Avalable onlne at www.scencedrect.com Energy Proceda 4 (011) 179 186 Energy Proceda 00 (010) 000 000 Energy Proceda www.elsever.com/locate/xxx www.elsever.com/locate/proceda GHGT10 Solublty of Carbon Doxde n Aqueous.5 M of Dethylenetramne (DETA) Soluton Hartono, A. 1*, Hoff,. A., Mejdell, T., and Svendsen, H. F. 1 1 Department of Chemcal Engneerng, Norwegan Unversty of Scence and Technology (NTNU), Trondhem, Norway, Sem Saeland Ve 4, NO7045 SINTEF Materal and Chemstry, Trondhem, Norway Elsever use only: Receved date here; revsed date here; accepted date here Abstract Thermodynamc modelng of the solublty of carbon doxde ( ) vapor/lqud equlbrum nto aqueous solutons of.5 M dethylenetramne (DETA) was developed. The DeshmukhMater (DM) model was mplemented to ths system for predctng the partal pressures aganst loadng for dfferent temperatures. A two step procedure, n general, s needed to ft the parameters n the DM model. However, n ths work only the frst step (the DebyeHückel part) was executed because, wth the complexty of the DETA system, the avalable expermental data materal was not large and vared enough to mert gong nto ths procedure. The model fts the partal pressure data reasonably well up to loadng of about 1.0 whle the predcted sothermal heat of absorpton based on the equlbrum agree reasonably well for loadngs n the range 0.6 to 1.. The fttng clearly ndcated that the regresson parameters need to be further mproved and that bnary nteracton parameters should be taken nto account. c 011 010 Elsever Publshed Ltd. byall Elsever rghts reserved Ltd. Open access under CC BYNCND lcense. eywords: VLE; NMR; DETA; DeshmukhMater; Specaton. * Correspondng Author. Telephone: (47) 73550338, Fax: (47) 735904080 Emal address: hartono@nt.ntnu.no, do:10.1016/j.egypro.011.01.039
180 A. Author Hartono name et/ al. Energy / Energy Proceda Proceda 00 (010) 4 (011) 000 000 179 186 1. Introducton A potental alternatve solvent for capture should have hgh cyclc capacty, fast absorpton rate, low heat of absorpton, and a hgh equlbrum temperature senstvty. Dethylenetramne (DETA), a polyamne, has showed to be a possble absorbent for capture,.e.: has hgher mass transfer rate [1], cyclc capacty [], and sgnfcant lower heat of absorpton [3] than the monoamne monoethanolamne (MEA). However, on the negatve sde DETA has been shown to degrade rapdly under certan condtons [4]. Determnng absorpton equlbrum nvolves the measurement of correspondng values of lqud loadng and gas phase partal pressure at dfferent temperatures. The equlbrum data for.5 M DETA were reported [] and were measured n two apparatuses, one for low temperatures (up to 80 C) and pressure (up to atmospherc) condtons typcal for absorber condtons, and one for hgh temperatures (up to 130 C) and medum pressures (up to 700 kpa) for desorber condtons. The expermental results were obtaned for loadngs up to 1.4 and for partal pressures of rangng from 0.01 450 kpa. Modelng the equlbrum data, furthermore, provdes more nformaton relatng to the cyclc capacty, temperature senstvty, lqud specaton and predcton of the heat of absorpton respectvely. The DETAH O system has many speces due to the three amne functonaltes. It was reported [5] that a total of speces mght exst n the system as shown n Fg 1. Fgure 1. Specaton n the DETAH O system The absorpton of nto aqueous solutons of DETA s governed by physcal absorpton, whch can be represented as a Henry s law constant, and a seres of chemcal reactons correspondng to the chemcal equlbrum for all speces s summarzed n Table 1. These can be wrtten n terms of actvty coeffcents ( ) and mole fractons ( x ) as: j a ( x) j 1,,3,... R (1) Where vj s the stochometrc coeffcent for component n reacton j and R s the number of chemcal reactons.
