Fluid Phase Equilibria 325 (2012) 9099

Contents lists available at SciVerse ScienceDirect

Fluid Phase Equilibria

journal homepage: www.elsevier.com/locate/fluid

Quantitative analysis of the liquid phase by FT-IR spectroscopy in the system CO2 /diethanolamine (DEA)/H2 O
Farah Diab a,b , Elise Provost a, , Nicolas Lalou b , Pascal Alix b , Vincent Souchon b , Olivier Delpoux b , Walter Frst a
a b

UCP/ENSTA-ParisTech, 32 Bd Victor, 75015 Paris, France IFP Energies Nouvelles, Rond-point de lchangeur de Solaize. BP 3, 69360 Solaize, France

a r t i c l e

i n f o

a b s t r a c t
FT-IR spectroscopy was used to measure the liquid phase speciation in waterdiethanolamine (DEA)CO2 system, the CO2 loading varying from 0.04 to 0.9. The DEA mass fractions in the unloaded solution were 0.2, 0.3 and 0.4, and the temperature varied from 298.1 K to 333.1 K. The concentrations of carbamate and molecular CO2 have been experimentally determined and the other species concentrations (DEA, protonated DEA, bicarbonate) have been deduced from mass balances and electroneutrality relation. These values, combined with the measurement of CO2 partial pressure in equilibrium with the liquid phase, allow the determination of apparent CO2 Henry constant (HCO2 )DEAsolution original values. These results show that (HCO2 )DEAsolution is independent of the CO2 loading. Moreover, it is shown that the increase of temperature results in a decrease of (HCO2 )DEAsolution . 2012 Elsevier B.V. All rights reserved.

Article history: Received 28 October 2011 Received in revised form 5 April 2012 Accepted 21 April 2012 Available online 30 April 2012 Keywords: CO2 DEA Speciation FT-IR VLE Henry constant

1. Introduction Aqueous alkanolamine solutions are widely used for the capture of acid gases found in natural gas or industrial ue gas. To design and optimize the process, an accurate reactive mass transfer model is required. For this purpose, representation of the liquid phase speciation through a realistic thermodynamic model is crucial for the evaluation of the driving forces as well as for the determination of a precise kinetic model all along the absorption column. Generally, the estimation of thermodynamic parameters is based on solubility data, e.g. the acid gas partial pressure as a function of the loading ( is the acid gas mole number divided by the amine mole number). However, due to the complexity of the system with up to 9 species (in the case of primary or secondary amines), the database has to contain liquid phase speciation data to ensure a precise and reliable representation of the liquid phase. For the last few years, many efforts have been made to obtain and publish such data. For instance, Benamor and Aroua have reported carbamate concentrations in CO2 loaded diethanolamine (DEA) or methyldiethanolamine (MDEA) aqueous systems, using a titration method [1]. Nevertheless, the speciation determination is most often performed by spectroscopic methods. 1 H and 13 C NMR spectroscopy were the rst analytical methods used to char-

Corresponding author. Tel.: +33 1 45 52 54 70; fax: +33 1 45 52 83 22. E-mail address: elise.provost@ensta-paristech.fr (E. Provost). 0378-3812/$ see front matter 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.uid.2012.04.016

acterize the liquid phase. Thanks to 1 H or 13 C NMR measurements, Barth determined the ratio of the carbamate concentration to the concentration of carbonates and bicarbonates in aqueous solutions of diisopropanolamine (DIPA), DEA, monoethanolamine (MEA) and diglycolamine (DGA) [2]. The corresponding equilibrium constants related to the formation of carbamate were calculated. Suda et al. [3], Polpsteinova et al. [4], Bttinger et al. [5,6], Ermatchkov et al. [7] and Yang et al. [8] also used 13 C as well as 1 H NMR and observed different species: amine, protonated amine, bicarbonate, carbonate, carbamate and molecular CO2 . Quantitative results have been obtained, even though no difference can be made between the species that differ only by a proton (amine and protonated amine or carbonate and bicarbonate). Online analysis of the liquid phase was even achieved by Bttinger et al. [5,6] with a specically designed experimental device. Besides, in order to solve the issue of the quantication of the species which can not be distinguished by NMR spectroscopy, several authors turned to vibrationnal spectroscopy. Souchon et al. recently published speciation data determined by in situ Raman spectroscopy in the case of the H2 O/MDEA/CO2 system. Carbamate stability constants for systems containing primary or secondary amines were also calculated from these measurements [9]. However, infrared spectroscopy is the most widely used vibrational spectroscopic method to determine the liquid phase composition of CO2 loaded aqueous alkanolamine solutions. The rst study was carried out by Rogers et al. [10,11]. In its experimental device, both the liquid and the gaseous phases were analyzed by FT-IR,

