Removal and recovery of SO 2 and NO in oxy-fuel combustion flue gas by calcium-based slurry

. This study investigates the use of calcium-based slurry for simultaneous removal NO and SO 2 from oxy-fuel combustion flue gas, and recovery of the sulfur and nitrogen species in resulting solutions. The experiments were performed in a bubbling reactor in a transient mode under the pressure of 20 bar. The various influencing factors including the CaO amount, carrier gas (N 2 /CO 2 ), and absorption time on the simultaneous NO and SO 2 removal process, and the solution products were studied comprehensively. The results show that the NO 2 removal efficiency can be improved by the presence of CO 2 , and the gas phase HNO 2 produces in this process. The addition of CaO has positive effects not only on the NO 2 removal efficiency but also on the formation of stable HNO 3 . With the presence of CO 2 , CaCO 3 is formed in a solution initially. With the decrease of pH, CaCO 3 is gradually converted to CaSO 4 , and in particular CaCO 3 can be fully avoided through decreasing the pH of an absorption solution to 1.14. At the same time, the formation of unstable S(IV) and NO 2  can be prevented when the solution pH is lower than 1.37. The nitrogen and sulfur compounds in the absorption solution (at pH 1.14) were further separated by the addition of different amounts of CaO. In particular, 95% of SO 42  finally can be recovered in the form of CaSO 4 ∙ 2H 2 O with nitrogen in solution existing as NO 3- by controlling the Ca/S ratio at 4.70. The effectiveness of calcium-based slurry on the removal and recovery of SO 2 and NO


Introduction
In order to reduce the greenhouse gas CO 2 emission, various pre-combustion, post-combustion (e.g., carbon capture and storage, CCS) and combustion enhanced (e.g., oxy-fuel combustion) technologies have been proposed [1]. Oxy-fuel combustion technology is considered to be one of the most promising ways for carbon capture [2]. This technology not only effectively enriches flue gas with high CO2 concentration [3], but also effectively reduces NO and SO 2 emissions per unit mass of fuel [4]. However, the NO and SO 2 concentrations are higher in the oxy-fuel combustion flue gas than those in conventional aircombustion flue gas [5]. Therefore, these acid gases (mainly includes NO and SO 2 ) must be removed from CO 2 stream considering the safety of transport pipelines and sequestration sites.
Currently, the techniques used in existing power plants to remove SO 2 and NO are often independent. The selective catalytic reduction (SCR) and selective noncatalytic reduction (SNCR) technologies are adopted for NO removal [6], while the wet flue gas desulfurization (WFGD) technology using calcium based solution is adopted for SO 2 removal [7]. In the flue gas, NO accounts for more than 90% of the NO x [8], and is insoluble in aqueous solution. Thereupon, oxidizing insoluble NO to soluble NO 2 before wet scrubbing method is necessary. In this oxidizing process, many oxidation-absorption combined processes have been studied extensively to oxidize NO to NO 2 for the purpose of simultaneous removal of NO and SO 2 , including strong oxidizing agent injection [9], selective catalytic oxidation [10,11], and photo-catalytic oxidation [12].
Apart from conventional technologies for NO and SO 2 removal, many new approaches have been proposed specifically for cleaning oxy-fuel combustion flue gases, mainly including two-stage and one-stage methods. Air Products proposed a two-stage scrubbing process with the low pressure (1.5 MPa) and high pressure (3 MPa) stage to remove SO 2 and NO separately in the forms of H 2 SO 4 and HNO 3 [13]. Similarly, Linde also proposed a twostage method with the first stage (atmospheric pressure) to remove SO 2 as CaSO 4 , and second stage (1.8 MPa) to remove NO as NH 4 NO 3 [14]. In addition, Air Liquide proposed using a sodium solution to remove SO 2 in the atmospheric caustic scrubber, and using a four-stage compression process (at a final pressure of 2.4 MPa) to remove NO in the condensates in the form of HNO 3 [15]. Previous methods generally use the two-stage method for the removal of SO 2 and NO separately, and this method increases the infrastructure cost, while using the one-stage method to remove both SO 2 and NO is limited. Although the costs for infrastructure and operation for the one stage scrubbing method are significantly reduced, the products are generally mixed in compounds containing sulfur and nitrogen [16], which is detrimental for recovering the sulfur and nitrogen compounds.