A. Hartono Author et name al. / Energy Proceda00 4 (011) (010) 000 000 179 186 181 3 Table 1. The physcal and chemcal equlbrum reactons n the DETAH O System No Reacton Remark 1 H g Physcal absorpton of nto lqud phase w HO H3O OH Dssocaton of water 3 HO H 3 H3O Dssocaton of carbon doxde 4 3 H 3 3 H3O H H O Dssocaton of bcarbonate on 5 HO 3O (p) Dssocaton of protonated DETA 6 (pp) HO H3O (p) (pp) Dssocaton of dprotonated DETA 7 8 9 3 3 3 3 HO H3O (pp) In order to fully determne the speces n the DETAH O system from the equatons of chemcal reacton equlbrum, addtonal equatons are requred.e.: mass balances for DETA,, total mass, and electroneutralty.. Thermodynamc framework At phase equlbrum, the sofugacty crteron holds between vapour and lqud. The equlbrum for volatle solutes (.e. ) and for the solvent (water and possbly amne) was calculated based the followng Eqs. and 3: ( PP ) y exp w P x H RT () ( PP ) y exp w w wwp xwwpw w RT (3) To calculate the fugacty coeffcents ( ), the PengRobnson equaton of state [6] was mplemented. 3. Actvty Coeffcent Model Dssocaton of trprotonated DETA DETA (p) DETA HO DETA H3O (p) DETA (s) DETA HO DETA H3O (s) 10 DETA (pp) DETA (p) HO DETA H (pp) 3O 11 DETA (ps) DETA (s) HO DETA H (ps) 3O 1 (p) (p) HO DETA H3O (p) (p) (p) 13 (s) (p) HO DETA H3O (p) (s) (p) 14 (p) (s) HO DETA H3O (s) (p) (s) 15 16 17 18 19 (ps) (p) H O (p) H3O (p) (ps) (p) (pp) (s) H O (p) H3O (s) (pp) (s) ( ) ( ) (s) (pp) HO DETA H3O (pp) (s) (ps) (p) ( ) (ps) H O DETA (ps) H O 3 (p) (pp) Formaton of carbamates on prmary amne group Formaton of carbamates on secondary amne group Formaton of dcarbamates on prmary amne group Formaton of dcarbamates on secondary amne group Dssocaton of protonated carbamates on prmary amne group Dssocaton of protonated carbamates on secondary amne group Dssocaton of protonated carbamates on secondary amne group Dssocaton of dprotonated carbamates on prmary amne group Dssocaton of dprotonated carbamates on secondary amne group Dssocaton of protonated dcarbamates on prmary amne group Dssocaton of protonated dcarbamates on secondary amne group DETA DETA DETA( ) 3 3 3 DETA( ) H O 3 3 Formaton of trcarbamate (pp) (ps) The DeshmukhMather model [7] was chosen to study the specaton of ths system and can be formulated as:.303a z I ln (4) jm j 1Ba I jw
418 A. Author Hartono name et/ al. Energy / Energy Proceda Proceda 00 (010) 4 (011) 000 000 179 186 The frst term expresses the DebyeHückel lmtng law [8] and represents the electrolyte long range forces. The second term takes nto account shortrange Van der Waals forces. The onc strength ( I) s calculated wth the followng equaton as: 1 I mz (5) where m and z represent the molalty and the charge number of the on. In the DeshmukhMather model, only water s consdered to be a solvent and not the amne. A transformaton between molalty base and molarty base s requred to fulfll the consstency n the equlbrum constant. 4. Thermodynamc parameters The thermodynamc parameters needed for the model are all the chemcal equlbrum constants ( j ) Henry s law constant for n pure water ( H ), the parameters n the fugacty coeffcent ( ) (, H O) and the parameters n the actvty coeffcent ( ) all the chemcal equlbrum constants ( ) j, the for vapor phase for all solute speces. The temperature dependency for and the Henry s constant ( H ) can be expressed as: ln c c / T c lnt c T (6) 1 3 4 Among the 18 chemcal equlbrum constants, 4 equlbrum constants were taken from the lterature [9, 10, 11] (reactons 1,, 3 and 4), 3 equlbrum constants were determned separately expermentally (reacton 5, 6 and 7) and the rest of the equlbrum constants were determned durng the fttng procedure based on the partal pressures of at dfferent loadngs and temperatures gven by the VLE measurement for the DETAH O system. 5. Parameter regresson In the DeshmukhMather model, the parameters were regressed n two steps [7, 1, 13],.e. frst the temperature dependences of the equlbrum constants for the reactons Eqs. 8 to 19 were determned whle all ( j) were set to zero. The second step was to determne the ( j) parameters, whch showed sgnfcant senstvty, whle keepng the equlbrum constants obtaned from the frst step. The regresson was performed by usng Modft [14], an nhouse MATLAB computer program for parameter estmaton. The regresson method used was the LevenbergMarquardt mnmzaton wth the normalzed objectve functon as gven by exp n cal p p F exp p (7)