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099

91

but no clear exploitation of FT-IR spectra was presented. Lichtfers and Rumpf [12] have carried out experiments in order to make a speciation on the waterNH3 CO2 system by infrared spectrometry. Derks et al. have also determined the speciation in the system CO2 /MDEA/H2 O with a CaF2 circulating cell [13] and considered the measurement of molecular CO2 concentration as an objective for further development. Jackson et al. have built an apparatus allowing the in situ analysis of the liquid phase, using a diamond ATR probe but only qualitative results are reported [14]. Sidi-Boumedine [15] have developed the same kind of experimental set-up as Rogers et al. [10] in which an external loop allows the IR analysis of the liquid phase for DEA and MDEA loaded solutions using an ATR accessory. The peak corresponding to molecular CO2 has been identied and the concentration of molecular CO2 and the carbamate evolution have been determined. Using the same apparatus, Archane et al. [16] have published speciation data for the H2 O/DEA/Methanol/CO2 system. These authors have also developed an original device using an ATR diamond accessory coupled to the FT-IR spectrometer in order to determine in situ the composition of the liquid phase together with VLE data. Similar speciation data were also published [17] (molecular CO2 concentration and carbamate qualitative evolution) for the system H2 O/DEA/PEG400/CO2 . In this study, the same apparatus as the one designed by Archane et al. [17] was used. IR spectral data have been treated in order to determine in situ the composition of the liquid phase in CO2 loaded aqueous solutions of DEA together with VLE measurements. VLE and liquid phase speciation measurements were performed at 298.1 K, 313.1 K and 333.1 K for DEA mass fractions ranging from 0.20 to 0.40. In the whole paper, the mass fraction of DEA, noted wDEA , is relative to unloaded DEA/H2 O mixture. In order to increase the database used for thermodynamic modeling, VLE data were also acquired at 298.1 K for wDEA fraction from 0.3 to 0.41. The aqueous solutions were loaded up to 0.9 mole CO2 per mole of DEA. 2. Experimental 2.1. Experimental set-up A detailed description of the experimental device is presented in the paper of Archane et al. [17]. The apparatus is presented in Fig. 1 and consists in a stainless steel double-jacketed cell (Vcell = 294.2 0.1 106 m3 ), equipped with a Pt100 temperature probe (precision 0.1 K) and a pressure sensor (Drck PTX 611, 01000 kPa, 1 kPa). The lid of the cell contains tubes and valves, for the introduction of the liquid and the gas. The temperature of the cell is monitored through the circulation of silicon oil in the double-jacket and the seal is overheated ( T = 3 K) to avoid any condensation phenomena. The liquid phase and the vapour phase are stirred by two four blades impellers. The bottom of the cell is drilled, and is xed to an Attenuated Total Reectance accessory (Durasample cell) put in the IR spectrometer (Bruker Tensor 27). It allows the in situ analysis of the liquid phase. As our aim is to determine the true concentration of all species, including molecular CO2 , interference with the atmospheric CO2 has to be avoided. This is obtained by a circulation of CO2 - and water-free air inside the spectrometer. 2.2. Experimental procedure CO2 (Air Liquide, 99.995%) was used and water was freshly distilled prior to the preparation of amine aqueous solutions. DEA (Sigma Aldrich, 99%), NaHCO3 (Sigma Aldrich, 99.5%) and aqueous hydrochloric acid (Sigma Aldrich, 36.538.0% mass) were used as received.

The solvents are individually degassed. For this purpose, water is freezed using liquid nitrogen and pumped under vacuum during 30 min, this procedure being repeated three times. DEA is heated at 323 K under vacuum during 5 h. The solutions are then prepared by weighing. The balance used in this work is the model XP 1203S, from Mettler Toledo. Typical weighed masses are 200 g for solvent (uncertainty = 0.001 g) and 1 g for gas (as the mass of the cylinder containing the CO2 is about 900 g, the uncertainty associated to the gas measurement is 0.005 g). At the beginning of each experiment, the cell is pumped under vacuum until a primary vacuum is obtained (P < 1 kPa). The solution containing DEA and water is then transferred inside the cell. When sat correspondthe target temperature T is reached, the pressure Pv ap ing to the solvent vapour pressure and the IR spectrum are both recorded. CO2 is then introduced in the reactor. Its amount is measured by weighing the gas vessel before and after the addition. The pressure inside the cell and the IR spectrum are monitored and the equilibrium is supposed to be reached when no signicant change is observed on both pressure and spectra within 1 h. After recording the spectrum and Ptot , a new addition of CO2 is realized to get measurements at higher loadings. 2.3. CO2 partial pressure and loading calculation The CO2 partial pressure PCO2 is estimated from the measured total pressure Ptot by applying the Raoults expression:
sat PCO2 = Ptot Pv ap (1 xCO )
2

app

(1)

where xCO is the apparent molar fraction of CO2 , which is the 2 sum of the molar fraction of molecular carbon dioxide, carbamate (DEACOO ), bicarbonate (HCO3 ) and carbonates (CO3 2 ). In Eq. sat is the vapour pressure of water. Indeed, (1), the main part of Pv ap the vapour pressure of DEA varies from 0.05 Pa at 298.15 K to 2.43 Pa at 333.15 K [18] and DEA can be considered as a non volatile solvent. We have conrmed in a previous work [16] that Eq. (1) leads to a precise estimation of PCO2 in a similar system. The difference between PCO2 estimated from Eq. (1) and PCO2 measured by gas chromatography turned out to be less than 0.5%. The loading is dened as the ratio between the absorbed CO2 mole number and the amine mole number in the liquid phase. According to Archane et al. assumptions and mass conservation considerations [16], is expressed as follows: = nCO2 ,in [PCO2 CO (VCell Vsolution )/RT ]
2

app

vap

nDEA,in

(2)

where nCO2 ,in is the total mole number of CO2 introduced, nDEA,in the DEA mole number introduced, VCell the volume of the cell, vap Vsolution the volume of the liquid phase, CO the fugacity coefcient of CO2 and R the ideal gas constant. CO has been calculated using the virial equation of state: CO = exp(B/RT ) with B the CO2 virial 2 second coefcient from van Huff [19]. The vapour phase correction implies a variation on the CO2 loading less than 0.4%, considering Vsolution /Vvapour 0.4, where Vvapour is the volume of the vapour phase. 3. Quantitative analysis of the liquid phase When CO2 is added to an aqueous solution of DEA, the following equilibrium reactions take place: Amine protonation : DEA + H3 O+ DEAH+ + H2 O CO2 + H2 O + H2 O HCO 3 + H3 O
+ +

vap

vap

(3) (4) (5)

Carbon dioxide hydrolysis : Bicarbonate hydrolysis :

HCO 3

CO2 3

+ H3 O

92

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099

Fig. 1. Experimental set-up for the acquisition of IR spectra.