Currently, the one-stage method utilizing water solvents to remove both SO2 and NO at a high pressure is extensively studied [17][18][19][20]. These studies have reached the following consensus: (1) the liquid products mainly include H2SO4, HNO3, H2SO3, and HNO2 in this process [18]; (2) this process is heavily dependent upon the pH level [19]; (3) the reaction between HNO2 and H2SO3 is critical for the absorption rates of NOx and SOx from the gas to the liquid phase [20]. Moreover, the modeling study mentioned that gas phase HNO2 (HONO) can be produced in this process [20], but it is not identified in the experimental process [21]. Since Platt et al. [22] first measured the HNO2 concentration in the atmosphere by differential absorption spectroscopy (DOAS), the importance of HNO2 in atmospheric chemistry has gradually become known. At the same time, the modeling method can easily obtain the dynamic changes of unstable H2SO3 and HNO2 in solution [23], but these products have not been systematically studied through experiments. Therefore, the quantification of the gas phase HNO2, the liquid phase H2SO3, and HNO2 is critical for the understanding the simultaneous absorption of SO2 and NO and implementing simultaneous removal technique in the practical applications.
In this work, the one-stage method utilizing calciumbased slurry was adopted to remove NO and SO 2 in the oxy-fuel combustion flue gas, and also to recover the sulfur and nitrogen compounds in the forms of CaSO 4 •2H 2 O precipitate and NO 3 -. This approach can potentially reduce the operation cost through the utilization of the low-cost absorbents and the recovery of the end products. The effects of the CaO amount and absorption time on the removal efficiency of SO 2 and NO as well as the effectiveness of the recovery of H 2 SO 4 and HNO 3 are discussed. In addition, the mechanisms for simultaneous removal, and recovery processes are elucidated.

Experimental apparatus and method
As shown in Fig.1, the experimental apparatus mainly included a high pressure bubbling reactor, four high pressure mass flow controllers, a light source, an optical lens, an absorption cell, a spectrograph and an MRU delta 2000 flue gas analyzer. The high pressure mass flow controllers were used to control the flow rate of the simulated flue gas into the high pressure bubbling reactor, including O2 (purity > 99.999%), N 2 /CO 2 (purity > 99.999%), 1% (NO in N 2 ) and 3% (SO 2 in N 2 ). The 316 stainless steel tube was used as the gas transmission line, and the total flow rate of the simulated flue gas was 2 L/ min. The simulated flue gas was introduced to the cylinder reactor from top of the reactor to nearly the bottom of the reactor through a 1/8 inch tube to maximize the residence time of gas mixture in the reactor. The volume of the reactor was 1 L with half-filled with absorption solution. The average residence time was obtained by dividing the volume of reactor by the gas flow rate. Under the pressure of 20 bar, the residence time of gas mixture in liquid phase is 300 s. During the contact between gas and liquid, soluble gases in the gas mixture were absorbed into absorption solution. Above the absorption solution, gases containing water vapor flowed upwards to the outlet of the cylinder reactor. The pressure inside the reactor was controlled by the pressure regulator. Following the regulator, the simulated flue gas flowed to the absorption cell, where the NO2 and HNO 2 concentrations were measured online by the DOAS. After the absorption cell, the simulated flue gas was partially drawn to the MRU delta 2000 flue gas analyzer for the measurement of the SO 2 , NO, and O 2 concentrations, and the residual flue gas was washed before being vented to atmosphere.