A. Hartono Author et name al. / Energy Proceda00 4 (011) (010) 000 000 179 186 183 5 6. Modelng the VLE data 6.1. Determnaton of the dssocaton constant of DETA (pa) DETA s a tramne wth two prmary and one secondary amne group. It wll, therefore, have three equlbrum ponts correspondng to the formaton of mono, d and trprotonated DETA (reacton 5, 6 and 7). Potentometrc ttraton wth a glass calomel electrode was used and the pa values were determned from the observed end ponts. Fg. shows the ttraton curves where only two end ponts are clearly vsble. The reason for ths behavor could be explaned by the symmetrcal structure of DETA [5]. By havng a symmetrcal structures of two prmary amne groups, the frst two end ponts wll be close to each other. The frst and second dssocaton constants are assocated wth the frst end pont (prmary amne groups) and the thrd dssocaton constant s assocated wth the second end pont (secondary amne group). 14 1 10 8 ph 6 4 0.1M 0.5M 1.0M.5M 0 0 0.5 0.5 0.75 1 1.5 1.5 H SO 4 / Soluton (cm 3 g 1 ) Fg.. Ttraton curves for determnaton the dssocaton constants of DETA for dfferent concentratons and temperatures (,0.1 M ;, 0.5 M ;, 1.0 M;.5 M; Blue, 0C; Red, 30C; Green, 40C; Black, 50C, Brown, 60C). To calculate the frst end pont, t was assumed that the frst clear end pont seen n Fg. was a combnaton of the two prmary amne groups. Ths s reasonable when judgng from the acd amount needed for the secondary amne group. The pa values for the dfferent concentratons were evaluated and the values at nfnte dluton were determned by extrapolatng to zero concentraton and can be expressed as temperature functons as seen n Table. Table. Thermodynamc coeffcent for the protonaton of DETA at nfnte dluton Reacton Parameter c 1 c c 3 c 4 T (C) 5 6 (pp) 7 3 3 4.815 5510.0 0 0 060 3.0355 5710.9 0 0 060 0.0699 3509.4 0 0 060
6184 A. Author Hartono name et/ al. Energy / Energy Proceda Proceda 00 (010) 4 (011) 000 000 179 186 6.. VLE data regresson The equlbrum constants n Eqs. 8 to 19 can not be determned ndvdually from the experments. The values were determned durng the fttng of the VLE equlbrum data usng Eqs. to 4 and the bnary nteracton parameters ( ) were set to zero. Each equlbrum constant represents one parameter to be estmated. However, the j equlbrum constants are temperature dependent and ths mght be represented as frst order, second or even hgher order functons of temperature, as gven n Eq. 6. Takng nto account the temperature dependences ncreases dramatcally the number of parameters to be determned. The ntal step was done to regress the parameters to each partal pressure sotherm (from 40 C to 10 C). The ntal regresson parameters were then brought to ft wth the VLE expermental data of partal pressure of aganst loadng. The regresson adjustment had to be done n steps to mprove the ft and was off course tme consumng. The regresson parameters obtaned are tabulated n Table 3. As seen n Table 3, 1 regressed equlbrum constants were proposed, and wth a frst order temperature dependency ths lead to 4 coeffcents to be determned. Ths gves a very large number of possbltes to offer a god ft to the expermental VLE data. Table 3. The ftted parameters for the equlbrum constants n the DETAH O system Reacton Parameter c 1 c c 3 c 4 T (C) 8 DETA (p) 31.0591 8689.9 0 0 4010 9 DETA (s) 6.573 1993.5 0 0 4010 10 DETA (pp) 3.85 8360. 0 0 4010 11 DETA (ps) 38.114 786.5 0 0 4010 1 (p) (p) 13 (s) (p) 14 (p) (s) 15 (ps) (p) 16 (pp) (s) 17 (s) ( ) (pp) 18 (p) ( ) (ps) 19 DETA( ) 3 3 53.54 13833.7 0 0 4010 38.114 906.5 0 0 4010 9.8 407.1 0 0 4010 3.386 6053.5 0 0 4010 6.601 969. 0 0 4010 5.31 41.6 0 0 4010.679 518. 0 0 4010 11.754 496.0 0 0 4010
A. Hartono Author et name al. / Energy Proceda00 4 (011) (010) 000 000 179 186 185 7 The expermental VLE data gathered [] at dfferent temperatures were plotted n Fg. 3 together wth the model predctons usng the coeffcents n Tables 3. The model represents the expermental data reasonably well up to a loadng of about 1.0, however at loadngs above 1.0 the dscrepances are large. Between 1.0 and 1. n loadng the model seems to under predct the partal pressures, whereas for loadngs hgher than 1. the model overpredcts the pressures. The fttng clearly ndcates that the regresson parameters need to be further mproved and that bnary nteracton parameters ( ) may have to be taken nto account. j 10 6 10 5 10 4 p (Pa) 10 3 10 40C 10 1 60C 80C 100C 10 0 10C 5C 10 1 0 0. 0.4 0.6 0.8 1 1. 1.4 1.6 (mol /mol DETA) Fgure 3. Representaton of data by model based on the ftted parameters (Sold lnes, model; Blue dashed lne, extrapolated curve for 5C) The GbbsHelmholtz equaton was mplemented to estmate approxmately the dfferental heat of absorpton from the VLE model as: ln p H Abs (1 / T) (8) R x The estmated value for the dfferental heat absorpton of aqueous.5 M DETA soluton can be seen n Fg. 4. It s seen that the estmated value of the heat of absorpton n.5 M DETA s far too hgh n the low loadng range up to a loadng of about 0.6. The agreement s reasonable for loadngs between 0.6 and 1.. Ths can be explaned from the VLE experment data, only few ponts were measured for lower loadng. For loadngs hgher 1., the estmated values are seen to be underpredcted due to the overpredcton n the partal pressure of. In order to mprove the ft, the expermental heat of absorpton data could be used to estmate the overall temperature senstvty. H Abs (kj/mol ) 150 140 130 10 110 100 90 80 70 40 o C 80 o C 10 o C VLE data 60 50 0 0. 0.4 0.6 0.8 1 1. 1.4 1.6 Loadng (mol / mol DETA) Fg. 4. The estmated value for the absorpton heat of.5 M DETA from the VLE model.
