Aqueous Ammonia (NH 3 ) Based Post Combustion CO 2 Capture: A Review

Re´sume´ — Capture de CO 2 en postcombustion par l’ammoniaque en solution aqueuse (NH 3 ) : synthe`se — L’ammoniaque en solution aqueuse (NH 3 ) est un solvant novateur, prometteur et un challengeur pour la capture du CO 2 e´mis par les centrales e´lectriques au charbon et au gaz et autres sources industrielles. Au cours des dernie`res anne´es, d’intenses activite´s de recherche ont e´te´ mene´es pour tenter de comprendre les proprie´te´s thermochimiques et physiques du syste`me CO 2 -NH 3 -H 2 O, ainsi que la chimie et la cine´tique associe´es, aﬁn d’obtenir plus d’informations sur le taux d’absorption, la perte en ammoniaque, les e´nergies de re´ge´ne´ration et les proble`mes pratiques, et aussi pour e´valuer la faisabilite´ technique et commerciale des technologies de capture en postcombustion par l’ammoniaque en solution aqueuse. Cet article offre une synthe`se de ces activite´s de recherche, apporte des comparaisons avec le Abstract — Aqueous Ammonia (NH 3 ) Based Post Combustion CO 2 Capture: A Review — Aqueous ammonia (NH 3 ) is an emerging, promising and challenging solvent for capture of CO 2 emitted from coal and gas ﬁred power stations and other industrial sources. Over the last few years, intensive research activities have been carried out to understand the thermo-chemical and physical properties of the CO 2 -NH 3 -H 2 O system and the chemistry/kinetics involved, to obtain information on absorption rate, ammonia loss, regeneration energies, and practical issues, and to assess the technical and economical feasibility of the aqueous ammonia based post combustion capture technologies. This paper reviews these research activities, makes comparisons with MonoEthanolAmine (MEA, the benchmark solvent) and identiﬁes the research opportunities for further improvements.


INTRODUCTION
With the rapid industry development globally, excessive amounts of the greenhouse gases in particular CO 2 have been released to the atmosphere, which is believed to be the major contributor to global warming. It is generally agreed that Carbon Capture and Storage (CCS) is the only technology available to make deep cuts in greenhouse gas emissions while still using fossil fuels and much of today's energy infrastructure. Developing an economic and efficient method to capture carbon dioxide is an important objective for successful deployment of effective CCS process chains.
Post Combustion Capture (PCC) is a process that uses an aqueous absorption liquid incorporating compounds such as ammonia or amine to capture CO 2 from power station flue gases and many other industrial sources. It is the leading capture technology as a result of the potential benefits, such as, -it can be retrofitted to existing power plants or integrated with new infrastructure to achieve a range of CO 2 reductions, from partial retrofit to full capture capacity; -it has a lower technology risk compared with other competing technologies; -renewable technologies can be integrated with PCC, for example, low cost solar thermal collectors can provide the heat required to separate CO 2 from solvents; -PCC can be used to capture CO 2 from a range of sources -smelters, kilns and steel works, as well as coal-and gas-fired power stations. Currently, the commercially available PCC technology is mainly based on alkanol/alkyl amine solutions. A study by Dave et al. (2009) shows that retrofitting a MEA based PCC plant to the existing/new mechanical draft water cooled black coal fired plants will reduce the power plant efficiency by 10 absolute percentage points and involves significant capital investment costs. Moreover, there is some concern about the formation of carcinogenic nitrosamines from the use of amines in PCC and their possible spread into the environment (MacDowell et al., 2010).
Aqueous NH 3 , as a promising solvent for CO 2 capture, has been receiving increasing attention recently. Compared to traditional amines, NH 3 is a low cost solvent that is not easily decomposed by other gas present in flue gases. Moreover, it has a high CO 2 removal capacity, low absorption heat and low regeneration energy and is less corrosive to instrument. It also has the potential of capturing multiple flue gas components (NO x , SO x and CO 2 ) and producing value added chemicals, such as ammonium sulphate and ammonium nitrate, which are commonly used as fertilizers. However, ammonia is volatile and hence it slips to flue gas and CO 2 product, thus requiring additional energy and capital costs to reduce ammonia evaporation and/or recover ammonia. This could offset the advantages offered by ammonia.
Over the last few years, there has been a significant amount of research work on the aqueous ammonia based CO 2 capture technologies including experimental investigations at various scales and process modeling to evaluate energy consumption and economics. This paper will briefly review these research activities and make comparisons with other solvents in particular MonoEthanolAmine (MEA), the benchmark solvent, and identify the research opportunities for further improvements.

