Controllable growth of Na 2 CO 3 ﬁ bers for mesoporous activated alumina ball modi ﬁ cation towards the high-e ﬃ ciency adsorption of HCl gas at low temperature

To fundamentally solve the problem of chlorine corrosion in blast furnace gas (BFG) systems, a supported adsorbent is used for HCl removal at low temperature. In this paper, three alkaline substances, Ca(OH) 2 , NaOH, and Na 2 CO 3 , were separately coated on the surface of activated alumina balls (AABs) by wetness impregnation method, using the as-active components. These materials were measured by XRF, XRD, SEM, TEM, and N 2 adsorption to characterize textural properties. The modi ﬁ cation experiments indicated that Na 2 CO 3 ﬁ bers can be obtained on the surface of AABs by adjusting the loading amounts, impregnation time, and drying conditions. The ﬁ bers' structure contributes to the gas phase di ﬀ usion in the product layer, greatly improving the conversion of Na 2 CO 3 (>0.98). The highest value of HCl adsorption capacity reaches 3.56 mmol g (cid:1) 1 when the Na 2 CO 3 loading amount is 20 wt%, ﬁ ve times the adsorption capacity of pure AABs. The kinetics of HCl removal by the Na 2 CO 3 ﬁ ber-modi ﬁ ed AABs is controlled by the interfacial chemical reaction. amount of Na 2 CO 3 increased synchronously


Introduction
It has been widely accepted that blast furnace gas (BFG) is a reusable derivative energy source due to the large heat and high pressure in the process of blast furnace production. 1 In the process of recycling the blast furnace gas, the dry-dedusting process is adopted to make the gas purication efficient. However, serious corrosion has appeared in the BFG pipeline in the short term-a large amount of salt blocks the Blast Furnace Top Gas Recovery Turbine Unit (TRT), and even the refractory bricks in the hot blast stove have been partially broken. 2 According to the reports, hydrogen chloride (HCl) in the gas stream is the only explanation for equipment corrosion. 2,3 Normally, the temperature of BFG is in the range of 380-423 K. BFG steam consists of 23% CO, 20% CO 2 , 55% N 2 , and a small amount of methane; simultaneously, there is about 300 ppm hydrogen chloride (HCl). Although the content of HCl in the BFG is very small, the blast furnace still generates a heavy chlorine burden attributed to the large amount of blast furnace gas production. For example, a 2650 m 3 BF produces 8 million m 3 of BFG per day, containing about 3 tons of HCl, which is a huge threat to the gas pipeline and hot blast stove. Therefore, nding an effective adsorption material to remove HCl from the BFG is urgently necessary.
At present, more studies have paid attention to the removal of acid gases from the hot ue gas (673-1073 K) of the coal industry. [4][5][6][7][8][9][10][11] The reaction mechanism mainly depends on the neutralization of acid and alkali to capture acid gas in the process of gas stream cleaning. Duo et al. 5,12 reported that the higher HCl capacity of sorbents correspond to the higher temperature from 573 to 873 K. Na sorbents showed optimal HCl sorption performance in the temperature range of 673-773 K. Dou et al. 13,14 found that Na 2 CO 3 , Ca(OH) 2 , or CaO were suitable to achieving the tolerance limit of 1 ppmv HCl from hot gas. The experimental results showed that the HCl capacity of alkali materials at 473 K is less than 20% (3.5 mmol HCl g À1 ) of that at 773 K, suggesting that the chemical reaction of adsorbents is limited at low temperature. 15 Besides, with poor structure, the conversion of the active materials is less than 50%, 16 which undoubtedly increases the cost of dechlorination.
