1. VLE Modeling of Aqueous Solutions of Unloaded and Loaded Hydroxides of Lithium, Sodium and Potassium Shahla Gondal, Muhammad Usman, Juliana G.M.S. Monteiro, Hallvard F. Svendsen, Hanna Knuutila 8th Trondheim Conference on CO 2 Capture, Transport and Storage (TCCS-8) 16 - 18 June 2015

2. 2 Contents  Introduction  VLE and Apparent Henry’s Law Constant Modeling  Electrolyte-Non Random Two Liquid (e-NRTL) Model  Parameter fitting in the (e-NRTL) Model  Experimental data used for modeling  Experimental data used for Equilibrium modeling of Li +  Experimental data used for Equilibrium modeling of Na +  Experimental data used for Equilibrium modeling of K +  The Equilibrium Model  Constants used in the Model  Results  Parity plots  Summary of the Statistics of Results  Conclusions

3. 3 Introduction  The process of absorption of carbon dioxide (CO 2 ) into aqueous hydroxide and carbonate (loaded hydroxide) solutions has regained great interest during the last decade;  Firstly, the reaction between carbon dioxide and hydroxide ions resulting in production of bicarbonate and carbonate is of special interest as it occurs in all alkaline solutions  Secondly, these solutions do not degrade and are environment friendly as compared to organic solvents used for carbon capture  Promotion of bicarbonate formation, by e.g. carbonic anhydrase can make these systems more reactive

4. 4 VLE and Apparent Henry’s Law Constant Modeling  For the designing of an absorption column and/or stripper in the CO 2 capture system, we need to predict;  The composition of vapor and liquid phases in the columns  The temperature and pressure profiles in the columns  Energy requirements for stripping  An equilibrium model gives a reasonable representation of the system behavior  The equilibrium model needs modeling of both the Vapor-Liquid- Equilibrium (VLE) and the Henry’s Law constant  In this work, experimental data for VLE and the Henry’s law constant are regressed simultaneously  The activities calculated by using this model would be consistent with the Henry’s law constant

5. 5 The e-NRTL (Electrolyte-Non Random Two Liquid ) Model  The predictive equilibrium model must include corrections for non-idealities in both liquid and vapor phases  Accurate calculation of activities of involved species over a wide range of temperatures, pressures and concentrations are required  The e-NRTL model provides a general framework with which experimental data of electrolyte systems can be satisfactorily represented with binary parameters only  The e-NRTL model has been used successfully to model many important industrial electrolyte systems, among which are the hot carbonate CO 2 removal system, the sour water stripper system, and flue gas desulfurization

6. 6 Parameter fitting in the e-NRTL Model

7. 7 Reference *Conc. as LiOH [wt. % ] *Loading [mol CO 2 /mol Li + ] Temp. [°C] No. of data points (Aseyev, 1999) and This study (Ebulliometric data)Aseyev, 19990.58 – 462.40.25 – 1000 – 15043 (Walker et al., 1927)Walker et al., 19270.03 – 0.040.013–0.960.51 – 0.9225 – 3727 (Gondal et al., 2014)Gondal et al., 20144.25 – 20.20.24 –4.66025 – 8042 Total0.03 – 462.40.013 – 100 – 0.920 – 150112 Experimental data used for equilibrium modeling of Li + * The concentrations of Li 2 CO 3 solutions are recalculated as LiOH solutions with 0.5 loading [mol CO 2 /mol Li + ]. **The physical solubility for CO 2 was calculated from N 2 O solubility data by using N 2 O analogy.

