Insights into the Effect of Adsorption-Desorption Cycles on SO 2 Removal over an Activated Carbon

The use of an activated carbon (AC) with high ash content (15.92%) for SO2 removal was investigated during adsorption-desorption cycles. Significant deterioration in both dynamic and equilibria adsorption processes during the cycles was observed. To investigate the causes of deactivation, SO2 temperature-programmed desorption (SO2-TPD) experiments were conducted. The results indicated that most of the stored sulfur-containing species were released in the form of SO2 when the temperature was below 400°C. In addition to SO2, traces of CO were detected, but higher temperatures were required for the abundant release of CO. A Fourier-transform infrared spectrometry (FTIR) experiment was used to investigate changes in the oxygen-containing groups, and the results confirmed the formation of stable C-O complexes. These formations were tentatively attributed to the CO precursor’s occupation of active sites. Based on the formation of C-O complexes, two deactivation pathways in the cycles were proposed. The adsorption-desorption cycles also affected the AC ash. The formation of sulfur-containing species in the ash was confirmed through thermodynamic calculation and powder X-ray diffraction.


INTRODUCTION
SO 2 resulting from fossil fuel combustion is a major source of air pollution, which contributes to the formation of acid rain.Several methods have been used commercially for the reduction of SO 2 emissions, such as wet flue gas desulfurization (WFGD), semidry flue gas desulfurization (semidry FGD), and dry flue gas desulfurization (dry FGD; Wang et al., 2017;Ni et al., 2018).The activated carbon (AC)-based desulfurization process is categorized as dry FGD, which has attracted research attention because of its potential for use in converting SO 2 into by-products of higher value, such as H 2 SO 4 or S.Moreover, this process requires relatively little or even no water, in contrast to conventional WFGD or semidry FGD; thus, it is a suitable process for water-deficient areas (Gao et al., 2011a;Guo et al., 2015;Li et al., 2016).
The following reactions govern the desulfurization process (Rosas et al., 2017): These reactions may occur at different locations and be influenced by various properties of AC.It is generally accepted that oxidation of SO 2 is limited to the vicinity of active sites, and hydration of SO 3 usually occurs within the pores of AC, which offer a relatively large amount of space for reaction (Mochida et al., 1997;Ma et al., 2003;Gaur et al., 2006;Sun et al., 2015).Further research should be conducted to identify the active sites in SO 2 oxidation.Although Zhang et al. posited that C-O-C in ethers and C=O in quinones are the main active sites (Zhang et al., 2017), most researchers have inferred that bare carbon atoms located in edge positions are responsible for SO 2 adsorption (Lizzio and DeBarr, 1996;Mochida et al., 1997;Mochida et al., 1999;Yang and Yang, 2003;Liu et al., 2016;Sun et al., 2016).Regarding hydration of SO 3 , the pore structure of AC may play a critical role, details about which have been reported based on a rigorous, long-term study.Initially, Lizzio and DeBarr (1996) demonstrated that SO 2 adsorption does not exhibit a relationship with a specific area of AC, whereas Rubio and Izquierdo (1997) and Rubio et al. (1998) suggested that optimizing the surface area could promote SO 2 adsorption.Until very recently, hierarchical structures, such as micro-, meso-and macropores, could be classified (Gao et al., 2011b;Yu et al., 2018).Based on these preliminary findings, Sun et al. (2015) suggested that meso-and macropores serve in H 2 SO 4 transportation, facilitating the overall process.Researchers have also discovered a linear relationship between SO 2 uptake and volumes of ultramicropores on special CO 2activated carbons (Zhu et al., 2012).
Besides these findings, more complex topics are the roles of AC chemical properties in SO 2 removal.The basicity of a surface may be assumed to promote adsorption of acidic SO 2 (Atanes et al., 2012;Zhu et al., 2012;Ren et al., 2017).However, a promotion effect from acidic oxygen-containing groups such as carboxyl groups has also been reported (Liu et al., 2016;Cui et al., 2017).
Desulfurization is a continuous process, and ACs are degraded in adsorption-desorption cycles, which cause changes in their physical and chemical properties; these changes, in turn, affect SO 2 removal.Additionally, the cost of an AC is a critical concern for process commercialization (Mochida et al., 2000).In previous studies, cheap precursors, such as low-grade coals, have been used for AC preparation, resulting in considerable amounts of ash in the ACs (Lizzio and DeBarr, 1996;DeBarr et al., 1997;Rubio andIzquierdo, 1997, 1998;Rubio et al., 1998;Izquierdo et al., 2003;Liu et al., 2003a;Liu et al., 2003b).However, to the best of our knowledge, few studies have discussed the conversion of ash in the AC desulfurization process.
This investigation focused on the physical and chemical changes of AC during adsorption-desorption cycles.Specifically, variations of texture, oxygen-containing groups, and ash were evaluated.The findings revealed the effects of adsorption-desorption cycles on SO 2 removal.