Mesoporous materials such as activated carbon (AC), activated alumina balls (AABs), molecular sieves, etc. have been developed as potential adsorbents or catalyst supports for the removal of acidic gases due to their large surface area and abundant porous structure. [17][18][19][20][21] Taguchi et al. [22][23][24] pointed out that the reaction mechanism of gas on the surface of these porous materials relies on physical adsorption, which contributes to regeneration and recycling. ZnO supported by 3-D structural mesoporous molecular sieves had been synthesized for H 2 S removal from gas stream, reported by Li,25 suggesting that the mesoporous structure beneted the dispersion of active phase. Among them, activated alumina balls (AABs) are well known as a potential carrier material for the adsorption of acid gases at low temperature due to their receptivity to modication with alkaline substances. [26][27][28][29] For gases at room temperature, alkaline substances are usually used as active materials, supported by impregnation method, to remove acid gas for feeding molten carbonate fuel cells (MCFCs). [30][31][32][33] Lee 34 showed that the HCl capacity of NaOH/AC is about 6 times that of commercial alumina. However, Micoli et al. 35,36 investigated the HCl adsorption performance of activated carbon impregnated with KOH, NaOH or Na 2 CO 3 , respectively. They suggested that the best adsorption performance for HCl adsorption was obtained by impregnating Na 2 CO 3 . Due to temperature limits, the chlorine capacity of the sorbent was lower than 0.4 mmol HCl g À1 . Zhao et al. [37][38][39] suggested that K 2 CO 3 / Al 2 O 3 and Na 2 CO 3 /Al 2 O 3 had the potential to be employed as excellent sorbents for CO 2 uptake due to their rapid reaction rate and high CO 2 capture capacities. Zhao 40 investigated the effect of Na 2 CO 3 loading on the adsorption capacity of the acid gases, which reached a maximum at the dispersion threshold. The paper of Dong 41 showed that there was a change of surface structure aer the modication of g-Al 2 O 3 with Na 2 CO 3 , unfortunately without any further analysis. It is worth mentioning that the change of HCl concentration in mixed gas has little effect on the adsorption capacity in the adsorption test. 5,34 Therefore, many experiments to simulate the removal of HCl from gas tend to adopt an accelerated aging test with high HCl concentration.
In general, there is a lack of literature on HCl removal from BFG in the steel industry for its extreme conditions, namely, selective removal of HCl at the lower temperature of 380-423 K. For the purpose of environmental protection and energy recovery, the importance of HCl removal from blast furnace gas is self-evident. In this work, three modiers, Na 2 CO 3 , Ca(OH) 2 and NaOH, were coated on AABs by impregnation method. The structure properties of the initial, impregnated and exhausted materials have been characterized. The adsorption activity of the prepared materials had been measured using the accelerated aging test with a HCl concentration of 50% at the low temperature of 423 K; accordingly, the reaction mechanism of HCl and sorbents are also proposed.

Materials
Commercial activated alumina balls (AABs) were purchased from Shandong province in China without further purication. Calcium hydroxide (Ca(OH) 2 ), sodium carbonate anhydrous (Na 2 CO 3 ), and sodium hydroxide (NaOH) were analytical grade and used as received without further purication. The basic physical properties of AABs are shown in Table 1.

Preparation of modied sorbents
The modied sorbents in this study were prepared by the impregnation method. The preparation process consisted of two stages, namely, the selection of modiers and optimization of impregnating conditions.
2.2.1 Alkaline modier selection. The impregnation solutions were obtained by dissolving 0.34 g Ca(OH) 2 powder, 20 g NaOH powder and 20 g Na 2 CO 3 powder in 200 mL of deionized water, respectively, to form three kinds of impregnation solutions. Then, 40 g AABs were severally added into the impregnation solutions, which were then subjected to ultrasonic shocking at room temperature for 1 h, immersion for 17 h and nally drying at 473 K for 24 h. The obtained samples were named Ca(OH) 2 -AAB, 10NaOH-AAB, and 10Na 2 CO 3 -AAB, respectively; 10 represents the designed loading amounts of NaOH and Na 2 CO 3 , whereas the loading amount of Ca(OH) 2 is the maximum value for its lower solubility. The pure AABs without alkaline modication were used for comparison. The actual loading amounts of NaOH and Na 2 CO 3 on the sorbents were determined by X-ray uorescence (XRF).