AQUEOUS AMMONIA BASED CO 2 CAPTURE PROCESSES
A number of pilot and demonstration plants has been constructed and operated to evaluate the technical and economic feasibility of NH 3 based PCC processes to remove CO 2 from flue gas being emitted from various industrial sources including both coal and natural gas fired power stations. Alstom, Powerspan, CSIRO and KIST (Korean Research Institute of Industrial Science & Technology) are the major players in this area and have tested the technologies at pilot-plant scale (McLarnon and Duncan, 2009;Rhee et al., 2011;Telikapalli et al., 2011;Yu et al., 2011a).
Powerspan developed an aqueous NH 3 based CO 2 capture process, called ECO 2 Ò , in which the absorption takes place at ambient temperatures and no slurry is involved in the absorber (McLarnon and Duncan, 2009). The ammonia concentration used was not given and believed to be between 5 and 15%. Desorption occurs at low pressures. An illustration of how the ECO 2 system is integrated with the ECO multi-pollutant control system is shown in Figure 1. The ECO system uses a barrier discharge reactor, wet scrubber and wet electrostatic precipitator to remove high levels of NO x , SO 2 , mercury and fine particulate matter. The wet scrubber uses aqueous ammonia for SO 2 scrubbing, producing ammonium sulphate fertilizer as a co-product. Powerspan initiated a pilot test program with FirstEnergy at the R.E. Burger Plant to demonstrate CO 2 capture through integration with the ECOÒ multi-pollutant control process. A 1-MW pilot demonstration was designed to produce approximately 20 metric tonnes of sequestration ready CO 2 per day. However pilot plant results have not been reported in public domain and there is no further development of this technology reported in the literature.
Alstom has been playing a leading role in developing and advancing a Chilled Ammonia Process (CAP). In this process as described in the open literature (Darde et al., 2009;Telikapalli et al., 2011), CO 2 is absorbed in a highly ammoniated solution at low temperatures (0-10°C), producing a slurry containing ammonium bicarbonate. In the stripper, ammonium bicarbonate is converted to ammonium carbonate at temperatures above 100°C and pressures of 20-40 bar. Figure 2 shows the simplified chilled ammonia process flow diagram.
Alstom conducted a number of field tests of CAP and completed the demonstration project at American Electric Power's Mountaineer Plant at end of May 2011 after a 21 month period (Jo¨nsson and Telikapalli, 2012). The technology is ready for the next major validation on a gas turbine exhaust as well as industrial (refinery) off-gas with the installation of demonstration plant at Technology Centre Mongstad (TCM). TCM is the world's largest facility for evaluating carbon capture technologies.
The processes investigated by CSIRO and under development at KIST are similar to Powerspan's ECO 2 : they both operate at close to ambient temperatures and use ammonia concentrations below 10%.
The major difference is that in the CSIRO research, the desorption takes place under medium pressure and the process combines SO 2 removal and ammonia slip reduction (Yu et al., 2011a(Yu et al., , 2012a while RIST is developing technology for CO 2 capture from Blast Furnace Gas (BFG) generated from iron and steel making process. The process utilises the unrecovered waste heat from a steelmaking process for solvent regeneration and ammonia recovery at low temperatures (Rhee et al., 2011).

PROCESS CHEMISTRY
Thermo-chemical properties for the CO 2 -NH 3 -H 2 O system have been reasonably understood thanks to intensive experimental and modeling studies over the years, in particular the last few years. These studies include experimental determination of vapour-liquid-solid equilibrium, speciation, heat of absorption, reaction kinetics, and the thermodynamic model development. Research work in this area has been reviewed by Maurer (2011) and Zhao et al. (2012). This section only briefly reviews vapour-liquid-solid equilibrium and reaction mechanism involved in the liquid phase.  Figure 1 Incorporation of the ECO 2 scrubbing process with Powerspan's commercially available multi-pollutant control ECO process.

Vapour-Liquid-Solid Equilibrium (VLSE)
The CO 2 -NH 3 -H 2 O equilibria relevant for the application of aqueous ammonia based CO 2 capture processes are described in Figure 3 ( Thomsen and Rasmussen, 1999). As shown in Figure 3, despite the fact that ammonia can be regarded as the simplest amine, its interaction with CO 2 is quite complex involving the evaporation of ammonia to the gas phase, the formation of carbonate, bicarbonate and carbamate in the liquid phase and solid precipitation. Ammonium bicarbonate is believed to be the major specy in the solid, but not all authors agreed on the possible formation and combination of the salts made of (NH 4 ) 2 CO 3 , NH 2 COONH 4 , 2NH 4 HCO 3 . The complex nature of the CO 2 -NH 3 -H 2 O system makes ammonia an intriguing solvent and it imposes some challenges for its application on CO 2 capture, as discussed in the following sections.   Vapour liquid solid equilibria in the CO 2 -NH 3 -H 2 O system.
