Yu Dong,Tiantian Ping,Shufeng Shen

School of Chemical and Pharmaceutical Engineering,Hebei University of Science and Technology,Shijiazhuang 050018,China

Keywords:Solubility CO2 capture Nonaqueous absorbent 2-(Butylamino)ethanol 2-Butoxyethanol Model

ABSTRACT Nonaqueous amine-based system is an attractive solution to overcome high-energy-intensive CO2 capture process using the conventional aqueous amines.Advanced nonaqueous absorbent of 2-(butylamino)ethanol (BAE) with 2-butoxyethanol (2-BE) has been recently proposed for low-energyconsumption CO2 capture.In this work,Henry’s law constants of CO2 in the BAE/2-BE blend were obtained by N2O/CO2 analogy,and correlated in the temperature range of(283–333)K.Vapor-liquid equilibrium(VLE)data for the BAE+CO2+2-BE system at 65.4%(mass)BAE were also determined in a stirred equilibrium cell at temperatures of(313–393)K and CO2 partial pressures up to 275 kPa.A single apparent equilibrium constant was proposed for this system and correlated as a function of temperature,carbonated degree of amine and CO2 loading.Solubility data were well represented by the modified Kent-Eisenberg model with an average absolute relative deviation (AARD) of 13%.

1.Introduction

Carbon dioxide (CO2) capture,utilization and storage (CCUS) is one of the feasible pathways to mitigate CO2emissions from industrial activities [1].Aqueous 30% (mass) monoethanolamine (MEA)is the most used absorbent in the most close-to-market solvent absorption.However,major disadvantages such as high energy consumption and severe solvent degradation were reported in the practical applications [2–4].It has been noted with concern that the use of water as cosolvent is the weakest point in terms of regeneration energy consumption for the aqueous amines,because of its large specific heat capacity and enthalpy of vaporization,as well as high pressure ratio of water vapor to CO2during solvent regeneration [5,6].Using organic solvent replacing water in amine-based absorbents has recently proved to have significant advantages as for reducing regeneration energy and potentially reducing corrosiveness and degradation [7,8].

In the reported nonaqueous systems,organic solvents include volatile alcohols (e.g.methanol,ethanol),the low volatile alcohols and glycols (e.g.ethylene glycol),glymes (e.g.2-methoxyethanol,2-butoxyethanol),dimethyl formamide (DMF) and specific ionic liquids [8–12].Among these solvents,it seems to be impractical to use high volatile alcohol in the industrial process due to possible huge solvent loss.Although the glycols and ionic liquids have negligible vapor pressures at absorption temperature,the dramatic increase in viscosity might result in poor mass transfer performance.In our recent work [13],secondary alkanolamine-based nonaqueous absorbents using 2-butoxyethanol as cosolvent were found to significantly reduce the regeneration energy compared with aqueous MEA using a rigorous evaluation method.Specifically,the regeneration energy consumption of nonaqueous absorbent of 2-(butylamino)ethanol (BAE) with 2-butoxyethanol (2-BE) was reported to be about 1.73 MJ·kg-1CO2,which is about 55%lower than aqueous MEA under the same operating conditions.In addition,2-BE has low volatility,low specific heat capacity,low viscosity and environmentally friendly features [14,15].Nonaqueous BAE/2-BE absorbent has shown great potential in energyefficient CO2capture process.However,thermodynamics data for this system are rare in the open literature.

Solubility data of CO2in an absorbent are very important for establishing a rigorous thermodynamics model in process modeling and plant design [4,6,7].In addition,Henry’s law constant is also an essential parameter for determining the reaction kinetics[16,17].Due to the reactive nature of amines with CO2,physical solubility of CO2in amines or their blends cannot be measured directly.N2O/CO2analogy is widely used and accepted due to their similarities in molecular volume,configuration and electronic structure [18–20].In the present work,vapor–liquid equilibrium for the system of CO2+BAE+2-BE,using the similar experimental methods in our previous paper [21].Data were interpreted using the modified Kent-Eisenberg model [22,23].The apparent equilibrium constant was correlated as a function of temperature,CO2loading and carbamate concentration.