DEA + HCO 3 DEACOO + H2 O

Carbamate formation :

(6)

Table 1 shows the species that can be found in the liquid phase and their characteristic IR vibrationnal bands. The spectra of a DEA aqueous solution loaded with increasing amounts of CO2 are shown in Fig. 2. With increasing loadings, there are signicant changes in the IR spectra in the 12001600 cm1 range, where peaks corresponding to DEACOO and HCO3 are present [16,17]. A peak corresponding to molecular CO2 is located at 2340 cm1 . Finally, CO3 2 is not observed and this species will thus be neglected as its concentration is very low, which is a common assumption in this kind of study [4]. 3.1. Assumptions The quantication of DEA, DEAH+ , DEACOO , HCO3 and CO2 in the liquid phase may be achieved with the following hypotheses: Hypothesis 1. The solution is supposed to be an ideal mixture from a spectroscopic point of view, which means that the species have the same spectroscopic features in the mixture as if they were

alone in aqueous solutions. As a consequence, the infrared spectrum of the mixture is considered as the sum of the infrared spectra of the lone species in aqueous solutions, balanced by their concentrations. Thus, the equation below can be written for each wave number :
= Itotal i

Ii

(7)

and Ii are the intensities of the infrared spectrum of the mixture and of the pure species i at wave number respectively. This hypothesis, assuming that spectral properties of individual species are not affected signicantly by ionic interactions generated by highly concentrated CO2 -loaded aqueous DEA solutions, is necessary in this study. This assumption was validated in the case of molecular CO2 , for which the dependence of its spectral features towards its chemical environment was found small, especially if compared to experimental deviations [16]. This hypothesis will be discussed in Section 4. Hypothesis 2. Ii is directly proportional to the concentration of pure species i and it has been checked in previous studies [15]. In the same way as above, the equation below can be written: Ii = ki Ci (8)

where Itotal

where ki is the response factor of species i at the wave number and Ci its concentration. By combining these two hypotheses it leads to the BeerLambert type equation:
Itotal = i

ki Ci

(9)

where the response factor ki is independent of the medium in which the species i is diluted according to the rst hypothesis. 3.2. Determination of the response factors (ki )
Fig. 2. IR spectrum evolution of a wDEA = 0.4 aqueous solution at 313.1 K, with increasing loadings.

DEA, DEAH+ , DEACOO , HCO3 , molecular CO2 and water are the main species in the liquid phase of CO2 loaded aqueous

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099 Table 1 Species that can be found is the system DEA/CO2 /H2 O and their IR vibrationnal bands. Species H2 O CO2 HCO3 max (cm1 ) 1625 2342 1300 1360 1625 Not observed 1334 1446 1460 1244 1410 1475 1530 Type H O H deformation d C O asym. streching as C OH deformation d Sym. C O stretching sy Asym. C O sterching as Asym. C O sterching as Asym. C O sterching as CH d + OH d CH d N H deformation d Sym. C O stretching Asym. C O sterching
sy Lit max (cm1 )

93

Source Nakamoto [20] Nakamoto [20] Nakamoto [20] Davis and Oliver [21] Kurtaran et al. [22] Lichtfers and Rumpfs [12] Lichtfers and Rumpfs [12] Frasko [23]

CO3 2 DEA DEAH DEACOO

+

1640 2349 1300 1355 1620 1380 1436 1350 1456 1430 1400 15251545

as

solutions. The response factors were evaluated for each temperature and for all the components. The following experiments were carried out: Water: The response factor of water is determined by dividing the absorbance of the spectrum of pure water by the concentration of water expressed in mol L1 at the given temperature. Molecular CO2 : The peak corresponding to the antisymmetric stretching of CO2 is located at 2340 cm1 where no other peak is observed. As a consequence, its intensity can be easily estimated. To determine CO2 response factor at 2340 cm1 , CO2 has been added to pure water, and the peak height above the baseline has been represented versus the CO2 concentration estimated from mass balance measurements (see Fig. 3). A straight 2340 . Molecular CO can thus line is observed, whose slope is kCO 2 2 be detected and measured using diamond accessories. It has to be pointed out that this determination has not been achieved in most of the papers devoted to speciation of CO2 -amines systems [414]. In this study, it has been observed that molecular CO2 can be measured for concentrations larger than 0.005 mol L1 . DEA, DEAH+ and HCO3 : Aqueous solutions of each of these species with known concentrations have been prepared and analyzed by IR spectroscopy. For DEAH+ solution, HCl has been added in excess compared to DEA (1.1 equivalent) in order to ensure that DEA is totally protonated. HCl was chosen since Cl is not spectroscopically active. The response factors have then been determined by linear regression by plotting the measured IR absorbance corrected from the contribution of water as a function of the concentration of the species of interest.

Carbamate: It is not possible to isolate DEACOO in an aqueous solution because of equilibrium 6. To evaluate its response factor, we have considered a CO2 /DEA system. In this case, addition of CO2 to pure DEA leads to carbamate formation with concomitant protonation of DEA. Formation of HCO3 does not occur and molecular CO2 concentration remains sufciently low to neglect this species. The introduced CO2 is then fully transformed into carbamate according to: 2DEA + CO2 = DEACOO + DEAH+ (10)

We have limited this calibration to low loadings ( < 0.2) since the concentration of the carbamate produced at a loading of 0.2 (2 M) is larger than the maximum carbamate concentration determined in this study (cf. Tables 16). On the spectra, molecular CO2 is not observed, which conrms that its concentration can be neglected regarding the concentration of DEACOO . Knowing the loading of the studied solution, DEACOO concentration is also known, as well as the concentration of DEA and DEAH+ when considering electroneutrality and total amine mass balance. The spectrum corresponding to pure DEACOO was then calculated by subtracting the contribution of DEA and DEAH+ to the overall IR spectrum, with the help of the response factors previously determined. The main band of the resulting spectrum is located at 1548 cm1 and corresponds to the antisymmetric stretching of the COO group. In Fig. 3 is presented the calibration line for = 1548 cm1 , which is used to determine the response factor of the carbamate species k1548 .
DEACOO

3.3. Determination of concentrations The calibration coefcients determined in water (Section 2.2) will then be used for the study of the system CO2 /DEA/H2 O.

