© A. Mehassouel et al., published by IFP Energies nouvelles, 2018

1 Introduction

Human activities have gradually increased the atmospheric concentration of greenhouse gases such as CO₂, methane, nitrous oxide and chlorofluorocarbons over the past century. According to the 5th report of the Intergovernmental Panel on Climate Change (IPCC, 2013), the Earth will continue to warm up as a result of human activities; the temperature will rise by 0.3–4.8 °C. There are various methods for capturing CO₂ (Kanniche et al., 2010) such as, the chemical absorption method which is the most used for CO₂ capture using a usually amine-based chemical solvent. There are different types of amines:

• the primary amines such as monoethanolamine (MEA) and diglycolamine (DGA), which are characterized by their very large reactivity and low absorption capacity;

• secondary amines such as diethanolamine (DEA) and diisopropylamine (DIPA), which are less reactive than the primary amines; the reaction of these amines (primary and secondary) with CO₂ forms a carbamate;

• tertiary amines such as N-methyldiethanolamine (MDEA) and triethanolamine (TEA) which are characterized by their low reactivity and a high absorption rate.

For MDEA, one mole of amine is required to absorb one mole of CO₂, besides, it is very resistant to thermal and chemical degradation, the reaction of tertiary amines with CO₂ does not form a carbamate. Therefore, the energy required for the regeneration of these amines is low, which allows to minimize the cost of the capture process. Several studies have shown that the mixture of primary and tertiary amines combines the high absorption capacity of tertiary amines with the high reactivity of primary and secondary amines (Paul et al., 2009; Zoghi et al., 2012; Toro-Molina and Bouallou, 2013). The regeneration costs of mixtures can be reduced compared to those of MEA or DEA. Thus, the use of amine mixtures can improve the efficiency of gas treatment processes. However, these processes always require lots of energy during desorption. Therefore, the key problems are, how to reduce the energy penalty, and how to reduce the capture cost. Recently the demixing solvents (DMX^TM) have been developed by the IFP Energies nouvelles to reduce the operating cost by phase splitting in decanter before steam stripping (Lecomte et al., 2009; Aleixo et al., 2011; Raynal et al., 2011a, b; Xu et al., 2012). The process consists of using aqueous amine solutions that are capable to form two immiscible liquid phases at a given temperature and concentration in CO₂.

The principle of these demixing solvents is to absorb the carbon dioxide in the aqueous solution of demixing amine. A monophasic solution loaded with CO₂ is then obtained. After heating, the formation of two distinct phases is observed, one poor in CO₂ and rich in amine, the other rich in CO₂ and poor in amine, the phase that is poor in CO₂ can directly be recycled to absorber. Demixation can be controlled by temperature, it occurs above a given temperature value (lower critical solution temperature), above which the two phases are not miscible. The other amines that have emerged in recent years are lipophilic amines, hybrid molecules with hydrophilic and hydrophobic functional groups; they form two phases during regeneration. The organic phase acts as an extractive agent (Tan, 2010), this characteristic allows capturing CO₂ at a low cost. Several amines with phase change have been studied. Zhang et al. (2011) studied the absorption of CO₂ by a mixture of DMCA + DPA and found a precipitation of DPA at high CO₂ loading, they found that this mixture is more efficient than the monoethanolamine (MEA 30 wt.%), because it allows to minimize the regeneration temperature of 120 °C for MEA 30 wt.% at a temperature below 80 °C, which allows to minimize the energy required for regeneration. Zhang et al.(2012a, b) investigated different lipophilic amines for improving the energy efficiency of post combustion, they classified them into two categories: activators as hexylamine (HA) and dipropylamine (DPA) and regeneration promoters such as N-ethylpiperidine (EPD), N,N-dimethylcyclohexylamine (DMCA), N-methylpiperidine (MPD) and dibutylamine (DBA). The decrease of the aqueous amine solubility while the temperature increases causes phase separation above the Lower Critical Solution Temperature (LCST). Zhang et al. (2013) studied the mixture of methylcyclohexylamine (MCA) and dimethylcyclohexylamine and 2-amino 2-methyl propylamine (AMP) as solubilizer, the addition of AMP increases the solvent efficiency as well as LCST. The percentage of the solubilizer should not exceed 15 wt.%.