A number of thermodynamic models has been developed for the NH 3 -CO 2 -H 2 O system in particular models developed by Thomsen and Rasmussen (1999) and Que and Chen (2011) which satisfactorily predict the vapour-liquid-solid equilibria over a wide range. Thomsen and Rasmussen developed a thermodynamic model which includes the aforementioned equilibrium processes (Fig. 3) under the 0-110°C range of temperature and 1-100 bar of the pressure, the ammonia concentration up to 80 molal. This model is not only for the vapour-liquid system but also for vapourliquid-solid equilibrium including formation of ammonium bicarbonate: NH 4 HCO 3 ; ammonium carbonate: (NH 4 ) 2 CO 3 .H 2 O; ammonium carbamate: NH 2 COONH 4 and ammonium sesqui-carbonate: (NH 4 ) 2 CO 3 Á2NH 4 HCO 3. The thermal properties of the aqueous electrolyte solution are calculated on the basis of the extended UNIQUAC model and the Soave-Redlich-Kwong equation of state. Que and Chen (2011) developed the electrolyte NRTL (Non-Random Two Liquid) activity coefficient model which can satisfactorily represent the thermodynamic properties of the NH 3 -CO 2 -H 2 O system at temperatures up to 473 K, pressures up to 7 MPa, NH 3 concentrations up to 30 wt%, and CO 2 loadings up to unity. However, the model considers ammonium bicarbonate as the only species in the solid.
The availability of these models allows the reliable calculation of thermochemical properties of the CO 2 -NH 3 -H 2 O system under various conditions, assessment of the energy performance of the capture process and even the definition of the best range of operation conditions. For example, Figure 4 shows the predicted solubility curve for ammonium bicarbonate as a function of ammonia concentration and CO 2 loading at various temperatures (Yu et al., 2012a). In the region above each curve, ammonium bicarbonate precipitates from the liquid phase. To avoid solid precipitation in the absorber, the ammonia concentration, CO 2 loading and absorption temperatures should be considered carefully.

Reaction Mechanism/Kinetics in the Liquid Phase
CO 2 absorption in aqueous ammonia is a relatively slow process and limited by the mass transfer in the liquid phase (Darde et al., 2011;Liu et al., 2011;Qi et al., 2013). So it is critical to have a detailed understanding of the reaction mechanism/kinetics involved in the absorption chemistry in order to develop a process model for process evaluation, optimisation and scale-up. A number of researchers investigated the reactions involved in the absorption process and put forward different reaction mechanisms. Similar to the case for the reaction of CO 2 with primary amines, the most important reaction in the presence of free ammonia is the reaction of NH 3 with CO 2 . The reaction mechanisms proposed include the zwitterion mechanism (Derks and Versteeg, 2009) and the ter-molecular mechanism (Qin et al., 2010). A more detailed model was developed by Wang et al. (2011) who studied the kinetics of the reversible reaction of CO 2 with ammonia (instead of irreversible reaction proposed by other researchers) in aqueous solution using stopped-flow techniques and established a complete reaction mechanism, as shown in Figure 5  .
The interactions of CO 2 (aq) with water and hydroxide ions are indicated in dashed lines. Equations (1-6) summarize this part of the reaction scheme as well as the protonation equilibrium of ammonia: The formation reactions of the carbamate/carbamic acid included in the mechanism are shown in Equation (7), Predicted ammonium bicarbonate solubility operation line.
The temperature dependence of all rate and equilibrium constants including the protonation constant of NH 3 between 15 and 45°C were reported by Wang et al.
In a CO 2 capture process aqueous ammonia reacts with CO 2 (aq) to form carbamic acid which deprotonates instantly to give a carbamate. Due to the pH changes with continued absorption of CO 2 , the equilibrium is reversed and the carbamate eventually decomposes to generate protonated ammonia and bicarbonate. Table 1 shows the carbamate equilibrium constants and pKa of amines including ammonia. The carbamate equilibrium constants for MEA is much higher than that for ammonia, suggesting that ammonia-derived carbamate is formed with a lower yield than the equivalent Mono-EthanolAmine carbamate, which contributes to the higher CO 2 loading capacity for aqueous ammonia compared to MonoEthanolAmine.