2.2.2 Impregnation parameters selection. In order to optimize the microstructure of sorbents, factors such as loading amounts of Na 2 CO 3 , impregnation time and drying time have been investigated; the details of inuencing factors are shown in Table 2. To prepare the impregnation solution, 10 g, 30 g, 40 g, 50 g and 60 g pure Na 2 CO 3 powder were separately placed in 200 mL of deionized water resulting in, respectively, 0.47 mol L À1 , 1.42 mol L À1 , 1.88 mol L À1 , 2.36 mol L À1 and 2.83 mol L À1 Na 2 CO 3 solutions, and ve sorbents of different Na 2 CO 3 loading amounts were obtained aer 40 g AABs had been impregnated in the corresponding Na 2 CO 3 solution. The designations of 5, 15, 20, 25 and 30 correspond to the percentage of Na 2 CO 3 loading amounts. The effects of the impregnation time (2 h, 10 h, 17 h and 24 h) and drying time (16 h, 20 h, 24 h and 28 h) on the microstructure of sorbents were subsequently investigated. The actual loading amounts of Na 2 CO 3 on the sorbents were determined by XRF.

Activity test of HCl removal
Aer the preparation of modied sorbent loaded with Na 2 CO 3 , its HCl removal activity was subsequently tested in a xed bed reactor with the reaction temperature of 423 K under normal atmosphere. The experimental apparatus is shown in Fig. 1. The sorbents were rstly treated at 423 K for 0.5 h in N 2 ow (20 mL min À1 ) before the test. High-purity HCl and N 2 , whose ow rates were similarly controlled by mass ow meters at 20 mL min À1 , were mixed to pass through the sorbents (about 23 g) in the heating furnace. Because the work temperature of BFG is in the range of 380-423 K, the experimental temperature for HCl adsorption was controlled at 423 K to simulate the real temperature of BFG. Aerwards, the escaped HCl was absorbed by 1 L deionized water in the adsorption bottle, and the concentration C t of HCl in the outlet gas at that moment was ascertained synchronously by detecting the change of the pH value in the adsorption solution. 35,36,42 Finally, the gas stream owed through the treatment bottle, ensuring no pollution to the environment before the vent. The test was stopped immediately when the pH value decreased sharply in the adsorption bottle; this moment was named the breakthrough time t b . The exhausted sorbents were taken out of the reactor and desorbed in the deionized water, and the concentration C e of chloride ion in the desorption solution was determined by a chloride ion concentration meter.
The HCl adsorption capacity q e (mmol g À1 ) of the exhausted sorbents can be calculated from the following eqn (1): where m 0 (g) is the initial mass of sorbents, V (L) is the volume of the desorption solution, and C e (mol L À1 ) is the concentration of Cl À in the desorption solution at equilibrium. The adsorption efficiency h of the sorbents at every moment can be calculated using the following eqn (2): where C 0 (constant, 20 mL min À1 ) represents the concentration of HCl in the inlet gas, and C t (mL min À1 ) represents the concentration of HCl in the outlet gas at time t.

Characterizations
The actual loading amounts of Na 2 CO 3 supported on the sorbents were determined by X-ray uorescence spectrometer (XRF-1800, Japan). The eld emission scanning electron microscope (FESEM) (Zeiss supra 55), operated at 10 kV, coupled with energy dispersive X-ray analysis (EDS) was used to investigate the morphologies and element distribution of sorbent nanostructures. The X-ray diffraction (XRD) analysis, carried out on a powder X-ray diffractometer (Rigaku Dmax-2500 diffractometer using Cu Ka radiation), was used to determine the phase composition of the sorbents. Further structural characterization of the Na 2 CO 3 bers coated on the AABs substrate was performed by applying high-resolution transmission electron microscopy (HRTEM, Tecnai F20) operated at 200 kV. Surface area measurements were carried out by N 2 adsorption at 77 K using a Micromeritics ASAP2020 instrument. The pH value of the adsorption solution was monitored by pH electrode 201T-M connected to the pH meter (MP520, Shanghai, San-Xin). The chloride ion concentration in the HCl desorption solution was measured by chloride ionselective electrode CL502 connected concentration meter (MP523, Shanghai, San-Xin).

