Experimental Measurements of Amine-Functionalized Anion-Tethered

Dec 2, 2010 - Furthermore, the CO2 absorption isotherms of [P66614][Gly], [P66614][Ile], [P66614][Sar], and [P66614][Ala] were measured using a volume...
17 downloads 13 Views 613KB Size
Ind. Eng. Chem. Res. 2011, 50, 111–118

111

Experimental Measurements of Amine-Functionalized Anion-Tethered Ionic Liquids with Carbon Dioxide Brett F. Goodrich, Juan C. de la Fuente,*,† Burcu E. Gurkan, David J. Zadigian, Erica A. Price, Yong Huang, and Joan F. Brennecke* Department of Chemical and Biomolecular Engineering, UniVersity of Notre Dame, Notre Dame, Indiana 46556, United States

Amine-functionalized anion-tethered ionic liquids (ILs), trihexyl(tetradecyl)phosphonium glycinate [P66614][Gly], alanate [P66614][Ala], sarcosinate [P66614][Sar], valinate [P66614][Val], leucinate [P66614][Leu], and isoleucinate [P66614][Ile], were synthesized and investigated as potential absorbents for CO2 capture from postcombustion flue gas. Their physical properties, including density, viscosity, glass transition temperature, and thermal decomposition temperature, were determined. The influence of changing the anion and, more specifically, the length of the alkyl chain is discussed. Furthermore, the CO2 absorption isotherms of [P66614][Gly], [P66614][Ile], [P66614][Sar], and [P66614][Ala] were measured using a volumetric method, and the results were modeled with two different Langmuir-type absorption models. All four ILs reached greater than 0.5 mol of CO2 per mole of IL at CO2 pressures of less than 1 bar. This indicates the predominance of the 1:1 mechanism, where the CO2 reacts with one IL to form a carbamic acid, over further reaction with another IL to make a carbamate (the 1:2 mechanism). The chemical absorption of CO2 dramatically increased the viscosity of the IL, but this can be mitigated to some extent by decreasing the number of hydrogens on the anion available for hydrogenbonding. CO2 + R+NH2 T R+N+H2CO2

Introduction Coal provides just under half of the electricity consumed by Americans1 and makes even greater contributions to power production in China and India. Existing coal and natural gas power plants will need a postcombustion carbon capture system to limit the pollution we leave behind for future generations and to satisfy anticipated future regulations on carbon dioxide (CO2) emissions. The current leading technology involves the use of aqueous amine solutions (typically 30% amine by volume), which will chemically absorb CO2.2-4 Near-term implementation of carbon capture for postcombustion flue gas would most certainly take advantage of this technology.5 However, conventional aqueous amine solutions have many documented disadvantages including a high parasitic energy load, degradation in the presence of oxygen, and corrosion. Moreover, the loss of volatile amines causes environmental pollution, as well as raises the cost of operation through solvent replacement.6 Ionic liquids (ILs) are an attractive replacement because they have negligible vapor pressure, which eliminates the volatilization problem. They are liquid over a wide range of temperatures, have decomposition temperatures as high as 300-400 °C, and possess virtually limitless chemical tunability.7 Bates et al. tethered an amine to the cation of an IL and demonstrated that amine-functionalized ILs will chemically react with CO2 according to the reaction mechanism in eqs 1 and 2.8 The CO2 first reacts with the amine functionality on the IL (eq 1), which then reacts with another amine (on another IL) to form a neutral carbamate with an ammonium dication. Of course, the charge is balanced by the anions of the two ILs, which are not shown in eqs 1 and 2. * Corresponding author. Tel: (574) 631-5847. Fax: (574) 631-8366. E-mail: [email protected]. † Present address: Departamento de Ingenierı´a Quı´mica y Ambiental, Universidad Te´cnica Federico Santa Marı´a, Valparı´so, Chile.

(1)

+ + + + R+N+H2CO2 + R NH2 T R NHCO2 + R N H3

(2) This reaction mechanism results in a maximum of one CO2 being captured for every two ILs (a 1:2 mechanism), which is akin to that observed for most aqueous amines. Although this results in CO2 capacity by ILs significantly better than physical absorption,9-11 it is not particularly efficient reaction stoichiometry. Theoretical studies suggested that a 1:1 CO2:IL molar ratio is achievable when the amine is tethered to the anion instead of the cation,12 and we have demonstrated this experimentally for trihexyl(tetradecyl)phosphonium methioninate ([P66614][Met]) and prolinate ([P66614][Pro]).13 The theoretical calculations show that the dianion that would be formed in the second reaction, eq 4, is relatively chemically unstable, and therefore one would expect the reaction to terminate at eq 3 upon addition of CO2 to amine-functionalized anion-tethered ILs. CO2 + R-NH2 T R-N+H2CO2

(3)