PERFORMANCES OF AQUEOUS AMMONIA BASED POST COMBUSTION CAPTURE PROCESSES
Pilot plant trials and demonstration projects have confirmed the technical feasibility of the aqueous ammonia based CO 2 capture processes and some of the anticipated benefits (Jo¨nsson and Telikapalli, 2012;Rhee et al., 2011;Yu et al., 2011b). The benefits include high CO 2 removal efficiency and high purity of CO 2 product; high stability; high CO 2 loading capacity and high effectiveness for capturing SO 2 (Yu et al., 2011b). This section focuses on absorption rate, regeneration energy, ammonia loss and overall performance.

CO 2 Absorption Rate
The CO 2 absorption rate is an important parameter which can determine the size of column and hence determine capital cost. It is related to the mass transfer coefficient, the effective surface area for mass transfer and the driving force as described in Equation (9): where N CO 2 is the CO 2 absorption rate (mol/s), K G is the overall gas phase mass transfer coefficient (mol/(s.m 2 .kPa)), P CO 2 is the partial pressure of CO 2 in the flue gas (kPa), P Ã CO 2 is the CO 2 equilibrium partial pressure in the solvent (kPa) and A int is the effective interfacial surface area of packing (m 2 ).
Mass transfer coefficients are used to assess the CO 2 absorption rates in the solvents. The bench scale studies using wetted wall column and pilot plant studies all suggest that CO 2 mass transfer coefficient in aqueous ammonia are lower than those in MEA (Yu et al., 2011a(Yu et al., , 2012b. Figure 6 shows the comparison of overall Reaction scheme, including all reactions between ammonia, the CO 2 /carbonate group, and protons . AMP: 2-amino-2-methyl-1-propanol; DEA: Diethanolamine; PIPD: Piperidine; 4-PIPDM: 4-piperidinemethanol; 4-PIPDE: 4-piperidineethanol. gas phase mass transfer coefficient as a function of CO 2 loading in aqueous ammonia and the standard 30 wt% MEA (Yu et al., 2012b). Good agreement between results from pilot plant and wetted wall column experiments for the ammonia solvent was observed. This suggests that the mass transfer coefficients obtained from the wetted wall column can be used to estimate those in large-scale packed columns, at least under the conditions reported here. The mass transfer coefficients for CO 2 in 3 M aqueous ammonia alone are 3 to 5 times lower than those in MEA solutions at the CO 2 loading range of 0.2-0.4 mole CO 2 per mole amine. The low values for the mass transfer coefficients are due to the low rate constant for the reaction of NH 3 and CO 2 in comparison with that for the reaction between MEA and CO 2 . This reaction is the key step for CO 2 absorption in aqueous ammonia . The reaction of CO 2 + OH -, can contribute to CO 2 absorption but its contribution is small due to the low concentration of OH -. According to surface renewal theory of mass transfer with chemical reactions, for a pseudo first order reaction and assuming resistance in the gas phase is negligible, the liquid side mass transfer coefficient can be estimated by Equation (10) (Yu et al., 2012b): k is the rate constant for forward reaction of CO 2 with NH 3 , C NH 3 is the concentration of free ammonia in the liquid bulk, and D CO 2 and H CO 2 are diffusivity coefficients and Henry's law constants in aqueous ammonia, respectively. k values are typically around 200 M À1 Ás À1 at the temperature of 15°C . The values are significantly lower than the rate constants for the reaction of MEA with CO 2 at 40°C which are more than 10 000 M À1 Ás À1 (Versteeg et al., 1996). Both temperature and ammonia concentration can be increased to enhance K G , but the ammonia loss would increase as well.
There is no publicly available report on the CO 2 absorption rate in the chilled ammonia process. Recently, Alstom power published a paper in which described the rate based model for the chilled ammonia process (Lia et al., 2012). The model used the zwitterion mechanism to explain the reaction between CO 2 and NH 3 and the kinetics parameters in equations were derived by fitting them against experimental data reported by Derks (Derks and Versteeg, 2009). So it is speculated that the mass transfer coefficients for CO 2 in the chilled ammonia process are low, if no promoter/additive is used. Table 2 lists the solvent regeneration energies reported in the literature. The regeneration energies refer to the heat requirement for regeneration of solvents (reboiler duty) in the stripper. The values are quite different and vary from less than 1 000 KJ/kg CO 2 to more than 4 200 KJ/kg CO 2 . The significant discrepancy is due to under-estimation of heat of desorption in the previous studies and the large variation in operational conditions in particular the solvent concentrations.

