Densities, Viscosities of Pure 1-(2-Hydroxyethyl) Pyrrolidine, 3-Amino-1-Propanol, Water, and Their Mixtures at 293.15 to 363.15 K and Atmospheric Pressure

Densities and viscosities of pure 1-(2-hydroxyethyl) pyrrolidine, 3-amino-1-propanol, water, and their blends’ data are reported from 293.15 to 363.15 K and at ambient pressure. Densities of pure water show higher values than that of 3-amino-1-propanol and 1-(2-hydroxyethyl) pyrrolidine, whereas pure 3A1P is more viscous than 1-(2-hydroxyethyl) pyrrolidine and water. The excess molar volumes and viscosity deviations from the data are correlated to the Redlich–Kister equation. The shape and value for the excess molar volumes and viscosity deviations could explain the intermolecular interaction between the molecules.


INTRODUCTION
Lowering carbon dioxide emissions into the atmosphere is an urgent task to deploy. The power sector and cement industry, followed by the refinery sector, are the three biggest CO 2 emission contributors. 1 The absorption process with a chemically reactive solvent is a promising, mature, and viable technology to reduce CO 2 emissions. A temperature swing absorption−desorption process is often used. In the process, gaseous CO 2 contacts and reacts with an amine solvent to produce carbonated solvent in an absorber column. The carbonated solvent is regenerated in a stripper column to reverse the absorption reactions and to produce a pure gaseous CO 2 stream by adding heat. Steam provides heat to reverse the chemical reactions and maintain the pressure in the stripper, making this technology energy-intensive. 2 Finding better solvents is crucial. Solvent candidates should have high CO 2 absorption rates, cyclic capacity, and equilibrium temperature sensitivity. They should be stable under operation with only a few side reactions forming degradation compounds and environmentally friendly.
1-(2-Hydroxyethyl) pyrrolidine is a strong bicarbonateforming solvent for CO 2 capture. 3−5 However, 1-(2-hydroxyethyl) pyrrolidine, a tertiary amine, absorbs CO 2 slowly. Therefore, a promoter (primary amines) is needed to increase the absorption rate. 6 Basic physical properties (like density and viscosity) are essential to comprehensively understand the solvent performances. This work presents experimental density and viscosity data for 3-amino-1-propanol-promoted 1-(2hydroxyethyl) pyrrolidine as a function of concentration and temperature. The measured data were used to calculate excess molar volumes and viscosity deviations. These two properties were then correlated with a Redlich−Kister (RK) model, 7 representing the trends over different temperatures and concentrations. The generated correlations can be used when modeling absorption kinetics and vapor−liquid equilibrium. Table 1 shows an overview of the literature on the density and viscosity of the two amines. For 3-amino-1-propanol data, experimental data for pure and aqueous solutions are available, but for aqueous 1-(2-hydroxyethyl) pyrrolidine solution, only limited data was found. 4 Neither pure data were reported nor blends of these amines. Table  2 were used directly without any further purification. Seven different concentrations of aqueous 1-(2-hydroxyethyl) pyrrolidine solutions, nine of aqueous 3-amino-1-propanol solutions, and five mixtures of 1-(2-hydroxyethyl) pyrrolidine and 3-amino-1-propanol were made. Approximately 50 g of the solutions was prepared gravimetrically using an analytical balance Mettler Toledo ME204. The capacity of the balance was from 0.16 mg to 220 g. The estimated combined expanded uncertainties were (i) for the scale U m ( ) 0.6 mg C i = and (ii) for the prepared solutions U w ( ) 0.0002 C i = in mass fraction and u x ( ) 0.0002 i = in mol fraction. A DMA 4500 density meter coupled with a Lovis 2000ME viscosity meter, shown in Figure S1 (see in the Supporting Information), was used to measure the density and viscosity simultaneously. The DMA 4500 was calibrated by air and ultra-pure H 2 O supplied by the vendor at 298.15 K. The calibration was valid when the difference between the measured and reference was less than 0.03 kg·m −3 . As the calibration at higher temperatures agreed well with the literature, no other temperature calibrations were done. The pressure was measured using DPI520 (Druck, Germany). The detailed procedure for the density measurement can be found in our previous work. 17 The Lovis 2000ME micro-viscometer is based on a rolling ball principle, and operation conditions can be set from 278.15 to 373.15 K. A capillary borosilicate glass (with an internal diameter = 1.59 mm) is filled with a gold-coated stainless-steel ball to cover the viscosity measurements from 0.2 to 65 mPa·s. Ultra-pure H 2 O was used for calibration and was considered successful when the difference between the measured and reference was within 0.05% for at least five measurements.