Preparation of the Activated Carbon
Commercial AC obtained from Xinhua Chemical Company (Taiyuan, China) was used as the precursor in this study.The results of elemental and proximate analyses of the as-received carbon are presented in Table 1.To evaluate the effect of adsorption-desorption cycles, the asreceived carbon was treated using various cycle times.The nomenclature of each sample included the precursor (A), treatment method (R), and times.The following sample was also prepared for comparison: A-H10 (treated under nitrogen at 400°C for 10 h).

Activity Tests of the Activated Carbon
All of the tests were performed in an experimental installation constructed for this study (Fig. 1).The activities of SO 2 removal were investigated using various procedures, described as follows: (i) In dynamic tests, the feed gas was passed through the fixed bed at a constant temperature.Concentrations of SO 2 and O 2 in the inlet and the outlet of the reactor were measured online using a flue gas analyzer (Testo 350, Germany).Measurements were stopped at a breakthrough concentration of 4000 ppm.(ii) In equilibria tests, a flow was admitted into the reactor for 1 h, and the samples were then heated at 10°C min -1 under a flow of N 2 up to 400°C and maintained for 1 h.Adsorbed and desorbed SO 2 were detected using a slightly modified version of the procedure proposed by Davini (2001).After passing through the reactor, the adsorbed or desorbed gas flowed through a scrubbing bottle that contained 250 mL of hydrogen peroxide (5 vol.%).The effluent gases were assumed to contain SO x (SO 2 and SO 3 ), CO, CO 2 , N 2 , and O 2 , among which only SO x and CO 2 could be trapped by the H 2 O 2 solution and converted to H 2 SO 4 and H 2 CO 3 , respectively.The reacted solution was separated into two parts: The first part was titrated using a 0.01 M NaOH solution, and the second part was diluted and then evaluated using ion chromatography (792 Basic IC, Metrohm, Switzerland).These two evaluation methods guaranteed the validity of the results.The experimental error is discussed in a later section.The amount of adsorbed SO 2 was determined based on the difference between the results without and with carbon.For all of the adsorption processes, the gaseous mixture was 1 vol.%SO 2 , 7 vol.%O 2 , and 12 vol.%H 2 O in N 2 and at 120°C before the AC bed was introduced.Water vapor was fed by means of a saturator through which part of the mixture bubbled before entering the reactor.The amount of vapor fed was controlled by the temperature of the bath in which the saturator is immersed, assuming vapor-liquid equilibrium followed Antoine's law.The loaded ACs were cylindrical particles with a 5 mm diameter.The total flow rate of the mixture was maintained at 1.1 L min -1 using flow meters, resulting in a space velocity of 800 h -1 .Consecutive adsorption-desorption cycles were performed, and the characteristics of the samples were assessed at the end of the cyclic processes.