Regeneration Energy
During the solvent regeneration, heat is required for: -CO 2 desorption reaction; -steam regeneration; -sensible heat.
Due to the different process conditions, the contribution of the above three parts of energy is various, resulting in the various heat requirements in the CO 2 capture process.
During the stripper process, CO 2 can be desorbed via many pathways including the following overall reactions (Jilvero et al., 2012).
The previous research reported a low regeneration energy requirement based on the assumption that the regeneration only involves the decompose reaction of ammonia bicarbonate to CO 2 and ammonia carbonate (R1) and the heat of reaction for this reaction is low (DH = 26.88 kJ/mol). The equilibrium based modeling suggested that one cannot single out only one particular pathway for the solvent regeneration. The experimental and modeling investigations show that the heat of absorption is a function of CO 2 loading and a weak function of temperature and ammonia concentration. Figure 7 shows the experimental and predicted heat of absorption of CO 2 in 5% ammonia solution at the temperature of 40°C (Qi et al., 2013). The predicted results were obtained using the thermodynamic model for CO 2 -NH 3 -H 2 O developed by Aspentech (Que and Chen, 2011). Over the large CO 2 loading range (0-0.6), the heat of absorption is more than 60 kJ/mol, suggesting that the overall absorption/decomposition pathway  Predicted and experimental heat of CO 2 absorption in 5 wt% NH 3 at T = 40°C. is mainly via the formation of ammonium bicarbonate (R2) and carbamate (R3) which is consistent with the conclusion drawn by Jilvero et al. (2012). Through the thermodynamic analysis of the speciation distribution in both lean and rich solvent at NH 3 concentrations 5 and 15% and lean loading of 0.2 to 0.5, it has been found that a large proportion of the CO 2 is absorbed through reactions with free ammonia. As shown in Figure 8, for lean and rich solvents, there are negligible differences with respect to the fractions of carbamate and the carbonate, while the level of free ammonia is decreasing and the amount of bicarbonate is increasing. This suggests that the main reaction pathway for the absorption of CO 2 was identified as the formation of bicarbonate through the reaction of ammonia with CO 2 . In other words, the decomposition of bicarbonate to ammonia and CO 2 is the main reaction for the CO 2 desorption. The heat of reaction required to desorb CO 2 of ammonia was similar to that required for MonoEthanolAmine (MEA). The simulation results show that the heat requirement of ammonia regeneration could be less than 2 500 kJ/kg CO 2 captured. (Jilvero et al., 2012). The main reason for the lower heat requirement for ammonia compared to MEA is that the thermodynamic properties of the NH 3 -CO 2 -H 2 O system allow for pressurized regeneration, which results in significantly lower water vaporization.
All investigations based on equilibrium-based thermodynamic process modeling are in agreement regarding the potential to achieve a regeneration heat requirement that ranges between 2 000 and 3 000 kJ/kg CO 2 for an aqueous ammonia process. The recent publication from Alstom The CO 2 -lean solution (a and c) and the CO 2 -rich solution (b and d) liquid composition at 15% ammonia (a and b) and 5% ammonia (c and d) are shown as a function of the lean CO 2 -loading. The figures exclude water and trace element, such as N 2 , OH -, H + , and CO 2 . The process is conducted at a temperature of 10°C and at atmospheric pressure. The figure is copied from Jilvero et al. (2012).
shows a heat requirement of 2 200 KJ/kg CO 2 captured for the chilled ammonia process (Jo¨nsson and Telikapalli, 2012). This result is in line with the modeling prediction. These recent studies show that the aqueous ammonia based CO 2 capture process requires much less heat for the solvent regeneration than MEA based capture processes (Jo¨nsson and Telikapalli, 2012;Jilvero et al., 2012;Valenti et al., 2012;Darde et al., 2012).
The pilot plant trials conducted by CSIRO and Delta Electricity at Delta Electricity's Munmorah power station showed a much higher heat requirement (Yu et al., 2011b). One reason for the high heat requirement is that a diluted ammonia solution was used and the amount of CO 2 captured/released is low. As a result, more than 50% of energy is used to heat up the solvent (sensible heat). The results also show the heat of desorption is around 70-80 kJ/mol, suggesting under the pilot plant conditions where CO 2 loading in stripper varied between 0.4 and 0.1, the major pathway is the formation of carbamate via the reaction of CO 2 with free ammonia.