8186 A. Author Hartono name et/ al. Energy / Energy Proceda Proceda 00 (010) 4 (011) 000 000 179 186 7. Conclusons The dssocaton constant of DETA was determned expermentally and t was assumed that the frst clear end pont seen was a combnaton of the two prmary amne groups. The equlbrum constants whch can not be determned ndvdually from the experments were obtaned from a fttng procedure to all partal pressure data. Only the DebyeHückel part of the DesmukhMather model was mplemented n the VLE model of.5 M DETA. The model represents the expermental data reasonably well up to a loadng of about 1.0 and the estmated value for the dfferental heats absorpton was calculated from the GbbsHelmholtz equaton. The fttng clearly ndcates that the regresson parameters need to be further mproved and that bnary nteracton parameters also should be taken nto account. Acknowledgement The fnancal support from the CCERT project s greatly apprecated. The CCERT project s supported by the Research Councl of Norway (NFR 18607), Shell Technology Norway AS, Metso Automaton, Det Norske Vertas AS, and Statol AS References [1] Hartono, A., da Slva, E. F., and Svendsen, H. F., netc of carbon doxde absorpton n aqueous soluton of dethylenetramne (DETA), Chem, Eng, Sc., 009, 64, 305313. [] Hoff,. A., Svendsen, H. F., Mejdell, T., Characterzaton and selecton of solvents, CASTOR Report STF80MIF058, SINTEF Materal and Chemstry, Trondhem, 005. [3] m, I., Heat of reacton and VLE of post combuston absorbent, PhD Thess, Norwegan Unversty of Scence and Technology, Trondhem, Norway, 009. [4] Davs, J., Amne Thermal Degradaton, Lumnant Carbon Management Program; Research Revew Meetng, November 01, Washngton DC, USA, 008. [5] Hartono, A., da Slva, E. F., Grasdalen, H., Svendsen. H. F., Qualtatve Determnaton of Speces n DETAH O System Usng 13 C NMR Spectra, Ind. Eng. Chem. Res., 007, 46(1), 4954. [6] Peng, D. Y., and Robnson, D. B., A New Two Constant Equaton of State, Ind. Eng. Chem. Fundam., 1976, 15, 5964. [7] Deshmukh, R. D., and Mather, A. E., A Mathematcal model for equlbrum solublty of hydrogen sulfde and carbon doxde n aqueous alkanolamne soluton, Chem. Eng. Sc., 1981, 36, 35536. [8] Debye, P., and Hückel, E., Zur theore der electrolyte, Physk. Z., 193, 4, 9, 18506. [9] Chen, C. C., Brtt, H. I., Boston, J. F., Evans, L. B., Extenson and Applcaton of the Ptzer equaton for vapor lqud equlbrum of aqueous electrolyte system wth molecular solutes, AIChE Journal, 1979, 5, 80831. [10] Olofsson, G., Hepler, L. G., Thermodynamc of Ionzaton of Water over Wde Ranges of Temperature and Pressure, J. Soluton Chem., 1975, 4 (), 17143. [11] Edward, T. J., Maurer, G., Newman, J., and Prausntz, J. M., Vaporlqud Equlbra n Multcomponent Aqueous Soluton of Volatle Weak Electrolyte, AIChe Journal, 1978, 4, 966976. [1] Weland, R. H., Chakravarty, T., Mather, A. E., Solublty of Carbon Doxde and Hydrogen Sulfde n Aqueous Alkanolamnes. Ind. Eng. Chem. Res., 1993, 3, 14191430. [13] Ma mun, S., Jakobsen, J. P., Svendsen, H. F., and Julussen, O., Expermental and modellng study of the solublty of n aqueous 30 mass% (amnoethylamno)ethanol soluton, Ind. Eng. Chem. Res., 006, 45(8), 50551. [14] Hertzberg, T. and Mejdell, T., MODFIT for Matlab: Parameter Estmaton n a General Nonlnear Multresponse Model, 1998.