Characterization of the Activated Carbon
N 2 adsorption was performed at 77 K using an Autosorb-1-C installment (Quantachrome Instruments).The specific surface areas were calculated employing the Brunauer-Emmeet-Teller (BET) equation, while the total pore volumes and micropore volumes were decided by single point adsorption quantity at P/P 0 = 0.95 and t-plot method, respectively.Prior to the N 2 sorption, the sample was preheated at 250°C for about 3 h to remove physically sorbed material from narrow pores.Temperature-programmed desorption (TPD) experiments were carried out in the same apparatus as the activity test.In a typical run, a sample was heated under a flow of N 2 (1.1 L min -1 ) from room temperature to 750°C at a rate of 5°C min -1 to achieve nearly complete desorption of SO 2 .During the desorption, traces of SO 2 and CO were monitored.
Fourier-transform infrared spectrometry (FTIR) tests were carried out on a Nicolet 380 Spectrometer (Thermo Electron) at ambient temperature.The samples were mixed with KBr at a mass ratio of 1:100, grounded and then pressed into a thin slice.Thirty-two scans were made to yield a spectrum with a resolution of 4 cm -1 spanning 400-4000 cm -1 .
Powder X-ray diffraction (XRD) patterns were collected with an X'Pert PRO diffractometer using a Cu Kα radiation source.X-ray photoelectron spectroscopy (XPS) spectra were obtained with a Thermo ESCALAB 250Xi using Al Kα X-ray source (hν = 1486.6eV).The C 1s spectrum peaked at 284.8 eV was taken as the reference.The pH pzc values of carbon were determined by contacting 0.15 g of the sample with 2 mL distilled water for 3 days.It has been confirmed that in this case the pH value of solution was equal to the pH pzc value of carbon (Lizzio and DeBarr, 1996;Martin et al., 2002).
The ash composition of the as-received carbon was detected using an X-ray fluorescence (XRF) method in a Rigaku ZSX100e.

Thermodynamic Calculations
For the analysis, mineral impurity in AC was considered.Gaseous SO 2 , H 2 O and O 2 as well as the solid mineral and carbon were regarded as reactants in the process of adsorption at 120°C, while H 2 SO 4 instead of gaseous components was involved in the process of desorption at 400°C.The software package FactSage version 5.2 was used for the thermodynamic calculations.

Dynamic Tests
Changes in the SO 2 adsorption profiles for various samples are denoted in Fig. 2. For sample A-R2, little change was observed.However, with the increase in the number of cycles, the slopes of the profiles became steeper.Thus, for sample A-R10, the outlet SO 2 concentration reached approximately 38% of the amount at the inlet within 12 min, which can be compared with the 30 min required for the original AC (A-R0).

Equilibria Tests
Fig. 3 indicates the amounts of sulfur dioxide that were adsorbed and desorbed over several consecutive cycles.As mentioned, ion chromatography and acid-base titration were used for mutual verification.During the adsorption process, two broad peaks were evident.The uptake amount increased up to two cycles and decreased to a lower value after four cycles.Subsequently, after a slight ascent, the amount gradually decreased after six cycles.The amounts detected using the two methods were in good agreement, implying that the results were accurate.A clear reduction in SO 2 adsorption was observed.After 10 cycles, 15% loss was confirmed according to the ion chromatography method, and acid-base titration method indicated 10% loss.Regarding desorption, similar features were observed, and  two broad peaks were clear.However, compared with the descending trend in the adsorption results, the desorption data notably fluctuated at approximately 21 mg SO 2 g -1 AC.Additionally, a small difference between the amounts detected using the two methods was observed.This difference was attributed to the measurement principle.For the ion chromatograph, only SO 4 2-in solution was included in calculations of SO 2 amount, whereas in cases of acidbase titration, H + derived from both H 2 SO 4 and H 2 CO 3 in solution was used in the calculations.Consequently, during the desorption process, release of CO 2 from the AC may have affected calculation of SO 2 desorption amount in cases of acid-base titration.However, although differences between the results from the two methods were observed in each adsorption-desorption cycle, the ratios of average desorption amount to average adsorption amount were the same, at 99.4%.Thus, the diminishment of desulfurization may have been partially due to the residual SO 2 from previous cycles.