Ammonia Loss
Ammonia is very volatile and characterised by high equilibrium pressure even in the CO 2 loaded aqueous ammonia solution. Figure 9 shows predicted partial and total pressure at the ammonia concentration of 28% and temperature of 8°C; Figure 10 predicted equilibrium CO 2 and NH 3 partial pressure as a function of CO 2 loading at the ammonia concentration of 5 wt% and the temperature of 10, 20 and 30°C; Figure 11 predicted equilibrium NH 3 and CO 2 partial pressure as a function of CO 2 loading at the ammonia concentration of 2.5 and 5% and the temperature of 10°C. The conditions for Figure 9 are relevant for the chilled ammonia process in which precipitation of products is taken into account while the conditions for Figures  10 and 11 are for processes at ambient temperatures, used by Powerspan, CSIRO and KIST at lower ammonia Predicted partial and total pressure at the ammonia concentration of 28 wt% and the temperature of 8°C (Darde et al., 2009).
CO2 loading Equilibrium NH 3 and CO 2 pressure (bar) NH 3 P at 10°C NH 3 P at 20°C NH 3 P at 30°C CO 2 P at 10°C CO 2 P at 20°C CO 2 P at 30°C Predicted equilibrium CO 2 and NH 3 partial pressure as a function of CO 2 loading at the ammonia concentration of 5 wt% and the temperature of 10°C. Aspen Plus Ò V7.3 with a built-in thermodynamic model for CO 2 -NH 3 -H 2 O was used to obtain equilibrium pressure in Figure 10 and Figure 11 (AspenTechnology, 2012; Que and Chen, 2011).
CO2 loading of solvent Equilibrium NH 3 and CO 2 pressure (bar) NH 3 P at 2.5wt% NH 3 NH 3 P at 5.0% NH 3 CO 2 P at 2.5% NH 3 CO 2 P at 5% NH 3 Predicted equilibrium NH 3 and CO 2 partial pressure as a function of CO 2 loading at the ammonia concentration of 2.5 and 5% and the temperature of 10°C.
concentration in which precipitation did not take place. It is clear in all cases, that NH 3 vapour pressures are significant at CO 2 loading below 0.5. As a result, ammonia will inevitably slip to flue gas in the absorber and to the CO 2 product stream from the stripper in the capture process and measures will have to be taken to recover the ammonia. The operational and capital costs for an ammonia recovery process will be determined by the amount of ammonia slip and the impact of the operational parameters. Despite the fact that it is widely acknowledged that ammonia loss is a big challenge to the aqueous ammonia based CO 2 capture process, there are limited studies on the ammonia loss rate in the literature.
For the chilled ammonia processes, the results available in the literature are all from the modeling work. Mathias et al. (2010) studied the effect of ammonia concentration in the gas phase at the outlet of absorber. They found out that the NH 3 slip from the absorber is only weakly dependent on the NH 3 concentration in the range of 15-26% and depends mainly on temperature. At the absorber temperature of 10°C and atmospheric pressure, the NH 3 concentration is approximately constant at 2 230 ppmv at ambient pressure, the ammonia concentration of 26% and CO 2 loading of ca. 0.4 in the lean solvent and reduced to 242 ppm if absorber temperature can drop to À1.1°C. The equilibrium NH 3 partial pressure for a solvent at the ammonia concentration of 28% and CO 2 loading of 0.4 and 8°C is more than 0.05 bar equivalent to 50 000 ppm at atmospheric pressure (Fig. 9), according to Darde et al. (2009). The large discrepancy between the two references is most likely indicative of the uncertainty around the thermodynamics of liquid-solid systems.