Xu et al. (2013) investigated several lipophilic amines alone and mixed with different concentrations in order to select the best lipophilic solvent. Ye et al. (2015) studied several mixtures of lipophilic amines as activators for dimethylcyclohexylamine (DMCA) on one hand and with diethylethanolamine (DEEA) on the other hand.

Wang et al. (2017) investigated the kinetics of CO₂ absorption with the solvent N,N-diethylethanol amine (DEEA) and N,N-dimethylbutylamine (DMBA), the mixture forms two phase during absorption; it has a high absorption rate and low energy consumption compared to the conventional MEA solvent. Ye et al. (2017) studied the mechanism of absorption and desorption of CO₂ with phase transitional diethylene triamine (DETA) and pentamethyl diethylenetriamine (PMDETA). Knuutila and Nannestad (2017) studied the effect of the concentration of 3-(methylamino)-propylamine (MAPA) on the heat of CO₂ absorption in DEEA/MAPA blend.

Mehassouel et al. (2016) used the hexylamine (lipophilic amine) to activate methyldiethanolamine (MDEA). Pinto et al. (2013) studied the CO₂ absorption with blend 5M (DEEA) and 2M of 3-(methylamino)-propylamine (MAPA).

In parallel in recent years studies have also focused on reducing the concentration of CO₂, its conversion and its use for the production of high value products. Among these works, Tursunov et al. (2017) studied the conversion of CO₂ to methanol by hydrogenation using over copper and iron based catalysts as well as the influence of the different parameters on the conversion rate and the reaction mechanism. Bashipour et al. (2017) investigated the production of sodium sulphate (Na₂S) by the absorption of H₂S in NaOH in a spraying column. Na₂S has several applications in the chemical industry, the authors have used the Response Surface Methodology (RSM) to design and optimize experiments based on Central Composite Design (CCD). A model of Artificial Neural Network has also been used to predict the percentage by weight of Na₂S. Passalacqua et al. (2015) studied the possibility of converting CO₂ into fuel in the presence of solar energy and water in a direct photoreactor, the product obtained is a mixture of methane, methanol and formic acid.

Concerning the simulation part, Sharifi and Omidbakhsh (2017) simulated the capture of CO₂ by solvent MDEA using different types of absorption and regeneration columns, to select the best type in terms of absorption capacity. The cement industry is one of the most emitting industries of CO₂. It represents about 5% of the total emissions. Cement plant flue gas has a relatively high CO₂ concentration, typically about 25 mol% compared to about 14 mol% for a coal fired power plant (IEA, 2008) the emission of CO₂ comes from fossil fuel combustion in the kiln process and de-carbonation of limestone (CaCO₃) to Calcium Oxide (CaO) (Bosoaga et al., 2009). Among the authors who studied the simulation to capture CO₂ coming from a cement plant; Nazmul (2005) used the solvent MEA (30 wt%).

The aim of our work is to study the kinetics of CO₂ absorption by a mixture of methyldiethanolamine (MDEA) activated by hexylamine in different concentrations, the selected solvent is used to simulate the CO₂ capture coming from a cement plant.

2 Experimental

2.1 Chemicals

All the solutions have been prepared with deionized water. MDEA was obtained from Sigma Aldrich with 98% mass purity, HA was obtained from Acros Organics with 98% mass purity, CO₂ was provided by Air Liquid with a certified purity of 99.99 vol%.

2.2 Density and viscosity measurement

The density and viscosity of all the solutions prepared MDEA 33 wt.% + HA 7 wt.%, MDEA 35 wt.% + HA 5 wt.%, MDEA 37 wt.% + HA 3 wt.% and MDEA 40 wt.%, were measured by a Anton Paar densimeter in different temperatures 298 K, 313 K and 333 K (Tab. 1).