The operation procedures are given as follows. 0.5 L of deionized water with a certain amount of CaO was first introduced into the reactor. The bypass valve was turned on, and then the concentrations of O 2 , N 2 /CO 2 , NO, and SO 2 were set to the required concentrations. After 10 minutes stabilization, all the gas bottles were turned off and N 2 /CO 2 gas bottle was left to rinse the gases of NO and SO 2 in the system. After that, the bypass valve was turned off, and the N 2 /CO 2 gas was continuously introduced into the reactor to pressurize the reactor to the set pressure, and then the simulated flue gas was continuously introduced into the reactor. In all experiments, the inlet concentrations of O 2 , N 2 /CO 2 , NO, and SO 2 were set at 5%, 75%/75%, 1000 ppm, and 2000 ppm, respectively, and the pressure selected was 20 bar. In addition, the pH of the absorption solutions after experiments was determined by a pH meter. The concentrations of SO 4 2 and NO 3  in solution were measured by the UV-visible spectrophotometry [24], while the concentrations of S(IV) (i.e., HSO 3  , SO 3 2 , and SO 2 ) and NO 2  in solution were determined by the acid desorption method. The solid by-products of the absorption solutions were identified by the X-ray diffraction (XRD-6100). Above measurements were sampled after the experiment. The MRU delta 2000 flue gas analyzer was used to measure the SO 2 and NO concentrations, whereas the NO 2 and HNO 2 concentrations were measured online by the DOAS. The absorption characteristics of different gases are different in spectral bands because of the different structures of gas molecules. The measurement of different gases by the DOAS utilized the selective absorption of light from the ultraviolet to near-ultraviolet bands by gases. When a light source with an incident light intensity of I 0 passes the measuring medium, and then the emitted light intensity is I due to the radiation absorption of different gases. The Lambert Beer's law expressed by Eq. (1) describes the relationship between I and I 0 .
where α is the absorption coefficient; I 0 and I are the incident and emitted light intensity, respectively, candela; σ i is the absorption cross section, m 2 ; L is the absorption cell length, 0.35 m; C i is gas concentration, ppm. NO 2 absorbs the lights with the spectral range from 340 nm to 400 nm [25]. The near-ultraviolet absorption peak of HNO 2 is 341.8 nm, 354.2 nm, and 368 nm, respectively [26]. The center wavelength of the light source used in the experiment was 355 nm, and the full width at half the maximum light intensity was 15 nm. Therefore, in the spectral range from 340 nm to 370 nm, the least-squares fitting method was adopted for the calculations of the HNO 2 and NO 2 concentrations. Fig.2 shows the spectral fitting results for a single-point simultaneous measurement of both the NO 2 and HNO 2 [27]. By comparing the magnitudes of the absorption coefficients for fitting results with these for residuals, the absorption coefficients for NO 2 is two magnitudes higher than residuals while the absorption coefficients for HNO 2 is the same magnitude with the residuals. Therefore, the measurement for NO 2 may be more accurate than the measurement for HNO 2 . The concentrations of NO 3  and SO 4 2 in solution were measured by UV-visible spectrophotometry, which used the specific wavelength to measure the corresponding substance. Before the measurement, the standard curve for the standard solutions versus the corresponding absorbance was obtained by experiments. The substance concentrations at different absorbance could be regressed from the standard curve. When the light source passes through the solution to be tested, the absorbance of the solution is proportional to the concentration of the substance in the solution. The relationship is expressed by Eq. (2).
where A is the absorbance of the absorption solution; I and I 0 are the incident and emitted light intensity, respectively, candela; K is the molar absorption coefficient; L is the length of the cuvette, 50 mm; C is the concentration of the substance, mol/L. Based on the above principle, the NO 3  concentration was measured at the specific wavelength of 220 nm. The SO 4 2 concentration was measured by the UV-visiblemethod of barium-chromate (BaCrO 4 ). Under acidic conditions, the reaction between BaCrO 4 and SO 4 2 forms BaSO 4 precipitates and the CrO 4 2 ions which have the maximum absorption at 420 nm. After filtering the BaSO 4 precipitates, the CrO 4 2 concentration could be obtained by measuring the absorbance of the solution, so the SO 4 2 concentration can be obtained because the CrO 4 2 concentration is equal to the SO 4 2 concentration.
The S(IV) and NO 2  concentrations in solution were measured by acid-promoted desorption method. In order to completely convert S(IV) to SO 2 , a certain amount of HCl (1 ml 2.5 mol/L HCl ) was added to the absorption solution (5 mL), and then the absorption solution was purged with 2 L/min of N 2 with the discharged gas measured by the MRU delta 2000 flue gas analyzer for the SO 2 concentration. When the SO 2 concentration was zero, the N 2 purging process was terminated. The total S(IV) concentration in a solution was calculated by integrating the SO 2 concentration against time within the desorption period. In order to verify the accuracy of this method, the standard NaHSO 3 solutions with four concentrations of 0.0025, 0.005, 0.01, and 0.02 mol/L were measured with relative errors for these measurements being less than 3%.