8. 8 Experimental data used for equilibrium modeling of Na + Reference *Conc. as NaOH [wt. % ] *Loading [mol CO 2 /mol Na + ] Temp. [°C] No. of data points (Don and Robert, 2008), (Knuutila et al., 2010a), (Don and Robert, 2008) and (Taylor, 1955)Don and Robert, 2008Knuutila et al., 2010aDon and Robert, 2008Taylor, 1955 0.587 – 190.654.8 – 37.5020 – 105169 (Walker et al., 1927), (Hertz et al., 1970), (Mai and Babb, 1955), (Ellis, 1959) and (Knuutila et al., 2010a)Walker et al., 1927Hertz et al., 1970Mai and Babb, 1955Ellis, 1959Knuutila et al., 2010a 0.031 – 108.90.02 – 9.80.55 – 0.9820 –197165 (Rumpf et al., 1998) and (Lucile et al., 2012)Rumpf et al., 1998Lucile et al., 201212.7– 101633.69 – 3.840 – 2.1120 – 160102 (Gao et al., 1997), (Wong et al., 2005) and (Han et al., 2011)Gao et al., 1997Wong et al., 2005Han et al., 2011 100 – 576000.2 – 4.21.04 – 10.285 – 130148 (Knuutila et al., 2010b) and (Gondal et al., 2015)Knuutila et al., 2010bGondal et al., 20154.29– 75.560.4 – 16.50 - 0.525 – 8062 Total0.031 – 576000.02 – 37.50 – 10.215 – 197647 * The concentrations of Na 2 CO 3 solutions are recalculated as NaOH solutions with 0.5 loading [mol CO 2 /mol Na + ] and those of NaHCO 3 solutions are recalculated as NaOH solutions with 1 loading [mol CO 2 /mol Na + ]. **The physical solubility for CO 2 was calculated from N 2 O solubility data by using N 2 O analogy.

9. Reference *Conc. as KOH [wt. % ] *Loading [mol CO 2 /mol K + ] Temp. [°C] No. of data points (Pérez-Salado Kamps et al., 2007)Pérez-Salado Kamps et al., 2007267.2 – 92374.61 – 16.550.84 – 2.2940 – 12041 (Tosh et al., 1959)Tosh et al., 195923.86 – 979.117.34 – 37.20.5 – 0.8970 – 140148 (Walker et al., 1927), (Tosh et al., 1959), (Park et al., 1997) and (Jo et al., 2012)Walker et al., 1927Tosh et al., 1959Park et al., 1997Jo et al., 2012 0.03 – 22300.03 – 37.20.5 – 1.0125 – 120217 (Gondal et al., 2015) and (Knuutila et al., 2010b)Gondal et al., 2015Knuutila et al., 2010b4.2– 39.650.5 – 26.930 - 0.525 – 8043 Total0.03 – 92370.03 – 37.20 – 2.2925 – 140449 Experimental data used for equilibrium modeling of K + * The concentrations of K 2 CO 3 solutions are recalculated as KOH solutions with 0.5 loading [mol CO 2 /mol K + ]. **The physical solubility for CO 2 was calculated from N 2 O solubility data by using N 2 O analogy.

10. 10 The Equilibrium Model

11. 11

12. 12

13. 13 Constants used in the Model Reference ReactionABC 132.899-13445.9-22.4773(Edwards et al., 1978)Edwards et al., 1978 231.465-12092.1-36.7816(Edwards et al., 1978)Edwards et al., 1978 216.049-12431.7-35.4819(Edwards et al., 1978)Edwards et al., 1978 ABCDE 73.649-7258.2-7.30374.1653E-062(DIPPR, 2004)DIPPR, 2004 AB×10 -4 C×10 -6 D×10 -8 -6.83461.2817-3.76682.997(Carroll et al., 1991)Carroll et al., 1991 ABCD 492.03520.084942-18560.2-78.9292(Jou et al., 1992)Jou et al., 1992

14. Results This stu dy

15. 15 All data (647 data points) regression with 21.14% AARD Selected data (432 data points) regression with 4.13% AARD

16. All data (647 data points) regression with 29.53% AARD Selected data (432 data points) regression with 10.14% AARD

17. 17 All data (449 data points) regression with 7.92% AARD Selected data (354 data points) regression with 5.22% AARD

18. All data (449 data points) regression with 23.58% AARD Selected data (354 data points) regression with 19.31% AARD

19. LiOH with 1.76% AARD 354 selected data points for K + with 3.89% AARD 432 data points for Na + with 4.84% AARD

20. 20 CationsRegressed Property% AARD (Average Absolute Relative Deviation) All dataSelected data Li + 1.76- 5.75- 1.76 - Na + 21.144.13 29.5310.14 4.77 4.84 K+K+ 7.925.22 23.5819.31 4.773.89 Summary of the Statistics of Results

21. 21 Conclusions

22. Thank you ! Questions and Comments: shahla.gondal@ntnu.no