Cmolecular CO2 / kmol.m -3

0 0.02 0.04 0.06 0.08 0.02

0.3

I molecular CO2

0.015

I DEACOO -

3.3.1. Determination of the molecular CO2 concentration Molecular CO2 being the only species with signicant absorbance at 2340 cm1 , its concentration is obtained by dividing 2340 . the intensity at 2340 cm1 by kCO
2

0.2 0.01

0.1

0.005

0 0 1 2 3

CDEACOO- / kmol.m -3
Fig. 3. IR absorbance of DEACOO at 1548 cm1 in CO2 -loaded DEA solutions ( ) and of molecular CO2 at 2340 cm1 in water ( ) according to their concentrations.

3.3.2. Determination of the carbamate concentration DEA, DEAH+ , water, HCO3 and DEACOO are the species with signicant contributions to the spectrum in the wave number range 12001600 cm1 . However, at 1530 cm1 , the main contributor to the IR spectrum is the carbamate species. Indeed, the growing peak upon the addition of CO2 in aqueous DEA solutions (between = 0 and = 0.5) is mainly due to the DEACOO antisymmetric stretching. It is slightly shifted to the shorter wave numbers compared to experiments carried out in pure DEA where the maximum peak intensity is observed at 1548 cm1 . For the following calculations,

94 Table 2 Experimental equilibrium data for wDEA = 0.201 at 298.1 K. CO2 loading 0.000 0.155 0.001 0.235 0.002 0.335 0.002 0.470 0.003 0.591 0.004 0.644 0.005 0.706 0.005 0.749 0.005 0.786 0.006 0.822 0.006 0.879 0.006 0.892 0.006 Ptot (kPa (1 kPa)) 3 3 3 4 6 12 19 33 50 76 91 153 174

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099

CDEACOO (kmol m3 ) 0.285 0.402 0.517 0.576 0.496 0.444 0.386 0.326 0.275 0.226 0.156 0.163 0.025 0.034 0.043 0.048 0.042 0.038 0.033 0.029 0.025 0.022 0.017 0.018

CCO2

molec

(kmol m3 )

DEAsolution HCO (kPa m3 kmol1 )

2

0.019 0.025 0.043 0.048

0.002 0.002 0.002 0.002

3842 3520 3465 3542

371 270 181 174

Table 3 Experimental equilibrium data for wDEA = 0.289 at 298.1 K. CO2 loading 0.000 0.121 0.001 0.221 0.002 0.329 0.002 0.426 0.003 0.519 0.004 0.596 0.004 0.659 0.005 0.735 0.005 0.800 0.006 0.846 0.006 0.868 0.006 Ptot (kPa (1 kPa)) 3 3 3 4 5 8 14 26 55 101 154 190 CDEACOO (kmol m3 ) 0.361 0.565 0.757 0.868 0.909 0.758 0.660 0.521 0.401 0.333 0.281 0.031 0.047 0.063 0.072 0.075 0.063 0.056 0.045 0.036 0.031 0.028 CCO2
molec

(kmol m3 )

DEAsolution HCO (kPa m3 kmol1 )

2

0.006 0.014 0.027 0.041 0.048

0.002 0.002 0.002 0.002 0.002

3833 3643 3731 3659 3875

1047 479 274 196 188

Table 4 Experimental equilibrium data for wDEA = 0.400 at 298.1 K. CO2 loading 0.000 0.135 0.001 0.215 0.002 0.326 0.002 0.428 0.003 0.543 0.004 0.630 0.004 0.700 0.005 0.739 0.005 0.762 0.005 0.787 0.006 Ptot (kPa (1 kPa)) 3 3 3 3 4 11 30 69 107 137 179 CDEACOO (kmol m3 ) 0.364 0.612 0.900 1.091 1.117 0.967 0.804 0.673 0.616 0.548 0.032 0.051 0.075 0.090 0.093 0.081 0.068 0.058 0.054 0.049 CCO2
molec

(kmol m3 )

DEAsolution HCO (kPa m3 kmol1 )

2

0.007 0.017 0.025 0.034 0.042

0.002 0.002 0.002 0.002 0.002

4500 3824 4333 3912 4268

1142 428 331 240 224

Table 5 Experimental equilibrium data for wDEA = 0.398 at 313.1 K. CO2 loading 0.000 0.057 0.000 0.104 0.001 0.156 0.001 0.210 0.001 0.266 0.002 0.313 0.002 0.365 0.003 0.420 0.003 0.459 0.003 0.513 0.004 0.550 0.004 0.574 0.004 0.596 0.004 0.626 0.004 0.656 0.005 0.683 0.005 0.720 0.005 0.745 0.005 0.774 0.005 0.800 0.006 Ptot (kPa (1 kPa)) 7 7 8 8 10 11 11 12 14 17 26 36 47 60 83 115 150 217 275 361 457 CDEACOO (kmol m3 ) 0.227 0.370 0.513 0.662 0.789 0.881 0.974 1.055 1.080 1.064 1.053 1.015 0.973 0.926 0.884 0.788 0.744 0.664 0.589 0.520 0.012 0.016 0.021 0.027 0.032 0.035 0.039 0.042 0.043 0.043 0.043 0.042 0.040 0.039 0.037 0.034 0.033 0.031 0.029 0.027 CCO2
molec

(kmol m3 )

DEAsolution HCO (kPa m3 kmol1 )

2

0.006 0.007 0.010 0.013 0.020 0.026 0.039 0.051 0.068 0.083

0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.003 0.003 0.003

5000 5714 5889 5846 5684 5720 5553 5255 5299 5434

1827 1823 1282 1001 652 529 344 259 206 182

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099 Table 6 Experimental equilibrium data for wDEA = 0.399 at 313.1 K. CO2 loading 0.000 0.034 0.000 0.085 0.001 0.642 0.004 0.668 0.005 0.679 0.005 0.700 0.005 0.733 0.005 0.754 0.005 0.772 0.005 0.802 0.006 Ptot (kPa (1 kPa)) 7 7 7 97 130 144 181 253 310 365 477 CDEACOO (kmol m3 ) 0.125 0.345 0.862 0.870 0.831 0.760 0.703 0.663 0.625 0.486 0.007 0.015 0.037 0.037 0.036 0.034 0.032 0.031 0.030 0.026 CCO2
molec