Versteeg and Rubin (2011) did a thorough analysis of effect of ammonia concentration (0-30%), CO 2 loading (0.25 to 0.67), absorber temperature (5 to 20°C) on ammonia slip in a chilled ammonia process in which rich solvent is not recycled to absorber. An increase in ammonia concentration and absorber temperature as well as a decrease in CO 2 loading can lead to an increase in ammonia slip. Darde et al. (2012) also modeled the chilled ammonia process at the low ammonia concentrations up to 12% and investigated the effect of ammonia concentration, CO 2 loading, absorber temperature and process configuration. In a configuration close to the configuration proposed by Alstom (Fig. 2), it has been found that with increase in ammonia concentration in the solvent from 4 to 11%, the ammonia concentration in the gas stream leaving the absorber varies between 4 500 and 19 000 ppm. In Darde's work, a murphy tray efficiency was introduced to account for deviation from the equilibrium. Budzianowski (2011), Niu et al. (2012 and Yu et al. (2012c) conducted experiments to investigate ammonia loss at low ammonia concentrations and relatively high absorber temperatures. All results from these studies observed the trends similar to the modeling studies described above, but the ammonia loss is much higher. For example, Niu et al. (2012) studies the absorption at the ammonia concentration of 1-11 wt%, the CO 2 loading of 0.12 and the temperature of 24°C and the ammonia concentrations at outlet were 1-14 vol%. Similar to research work by Niu et al., Yu et al. used a pilot scale research facility to investigate the effect of a number of parameters. The pilot plant trials show that the conditions which typically enhance CO 2 absorption rate such as an increase in ammonia concentration and absorption temperature and a decrease in CO 2 loading tend to increase ammonia loss significantly. Among many parameters which affect ammonia loss, temperature is one of the most sensitive parameters. As shown in Figure 12, with the increase in solvent temperature, CO 2 absorption rate in absorber 1 remains similarly but ammonia loss rate increases significantly. From the ammonia loss point of view, the absorption temperature should be as low as possible. However, it is expected that a considerable amount of additional energy is required to produce chilled water. Dave et al. (2009) simulated the aqueous ammonia based process by keeping the flue gas and lean solvent temperatures to the absorber at 10°C and solvent concentration at 2.5 and 5 wt%. Effect of temperature on the CO 2 absorption rate and NH 3 loss rate in absorber 1 under similar conditions. NH 3 wt% = 3.8-4.2; CO 2 loading in lean solvent = 0.22-0.25. Solvent flow-rate = 134 L/min; Gas flow-rate = 640-820 kg/h; CO 2 in the flue gas = 8-10 vol%. Inner diameter of absorber 1 = 0.6 m; packing height = 3.8 m (Yu et al., 2011a).
The electricity consumption is estimated to be more than 1 400 KJ/kg CO 2 .
In addition, the research by Budzianowski (2011) and Qi et al. (2013) suggested that the NH 3 desorption process is a fast process and controlled by the vapour liquid equilibrium which could guide future research on process modification to reduce ammonia slip in absorber.
Ammonia is also present in the CO 2 product and requires removal. Yu et al. (2012a) investigated the effect of operation conditions on ammonia concentration in the CO 2 product. To reduce ammonia concentration, the high pressure operation and low temperature in the over-head condenser will help reduce ammonia concentration in the CO 2 -product. However, this can cause solid precipitation in the stripper where solvent regeneration leads to generation of vapour which contains significant amount of ammonia and CO 2 gas apart from water vapour. When the temperature of vapour drops in the condenser/reflux line or even in stripper when the regeneration stops, part of vapour will condense, leading to formation of the solution/droplet in which ammonia concentration is very high (more than 10 wt%). Since partial pressure of CO 2 is high in stripper, this facilitates formation of a highly carbonated ammonia solution which reaches the ammonium bicarbonate solubility limit. The precipitation of solid will block the stripper condenser and reflux line, causing a shut-down of the plant. Blockage also occurs in instrument tubing, leading to false readings and affecting the operation. For ammonia processes which operate at higher ammonia concentrations, the blockage could also occur in the absorber packing.

Overall Performance
The aqueous ammonia based CO 2 capture process is quite complex as it involves flue gas cooling, CO 2 absorption and stripping, ammonia capture and recovery. The technology is still under development. So far there are no published results available from optimised pilot plant/demonstration plants, which prevent the rigorous economical assessment of the technologies. In the literature, the assessment of the overall performance was carried out using the equilibrium based process model with the estimated CO 2 absorption rates in the absorber.
Previous studies indicated significant improvements in performance over traditional amine technologies but the analysis was preliminary and based on limited understanding of the system (Bai and Yeh, 1997;Ciferno et al., 2005;Gal et al., 2011). As described above, the understanding of the system has improved significantly over the last few years. Based on improved understanding of the system thermodynamics model, Versteeg and Rubin (2011) modeled a chilled ammonia-based post-combustion CO 2 capture system processing flue gas from a supercritical coal-fired power plant. The estimated performance and cost were compared to an amine-based capture system. For the ammonia system the absorber CO 2 capture efficiency, NH 3 slip, and solids precipitation were evaluated for changes in lean solution NH 3 concentration, NH 3 /CO 2 ratio, and absorber temperature. For 90% CO 2 capture the levelized cost of electricity generation (annual revenue requirement) for the plant with ammonia-based capture was estimated at $US 105/MWh, which is comparable to the levelized cost of electricity generation for the plant with an amine-based capture system. The ammonia-based CO 2 system benefits from lower steam loads and reduced compressor power requirements, but the chilling loads and associated costs offset these benefits. Versteeg and Rubin (2011) suggest that the viability of ammonia-based CO 2 capture may be location dependent. A plant located in a characteristically cold climate, or with direct access to a large cooling sink such as a deep water lake, would have lower parasitic energy demands for the process chillers. Linnenberg et al. (2012) evaluated the impact of an ammonia-based post-combustion CO 2 capture process on a steam power plant. To study the influence of the cold end of the process, two power plants with different cooling water temperatures are analysed. Additionally, two different process configurations of the capture plant, with either one single absorber or two absorbers connected in series where the first absorber captures the majority of the CO 2 and the second limits the NH 3 slip, are evaluated. The study shows that the configuration of the process with absorption at low temperature (approximately 10°C) with or without precipitation of ammonium bicarbonate compounds leads to a lower net efficiency penalty than an MEA-based process, assuming that low temperature cooling water is available. An estimate of the size of the absorber shows that the absorber columns of an ammonia-based process are significantly higher than the ones required for an MEA-based process. The conclusions are consistent with those by Versteeg and Rubin (2011).