Similarly, the NO 2  concentration in solution was also measured by acid-promoted desorption. After adding a certain amount of HCl (1 ml 2.5 mol/L HCl ), NO 2  was converted to HNO 2 which could be desorbed in the forms of NO, NO 2 , and HNO 2 . Four standard NaNO 2 solutions (0.0025, 0.005, 0.01, and 0.02 mol/L) were used to validate the accuracy of NO 2 -measurement. Overall, the accuracy of acid-promoted desorption method was over 90%.

Effects of CaO amounts on the simultaneous NO and SO2 removal process
To investigate the influence of CaO amounts on the NO and SO 2 removal process, different amounts of CaO (0, 0.285, 0.570, and 0.855 g) were introduced into water to make up a total liquid volume of 0.5 L. The experiments were performed in the presence or absence of CO 2 at room temperature. As shown in Figs. 3-A, 3-A', 3-B, and 3-B', the concentrations of NO and NO 2 at the outlet of bubble reactor gradually increased with time. Both of them increased significantly initially within 10 min, and then increased slowly after that. The initial 10 min quick increase for both NO and NO 2 could be due to the retention time of the bubble reactor before the gas analyzer.
The slow increase of NO and NO 2 after the quick initial increase may come from the complex chemical reactions between the gas phase NO and NO 2 with water. NO is considered oxidized easily under high pressure to form NO 2 [28]. Through forming high soluble intermediate, the absorption of NO and NO 2 into water is greatly enhanced to form HNO 2 , and HNO 3 [29]. HNO 3 is quite stable in solution, while HNO 2 is easily decomposed to form HNO 3 and NO at room temperature [30]. The different proportions of HNO 2 and HNO 3 is strongly dependent on the solution pH [31]. The decomposition of HNO 2 to HNO 3 and NO was enhanced with the decrease of the liquid pH, which results in the increase of NO in the gas phase. In a similar way, this also results in increasing NO 2 in the gas phase due to the re-oxidation of NO to NO 2 [31].
By comparing with the case in the presence of CaO, it can be seen that the introduction of CaO into water markedly enhanced NO and NO 2 absorption. NO and NO 2 absorption increased with the increase in the amount of CaO. The reason for this result is that the CaO dissolves into water to increase the alkalinity of the solution, and the alkaline condition leads to slightly higher NO and NO 2 absorption efficiencies [32]. Furthermore, the NO 2 absorption is further promoted when CO 2 is present in the gas phase. This can be explained that in the presence of CO 2 , the Ca 2+ preferentially participates in the formation of CaCO 3 ; in the absence of CO 2 , the Ca 2+ preferentially participates in the formation of low solubility CaSO 3 , which is detrimental for the NO 2 absorption [33].
Figs. 3-C and 3-C' show that the outlet of HNO 2 concentration increased slightly over time, and the HNO 2 concentration was approximately 5 ppm. Both the presence of CO 2 and the addition of CaO had little effect on the production of HNO 2 . SO 2 was not observed at the outlet of bubble reactor for all the experiments indicating that all SO 2 can be absorbed because of the instantaneous reaction of SO 2 with water [30]. ) and HNO 2 ((C) and (C')) in the presence or absence of CO 2 (inlet NO concentration is 1000 ppm, CO 2 and N 2 stand for CO 2 and N 2 atmosphere, respectively) The dynamic changes of the concentrations of SO 4 2 , NO 3  , S(IV), NO 2  , and the solution pH are shown in Fig.4. As shown in Fig.4, CO 2 has a slight effect on the NO 3  , S(IV), NO 2  concentrations, and remarkable effect on the SO 4 2 concentration, the solution pH. Both the SO 4 2 and NO 3  concentrations increased with time, and NO 3  concentrations were quite similar. However, in the presence of CO 2 , the SO 4 2 concentration was much higher than that in the absence of CO 2 , and this is caused by the reaction between SO 4 2 and Ca 2+ to form CaSO 4 precipitate to consume SO 4 2 [33], while in the presence of CO 2 , the Ca 2+ exists in solution in the form of CaCO 3 . In the presence of CO 2 , the solution pH initially dropped more slowly compared with that in the absence of CO 2 . Because the formed Ca(OH) 2 can be rapidly consumed by HCO 3  to form CaCO 3 in the pressurization process. However, in the presence of CO 2 , the final pH ( pH = 2.79 ) (60 min) was much higher than that ( pH = 1.92 ) (60 min) in the absence of CO 2 , and this may be caused by the formed CaCO 3 , which may work as a buffer in liquid.