SO 2 -TPD
SO 2 -TPD experiments were conducted to further investigate the causes of deactivation.Fig. 4 illustrates the TPD profiles of SO 2 based on desorption of saturated Sample A-R0.
All of the SO 2 -TPD profiles exhibited desorption peaks at approximately 330°C.The TPD profile of SO 2 adsorbed with 12% H 2 O exhibited the least amount of desorption; the amount was even less than the amount desorbed under the condition of SO 2 adsorbed alone.This phenomenon clearly indicated that competitive adsorption between SO 2 and H 2 O occurred on the sample surface, as suggested by Liu et al. (2003b) and the other at 600°C, corresponding with the main peak and the shoulder peak under the condition of SO 2 adsorbed with O 2 and H 2 O. Therefore, two types of sulfur-containing species were inferred to have existed on the carbon surface.One part may have been adsorbed on the active sites of the carbon with strength equal to the adsorbed SO 2 .The others may have been stored in the pores of the carbon due to the elution effect of the water.The results may also be partly ascribed to sulfates, which are described in a later section.Fig. 5 illustrates the TPD profiles of CO that accompanied SO 2 desorption.A blank sample without any adsorption was used as a reference.CO desorption increased rapidly at higher temperatures (> 400°C).At lower temperatures (250-400°C), minor peaks for SO 2 adsorbed with O 2 and H 2 O as well as Sample A-R0 impregnated with H 2 SO 4 were observed.This temperature window corresponded with that of the main SO 2 desorption peak.These findings may be interpreted in terms of the temperature programmed superficial reaction described by Perrard and Joly (1989).According to the proposed mechanism, the large amounts of CO observed after SO 2 adsorption demonstrated that carbon was oxidized during TPD (Martin et al., 2002).The oxidizing agent was SO 3 derived from the decomposition of H 2 SO 4 .The reducing agent was carbon, and as a consequence of oxidation reactions, oxygen-containing groups formed, and SO 2 was released (Muniz et al., 1998).The oxygen-containing groups could then be thermally decomposed into CO or CO 2 at different temperatures depending on the group type.A minor evolution of CO occurred from 250 to 400°C as a result of unstable groups.A substantial increase in the amount of evolved CO at higher temperatures was caused by the decomposition of stable groups.Because the regeneration temperature was 400°C (Figs. 2 and 3), the CO-forming groups likely formed through a superficial reaction and remained on the AC surface until they reached their natural decomposition temperatures.Compared with the sample impregnated with H 2 SO 4 , the sample pretreated with SO 2 + O 2 + H 2 O released more CO between 500 and 650°C.This phenomenon may have resulted from the increase in unstable CO-forming groups produced in the reaction between carbon and H 2 SO 4 .These findings were consistent with the previous SO 2 -TPD results (Fig. 4(a)), which revealed two peaks over both the SO 2 + O 2 + H 2 O pretreated and H 2 SO 4 impregnated samples.After deconvolution using the Gaussian function, it is clear to see that the second peak of SO 2 shifted to a lower temperature over the SO 2 + O 2 + H 2 O pretreated sample (Fig. 4(b)), and this result agreed well with the increase in CO release between 500 and 650°C.