2.3 Liquid side mass transfer coefficient

The value of physical mass transfer coefficient (k_L) of CO₂ in aqueous solutions was obtained from mass transfer correlation between dimensionless numbers Reynolds (Re), Schmidt (Sc) and Sherwood (Sh). This correlation (Eq. 1) has been obtained by Amrarene and Bouallou (2004) based on measurements of physical absorption of N₂O in aqueous solutions of MDEA. (1) (2) (3) (4)

This correlation is valid for: µ = dynamic viscosity (Pa s), d_cell = inner diameter of the Lewis cell (m), D = diffusivity (m² s⁻¹), d_Ag = diameter of the Rushton turbine (m), ρ = solution density (kg m⁻³). The validity of this correlation has been tested for our experiments in different temperatures and for different solvents studied (see Appendix B).

The diffusion coefficient and the Henry constant are obtained by applying the correlation of Al-Ghawas et al. (1989), for MDEA compositions up to 50 wt.%. The influence of the Hexylamine concentration is considered negligible. The equations used are given in Appendix A, N = 100 rpm is the stirring speed of the liquid phase; it is maintained constant during the experiment.

2.5 Experimental setup and procedure

2.5.1 Experimental apparatus

The apparatus used for carrying out the kinetic measurements (Fig. 1) (Toro-Molina and Bouallou, 2013), is a double jacket cell reactor whose temperature is kept constant by means of a control bath. The cell is made of Pyrex glass of 63.3 × 10⁻³ m internal diameter, whose two extremities are sealed up by two metal flanges, joints ensuring the seal of the assembly. Four Teflon counter-blades are placed inside the cell, in order to avoid the formation of a vortex and also to maintain a stable horizontal ring. This ring allows to stabilize and to define a gas-liquid interfacial area that is equal to 15.34 × 10⁻⁴ m². The value of k_L remains constant in the various solutions tested. The liquid phase agitator is assured by Rushton turbine with six-bladed 42.5 × 10⁻³ m in diameter; the gas phase is stirred by four-bladed impeller 40 × 10⁻³ m in diameter. The agitators of the liquid and gas phase are driven by a variable speed motor. The upper flange is connected to a Druck pressure sensor thermostatically controlled to 373 K in order to avoid any phenomenon of condensation in its support which permits to measure pressures up to 2.5 × 10⁵ Pa. The lower flange is equipped with a temperature sensor and a non-rotating needle valve. The temperature sensor and the pressure sensor are connected to an absorption data acquisition unit.

Fig. 1 Experimental apparatus.

2.5.2 Procedure

The aqueous solutions were prepared under vacuum, the water and the amine were mixed by gravity according to the concentration of the required aqueous solution, and the mass of water and amine were measured by weight. The cell must be evacuated before introducing the solvent into the reactor when the working temperature and pressure are stabilized in the cell. The volume of the liquid phase is measured based on the mass of the solution introduced and its density, in this case, the contact between the solvent and the CO₂ is initiated and the acquisition of data begins. The reproducibility of the absorption rate is within 10%.

During the data acquisition, the pressure in the cell is recorded as shown in Figure 2. The pressure within the cell can be divided into four periods:

• solvent vapor pressure at equilibrium;

• injection of the gas into the cell, which results in an increase in pressure;

• absorption of the gas by the solvent;

• return to system balance.

By way of illustration, Figure 3 shows the CO₂ absorption by aqueous MDEA 37 wt.% + HA 3 wt.% solvent at 298 K.

Fig. 2 Data acquisition example for CO2 absorption in aqueous solution.

Fig. 3 CO2 absorption by aqueous MDEA 37 wt.% + HA 3 wt.% solvent at 298 K.

2.5.3 Results and discussion

The absorption measurements were carried out at three different temperatures 298 K, 313 K and 333 K for the three solvents MDEA 37 wt.% + HA 3 wt.%, MDEA 35 wt.% + HA 5 wt.% and MDEA 33 wt.% + HA 7 wt.%.

A material balance on the CO₂ gas phase of the reactor allows us to directly determine the absorption flux: (5) a is the interfacial area and V_g is the volume of gas.