Meanwhile, both the S(IV) and NO 2  concentrations first increased, and then decreased dramatically with time because strong acidic conditions accelerate the decomposition of both the S(IV) and NO 2  [34], so that their concentrations (60 min) became very low. XRD analysis was used to determine the crystal composition of precipitate after gas absorption. Both dried and wet precipitates were analyzed. In the presence of CO 2 , the only precipitate was CaCO 3 , whereas in the absence of CO 2 , the only precipitate was CaSO 4 •2H 2 O. In comparison, the most popular WFGD uses CaCO 3 /CaO to remove SO 2 and NO, and the by-product of this process mainly includes CaSO 3 [35,36]. The comparison indicates that in a high pressure scrubbing process, the generated CaSO 3 can be oxidized to CaSO 4 by dissolved NO 2 and O 2 effectively [37,38], and CaSO 4 further reacts with H 2 O to form CaSO 4 •2H 2 O [37]. From the aspect of economy, the byproduct CaSO 4 •2H 2 O has a higher recycling value than CaSO 3 . The utilization of by-product CaSO 4 •2H 2 O can reduce the operation cost of a scrubber. As CO 2 is dominate in the gas phase for oxy-fuel combustion flue gas, the formation of CaCO 3 cannot be avoided. However, CaCO 3 is reactive towards SO 2 [39,40], and therefore, CaCO 3 can be converted to CaSO 4 if enough reaction time is provided.

Effects of absorption time on the simultaneous NO and SO2 removal process
The absorption time was prolonged to avoid the calcium exist in the form of CaCO 3 . The experiments were carried out under the same amount of CaO (0.855g) in the presence of CO 2 with different absorption times including 5, 10, 20, 40, 60, 180, and 240 min. Fig.7 shows the dynamic changes of the concentrations of SO 4 2 , NO 3  , S(IV), NO 2  , and the solution pH. Based on Fig. 7-A, the solution pH had a slight increase (from pH 4.78 to pH 5.66) within 20 min, and this result may be due to the addition of excessive CaO. In this pH range, the S(IV), NO 2  , and NO 3  began to form, and in particular, both the S(IV) and NO 2  concentrations reached the maximum at a pH 5.66. From pH 5.66 to pH 1.37, both the concentrations of S(IV) and NO 2  sharply decreased to zero, while both the SO 4 2 and NO 3  were increasingly formed with time. In this pH range, following two processes may occur: (1) some S(IV) (mainly includes SO 3 2 and HSO 3  ) are gradually oxidized into SO 4 2 by the dissolved NO 2 and O 2 [41,42]; (2) some S(IV) react with Ca 2+ /CaCO 3 to form CaSO 3 . With the decrease of solution pH, NO 2  is gradually converted to HNO 2 which is more easily decomposed to HNO 3 and NO, and this result in the NO 2  decreasing to zero. From pH 1.37 to pH 1.14, both the SO 4 2 and NO 3  concentrations reached the maximum because the continuous absorption of SO 2 and NOx into water to form the SO 4 2 and NO 3  . Furthermore, when the absorption time was increased to 240 min (pH = 1.14), the only precipitate was CaSO 4 •2H 2 O. Above results indicate that when solution pH was above 4.57, the formed CaCO 3 can stably exist, but when the pH was reduced to 1.14, the formed CaCO 3 can be completely decomposed by the formed H 2 SO 3 , HNO 3 , and H 2 SO 4 . At the same time, the formed sparingly soluble CaSO 3 can be oxidized to CaSO 4 by the dissolved NO 2 and O 2 [37,38]. The reaction between CaSO 4 and H 2 O also takes place in this process. In practical situation, the controllable factor is the pH of the solution. Above results indicate that CaCO 3 can be avoided through decreasing the pH of absorption solution to 1.14.