Material and Methods. The chemicals listed in
The calibration results for water densities show good temperature and density repeatability (±0.01 K and ±0.03 kg·m −3 ) with the estimated expanded uncertainty U(ρ) of the measured density of about 0.5 kg·m −3 . For water viscosities also, good repeatability (±0.01 K and ±0.05 mPa·s) was shown, with the estimated expanded uncertainty U(η) of about 0.08 mPa·s. In addition, two experiments with pure 3A1P at 298.15 K were performed using an Anton Paar MCR 100 rheometer 17  . Viscosity deviations ( ) i as the difference between the pure substances and the (measured) viscosity of the blend are expressed as where x i is the mol fraction and η i and η m are the viscosities of pure substances and mixtures (mPa·s), respectively. The required densities and viscosities of pure components in eqs 1 and 2 were also measured in this work. The measured data and literature data were regressed using eq 3 for density and eq 4 for viscosity, respectively    The excess molar volumes (V i E ) and the viscosity deviations (ln ) i as a function of temperature and concentration can be then correlated by the RK correlation 7 and presented by eq 5 for the binary (Y 1,2 ) and eq 6 for the ternary (Y 1,2,3 ) systems as Temperature dependency of the RK equation is expressed as where Y represents the excess molar volume or viscosity deviation for the considered binary or ternary system, k i are the fitted parameters, n is the degree of correlation, and T is the temperature in Kelvin.

The parameters
The quality of the fit was also expressed by an average absolute relative (AARD) deviation as   where N is the number of the density and viscosity data. Y exp. is the excess molar volume (eq 1) and the viscosity deviations (eq 2), and Y model are the RK model results (eqs 5 to 7).

Setup Calibration and Verification.
The water calibration results are presented in the Supporting Information ( Figure S2). The agreement with the literature was good, with the AARD values of 0.02% for density and 1.2% for viscosity. To further verify the setup, densities and viscosities of 30 mass % of MEA x ( 0.1124) MEA = were studied (see Table S1 in the Supporting Information) and compared to the literature 17 (see Figure S3 for data comparison and Figure S4 for deviations). For viscosity, the data in this work are mainly from a Lovis 2000ME micro viscometer while the literature data are based on Anton Paar MCR 100 rheometer. Still, the agreement was good for 30 mass % of MEA, with AARD of 0.2% for viscosity. For density, the AARD was 0.03%.

