affinity for CO 2 adsorption. 26 It is noteworthy that the delocalization and conjugation effects from electrons of nitrogen, aromatic rings and other atoms have a signicant impact on the basicity of nitrogen-containing groups, therefore it is arbitrary to allege the relative importance of these groups in terms of their affinity to acid gases such as CO 2 . 56,57 For the quaternary and nitro-nitrogen on the other hand, they are  partially positive charged due to the adjacent oxygen and aromatic rings, therefore it is reasonable to assume a limited contribution for these functionalities to the adsorption of CO 2 . The relative fractions of the identied nitrogen functionalities are tabulated in Table 3. It can be observed that with the increase of amination temperature, the fraction of nitronitrogen decreased notably due to the removal and/or reduction of these functionalities, while the relative abundance of pyridone and quaternary nitrogen were less affected due to their higher thermal stability as compared with nitro-nitrogens. A higher fraction of pyridinic nitrogen can be seen for the samples aminated at higher temperatures, which is indicative of the enhanced surface amination by the formed N-radicals and further cleavage of the N-H bonds at higher temperatures as has been discussed in Section 3.1.1.

CO 2 adsorption behaviour
3.2.1 Adsorption capacity. The spherical carbon adsorbents, including both the parent and surface-modied carbon beads, were subjected to different test regimes to determine their performance for CO 2 capture. Table 4 presents the CO 2 uptakes of the samples measured in TGA at different CO 2 partial pressures, and signicant improvements in CO 2 adsorption capacity was obtained for the surface-modied carbon beads particularly at low CO 2 partial pressures. At 30 C and a CO 2 partial pressure of 0.15 bar, the oxAC sample, which has the lowest textural properties (Table 2), shows the highest CO 2 uptake (0.91 mmol g À1 , 25% increase as compared with AC). This can be ascribed to the strong affinity of the abundant heteroatom functionalities, oxygenate functional groups in particular, with the quadrupole moment of the CO 2 molecule. Similar effects have also been reported for carbons, MOFs and task-specic ionic liquids (TSILs). 28,58,59 When comparing adsorption capacity at 15 vol% CO 2 of the three aminated smaples, it can be observed that oxAC-300NH3 possessed the highest CO 2 uptake at both 30 and 45 C, which can be explained by the higher amount of N-H functionalities (Tables 1  and 3) because the binding between CO 2 and C-N-H groups is stronger than that of C]N-C. At a CO 2 partial pressure of 1 bar, the CO 2 uptake of all the aminated samples were higher than that of oxAC, suggesting the greater importance of textural properties on adsorption capacity. The sample oxAC-300NH3 showed the highest uptake (2.57 mmol g À1 ), most probably due to its highest narrow micropore volume (W 0 ). It has been reported that at atmospheric pressure, only pores with diameters lower than 5 times of the molecular size of the adsorbate are effective for gas adsorption. 18 As for CO 2 , the dynamic molecular diameter is 0.209 nm, it is therefore expected that the narrow micropores are the major locations that accommodate CO 2 at 1 bar. 39 The combinative effect of surface chemistry and textural properties on CO 2 adsorption of surface-modied carbon materials has been well-demonstrated in open literatures, 41,54 and it is believed that the surface chemistry, rather than the textural properties, governs the CO 2 capture performance at lower (partial) pressures, while the textural properties becomes more important at higher CO 2 pressures as the adsorption process becomes controlled by the micropore lling mechanism.
In fact, activated carbons in general are well known for their capability to achieve higher adsorption capacities at pressures higher than atmospheric, and this together with their moderate adsorption strength makes them the ideal adsorbents for CO 2 capture at high pressures (such as pre-combustion capture). 21,23,31 In this study, the high pressure CO 2 adsorption performance of the samples was measured using a high pressure volumetric method, and the obtained isotherms are presented in Fig. 5. Note that all the reported high pressure adsorption capacities are absolute loadings, calculated from the experimentally obtained excess loadings and V total by using Peng-Robinson equation of state as suggested by Myers and Wang. 60,61 For all the samples, CO 2 uptake increased with the increasing of pressure, and the highest uptakes of 8.64 and 7.93 mmol g À1 were achieved over sample oxAC-800NH3 at 30 and 45 C, respectively. These high uptakes together with the unique spherical form and good mechanical strength make the current carbon beads a promising option for CO 2 capture applications. Fig. 6 illustrates the adsorption capacities at a pressure of 20.2 bar as a function of the BET surface areas of the samples, and the obtained good linear relationship highlighted the vital importance of the above mentioned pore lling mechanism where textural properties are expected to be the key driver of CO 2 uptake at elevated pressures. Table 5 presented a comparison of the CO 2 adsorption capacity of the carbon beads with those phenolic resin-derived carbons reported in the literature. It can be seen that the adsorption capacity of samples reported here is only slightly lower than those non-KOH activated carbon adsorbents in a powder form. Together with the uniform spherical form, we thus believe these carbon beads may represent a potential option for practical CO 2 capture. We are currently working on the effect of KOH activation on the adsorption performance as well as the morphological integrity of the carbon beads.
To examine the effectiveness of the carbon beads as an adsorbent for pre-combustion CO 2 capture by pressure swing adsorption, the working capacity of the carbon adsorbents, which is dened as the difference in equilibrium capacities at given adsorption and desorption pressures (in the present case, 20 and 2 bar for adsorption and desorption, respectively), were obtained as shown in Table 6. Of all the samples examined, the sample oxAC-800NH3 showed the highest working capacity of 4.94 mmol g À1 , which is nearly 20% higher than that of the parent AC. It is also interesting to highlight that the working capacities of all the carbon beads at 45 C are slightly higher than those achieved at 30 C, this is a phenomenon that has rarely been reported in previous studies on activated carbons, but is fairly meaningful because carbon-based materials are normally regarded as a highly temperature-sensitive adsorbents that rapidly lost their capacities with increasing temperatures, and thus the current observations are persuasive to demonstrate the feasibility of using the current carbon-based materials for pre-combustion CO 2 capture by pressure swing adsorption.
3.2.2 Adsorption selectivity. To assess the selectivity of the carbon adsorbents for CO 2 over N 2 and H 2 , the N 2 and H 2 adsorption isotherms were measured and compared with those of CO 2 for some of the modied samples, namely oxAC and oxAC-300NH3. It can be seen from Fig. 7 that within the adsorption pressure range examined, the adsorption capacities of the samples for CO 2 are remarkably higher than those for both N 2 and, in particular, H 2 . Here, the ratio of adsorption capacity of different gases under 5.0 bar was used as an indicator for the adsorption selectivity, the CO 2 /N 2 selectivities thus obtained were 4.61 and 4.81 for oxAC and oxAC-300NH3,   respectively. There values are remarkably higher for CO 2 /H 2 (30.50 and 38.36). Although investigations are ongoing, it is expected that the selectivity for CO 2 can be even higher in a competitive adsorption process, given the higher surface affinity of the adsorbents to CO 2 than to N 2 and H 2 .