The influence of the kinetics of the chemical reaction on the CO₂ absorption is reflected by the enhancement factor E, in this case, the absorption flux (Eq. 5) is written assuming that the concentration of CO₂ in the liquid phase is very small compared to its concentration at the interface: (6)

The gas phase is assumed ideal and the concentration of CO_2int at the interface is connected with its pressure via the Henry law: (7)

is the pressure of CO₂: (8) where P is the total pressure in the cell and P_sol is the vapor pressure in the cell before introducing the solvent.

The initial absorption rates are measured within a pressure range of 10 kPa from the initial total pressure, for this small pressure drop, the concentration of CO₂ resulting from the absorption does not change much the composition of the solution so, k_L, HCO₂, and E remains constant with time. The integration of Equation (5) using Equations (6)–(8) give: (9) (10) β is the slop  (s⁻¹), the enhancement factor E is obtained from β using Equation (10).

The condition of the pseudo first order reaction regime between CO₂ and blend MDEA + HA is tested, in this case, we have: (11) (12) (13) (14) 3 < Ha < E_i/2, k_ov is the overall reaction rate constant, it is calculated using Equation (14), and Ha is the Hatta number that allows us to locate the place of the reaction:

• Ha < 0.3: in the liquid phase.

• 0.3 < Ha < 3: both in the liquid phase and in the film diffusion.

• Ha > 3 in the film diffusion.

The film theory suppose that the resistance of mass transfer is located in a thin film adjacent to the gas-liquid interface, the mass transfer on the liquid phase is carried out only by molecular diffusion (Whitman, 1923).

E_i is the instantaneous enhancement factor which is also determined according to Equation (15): (15)

Table 2 summarized the different parameters results for CO₂ absorption by different amine mixtures at different temperatures.

The results show that the reaction takes place in the film diffusion, the Hatta numbers are all greater than 3. The condition of pseudo first order reaction is satisfied for the mixture MDEA 40 wt.% at 298 K, and the mixture MDEA 37 wt.% + HA 3 wt.% at 298 K, 313 K and 333 K.

The condition of pseudo first order reaction is not satisfied for the mixtures MDEA 35 wt.% + HA 5 wt.% and MDEA 33 wt.% + HA 7 wt.%. This may be due to several factors: firstly, the mixtures form two phases in the working temperature range, because, hexylamine which is lipophilic amine, has very low aqueous solubility, and the miscibility changes with temperature and mass concentration of amine, phenomenon that is already studied by Bishnoi and Rochelle (2002) for the absorption of CO₂ by the mixture of MDEA + PZ, the piperazine which has a low aqueous solubility can give two phases. Secondly, the presence of hexylamine can lead to errors in the calculation of the physical properties of the solutions.

We can also note that the addition of a small amount of hexylamine substantially increases the kinetics of CO₂ absorption for a given temperature (Figs. 4 and 5), this is due to hexylamine which has very fast absorption (Tan, 2010), and high capacity, so, the CO₂ reacts firstly with hexylamine in the blend MDEA + HA, the CO₂ loading increases within a short time of absorption. The increase in the hexylamine concentration decreases the fraction of MDEA, result in a decrease in CO₂ loading capacity for aqueous MDEA 35 wt.% + HA 5 wt.% and MDEA 33 wt.% + HA 7 wt.%. The aqueous solubility decreases as the temperature increases causing phase separation, thus a decrease in the CO₂ loading capacity for the MDEA 37 wt.% + HA 3 wt.% solvent at 333 K (Fig. 5).

The calculation of the activation energy for the reaction of CO₂ with aqueous mixture MDEA 37 wt.% + HA 3 wt.% (Fig. 6) is carried out by applying the Arrhenius law. (16) where is the second order kinetics constant for the reaction of CO₂ with blend MDEA 37 wt.% + HA 3 wt.%.

The calculated activation energy (Ea) is 22. 25 Kj mol⁻¹, which is very small compared to MDEA 40 wt.% that is equal to 44.3 KJ · mol⁻¹ (Amann and Bouallou, 2009; Cadours and Bouallou, 1998). The overall rate law is: (17)

The diminution in the activation energy for the reaction of CO₂ with aqueous solution of MDEA 37 wt.% + HA 3 wt.% decreases the energy demand for solvent regeneration.