Results and discussion
3.1 Preparation of the alkaline modied sorbents 3.1.1 Selection of modier. Three alkali-modied sorbents, namely, Ca(OH) 2 -AAB, 10NaOH-AAB and 10Na 2 CO 3 -AAB, were characterized by SEM, XRD and BET. The loading amounts and BET results are summarized in Table 3. It can be observed that the loading amount of Na 2 CO 3 is the same as NaOH. However, the amount of Ca(OH) 2 impregnated was less than 1 wt%. The BET results showed that the specic surface area and pore volume of 10NaOH-AAB were greatly reduced. In contrast, the sample 10Na 2 CO 3 -AAB retained the specic surface area aer loading with Na 2 CO 3 solution.
The XRD patterns of the modied samples are shown in Fig. 2. It depicts that sample Ca(OH) 2 -AAB shows corresponding characteristic peaks of the substrate, whereas sample 10NaOH-AAB shows trace peaks of NaAlO 2 compared with the obvious peaks of Na 2 CO 3 phase in sample 10Na 2 CO 3 -AAB besides the substrate, suggesting that a complex reaction between NaOH and the AAB substrate has occurred.
The SEM images of the untreated and modied AABs with alkaline substances are shown in Fig. 2(b-e). From Fig. 2(b), it can be seen that there were a large number of uffy structures on the surface of the original AABs, which makes its specic surface area large and benets the alkaline adsorption. The modication of Ca(OH) 2 as presented in Fig. 2(c) generates some lamellar structures in the AABs. However, the modication of NaOH deteriorates the surface morphology of alumina balls; namely, the pores in the AABs are almost blocked (Fig. 2(d)). The microscopic morphology of sorbent 10Na 2 CO 3 -AAB is shown in Fig. 2(e); a layer of brous network coats the surface of AABs, indicating that the Na 2 CO 3 bers on the support help to preserve its high surface area. Considering the surface morphology and physical properties of the sorbent aer modication, Na 2 CO 3 has been selected as the optimal modier.
3.1.2 Effect of Na 2 CO 3 loading amount. The relationship curve of the actual loading amounts of Na 2 CO 3 on the sorbents with the designed loading amounts is shown in Fig. 3(a). The actual loading amount of Na 2 CO 3 increased synchronously with the increasing concentration of Na 2 CO 3 solution up to 20 wt%, and then it continued to increase, but slowly.
Moreover, the actual loading capacities are almost accorded to the concentration of Na 2 CO 3 solution when it is lower than 25 wt%, indicating that the loading amount of Na 2 CO 3 in the sorbents reached saturation herein. Therefore, the loading amounts ranging from 5 wt% to 25 wt% is suitable for further investigation.
The XRD patterns of different loading amounts are shown in Fig. 3(b). When the loading amount reaches 10 wt%, the slight characteristic peaks of Na 2 CO 3 can be detected in the XRD patterns. When the loading amount increases to 15 wt% and even higher, the characteristic peaks of Na 2 CO 3 are obvious. There is no new phase formation in the entire process, indicating that no chemical reaction happened between Na 2 CO 3 and AABs. The BET results of the samples with different loading amounts are listed in Table 4. It shows that the specic surface area of the sorbent decreases rst and then increases with the increasing loading amounts, and reaches the highest value of 150 m 2 g À1 when loading amount is 15 wt%. The specic surface area of the sorbent obviously reduces when the loading amount is further increased to 25 wt%, suggesting that there must be a great change in AAB microstructure.