Characterization Results
The textural characteristics of the samples were investigated, and the results are presented in Table 2. Compared with those of the original AC, the BET surface area (S N2 ), total pore volume (V p ) and micropore volume (V micro ) increased after the cycles, whereas average pore diameter (D avg ) decreased.Note that after 10 cycles, the ratio of V micro /V p dropped from 0.47 to 0.36, indicating a relative reduction of micropore volumes, which play an important role in SO 2 adsorption.The changes in textural properties were mostly caused by the increasing degree of burn-off.However, the textural changes were difficult to correlate with the results of the dynamic or equilibria tests because of the generally accepted proposal that welldeveloped pore structures prompt adsorption.
A more precise understanding of SO 2 adsorption behavior may require more detailed information about the oxygen-containing groups formed after adsorption-desorption cycles.A Fourier-transform infrared spectroscopy (FTIR) experiment was used to investigate changes of the oxygencontaining groups.In general, CO 2 is derived from acidic surface oxides, such as carboxyl and lactone groups, and  CO is produced by ketone, aldehyde, and phenol, which are basic surface oxides (Muniz et al., 1998;Mochida et al., 2000).In the recorded spectra (Fig. 6), the absorption band in the 3000-2700 cm -1 range could be assigned to the stretching vibrations of C-H in aldehyde.Bands at 2854 cm -1 and 2926 cm -1 were detected in the cases of A-R1 and A-R2, indicating that some aldehyde groups had appeared on the surface of the carbon.The presence of bands between 1200 cm -1 and 1000 cm -1 may have been caused by vibration of C-O in phenol (Shin et al., 1997).The peak at 1084 cm -1 considerably increased after the cycles, implying an increase in phenol groups.
Based on these findings, the surface basicity of the sample should have been higher after the cycles.This deduction was confirmed through investigation of the pH pzc values of the carbons, as presented in Table 2.The point of zero charge was defined as the pH where the net surface charge resulting from the adsorption of the potentialdetermining ions, H + and OH -, was zero.In aqueous solution, Brønsted acidic groups of the carbon surface tended to donate their protons to water molecules, whereas the Lewis bases adsorbed protons from the solution.Therefore, the reported pH pzc values represent the average chemistry of the carbon surface.All of the samples exhibited basic properties, which were enhanced after the cycles.In general, the larger adsorption capacity was likely related to the higher basicity of carbon, because the basic sites could adsorb SO 2 (Davini, 2002;Martin et al., 2002).However, the amount of SO 2 adsorbed decreased significantly along with the increase of sample basicity.
XPS was further used to investigate the chemical structure change over carbon surface.C 1s , O 1s and S 2p spectra were recorded for each sample, based on which the atomic ratios such as O/C and S/C were calculated and illustrated in Fig. 7.With the increase of adsorption-desorption cycles, the ascending trend for both O/C and S/C ratios was clear.Compared with original AC (A-R0), A-R10 showed a 65% increment related to S/C ratio, consistent with results of the equilibria tests that showed a sulfur loss in gas phase.Also, the higher O/C ratio after adsorption-desorption cycles gave direct evidences that more oxygen-containing species were formed.
Both the N 2 adsorption results and aforementioned FTIR results implied that pore structure and basic functional groups exhibited little effect on SO 2 adsorption and that other factors may have played more critical roles.We also observed significant increases of oxygen-and sulfur-containing species over the carbon surface after the adsorption-desorption process.Previous research has explicitly shown the negative effect of oxygen-containing species.Mochida et al. (1997) clarified that the surface of activated carbon fiber was most active when almost no oxygen-containing groups were present, and oxygen-containing groups were not considered to be active sites.In a subsequent paper, they attributed an increase in SO 2 adsorption capacity to an increase in the number of reactive carbon atoms, which had been generated by the evolved CO (Mochida et al., 1999).Lizzio and DeBarr (1996) also observed that the formation of stable C-O complexes over carbon surface occupied adsorption sites.Yang and Yang (2003) performed ab initio molecular orbital calculations to identify possible pathways of carbon-catalyzed oxidation of SO 2 by O 2 /H 2 O. Based on their theory and our findings, two deactivation pathways during the adsorptiondesorption cycles were proposed, as denoted in Fig. 8.The active sites responsible for SO 2 adsorption were the zigzag edge sites with free sp 2 electrons.During the adsorptiondesorption cycles, O 2 from the gas phase was involved in the formation of H 2 SO 4 or SO 3 , leaving one oxygen atom on the edge site to form a stable CO-forming group.This group was not removed under thermal desorption at 400°C, as indicated in Fig. 5. FTIR results also verified that the number of CO-forming groups increased after the cycles.Thus, the stable CO-forming groups may serve only to occupy the active sites, thereby inhibiting SO 2 adsorption and resulting in a significant reduction in the amount of SO 2 uptake at equilibrium.