2.Materials and Methods

2.1.Chemicals and materials

MEA (CAS No.141-43-5,99.12% GC purity) and 2-BE (CAS No.111-76-2,99.82%GC purity)were purchased from Aladdin reagent,China.BAE (CAS No.111-75-1,99.70% GC purity) was purchased from Sigma-Aldrich.All chemicals were used as received without further purification.The water content in the studied solvents was determined by automatic Karl Fischer moisture titrator(Shanghai Anting Electronic Instrument,ZSD-2).The amine-based solutions were prepared by mixing amine (i.e.MEA or BAE) in a specific solvent (water or 2-BE) in a volumetric flask at 298 ± 1.0 K.Specifically,aqueous 30.2% (mass) MEA,and 65.4% (mass) BAE in nonaqueous 2-BE solvent were made for vapor-liquid equilibrium experiments.Electronic analytical balance (OHAUS,NVE2001ZH) was used for weight measurements.Carbon dioxide(CO2,99.99% (mass)) and nitrous oxide (N2O,＞99.90% (mass))were obtained from Shijiazhuang Xisanjiao oxygen generation station and Dalian Data Gas Co.Ltd.,respectively.All chemicals were used as received without further purification.Details of chemical substances are given in Table 1.

Table 1 Chemicals used in this study

Table 2 Experimental data on density ρ,molality of gas(i)in solvent(j)mi,j,partial pressure Pi under total pressure in the reactor cell Ptot and Henry’s law constant of gas in solvent Hi,j at temperatures T

Table 3 Solubility of CO2 in the MEA (1)-CO2(2)-H2O(3) system with 30.2% (w2) MEA at 313 and 373 K

2.2.Henry’s constants of N2O and CO2 in solvents

The method for measuring physical solubility of gases in solvents were described in our previous work[16,21,24].The apparatus is shown in Fig.1.A brief procedure is presented here.In each experiment,a known mass (mS,about 0.45 kg) of solution was injected into the equilibrium cell (VR) and degassed.Then,the initial pressure P0in the cell was recorded when the VLE was obtained at a desired temperature (TR).Thirdly,the gas (N2O or CO2)was added into the cell from a buffer vessel (VV)with a pressure transducer at a constant temperature (TV),and the total amount of gas added into the cell (nadd) can be calculated.Lastly,the total pressure in the cell (Ptot) was recorded again after the VLE was reached again.The physical solubility of gas(i)can be then calculated:

where Hi,jis the Henry’s constants of gas (i) in solvent (j),kPa·m3·kmol-1.Piand Ci,jare the equilibrium partial pressure of gas (i) and its molar concentration in solvent (j),respectively.msand ρ are the mass and the density of gas-free solution at the investigated temperature,respectively.It is assumed that the effect of dissolved gas on density of solution is negligible.ngis the amount of gas in the vapor phase at vapor–liquid equilibrium.Density of gas-free samples was measured by a densimeter (DMA-4100M,Anton Paar) at temperatures of (283.15–333.15) K.

2.3.Vapor-liquid equilibrium experiments of the BAE+CO2+2-BE system

The apparatus for VLE measurement at temperatures of 313–393 K is shown in Fig.1(b).An equilibrium cell (Ve) was heated by an oil bath with a magnetic stirrer.In general,the operating procedure is similar to the measurement of physical solubility.Since a series of VLE data at various CO2content are required,the CO2gas needs to be added into the equilibrium cell step by step and each step the system needs to reach a new vapor–liquid equilibrium.The cumulative CO2amount can be calculated by mass balance of the buffer vessel (Vv).The CO2partial pressure,in the gas phase was calculated from the different between theand P0,Tat the measured temperature (Te).Then,the partial amount of CO2in the gas phase at the equilibrium,can be calculated using the Peng-Robinson equation of state assuming CO2as a pure component.The CO2loading in the liquid phase can be obtained by mass balance,expressed in mole CO2per mole BAE or mole CO2per kg CO2-free solution.Thus,a set of data points ofvs.α (or m2) can be obtained.

Fig.1.Experimental apparatus for vapor–liquid equilibrium measurement.(a)Physical solubility of N2O and CO2 gases,(b)CO2 solubility in nonaqueous BAE-based solutions.

3.Model Representation of Experimental Data

The experimental data of physical solubility of CO2and N2O with respect to temperature in a specific pure solvent can be represented by an exponential model [16,25,26],expressed as:

where Hi,jis the Henry’s constant of the specific gas(i)in pure 2-BE and water.kPa·m3·kmol-1;T is temperature,K;a1,a2are the adjustable parameters.