Adsorption kinetics.
Using the CO 2 adsorption data obtained from the high pressure volumetric adsorption analyser, the adsorption kinetics could be estimated by plotting the ratio of the CO 2 uptake at time t to the equilibrium uptake, (uptake) t /(uptake) eq , as a function of contact time t. The results are shown in Fig. 8. It is encouraging that regardless of the operation temperature, 90% of the equilibrium capacity could be achieved within less than 1 min and only approximately 2 min is needed to reach the full equilibrium capacity for all the samples, highlighting the desirable adsorption kinetics of the carbon beads for CO 2 capture.
3.2.4 Cyclic adsorption performance. The reusability of oxAC-300NH3 was evaluated by TGA as presented in Fig. 8. Within 20 cycles of adsorption and desorption, there are negligible changes on the CO 2 uptake (Fig. 9a), an inspection of the weight change curve (Fig. 9b) indicates that the fast adsorption/desorption kinetics is stable as well, which is Fig. 9 Reusability of oxAC-300NH3 (a) CO 2 uptake, (b) weight change curve.  important to enhance the cyclic working capacity of the samples at reduced circulating times. The superior cyclic adsorption performance together with other adsorption characteristics as mentioned above endows the novel spherical carbon adsorbents with great promise in practical pre-combustion capture applications using pressure swing adsorption.

Conclusion
The effectiveness of using surface amination treatments as a means to improve the CO 2 capture performance of carbon beads produced from a novel cost-effective hydrothermal process has been investigated. It was found that the surface modications hardly compromised the desirable spherical morphology and mechanical strength of the activated carbon beads while their adsorption performance for CO 2 was enhanced. In specic, HNO 3 oxidation and low temperature amination with NH 3 appear to be signicantly more effective in improving the low pressure CO 2 adsorption performance of the carbon beads, while the modication with NH 3 at high temperatures was found instead to be more effective to boost their capture capacity at high pressures. Comprehensive characterisations demonstrate that the CO 2 uptake capacity of the carbons at low CO 2 partial pressures was more related to their surface chemistry and narrow micropores, while the high pressure adsorption capacity was found to have a linear relationship with the BET surface areas, indicating a pore-lling dominated adsorption mechanism. The highest working capacity of ca. 4.9 mmol g À1 was achieved for the sample oxAC-800NH3 and little variation was observed with the adsorption temperatures examined in this investigation. The results demonstrate the great potential of the spherical carbon beads as CO 2 adsorbents particularly suitable for precombustion capture by pressure swing adsorption systems.