- + R-N+H2CO2 + R NH2 T R NHCO2 + R N H3

(4) In this work, we synthesized amine-functionalized aniontethered ILs, including trihexyl(tetradecyl)phosphonium glycinate ([P66614][Gly]), alanate ([P66614][Ala]), sarcosinate ([P66614][Sar]), valinate ([P66614][Val]), leucinate ([P66614][Leu]), and isoleucinate ([P66614][Ile]). The structures of these compounds, as well as trihexyl(tetradecyl)phosphonium methioninate ([P66614][Met]) and prolinate ([P66614][Pro]), are shown in Figure 1. The anions chosen are simple amino acids with only carboxylate and amino functional groups. The P66614 cation was chosen somewhat arbitrarily as a relatively inexpensive, readily available, thermally stable cation. It plays very little role in

10.1021/ie101688a  2011 American Chemical Society Published on Web 12/02/2010

112

Ind. Eng. Chem. Res., Vol. 50, No. 1, 2011

Figure 1. Structure of ILs used in this study.

determining CO2 uptake but does have a significant effect on physical properties, as will be discussed below. Here we only focus on the anion, which is the moiety reacting with the CO2. In future publications we will address how the cation can be modified to reduce the molecular weight and the heat capacity, which would have a significant impact on the mass of absorbent needed in an industrial process, as well as the regeneration heat load. We experimentally measure the CO2 absorption isotherms and model them to better understand the reaction mechanism occurring. We further characterize the effect of CO2 on the physical properties of this class of ILs. In particular, we are concerned about the viscosity, which increases dramatically in the presence of CO2. Experimental Methods Ionic Liquid Synthesis. [P66614][Br] (97% purity, Cytex Industries, Inc.) was diluted in methanol (ACS grade, Fisher Scientific) down to 2 M. The solution was then reacted to form [P66614][OH] by the addition of 2:1 mol equiv of an ion-exchange resin (DOWEX SBR LC NR (OH), Dow Chemical Company). This was done in two approximately equal additions of the resin, allowing 1 day of reaction for each addition. A 1.1 mol equiv of the appropriate amino acid (all >98%, Sigma Aldrich) was added to the filtered solution and stirred overnight. Excess methanol was removed with a rotary evaporator at 45 °C. Cold acetonitrile (ACS grade, Fisher Scientific) was used to crystallize any remaining neutral amino acid, which was removed by vacuum filtration. Solvent and water removal was achieved by pulling a moderate vacuum at 60 °C for approximately one week. 1 H NMR spectra were recorded on Varian 300 spectrometers, and these results are shown in Supporting Information. No impurities were identified from the NMR results. Bromide

impurity content was determined to be no greater than 6000 ppm using a Cole-Parmer bromide-selective ion conductivity probe, model EW-27504-02. All ILs are hygroscopic, so water content was measured by a Brinkman 831 Karl Fischer Coulometer. Water content prior to measurements was below 800 ppm. Density Measurements. Densities were measured at atmospheric pressure in a DMA 4500 Anton Paar oscillating U-tube densitometer, which includes an automatic correction for the viscosity of the sample, as described previously.14 The uncertainty was estimated to be (5 × 10-5 g · cm-3. Two integrated Pt 100 platinum thermometers provided good precision ((0.01 K) in temperature control internally. Glass Transition Temperature Measurements. The glass transition temperature was measured with a Mettler-Toledo differential scanning calorimeter (DSC), model DSC822e, and the data was evaluated using the Mettler-Toledo STARe software version 7.01, as described previously.15 Samples were dried in situ in the DSC. The presence of volatiles significantly affects the glass transition temperature; therefore, samples were dried repeatedly until the phase transition temperatures remained constant. The equipment has an accuracy of (0.3 K when tested with a standard solution, but the presence of trace impurities raises the uncertainty to (1 K. Decomposition Temperature Measurements. Decomposition temperatures were measured with a Mettler-Toledo TGA/ SDTA 851e/SF/1100 °C thermal gravimetric analyzer, as described previously.15 We report the onset temperature, which is the intersection of the baseline weight, either from the beginning of the experiment or after the drying step, and the tangent of the weight vs temperature curve as decomposition occurs. The samples were run in aluminum pans under a nitrogen atmosphere at a heating rate of 0.17 K · s-1. When we observed any weight loss from the evaporation of water from the sample, it was further dried in situ at 403 K for 30 min. Reproducibility was verified by running three replicates for each ionic liquid. The largest uncertainty is from manually determining the tangent point, which results in an uncertainty in the thermal decomposition temperatures of (2 K. CO2 Absorption Measurements. The CO2 absorption was measured using a calibrated stirred vessel and carefully measuring the CO2 introduced into that vessel from a calibrated chamber. The calibrated chamber (177 mL) was filled with CO2 (99.99% purity, Praxair) to a pressure of about 2 bar. The stirred vessel (129 mL for Setup A, 257 mL for Setup B) was loaded with a known amount of IL and then evacuated to a pressure of about 5 mbar. The stirred vessel was held at a constant pressure (tolerance 1 mbar) for a minimum of 1 day to ensure that there were no substantial leaks. Then the valve that connected the stirred vessel to the calibrated chamber was briefly opened. The amount of CO2 to enter the stirred vessel was calculated from the pressure drop in the calibrated chamber and verified by the pressure increase in the stirred vessel. A magnetic stir bar was activated to mix the IL, and the pressure was recorded every 4 to 8 h. When the pressure remained constant for a minimum of 12 h, the system was considered to be at equilibrium. The amount of CO2 absorbed into the IL was calculated from the pressure drop in the stirred vessel between the time when CO2 entered the vessel and when equilibrium was reached. Viscosity Measurements. The viscosity of the ILs was measured with a Brookfield DV-III Ultra (cone and plate) rheometer, as described previously.14 Although the nominal uncertainty of the instrument is (2%, the actual uncertainties are higher due to slight changes in the concentration of water