The FESEM images of modied AABs with different loading amounts of Na 2 CO 3 are shown in Fig. 4(a-d). Clearly, the surface morphology of the samples changes greatly with different loading amounts. When the loading amount is 5 wt%, the   surface of the sample shows a small number of bers growing along the edge of AAB pores, meanwhile blocking most of the uffy structures. Further enhancing the loading amount to 15-20 wt%, the bers on the surface of the samples reach their longest; that is, the bers have an average length longer than 20 mm, the average aspect ratio is more than 100, and they uniformly cover the surface of AABs, which greatly improves the active sites of sorbents. When the loading amount increased to 25 wt%, almost all the bers disappeared, and the surface of AABs are covered completely by a thick layer of Na 2 CO 3 . The SEM results can explain the BET variation. In order to analyze and understand the cause of formation and the phase composition of bers, XRD, SEM and TEM of modied AABs with 15 wt% loading were subsequently detected. The EDS result in Fig. 4(e) shows that bers consist of elements Na and Al, suggesting that Na 2 CO 3 may be adsorbed on the Al 2 O 3 substrate. Furthermore, detection of Al illustrates that the surface of the Al 2 O 3 substrate has not been completely enclosed, which benets the effective utilization of the porous structure of the Al 2 O 3 substrate. Fig. 4(g) shows a typical TEM image of the bers. The crystal plane distance measured by the lattice fringe is 0.237 nm or 0.238 nm in Fig. 4(h), which is similar to the (112) lattice spacing of Na 2 CO 3 (0.237 nm, PDF card no. 72-0628). Furthermore, it can be found that the Na 2 CO 3 ber shows a stronger diffraction peak of (112) in Fig. 4(f), suggesting that the growth of bers is oriented along the [112] direction.
The pore volume distribution of the six samples is shown in Fig. 5. It can be seen that the pore volume distribution of modied samples concentrates in the range of 5-10 nm and that the samples exhibit almost uniform mesoporous structure, which agrees well with the result of SEM observation.  Fig. 6. It shows that the surface morphology of the samples transformed obviously with prolonged impregnation time. When the impregnation time is less than 10 h, no Na 2 CO 3 ber appears on the surface of the sample. The Na 2 CO 3 bers achieve maximum and uniform coverage on the surface of AABs when the impregnation time reaches 17 h. When the   impregnation time is more than 24 h, however, the Na 2 CO 3 bers on the substrate surface bond together, suggesting that impregnation for too long a time inhibited the growth of Na 2 CO 3 bers. Hence, the Na 2 CO 3 bers on the surface of the modied AABs only appear with maximum and uniform coverage aer 17 h impregnation.

Effect of drying time.
Aer immersion in 20 wt% Na 2 CO 3 solution for 17 h, the sorbents were dried from 16 h to 28 h in an oven (423 K). Their microstructures are shown in Fig. 7. Similarly, Na 2 CO 3 bers show obvious differences under different drying time. When the drying time is 16 h, the bud of the Na 2 CO 3 bers just grew on the surface of the AABs. With prolonged drying time to 24 h, the Na 2 CO 3 bers are also observed on the surface of AABs. However, when the drying time is further extended, the Na 2 CO 3 bers stopped further growth, suggesting that the production of bers achieved equilibrium due to the complete exhaustion of moisture.
The wetness impregnation method was used to support Na 2 CO 3 on AABs to obtain the modied sorbents. Na 2 CO 3 bers are controlled to grow on the AAB surface by adjusting the loading amount, the impregnation time and the drying time. The optimal conditions for ber growth are a loading amount of 15-20 wt%, impregnation time of 17 h and drying at 473 K for 24 h.