OUTLOOK
As highlighted by Zhuang et al. (2012), three inherent hurdles with the aqueous ammonia based capture technology have not been fully overcome yet: low CO 2 absorption rate resulting in large absorption equipment; high ammonia vapour pressure leading to ammonia slip into the cleaned flue gas, and ammonium bicarbonate solid formation in the absorption tower (in stripper condenser as well) which may plug the tower packing, stripper condenser and reflux line. Further research efforts should be focused on these areas in order to improve the technical and economic feasibility of the technologies. Recent research has shown that introduction of additives can enhance CO 2 absorption rate or suppress ammonia vapour (Ahn et al., 2012;Gal et al., 2011;Salentinig et al., 2012;Yu et al., 2012b). For example, a PZ-promoted ammonia solution, which has been patented by Alstom (Gal et al., 2011), is a highly effective promoter of CO 2 absorption in ammonia, as shown in Figure 6. The study by Yu et al. (2012b) has shown that alternative additives, which are environmentally friendly, have the potential to significantly increase the CO 2 mass transfer coefficients to a level comparable with MEA at high CO 2 loadings relevant to industrial applications. However, it is not clear about possible side effect of introduction of additives. New processes such as high pressure absorption, rich solvent split, re-engineering of the industrial ammonium bicarbonate fertilizer production process and combined capture of CO 2 , SO 2 and other impurities could offer potential for improvement (Yu et al., 2011a;Zhuang et al., 2012). All these research efforts are the early stage of the development.

CONCLUSIONS
The aqueous ammonia based post combustion CO 2 capture processes have gained intensive attention over the last few years thanks to many advantages over amine based solvents. A number of pilot plants and demonstration plants have been constructed and used to assess the technical and economical feasibility of the processes and identify opportunities to further advance the technologies. This paper has reviewed the research work available in the public domain including understanding of the thermo-chemical and physical properties of the CO 2 -NH 3 -H 2 O system, and absorption chemistry/kinetics involved during the absorption, information on absorption rate, ammonia loss, regeneration energies, and practical issues, and assessment of the technical and economical feasibility of the aqueous ammonia based post combustion capture technologies.
Thermo-chemical properties for the CO 2 -NH 3 -H 2 O system have been well understood and an extended UNI-QUAC model, described by Thomsen and Rasmussen (1999) and an electrolyte NRTL activity coefficient model by Aspentech (Que and Chen, 2011) can satisfactorily represent the thermodynamic properties of the CO 2 -NH 3 -H 2 O system over a wide range. A comprehen-sive reaction mechanisms involved in the reaction of CO 2 with ammonia in the aqueous medium has been developed by Wang et al. (2011).
Bench scale and pilot plant scale investigations show that aqueous ammonia absorbs CO 2 at a slower rate than the standard MEA solution under their respective operational conditions. Process modeling and experimental results both indicate that ammonia losses can be substantial. All investigations based on equilibrium-based thermodynamic process modeling are in agreement regarding a heat requirement that ranges between 2 000 and 3 000 kJ/kg CO 2 . The recent publication from Alstom shows a regeneration heat requirement of 2 200 kJ/kg CO 2 captured for the chilled ammonia process. The assessment of the overall performance using the equilibrium based process models shows that the chilled ammonia process leads to a lower net efficiency penalty than an MEA-based process, assuming that low temperature cooling water is available. An estimate of the size of the absorber shows that the absorber columns of an ammonia-based process will be significantly higher than the ones required for an MEA-based process.