95

(kmol m3 )

DEAsolution HCO (kPa m3 kmol1 )

2

0.014 0.020 0.023 0.029 0.044 0.053 0.068 0.086

0.002 0.002 0.002 0.003 0.003 0.003 0.003 0.003

6500 6474 6227 6000 5721 5827 5358 5529

911 675 591 473 328 276 217 173

it was supposed that the response factor at 1530 cm1 for the carbamate species in the CO2 /DEA/H2 O system was the same as the one measured at 1548 cm1 in pure DEA. It corresponds indeed to the maximum peak intensity of the same vibration. At 1530 cm1 , where the main contribution is largely due to the DEACOO species and we can write:
1530 1530 IDEACOO = Itot j= / i 1548 Ij1530 = kDEACOO CDEACOO

1000
wDEA= 0.206 [24] wDEA= 0.206 [25] wDEA= 0.206 [26] wDEA= 0.206 [27] wDEA= 0.201 this work wDEA= 0.289 this work wDEA= 0.300 this work wDEA= 0.300 this work wDEA= 0.400 this work wDEA= 0.408 this work

100

(11)

PCO2 / kPa

10

The correction to the total intensity due to the other compounds can be the written as:
1530 1530 1530 1530 1530 +IDEAH + IH IDEA + +I O = kDEA CDEA HCO
3 2

1 0,2 0,4 0,6 0,8 1

1530 1530 1530 + kDEAH CHCO + kH + CDEAH+ + k O CH2 O HCO

3 3 2

(12)
0,1

CO2 loading

In the system, the following set of relations between the various concentrations is satised: Total CO2 mass balance :
0 CDEA = CDEACOO + CHCO
3

Fig. 4. Solubility of CO2 in aqueous DEA solutions at 298.1 K for various DEA mass fractions.

+ Cmolecular CO2
0 CDEA = CDEA + CDEAH+ + CDEACOO 0 CH 2O

(13)

DEA mass balance :

(14) (15)

xed loading, the CO2 solubility decreases as DEA mass fraction increases, the phenomenon being due to the salting out effect related to larger ionic concentrations. The evolution of the solubility with loading at various temperatures is presented in Fig. 5. As expected, the solubility decreases when the temperature increases. 4.2. Speciation data

Oxygen mass balance : Electroneutrality :

= CH2 O + CHCO
3 3

CDEACOO + CHCO = CDEAH+

(16) The DEACOO and molecular CO2 concentrations for the six experimental conditions under study with various wDEA and temperatures are shown in Tables 27. The evolutions of DEACOO and molecular CO2 concentrations with loading change are presented
600

As the response factors are all known, the combination of Eqs. (11) to (16) allows to determine the concentration of DEACOO from the measured IR absorbance at 1530 cm1 in CO2 -loaded aqueous DEA solutions for each temperature. Concentrations of the other species are then deduced from this result. 3.3.3. Precision The uncertainties of quantities deduced from measured intensities were estimated using the error propagation method. The peaks height (in arbitrary unit) is determined with a precision of 0.003. The uncertainty is reported for all the measurements in the corresponding tables. 4. Results and discussion 4.1. Solubility data The experimental solubility values are given in Tables 28. To ensure the accuracy of the experimental device, the measured CO2 solubility has been compared with the literature [2427] when available. Repeatability measurements have also been undertaken. As shown in Fig. 4, the values are self consistent and there is a good agreement with literature measurements. As usually observed, at

PCO2 / kPa

400

298.1 K wDEA=0.400 this work 300.0 K wDEA=0.408 this work 313.1 K wDEA=0.399 this work 313.1 K wDEA=0.398 this work 333.1 K wDEA=0.397 this work

200

0 0 0,2 0,4 0,6 0,8 1

CO2 loading
Fig. 5. Solubility of CO2 at various temperatures in DEA aqueous solutions for wDEA around 0.4.

96 Table 7 Experimental equilibrium data for wDEA = 0.397 at 333.1 K. CO2 loading 0.000 0.005 0.000 0.019 0.000 0.059 0.000 0.102 0.001 0.143 0.001 0.249 0.002 0.257 0.002 0.274 0.002 0.315 0.002 0.359 0.003 0.403 0.003 0.502 0.004 0.515 0.004 0.547 0.004 0.587 0.004 0.609 0.004 0.629 0.004 0.650 0.005 0.686 0.005 0.698 0.005 Ptot (kPa (1 kPa)) 17 17 17 17 18 19 21 22 23 25 29 30 74 83 114 168 210 252 306 428 470

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099

CDEACOO (kmol m3 ) 0.032 0.096 0.243 0.373 0.488 0.764 0.785 0.818 0.885 0.963 1.040 1.065 1.066 1.035 1.003 0.975 0.948 0.907 0.840 0.818 0.010 0.013 0.025 0.037 0.048 0.074 0.076 0.079 0.085 0.093 0.100 0.103 0.103 0.100 0.097 0.095 0.092 0.088 0.082 0.080

CCO2

molec

(kmol m3 )

DEAsolution HCO (kPa m3 kmol1 )

2

0.005 0.011 0.015 0.021 0.027 0.031 0.044 0.048

0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003

13,400 9700 10,857 9650 9038 9667 9364 9660

8067 2753 2123 1443 1064 999 710 670

Table 8 Evolution of the total pressure with loading for DEA aqueous solutions at 298.1 K for various DEA mass fractions (the uncertainty on Ptot is 1 kPa). wDEA = 0.300 CO2 loading (molCO2 /molamine ) 0.457 0.560 0.621 0.698 0.745 0.786 0.003 0.003 0.004 0.005 0.005 0.006 Ptot (kPa) 6 12 26 59 89 133 wDEA = 0.300 CO2 loading (molCO2 /molamine ) 0.538 0.626 0.705 0.757 0.810 0.870 0.896 0.003 0.004 0.005 0.006 0.006 0.006 0.006 Ptot (kPa) 12 29 61 105 170 300 389 wDEA = 0.408 CO2 loading (molCO2 /molamine ) 0.422 0.467 0.512 0.555 0.632 0.672 0.713 0.751 0.784 0.814 0.842 0.861 0.910 0.003 0.003 0.003 0.003 0.004 0.005 0.005 0.005 0.006 0.006 0.006 0.006 0.007 Ptot (kPa) 7 8 12 19 42 66 101 143 205 277 367 430 496