Physical solubility of N2O in 65.4% (mass) BAE/2-BE blend can be simply correlated using Eq.(5) and then physical solubility of CO2in this blend were calculated by the N2O/CO2analogy that the Henry’s constant ratio of CO2and N2O remains the same in the pure physical solvent(s)and the mixed blend at a specific condition [18,27,28].

In this work,a modified Kent and Eisenberg model was used to fit the experimental data.The non-ideality in the gas phase was neglected.All the liquid phase non-idealities are lumped into the unknown equilibrium constant,expressed as apparent equilibrium constant,by fitting it to the experimental VLE data.Thus,CO2equilibrium partial pressure(the total pressure minus the initial vapor pressure) can be related to the concentration of molecular CO2in the solution using Henry’s law relationship.

In anhydrous BAE(denoted as M)-based solution,CO2can react with BAE forming unstable zwitterions which can also react with an excess of amine to form the corresponding ionic couples,i.e.,BAE carbamate and protonated BAE,which was proved in our previous work [13,29].2-BE is considered as non-reactive diluents as components in the mixtures.The reactions are thermally reversible,thus allow obtaining the free amines and pure CO2at specific conditions.The overall reaction can be expressed as:

where MCOO-and MH+stand for the ionic couples.Thus,a single equilibrium constant of BAE carbamate dissociation can be described as follows:

where [i] and γiare the molar concentration and activity coefficient of species i in the solution,respectively.The following assumptions were made in the abovementioned system:(1) M is referred to the free amine BAE and can react with CO2and is considered as an electrically neutral;(2) MCOO-is a carbonated/carbamate species and has a single negative charge (anion);(3) MH+has a single positive charge(cation);(4)the concentration of molecular CO2in the liquid solution can be estimated from the measured partial pressureover the solution and the Henry’s constants of CO2[25,28].Therefore,mass balance and charge balance equations are given as below:

Fig.2.Solubility of N2O in water solvent compared with literature data[19,20,30].

Fig.3.Comparison of CO2 solubility in aqueous 30.2% (mass) MEA between this work and literature data [31–34].(a) 313.2 K；(b) 373.2 K.

The equilibrium constants (KCO2) should not be composition dependent,but the activity coefficients are.In the modified Kent and Eisenberg model,the non-idealities represented by activity coefficients have been lumped into the apparent equilibrium constant KCO2，appwhich is expressed to be dependent on temperature(T),carbonated degree of amineand CO2loading (α).

The obtained equilibrium constants from each data point were regressed by least square method.The expression foris then assumed to be:

where b1–b4are fitting parameters,which can be obtained by the fitting method of the Levenberg-Marquardt algorithm to minimize Reduced Chi-square value.Each proposed term representing the liquid non-idealities was determined by an F-test to see if it has a significant effect on the result,which decided by a p-value.If the p-value is less than 0.05,then the term is significant.

The average absolute relative deviation (AARD) between the experimental and the calculated values from the proposed correlations was used to assess the models.

where n was the number of data points,represent the experimental and the calculated values from the model correlation,respectively.

4.Results and Discussion

In order to validate the apparatus and the method in this work,Henry’s law constants of N2O in water and CO2solubility in 30.2%(mass) MEA were measured,as presented in Tables 2 and 3.The graphic comparison of HN2Obetween this work and the literature data [19,20,30] is presented in Fig.2.It can be observed that the measured data of HN2Oin water were in lines with the reported data with an AARD within 3%.Two set of CO2solubility data at 313 K and 373 K were compared with the most cited literature data[31–34],as shown in Fig.3.The experimental data of CO2solubility matched well with the published data,especially for CO2partial pressures above 10 kPa.Therefore,the apparatus and method in this work are reliable.

4.1.Physical solubility of N2O and CO2

Since 2-BE is a non-reactive solvent in the mixture of BAE and 2-BE,solubility data of N2O and CO2in 2-BE is necessary for N2O analogy method.Solubility data of N2O and CO2in 2-BE and65.4% (mass) BAE/2-BE blends at temperatures from (283–333) K are also listed in Table 2.

Table 4 Parameters of correlation model in Eq.(6) for Henry’s law constant of N2O and CO2 in different solutions

It can be found that Henry’s law constants of N2O in all solvents investigated show a clearly increase with an increase in temperature,indicating that intermolecular interactions weaken with the increasing temperature.It is noted that N2O has a little higher solubility in pure 2-BE than CO2.The addition of BAE into 2-BE can decrease the N2O solubility,probably due to the higher solvent polarity of BAE (dielectric constant ε ＞20) than 2-BE (ε=9.4)[14,35].As N2O has a low dielectric constant,the BAE molecule with polar groups has difficulty in dissolving nonpolar species.Intermolecular hydrogen bonding between BAE and 2-BE also results in aggregation of these unlike molecules,which has been proved by FT-IR results in our previous work [36].