Effects of the ratios of Ca/S on the recovery process
As can be seen from Fig.7

Proposing the mechanism for simultaneous removal and recovery of NO and SO2
The schematic diagram of the reaction pathway for simultaneous removal, and recovery of NO and SO 2 including the pressurization, absorption, and recovery processes is shown in Fig.10. The solution pH has a significant effect on the products formation. In the pressurization process, CaO firstly reacts with H 2 O to form Ca(OH) 2 (R1), and then it is rapidly consumed by HCO 3  to form CaCO 3 precipitate (R2, R3). Compared to the pressurization process, the absorption process is more complicated. With the introduction of the simulated flue gas, NO can be converted into NO 2 by O 2 under high pressure (20 bar) in the gas phase (R4) [20]. NO and NO 2 may also react with water vapor to form the gas phase HNO 2 (R5) [43]. In the liquid phase absorption process, from pH 4.78 to pH 5.66, the NO 2  , NO 3  and S(IV) begin to form due to the hydrolysis of SO 2 and NO 2 (R6-R9) [31]. However, some S(IV) (mainly includes SO 3 2 and HSO 3  ) can be oxidized into the SO 4 2 by NO 2 and O 2 (R10, R11) [41,42], and some S(IV) react with Ca 2+ /CaCO 3 (R12, R15) to form CaSO 3 which can be oxidized to CaSO 4 by the dissolved NO 2 and O 2 (R16, R17) [37,38], and with the decrease of solution pH, NO 2  is gradually converted to HNO 2 which is more easily decomposed to HNO 3 and NO (R18). From pH 5.66 to pH 1.37, both the NO 2  and S(IV) concentrations decrease to zero, whereas both the NO 3  and SO 4 2 increasingly form, and the formed CaCO 3 is gradually decomposed to form CaSO 4 .2H 2 O (R12-R14). From pH 1.37 to pH 1.14, the residual formed CaCO 3 can be completely decomposed to form CaSO 4 .2H 2 O by the formed HNO 3 and H 2 SO 4 . In this pH range, following two processes occur: (1) the residual formed Ca 2+ reacts with SO 4 2 to CaSO 4 (R19); (2) the formed CaSO 4 reacts with H 2 O to form CaSO 4 .2H 2 O (R20).
In the separation process, the absorption solution (at a pH of 1.14) only contains H 2 SO 4 and HNO 3 , and CaO was introduced into the obtained absorption solution to further separate H 2 SO 4 and HNO 3 . When the ratio of Ca/S is 4.70, the solution pH is 8.94, and these reactions only take place in solution to form CaSO 4 •2H 2 O and Ca 2+ , NO 3 -(R19-R22). When the ratio of Ca/S is 7.05, the solution pH is 11.27. As excess CaO is added to the solution, and CaO cannot be fully consumed by H 2 SO 4 and HNO 3 . Reactions take place in solution to form Ca(OH) 2 , CaSO 4 •2H 2 O, and Ca 2+ , NO 3 -(R1, R19-R22).

Conclusions
Wet-based method using calcium-based slurry was proposed for simultaneous removal of SO 2 and NO from oxy-fuel combustion flue gas at a high pressure while recovering products as the CaSO 4 •2H 2 O precipitate and NO 3 -. This process was developed and investigated through experiments. The following conclusions can be drawn: (1) NO is effectively oxidized to NO 2 at a pressure of 20 bar, and the gas phase HNO 2 is produced in this process. The presence of CO 2 in the simulated flue gas can enhance the NO 2 absorption. At the same time, the addition of CaO into water not only enhances the NO 2 absorption but also the formation of HNO 3 .
(2) In the presence of CO 2 in the simulated flue gas, the initial formed CaCO 3 can be completely converted into CaSO 4 •2H 2 O through decreasing the pH of absorption solution to the pH 1.14. Both the formation of S(IV) and NO 2  can be prevented when the solution pH is lower than 1.37.