HCl adsorption measurement
Aer the AABs were modied with the Na 2 CO 3 bers, the HCl removal experiments were carried out. In this study, the principle of HCl removal is an acid-base reaction, as shown in eqn (3).
The capture of HCl involved rstly the physical selective adsorption on the surface of Na 2 CO 3 bers, followed by the acid-base reaction, forming NaCl crystals on the sorbent surface along with CO 2 , which escapes out.
The total conversion rate X e of Na 2 CO 3 can be adopted to evaluate the reaction degree, which is depicted as eqn (4): where M N is the molecular weight of Na 2 CO 3 (106 g mol À1 ), A (wt%) is the initial loading amount of Na 2 CO 3 , q 0 (mmol g À1 ) is the HCl adsorption capacity of the AABs, and q e (mmol g À1 ) is the HCl adsorption capacity of the modied AABs. 3.2.1 Reaction product analysis. The surface morphology image of the exhausted sorbent with 20 wt% loading is obviously different from the original sorbents, as shown in Fig. 8(a). The elemental distribution of the energy spectrum, shown in Fig. 8(b), suggests that the phases on the surface of the sorbent are almost entirely NaCl and Al 2 O 3 (carbon had been sprayed for electrical conduction). The XRD patterns of exhausted sorbents are shown in Fig. 8(c); they contain the characteristic peak of NaCl besides the substrate, indicating again that the solidphase product of the reaction is only NaCl.
3.2.2 Adsorption efficiency analysis. The pH evolution curves and the adsorption efficiency h curves of the sorbents with different loading amounts are shown in Fig. 9. It can be seen that the pH value is almost kept constant during the whole adsorption process for the AAB sorbents before breakthrough ( Fig. 9(a)), indicating that the physical adsorption of HCl is efficient, with the breakthrough time of 37 min. However, the breakthrough time is signicantly delayed with the increase of the Na 2 CO 3 loading amounts ( Fig. 9(b)), suggesting that the adsorption performance has been improved by chemical adsorption, and the longest breakthrough time of 120 min is obtained with the loading amount of 20 wt% and 25 wt%. As shown in Fig. 9(c) of the enlarged curves, there is a clear  reduction of pH value in the modied adsorbents, at which the physical adsorption is suppressed; meanwhile, the chemical reaction increases slowly, and it is named as transition time t c . The breakthrough times t b and transition times t c are shown in Fig. 9(d).
The physical adsorption processes of these modied sorbents are far less than the AABs, suggesting that Na 2 CO 3 has blocked the original pores of AABs. When the neutralization reaction between HCl and Na 2 CO 3 is faster than the physical adsorption, the overall adsorption efficiency is only attributed to the chemical adsorption, where the sorbents completely capture the HCl. Aer the sorbent was consumed to a certain extent, HCl adsorption stopped, and the adsorption efficiency dropped dramatically ( Fig. 9(b)); the sorbents became ineffective.
3.2.3 Adsorption capacity analysis. The HCl adsorptive capacity q e and nal conversion ratio of Na 2 CO 3 X e were calculated by eqn (1) and (4). The curves of q e and X e with different loading amounts of Na 2 CO 3 are shown in Fig. 10. It can be observed that the highest value of q e reaches 3.56 mmol g À1 , which is ve times the adsorption capacity of AABs. Meanwhile, it keeps a conversion ratio higher than 0.98 before the sorbents have been exhausted. When the loading amount increased to 25 wt%, the conversion ratio of Na 2 CO 3 decreased from 0.98 to 0.90. The conversion of the Na 2 CO 3 obtained in this study is higher than that of the previously reported ones in literature, summarized in Table 5.
3.2.4 Kinetics of HCl adsorption. In order to elucidate the reaction mechanism of HCl adsorption by AAB modied with  Na 2 CO 3 bers, some models have been tted to explain the kinetics of HCl adsorption at 423 K. The process of HCl adsorption can be divided into the following four steps: (1) The HCl gas adsorbs on the surface of the Na 2 CO 3 bers.