in Figs. 6 and 7. Concentrations of HCO3 , DEA, and DEAH+ are also plotted, even though they are not directly deduced from the spectroscopic measurement but from mass balance equations. Among the various speciation determinations relative to CO2 /amines/H2 O systems only the data of Bttinger [5] and Benamor et al. [1] can be compared to ours. The speciation data for the system CO2 DEAH2 O (wDEA = 0.20) at 298.1 K have been compared with the data of Bttinger et al. [5]. As it is clear in Fig. 6,
2.5
cDEA + cDEAH+ [5] cDEACOO- [5] cHCO3- [5] cDEA + cDEAH+ this work cDEACOO- this work cHCO3- this work

data sets obtained using very different methods, are consistent with one another, which tends to prove that the assumptions made for the determination of DEACOO concentration from IR measurements are consistent. However, it can be seen in Fig. 7 that the DEACOO concentrations measured by Benamor et al. [1] in the case of wDEA = 0.4 solutions at 313 K are larger than the values obtained in this work. This can be explained by the use of a chemical titration method which may lead to the overestimation

4.5
cDEACOO- [1] cDEACOO- wDEA=0.398 cHCO3- wDEA=0.398 cDEA wDEA=0.398 cDEAH+ wDEA=0.398 c molecular CO2 wDEA=0.398

0.09

C / kmol.m -3

1.5

C / kmol.m -3

0.06

1.5

0.03

0.5

0 0 0.2 0.4 0.6 0.8 1

0 0 0.2

CO 2 loading
Fig. 6. Comparison of the speciation data from Bttinger [5] (wDEA = 0.2, 293 K) and those obtained in this work (wDEA = 0.201, 298.1 K).

CO2 loading

0.4

0.6

0 0.8

Fig. 7. Liquid phase speciation for CO2 /DEA/H2 O systems at 313.1 K for wDEA = 0.398 and wDEA = 0.399.

C molecular CO 2 / kmol.m -3

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099

1.2

97

CDEACOO - / kmol.m -3

0.8

0.4

wDEA=0.201 T=298.1K wDEA=0.289 T=298.1K wDEA=0.400 T=298.1K wDEA=0.398 T=313.1K wDEA=0.399 T=313.1K wDEA=0.397 T=323.1K

0 0 0.3

CO2 loading

0.6

0.9

Fig. 8. Variation of carbamate concentration with CO2 loading at various temperature and DEA mass fractions.

of DEACOO concentration due to the displacement of the equilibria during the titration by NaOH. Finally, CO2 DEAH2 O system (wDEA = 0.4) has been studied twice at 313.1 K to check the repeatability of our data. As it can be observed in Fig. 8, there is no signicant difference between both experiments and a good repeatability for these experiments is observed. The species concentration changes with increasing loading are characteristic of primary or secondary amines. At low loadings, the absorbed CO2 is mainly transformed into DEACOO . The DEACOO concentration increases quite proportionally to the loading up to = 0.4, and reaches a maximum around = 0.5. For higher CO2 loading, the DEACOO hydrolysis leads to the formation of DEAH+ and HCO3 . At loadings higher than 0.5, the molecular CO2 becomes sufciently concentrated to be detected by IR spectroscopy in DEA aqueous solutions with wDEA > 0.2. By comparing the DEACOO evolution at 298.1 K, 313.1 K and 323.1 K (see Fig. 8), it can be seen that, as expected, the higher the amine fraction, the higher the DEACOO concentration. There is no signicant temperature dependency. Furthermore, the determination of the concentration of all species also allows to calculate the apparent carbamate formation constant, (K6 )app , dened in the following equation (K6 )app = CDEACOO = (K6 )true CDEA CHCO
3

Fig. 10. Evolution of the Henry constant with loading and temperature. The values provided by Browning and Weiland [32] and those from this work are related to Henry constants of N2 O and CO2 , respectively, both measured in DEA aqueous solutions.

c,DEA c,HCO
3

c,DEACOO aH2 O

(17)

It appears that the apparent equilibrium constant has a somehow regular behaviour, at least for loadings larger than 0.2. In the case of lower loadings, (K6 )app increases signicantly, but the concentration of carbamate and HCO3 being low, their numerical values are less precise. It is worth notifying that the apparent constant depends on the loading and that this tendency is more signicant at lower DEA concentrations. Furthermore, a comparison of the values calculated from Bttingers study [5] with those obtained in this work seems to underline a larger dispersion for Bottingers values. The extrapolation of the constants to unloaded solution is difcult because of the existence of some dispersion in the calculated values. Furthermore, the extrapolated values correspond to an innite dilution in the waterDEA mixture reference state, instead of the innite dilution in water reference state used in modeling studies. To get an estimated of the carbamate formation constant in the reference state where water activity and the solute activity coefcients become equal to 1, the extrapolated value deduced from Fig. 9 have been expressed as a function of the DEA concentration. The extrapolation to wDEA = 0 allows to get an estimation of (K6 )true , the carbamate formation constant. We obtain (K6 )true = 11.8 m3 kmol1 . 4.3. Henry constant The solubility of a gas in a liquid can be represented by the Henrys law that relates the partial pressure of a gas in equilibrium with a liquid (Pi ) and the concentration of the gas in the liquid (Ci ) following the equation Pi = Hi Ci (18)

In Fig. 9 the evolution of the constant is plotted as a function of the loading for the three DEA concentrations studied in this work.
15

Kapp m .kmol

10

wDEA=0.201, 298.15K wDEA=0.289, 298.15K wDEA=0.400, 298.15K wDEA=0.200, 293.15K [5]

-1

Hi depends on the used temperature and reference state (nature of the solvent). If the solution differs signicantly from the reference system, Eq. (16) involves an apparent Henry constant which takes into account the activity coefcient of the i species. Since carbon dioxide reacts with alkanolamines, it is not possible to experimentally determine the Henry constant unless being able to measure the concentration of carbon dioxide in the liquid. Our measurements allow a direct determination of the CO2 Henry
1 1.2
DEAsolution = constant in DEA solutions, dened as HCO
2

0 0 0.2 0.4 0.6 0.8

PCO

CO2 loading
Fig. 9. Evolution of the apparent constant of carabamate formation (K6 )app with CO2 loading: comparison of the values calculated in this work at 298.15 K (wDEA = 0.201, wDEA = 0.289 and wDEA = 0.400) and the values from Bttinger [5] at 293.15 K (wDEA = 0.2).