The mixture MDEA 37 wt.% + HA 3 wt.% has very fast absorption rate and high capacity compared with MDEA 40 wt.% and compared to other solvents MDEA 35 wt.% + HA 5 wt.% and MDEA 33 wt.% + HA 7 wt.% at 298 K (Fig. 7). The explanations for this behavior can be either, the stability of the hexylamine carbamate for the reaction of CO₂ with aqueous solution MDEA 37 wt.% + HA 3 wt.% is very low or, there is any formation of the carbamate in the solution, resulting into a high amount of free amine in the solution available to react with CO₂.

We can also say that, the absorption which is an exothermic phenomenon is favored at temperature 298 K.

Fig. 4 CO2 loaded versus time for different solvents at 313 K.

Fig. 5 CO2 loaded versus time for different solvents at 333 K.

Fig. 6 Arrhenius law for the CO2 absorption in aqueous MDEA 37 wt.% + HA 3 wt.% solution.

Fig. 7 CO2 loaded versus time for different solvents at 298 K.

3 Simulation

The Aspen Plus^TM software was used to simulate the capture of CO₂ by aqueous MDEA with two different mass concentrations (MDEA 40 wt.% and MDEA 50 wt.%) and with MDEA 37 wt.% + HA 3 wt.%. The flue gas studied was coming from the cement plant, its composition before introducing to the absorber is shown in Table 3. It is in 0.12 MPa and 313 K, Figure 8 shows the alkanolamine process for CO₂ capture. CO₂ is washed by a countercurrent with the solvent. The rich solvent leaving the absorber is pumped by P1 to 0.21 MPa and sent to heat exchange where it is preheated by the regenerated solvent recovered at the bottom of the stripper. After regeneration, the lean solvent is pumped by P2 to 0.21 MPa and cooled at 0.12 MPa and 313 K before recycling to the absorber. The CO₂ stream at the top of the stripper is compressed to 15 MPa in a multi stage compressor.

We have used the ELECtrolyte Non-Random Two Liquid activity coefficients (ELECNRTL) thermodynamic model for the CO₂ capture simulation by the Aspen Plus^TM software, this model allows simulating non-ideal aqueous solutions, while Aspen Plus^TM has the KEMDEA inserts data package for the simulation with the solvent MDEA. For MDEA 37 wt.% + HA 3 wt.%, we modified the insert data package used for the simulation with MDEA (KEMDEA) by adding the CO₂ reactions with hexylamine (Eqs. 18 and 19). (18) (19)

The parameters of binary interactions between MDEA and hexylamine as well as electrolytic pair interactions and kinetic constants reaction were regressed from experimental data using the Aspen Plus^TM Data Regression System (DRS), module which was already used by lot of literature studies (Pinto et al., 2013; Aroua et al., 2002; Mudhasakul et al., 2013).

The study of sensitivity was carried out based on the number of the theoretical stages of the absorber and the stripper. It was conducted in the aim to minimize, the solvent flow rate in the absorber and the heat duty in the reboiler. For 85% CO₂ recovery, Figure 9 shows the evolution of energy regeneration with lean CO₂ loading and concentration for all solvent studied. It is clear that MDEA 50 wt.% gives a lower energy regeneration compared to MDEA 40 wt.%, the low energy for solvent MDEA 40 wt.% is equal to 3.45 GJ/tCO₂, whereas for lean CO₂ loading it is equal to 0.115.

For MDEA 37 wt.% + HA 3 wt.% solvent, the low energy is equal to 2.9 GJ/tCO₂ whereas for lean CO₂ loading it is equal to 0.2 (Tab. 4).

Fig. 8 Alkanolamine process for CO2 capture.

Fig. 9 Energy regeneration for differents lean CO2 loaded.