The experimental solubility data with respect to temperature were correlated by the empirical model.The fitted parameters for the systems abovementioned are listed in Table 4.Using N2O analogy,the physical solubility of CO2in 65.4% (mass) BAE/2-BE blends can be obtained,expressed as a correlation:

Table 6 Model parameters fitted in Eq.(15) for the apparent equilibrium constant KCO2，app

This correlation can be used to estimate the CO2concentration(Eq.(14)) in the liquids.

4.2.CO2 solubility in BAE/2-BE solvent

The vapor-liquid equilibrium data were determined for BAE+CO2+2-BE system with 65.4%(mass)BAE(about 5.0 mol·L-1at 298 K)at temperatures of(313–393)K over CO2partial pressure range of 5–275 kPa and presented in Table 5.Fig.4 show the graphical representation of VLE data in terms of CO2loading (α)against equilibrium CO2partial pressure.In all runs,the initial equilibrium pressure (P0,T) after degassing was recorded before injecting the CO2gas into the reactor cell.The total pressures in the equilibrium cell were not higher than 300 kPa.

Fig.4.CO2 solubility in the system of BAE+CO2+2-BE with 65.4%(mass) BAE.(a)comparison between the experimental and the calculated data from the modified KE model;(b) Parity plot.

It can be seen from Fig.4 that CO2solubility,expressed as CO2loading,generally increases with the increasing CO2partial pressures over the solutions for each temperature investigated.Equilibrium temperature has significant effect on CO2loading at a specific CO2partial pressure.For example,at constant CO2partial pressure 100 kPa,equilibrium solubility of CO2is about 0.47 mol CO2·(mol BAE)-1at 313 K,while lower than 0.10 at 373 K and only about 0.025 at 393 K.At relatively low temperature,e.g.313 K,the CO2partial pressure slightly increases with an increase in carbonated degree of amine BAE.When the CO2loading is greater than about 0.45,the CO2partial pressure shows steep rise,especially at CO2loading above 0.50.In this case,physical absorption of CO2dominates in this step,resulting in a sharp rise in CO2partial pressure with the increasing CO2loading.These phenomena can be explained by the stability of BAE carbamate which is related to the equilibrium constantThe theoretical CO2loading from the overall reaction is 0.50 mol CO2·(mol BAE)-1at all temperatures.However,physical solubility of CO2should be considered especially at low temperature and high CO2partial pressure.

It should be pointed out that the proposed correlation(Eq.(17))for physical solubility of CO2in 65.4% (mass) BAE/2-BE blends can be used at temperatures lower than 333 K.Since the solubility data at high temperatures for N2O analogy is currently unavailable due to the limited experimental conditions,the correlation is extrapolated to the whole temperature investigated in this work to calculate the CO2concentration in the liquid solvent.The fitted parameters for the apparent equilibrium constantin the modified Kent and Eisenberg model are presented in Table 6.Using the regressed parameters,CO2partial pressure can also be calculated for a given experimental condition.The calculated values from the models are presented as curved solid lines in Fig.4,along with the experimental data.Parity plot is also shown to describe the model accuracy.

5.Conclusions

Solubility data of N2O and CO2in 2-BE and BAE/2-BE blend were measured at temperatures from 283 to 393 K.Physical solubility of CO2in the system of BAE+CO2+2-BE at 65.4% (mass) BAE were well correlated by the N2O analogy,with AARDs within 0.8%.In anhydrous BAE/2-BE solution,an apparent equilibrium constantwas proposed as a function of temperature,carbonated degree of amine and CO2loading to consider the liquid phase non-idealities.The modified Kent-Eisenberg model can represent the experimental VLE data reasonably well,with an AARD about 13%.New VLE data and modeling are helpful for process demonstration and evaluation of novel nonaqueous absorbent.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by Natural Science Foundation of Hebei Province(B2018208154),Department of Education of Hebei Province,P.R.China (SLRC2019051) and Key Foundation of Hebei Provincial Department of Science and Technology,P.R.China(21373703D).