C DEAsolution
CO2

, where

DEAsolution measurements are of dissolved CO2 . The experimental HCO 2 presented together with the speciation data in Tables 27, and a graphical representation is presented in Fig. 10. In the literature,

DEAsolution CCO 2

corresponds to the concentration to the molecular form

98

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099

Table 9 DEAsolution DEAsolution and HCO for various temperatures and various DEA mass Values of HN O
2 2

5. Conclusion The thermodynamics of electrolyte solutions is, in many applications, characterized by the necessity of combining a chemical model and a model taking into account the interactions between ionic or molecular compounds in these systems where the excess properties cannot be neglected. This is the case of acid gasaqueous alkanolamine solutions where numerous ions are produced by chemical reactions. However data used in the numerous models for such systems have been restricted to phase equilibrium experimental values until very recently. This paper has extended previous studies [1517] that used IR spectroscopy to determine the speciation in CO2 wateramine systems. In previous studies, we had focused mainly on the determination of the concentration of molecular CO2 , a species that is very important, especially if the thermodynamics modeling has to be included in a kinetics model. In the present work, measurements on CO2 -loaded DEA aqueous solutions have been performed in various conditions with a specic device and an elaborated method for the calculation of carbamate concentrations at equilibrium has been applied. Speciation data were determined from in situ ATR-IR measurements for several experimental conditions assuming spectral hypotheses and that the carbonate concentration remains negligible. It is noticeable that our determinations of DEACOO concentration are consistent with the values determined by Bttinger [5], measured in similar conditions using NMR spectroscopy. This consistency justies the assumptions made in our method to calculate DEACOO and HCO3 concentrations. Furthermore, molecular CO2 was detected for loadings larger than 0.5, and the corresponding values of the apparent Henry constants have been calculated. It has to be pointed out that these values are not available in the literature, and that IR spectroscopy turns out to be more sensitive than NMR 13 C spectroscopy for the quantication of molecular CO2 . Thus, the solubilities of CO2 in loaded aqueous DEA solutions were not found loadingdependent as expected if the N2 O analogy is extrapolated to loaded aqueous amine solutions. List of symbols

fractions using the correlations presented by Versteeg et al. [28]; and Littel et al. [29]. wDEA T (K) 3 1 DEAsolution HN (kPa m kmol ) O
DEAsolution (kPa m HCO
2 2

0.2 298.1 4440 3268

0.3 298.1 4657 3427

0.4 298.1 4896 3603

333.1 8873 6001

kmol

DEAsolution in CO unloaded or experimental measurement of HCO 2 2 loaded aqueous solutions of DEA cannot be found. Generally, the N2 O non-reacting gas is used to determine CO2 properties such as Henry constant and diffusion coefcient in alkanolamine solutions, because of its similarities in terms of conguration, electronic structure and molecular volume. The Henry constant of the carbon dioxide is then deduced from the N2 O analogy presented in the following equation:

HCO2 HN2 O
DEAsolution

HCO2 HN2 O
water

(19)

Several authors (Versteeg et al. [28]; Littel et al. [29]; Haimour and Sandall [30]; Al-Ghawas et al. [31]) have developed correlations of the calculations of Henry constants, based on experimental measurements of the solubilities of CO2 and N2 O in water and of N2 O in aqueous alkanolamine solutions. They manage to estimate the Henry constant of CO2 but are limited to unloaded solutions. Indeed, as carbon dioxide reacts with the amine, the chemical composition of the liquid changes and the ionic strength of the solution increases. This phenomenon triggers off a decrease of gas solubility as it is conrmed by measurements of N2 O solubility in loaded DEA solutions performed by Browning and Weiland [32]. The experimentally-determined Henry constants for CO2 in this work are compared to the estimations deduced from the N2 O analogy using the correlations of Versteeg and van Swaaij [28] and Littel et al. [29] that estimate (HCO2 /HN2 O )water and the solubility of N2 O in DEA solutions respectively. The calculated HCO2 for unloaded solutions are presented in Table 9. The solubilities of N2 O in loaded DEA solutions measured by Browning and Weiland [32] at 298 K for a various CO2 loading and DEA concentrations are presented in DEAsolution . Fig. 9, together with our measurements of HCO
DEAsolution measured in this work increase with As expected, the HCO 2 temperature, traducing a decrease of the gas solubility in the soluDEAsolution is more tion. Nevertheless the temperature effect on HCO 2 pronounced in loaded solutions than in unloaded solutions (values estimated by N2 O analogy). Moreover it can be seen than the experDEAsolution determined in loaded solution ( imental HCO CO2 > 0.6) are 2 not loading dependent taking into account the accuracy of the Henrys constant experimental determination (from 30% at lowest loadings to 10% at highest loadings). This observation holds for the three temperature tested. Those observations are not in agreement with the N2 O solubilities measured in loaded aqueous DEA solutions by Browning and Weiland [32] at 298 K, for which the CO2 loading signicantly inuences the apparent Henry constant, as more as the concentration of DEA increases. An explanation can be that the N2 O analogy is used in the case of innite dilution and the salting out effect may not be the same for N2 O and CO2 . Moreover, the values determined by Browning and Weiland [32] correspond to loading lower than 0.5 where the main ionic species are DEAH+ and DEACOO . On the contrary our determinations have been obtained at loadings larger than 0.5 where the main ionic species are DEAH+ , DEACOO but also HCO3 as the ratio CHCO /CDEACOO increases with increasing loadings. This also may
2