Conversion of Ash
The original carbon contained a considerable amount of ash (15.92 wt.%), mainly composed of Si, Al, Fe, and Ca, and this ash may affect SO 2 adsorption.Table 2 presents the results of an ash composition analysis of the AC.Thermodynamic calculations at 120 and 400°C were conducted to evaluate ash conversion.
At 120°C, SO 2 reacted with Fe 2 O 3 and carbon: At the regeneration temperature of 400°C, H 2 SO 4 , Fe 2 O 3 , and carbon were identified as major reactants, and the complexity of the products correlated with the initial amount of H 2 SO 4 (Fig. 9).When the amount of H 2 SO 4 was below 10 mg g -1 AC, the H 2 SO 4 reacted with Fe 2 O 3 and carbon: As H 2 SO 4 increased, carbon reacted with excess H 2 SO 4 : Due to the generation of H 2 S, the inert atmosphere was converted into a reduction atmosphere.With the continuous increase of H 2 SO 4 , some of the FeS was converted into FeS 2 : The reaction continued to completion until all of the Fe appeared in the form of FeS 2 .These assumed reactions indicated that the SO 2 may fix on carbon as FeS or FeS 2 after the adsorption-desorption cycles.FeS and FeS 2 molecules accounted for part of the residual sulfur-containing species.
These assumed reactions were partly confirmed through X-ray diffraction (XRD) experiments (Fig. 10).All of the samples exhibited two broad peaks at approximately 2θ = 25° and 43°, corresponding with an amorphous carbon structure.In addition to these bands, XRD diffraction peaks corresponding to SiO 2 (at 2θ = 20.9°,26.6°, 36.7°,39.5°, and 50.1°) also appeared.A peak at 2θ = 31.4°was ascribed to Na 4 SiO 4 , and peaks at 2θ = 23.2° and 35.4° were caused by the presence of CaSO 4 crystals.Peaks at 2θ = 29.8°,33.3°, and 38.4° were associated with FeS, FeS 2 , and FeCO 3 , respectively.These results agreed with the X-ray fluorescence (XRF) findings (Table 3), which indicated that these crystals were the major mineral components in the samples.For Sample A-R10, a strong diffraction peak was observed at 2θ = 25.4°.This peak was attributed to Fe 2 (SO 4 ) 3 , which was most likely the intermediate product of the reaction among H 2 SO 4 , Fe 2 O 3 , and carbon.However, amorphous Fe 2 (SO 4 ) 3 may aggregate to form Fe 2 (SO 4 ) 3 crystals upon heat treatment, and it can be detected using XRD.To rule out this possibility, Sample A-H10 was prepared and investigated.The XRD pattern of A-H10 was the same as that of Sample A-R0.This finding further confirmed that the Fe 2 (SO 4 ) 3 crystals were a result of the reaction among H 2 SO 4 , Fe 2 O 3 , and carbon.

CONCLUSIONS
In the present study, the capacity of AC to remove SO 2 was assessed from a practical point of view.A significant reduction in the amount of SO 2 adsorbed after several adsorption-desorption cycles was observed.To determine the causes of deactivation, SO 2 /CO-TPD experiments were performed.Although most of the SO 2 was released at 400°C, higher temperatures were required for the release  of CO, and applying temperature-programmed superficial reaction theory to these results indicated that the carbon surface was oxidized after the cycles.Additionally, an FTIR experiment was conducted to investigate changes in the oxygen-containing groups, and the results confirmed the formation of stable C-O complexes, which may have occupied active sites.Based on the formation of these complexes, two deactivation pathways were proposed as being involved in the cycles.In addition to the change in oxygen-containing groups, the cycles resulted in carbon burn-off as well as subsequent changes in the physical properties of the carbon.An increase in BET surface area and a decrease in average pore size were observed.At last, the AC ash has been also changed in the cycles, during which the formation of sulfur-containing species was confirmed via thermodynamic calculation and powder XRD.

Fig. 5 .
Fig. 5. TPD profiles of CO from AC after adsorption tests.

Table 1 .
Elemental and proximate analyses of the as-received carbon.
a By difference.
Mochida et al. (1999) increased with the introduction of O 2 .At 330°C, two nearly coincident peaks TPD profiles of SO 2 from AC after adsorption tests.(a)Effect of pretreatment and (b) deconvolution results.wereobservedforprofiles of SO 2 adsorbed with O 2 and SO 2 adsorbed with O 2 and H 2 O.However, the TPD profile of SO 2 adsorbed with O 2 and H 2 O exhibited a shoulder peak at the higher temperature.Mochida et al. (1999)suggested that H 2 O increased SO 2 adsorption through oxidation and hydration of adsorbed SO 2 , resulting in the formation of adsorbed H 2 SO 4 .The profile of sample A-R0 impregnated with H 2 SO 4 exhibited two desorption peaks, one at 330°C

Table 2 .
Textural and acid-base characterization of the samples.