4 Conclusion

The aim of this study was to find the most favorable blend composition to capture CO₂. CO₂ absorption rates into MDEA-HA aqueous solutions are measured at three temperatures 298 K, 313 K and 333 K, and different mass concentrations MDEA 37 wt.% + HA 3 wt.%, MDEA 35 wt.% + HA 5 wt.% and MDEA 33 wt.% + HA 7 wt.%. HA is very reactive with CO₂, it was proposed to use this amine as an activator for an aqueous MDEA solution. The addition of HA leads to a significant enhancement of the absorption rates compared to an aqueous MDEA 40 wt.% solution. Results show that kinetics absorption of blended MDEA 37 wt.% + HA 3 wt.% is pseudo first order. Comparative simulations for CO₂ capture of flue gas for a cement plant shows that MDEA 37 wt.% + HA 3 wt.% solvent leads to lower energy consumption than those of MDEA 40 wt.%.

Nomenclature

a: interfacial area (m²)

C_i: concentration of component i (mol m⁻³)

DRS: data regression system

d_ag: Rushton turbine diameter (m)

d: density of the solution (kg m⁻³)

D_i: diffusivity of species i in liquid phase (m² s⁻¹)

d_cel: internal cell diameter (m)

E: enhancement factor

E_i: instantaneous enhancement factor

Ea: activation energy (KJ mol⁻¹)

ELECNRTL: electrolyte non-random two liquid activity coefficients

H_CO2: Henry's law constant (Pa m³ mol⁻¹)

Ha: Hatta number

IEA: international energy agency

k_L: liquid side mass transfer coefficient (m s⁻¹)

k_OV: overall rate constant (s⁻¹)

KEMDEA: insert data package for MDEA

k: equilibrium constant

n: number of mole (mol)

N: stirrer speed (rpm)

P: pressure (Pa)

R: ideal gas constant (8.314 J K⁻¹ mol⁻¹)

Re: Reynolds number

Sc: Schmidt number

Sh: Sherwood number

t: time (s)

T: temperature (K)

V: volume (m³)

χ_i: mole fraction of component i

Greek symbols

µ: dynamic viscosity (Pa s)

ϕ: chemical absorption rate (mol m⁻² s⁻¹)

β: slop (s⁻¹)

Subscripts and superscripts

Sol: solvent

Int: interface

L: liquid

0: initial

Abbreviations

CO₂: carbon dioxide

CaCO₃: limestone

CaO: calcium Oxide

DGA: diglycolamine

MDEA: N-methyldiethanolamine

DEA: diethanolamine

DIPA: diisopropylamine

DMCA: dimethylcyclohexylamine

DPA: dipropylamine

DEEA: 2-(diethyl amino)-ethanol

DETA: diethylenetriamine

DMBA: N.N.dimethylbutylamine

DMX^TM: demixing solvent

EPD: N-ethylpiperidine

HA: hexylamine

LCST: low critical solution temperature

MPD: N-methylpiperidine

MEA: monoethanolamine

MAPA: 3-(methylamino)-propylamine

MCA: methylcyclohexylamine

N₂O: nitrous oxide

PZ: piperazine

PMDETA: pentamethyldiethylenetriamine

TEA: triethanolamine

Appendix

Physicochemical properties for the mixture MDEA–H₂O–CO₂ (A1) (A2) (A3) (A4) (A5) (A6)

CO₂ diffusivity (A7) (A8) (A9) (A10)

MDEA diffusivity (A11) (A12)

Appendix

References

All Tables

All Figures

	Fig. 1 Experimental apparatus.
In the text

	Fig. 2 Data acquisition example for CO2 absorption in aqueous solution.
In the text

	Fig. 3 CO2 absorption by aqueous MDEA 37 wt.% + HA 3 wt.% solvent at 298 K.
In the text

	Fig. 4 CO2 loaded versus time for different solvents at 313 K.
In the text

	Fig. 5 CO2 loaded versus time for different solvents at 333 K.
In the text

	Fig. 6 Arrhenius law for the CO2 absorption in aqueous MDEA 37 wt.% + HA 3 wt.% solution.
In the text

	Fig. 7 CO2 loaded versus time for different solvents at 298 K.
In the text

	Fig. 8 Alkanolamine process for CO2 capture.
In the text

	Fig. 9 Energy regeneration for differents lean CO2 loaded.
In the text