Density and Viscosity
Results. The measured densities and viscosities of three binary and one ternary system at different concentrations and temperatures are reported in Tables 3−6 Figure 1 shows the density of the   pure amines and water, and the density correlation (eq 3). The fitted parameters are given in Table S2 in the Supporting Information As shown in Table 1, no density data for pure 1-(2HE)PRLD was found, but some data for pure 3A1P were available. Figure 1 shows good agreement of this work with the literature, except for one data set 11 where the reported data seem to be consistently higher. The chemical purity may explain the difference. Different correlations for the pure components are also presented. The literature data for 3A1P (solid blue line) and H 2 O (green dashed-dot line) 19 underpredict the data, but the proposed correlation by Jones and Harris 20 for H 2 O fits well with the measured data in this work.
Similarly, Figure 2 shows the viscosity of the pure amines and water, and the viscosity correlation (eq 4). The fitted parameters are given in Table S2 (in the Supporting Information). No literature data were found for pure 1-(2HE)PRLD, but some data for pure 3A1P were available. Figure 2 shows that the 3A1P results from this work agree well at higher temperatures (T ≥ 323 K) with literature data 10,14−16 but tend to be lower at lower temperatures (T < 323 K). The lowest temperature has the highest deviation. To verify our data, two experiments at 298.15 K were performed using a different apparatus (Anton Paar MCR 100 rheometer), and the results (■) fit better to this work (□/ ) than other literature data. Thus, it can be speculated if the chemical purity may explain the differences between this work and the literature. The viscosity correlations available in the literature for 3A1P (solid blue line) and H 2 O (magenta line) 19 are also presented. However, the 3A1P data are underpredicted by the literature while the agreement is better for water. The water correlations agree well with the proposed correlation by Kestin et al. 22 and with the results in this work.
The V E values in Tables 3−6 were regressed using the RK model (eq 5), and the fitted parameters are given in Table 7, along with the statistical information related to the fit (RMSE, AARD, uncertainties of the fitted parameter (u k ( ) i ) and pvalues). The p-value described the results from the statistical hypothesis test and can indicate whether the required correlations (eqs 5 and 6) are overparameterized (i.e., if the p-value ≥ 0.1). As given in Table 7 From a liquid theory, 24 the sign of the V E values can be negative or positive compared to their pure liquid component and can be attributed from physical, chemical, and structural characteristics. Physical contribution (like dispersion forces and non-specific physical interactions) is usually weak and gives a positive value. Chemical contribution and structural characteristics can both give positive and negative values. The positive value in the chemical contribution is associated with the breaking up in pure liquid components. However, negative values for specific interactions, such as the formation of hydrogen bonding, charge transfer, and dipole−dipole interactions, are common. 25 The positive and negative values for the structural characteristic are due to the favorable and unfavorable geometrical fitting of the molecular structure (e.g., size, shape, and volume).
The V E of the studied systems show negative values (relative to the ideal ones obtained from the corresponding mol fraction and density of pure components in eq 1). The negative V E indicates a volume contraction in the real mixture, which can be attributed to the chemical and structural characteristic    contributions. As alkanolamine, 1-(2HE)PRLD and 3A1P have two different functional groups (see Table 1 has an aminopropanol group. 27 It may be the consequence of the mixture being favorable for geometric filling due to steric hindrances at lower temperatures but then to be an unfavorable structural geometric at higher temperatures due to molecular motion (Figure 3c). The two binary aqueous solutions show the largest negative V E values, indicating the most efficient packing 28 in the system, but for blends of the two amines, the deviations were smaller. The representations of the RK model for the density data for the binary systems are presented in the Supporting Information (see Figures S6−S8). The reported densities data at low concentrations 13 seem to fit better to this work than the reported data from the literature 12 Table 8. The comparison of the model to literature data 14 for 3A1P(2)/ H 2 O(3) is also available in the Supporting Information (see Figure S10).
The Δη show positive values in the entire investigated temperature and concentration ranges and decrease with increasing temperatures. The positive value usually pairs with the negative excess of the molar volume where the chemical and structural characteristics dominate. A more positive value shows that the viscosity of the real mixture is higher than that of the ideal solution in which the molecules are getting closer to each other and have less space due to strong chemical interactions and a compact structure. The decreasing Δη value with increasing temperatures indicates that the chemical interaction becomes weaker due to faster molecule movement. The maximum Δη value indicates the strongest interaction between components.

Journal of Chemical & Engineering Data
pubs.acs.org/jced Article From the fitted parameters of the RK model, the viscosities of the considered system are shown in Figures S11−S13 in the Supporting Information and for ternary 1-(2HE)PRLD(1)/ 3A1P(2)/H 2 O(3) in Figures 6 and S14 in the Supporting Information The agreement between the model and the viscosity data is good, and three observations can be made: (i) when the total amine concentration is low (pink ), the viscosity of the solution is slightly higher than that of H 2 O.
(ii) When the total amine concentration (blue /red /green ) is high, the viscosity is similar to that of pure 3A1P.
(iii) Between these points with medium total amine concentration (black ), the viscosity is similar to that of pure 1-(2HE)PRLD.

CONCLUSIONS
A DMA4500 density meter coupled with a Lovis ME2000 viscosity meter was used to measure the densities and viscosities of pure 1-(2HE)PRLD, 3A1P, H 2 O, and their blends from 293.15 to 363.15 K at ambient pressure. The excess molar volumes show a negative volume due to a volume contraction for aqueous binary solutions. Intermolecular interactions and packing effects for different molecular sizes can explain this behavior. The negative values of the excess molar volume pair with the positive value of the viscosity deviations, indicating a positive deviation from its ideal mixture. The excess molar volume and the viscosity deviation are correlated to the RK model, and the fitted parameters are reported. The agreement between the data and correlations is good, and the correlations can be used to correlated viscosity and density as a function of temperature and concentration.