aH2 O Ci j Hi
Itotal

water activity concentration of the species i (kmol m3 ) Henrys constant of the species i in the solvent j (kPa m3 kmol1 ) total intensity of the spectrum at intensity of the species i at apparent formation constant of carbamate (m3 kmol1 ) true formation constant of carbamate (m3 kmol1 ) response factor of species i at total mole number of CO2 introduced in the cell total mole number of DEA introduced in the cell CO2 partial pressure (kPa) total pressure (kPa) solvent vapour pressure (kPa) ideal gas constant (J K1 J1 ) temperature (K) volume of the equilibrium cell (m3 ) volume of the solution (liquid phase) (m3 ) volume of the vapour phase (m3 ) DEA mass fraction apparent mole fraction of CO2

Ii (K6 )app (K6 )true ki nCO2 ,in nDEA,in PCO2 Ptot sat Pv ap R T Vcell Vsolution Vvapour wDEA app xCO
2

change the salting out effect.

Greek letters CO2 loading (mol CO2 / mol DEA)

F. Diab et al. / Fluid Phase Equilibria 325 (2012) 9099

99

vap CO

solute activity coefcients in the concentration scale wave number (cm1 ) fugacity coefcient of CO2 in vapour phase

Acknowledgment We want to thank IFP Energies Nouvelles for its nancial support. References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] A. Benamor, M.K. Aroua, Fluid Phase Equilib. 231 (2005) 150162. D. Barth, P. Rubini, J.-J. Delpuech, Bull. Soc. Chim. Fr. (1984) 78. T. Suda, T. Iwaki, T. Mimura, Chem. Lett. (1996) 777778. J. Poplsteinova Jakobsen, J. Krane, H.F. Svendsen, Ind. Eng. Chem. Res. 44 (2005) 98949903. W. Bttinger, M. Maiwald, M.H. Hasse, Fluid Phase Equilib. 263 (2008) 131143. W. Bttinger, M. Maiwald, M.H. Hasse, Ind. Eng. Chem. Res. 47 (2008) 79177926. V. Ermatchkov, A. Perez-Salado Kamps, G. Maurer, J. Chem. Thermodyn. 35 (2003) 12771289. Q. Yang, M. Bown, A. Ali, D. Winkler, G. Puxty, M. Attalla, Energy Procedia 1 (2009) 955962. V. Souchon, M. de Oliveira Aleixo, O. Delpoux, C. Sagnard, P. Mougin, A. Wender, L. Raynal, Energy Procedia 4 (2011) 554561. W.J. Rogers, J.A. Bullin, R.R. Davison, R.E. Frazier, K.N. Marsh, AIChE J. 43 (1997) 32233231. W.J. Rogers, J.A. Bullin, R.R. Davison, AIChE J. 44 (1998) 24232430.

[12] U. Lichtfers, B. Rumpf, Thermodynamic Properties of Complex Fluid Mixtures, Research Report, Deutsche Forschungsgemeinschaft DFG, in: G. Maurer (Ed.), Wiley-VCH, 2004, pp. 92119, ISBN: 3-527-27770-6. [13] P.W.J. Derks, P.J.G. Huttenhuis, C. van Aken, J.-H. Marsman, G.F. Versteeg, Energy Procedia 4 (2011) 599605. [14] P. Jackson, K. Robinson, G. Puxty, M. Attalla, Energy Procedia 1 (2009) 985994. [15] R. Sidi-Boumedine, Ph.D. Thesis Ecole des Mines de Paris, 2003. [16] A. Archane, L. Gicquel, E. Provost, W. Frst, Chem. Eng. Res. Des. 86 (2008) 592599. [17] A. Archane, W. Frst, E. Provost, J. Chem. Eng. Data 56 (2011) 18521856. [18] T.E. Daubert, R.P. Danner, Physical and Thermodynamic Properties of Pure Chemicals Data Compilation, Taylor and Francis, Washington, DC, 1989. [19] N.E. Van Huff, G. Houghton, J.J. Coull, J. Chem. Eng. Data 8 (1963) 336340. [20] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds, 4th ed., John Wiley and Sons, Inc., New York, 1986 (Part III). [21] A.R. Davis, B.G. Oliver, J. Solut. Chem. 4 (1972) 329339. [22] S. Kurtaran, I. Kaviak, G.S. Krkc oglu, Proceedings of the International Workshop on New Trends in Science and Technology, Ankara, Turkey, 34 November, 2008. [23] D.J. Frasko, J. Chem. Phys. 41 (1964) 21342140. [24] A.M. Bhairi, Experimental Equilibrium Between Acid Gases and Ethanolamine Solutions, Ph.D., Oklahoma State University, Stillwater, Oklahoma, 1984. [25] J.I. Lee, F.D. Otto, A.E. Mather, J. Chem. Eng. Data 17 (1972) 465468. [26] J.W. Mason, B.F. Dodge, Trans. Am. Inst. Chem. Eng. 32 (1936) 2747. [27] R.N. Maddox, E.M. Elizondo, Equilibrium Solubility of Carbon Dioxide or Hydrogen Sulde in Aqueous Solution of Diethanolamine at Low Pressures, Gas Processors Association, 1989 (Research Report-124). [28] G.F. Versteeg, W.P.M. van Swaaij, J. Chem. Eng. Data 33 (1988) 2934. [29] R.J. Littel, G.F. Versteeg, W.P.M. van Swaaij, J. Chem. Eng. Data 37 (1992) 4955. [30] N. Haimour, O.C. Sandall, Chem. Eng. Sci. 39 (1984) 17911796. [31] H.A. Al-Ghawas, D.P. Hagewiesche, G. Ruiz-Ibanez, O.C. Sandall, J. Chem. Eng. Data 34 (1989) 385391. [32] G.J. Browning, R.H. Weiland, J. Chem. Eng. Data 39 (1994) 817822.