(2) HCl and Na 2 CO 3 molecules react on the surface of Na 2 CO 3 bers, generating H 2 O, NaCl and CO 2 .
(3) The CO 2 molecules escape outwardly into the gas stream through the product layer; H 2 O molecules, combined with fresh HCl molecules, diffuse to the inside surface of unreacted bers.
(4) When the reaction rate of HCl and Na 2 CO 3 quickly reduces, the sorbents become exhausted.
In this study, the principle of HCl adsorption is shown in eqn (3). The conversion ratio with per-minute X t can be calculated to investigate the kinetics of HCl adsorption, as shown in eqn (5): where X e represents the maximum conversion rate, shown in Fig. 10, h (%) represents the adsorption efficiency of the sorbent at the time t, t b (min) is the breakthrough time of the sorbents, and t c (min) is the transition time of the sorbents. Considering the pore size distribution of the sorbent, the vast majority of the mesopores do not govern the entry or exit of the gas; that is, the external diffusion of the gas never becomes the rate-limiting link of the adsorption reaction. According to the previous reports, 42-46 the reaction kinetics of sorbents and HCl in the reaction of HCl adsorption at high temperature is mainly controlled by diffusion in the product layer or interfacial chemical reaction.
If reaction is controlled by the interfacial chemical reaction at the grain surface, it will follow eqn (6): 9,13,44 where t is the reaction time and K 1 represents the reaction rate constant. Fig. 10 The curves of adsorption capacity q e and conversion ratio of Na 2 CO 3 X e vs. loading amounts.  If the reaction is controlled by reactant diffusion through the product layer, it will follow eqn (7): where K 2 represents the reaction rate constant. Aer tting the experimental data to eqn (6) and (7), the calculated results are separately shown in Fig. 11(a) and (b). Compared with correlation coefficient, the gure shows that the model of the interfacial chemical reaction control could better explain the experimental results; especially, the linear tting coefficient R 2 of the modied AABs is larger than 0.980, proving that the adsorption reaction is completely controlled by the interfacial chemical reaction. On one hand, the Na 2 CO 3 bers did not block the uffy structures on the surface of original AABs as shown in Fig. 2(b) and maintained the large specic surface area of the AABs; on the other hand, the uniform pore distribution of Na 2 CO 3 bers provided many convenient channels for the gas inward and outward, both resulting in the easier diffusion of gas phase in the product layer. In this case, internal diffusion of gas phase will no longer be the rate-limiting step in the process of chemical adsorption.

Conclusions
In this paper, alkali-modied AABs have been prepared to evaluate their performance for HCl adsorption at low temperature (423 K). The results have demonstrated better modication ability of Na 2 CO 3 compared to Ca(OH) 2 or NaOH. A typical brous structure was obtained on the surface of the sorbent by loading Na 2 CO 3 . The loading amounts, impregnation time and drying time had great impact on the morphology of Na 2 CO 3 bers. When the loading amount was 15-20 wt%, the impregnation time was 17 h, and the drying time was 24 h, Na 2 CO 3 bers with good orientation and high aspect ratio could be obtained.
In addition, the HCl adsorption performance was investigated in a xed bed reactor at 423 K. Extremely high adsorption efficiency and the highest adsorption capacity of 3.56 mmol HCl g À1 were achieved when the loading amount was 20 wt%; the adsorption capacity of the modied sorbents is ve times that of AABs. The physical adsorption is faster than chemical adsorption in the initial process of HCl adsorption, and then the dominant reaction strongly depends on the chemical adsorption. The kinetics investigation indicates that the reaction is controlled by the interfacial chemical reaction; the diffusion of gas phase in the product layer has been greatly improved by the ber structure. We believe that Na 2 CO 3 ber-impregnated AABs are promising for the commercial application of HCl removal from BFG.

Conflicts of interest
There are no conicts to declare.