Ind. Eng. Chem. Res., Vol. 50, No. 1, 2011

113

Figure 2. Density of [P66614][Gly] (empty circles), [P66614][Ala] (filled diamonds), [P66614][Sar] (empty triangles), [P66614][Val] (stars), [P66614][Leu] (dashes), and [P66614][Ile] (empty squares) with temperature.

Figure 3. Natural log of density of [P66614][Gly] (empty circles), [P66614][Ala] (filled diamonds), [P66614][Sar] (empty triangles), [P66614][Val] (stars), [P66614][Leu] (dashes), and [P66614][Ile] (squares) with temperature.

in the samples, as described in the next paragraph. Repeat experiments showed a precision of (5%. Temperature was controlled with a precision of (0.1 K by means of a TC-602 bath thermostat with a Brookfield temperature controller unit attached. The measurement chamber for the rheometer had either nitrogen or CO2 flowing over the sample, depending on if the sample was neat or saturated with CO2. The sample was exposed to the atmosphere briefly during loading and unloading, which introduces an opportunity for the sample to absorb moisture from the atmosphere. Karl Fischer measurements on samples before and after the viscosity measurements typically showed a rise in water content for the neat ILs from about 200 ppm to about 1000 ppm. The CO2-saturated samples were too viscous to measure the water content using the Karl Fischer coulometer. However, the high viscosity of the samples would impede diffusion, so it was assumed that the CO2-saturated samples absorbed less water than their neat counterparts during transfer of the samples into the viscometer.

Table 1. Glass-Transition and Decomposition Temperatures

Results and Discussion Density. Density measurements of pure ILs were carried out at temperatures ranging from 10 to 70 °C at 1 bar. The experimental density data obtained is reported in Figure 2 for all of the pure ionic liquids studied. The density was observed to decrease as the alkyl chain length increased in the order [P66614][Gly] > [P66614][Ala] > [P66614][Sar] > [P66614][Val] > [P66614][Leu] > [P66614][Ile]. This agrees with reported trends from other groups.16,17 At 25 °C, the experimental densities of [P66614][Gly] (0.9039 g/mL) and [P66614][Ala] (0.9030 g/mL) are lower than the densities reported by Zhang et al.18 (0.9630 and 0.9500 g/mL) for the same anions with a smaller cation, tetrabutylphosphonium [P4444]. This supports the observation that increasing the alkyl chain length on either the cation or anion will decrease the density. The apparent linear relation observed in Figure 2 between density and temperature has already been reported by several authors.14,16,19,20 Further analysis by Rodrı´guez et al. revealed that there is a subtle curvature in the data, and a plot of the natural log of the density versus temperature can also be fit with a straight line with statistically significant improvement (Figure 3).14 Glass Transition and Thermal Decomposition Temperatures. Table 1 shows that the glass transition temperatures of the [P66614]+ amino acid ILs varied from -86 to -69 °C. This is lower than the values for the same anions with the

compound

glass transition temp, °C

decomposition temp, °C

[P66614][Gly] [P66614][Ala] [P66614][Sar] [P66614][Val] [P66614][Leu] [P66614][Ile]

-73 -69 -73 -73 -76 -86

319 336 329 332 308 340

1-ethyl-3-methylimidazolium [emim]+,21 tetrabutylphosphonium [P4444]+,7 and (3-aminopropyl)tributylphosphonium [aP4443]+ 16 cations. There is no consistent order for the [Ala]-, [Gly]-, [Val]-, [Ile]-, and [Leu]- anions when the cation is changed, but the range of the glass transition temperatures for these anions with the various cations is very narrow (just 11 to 17 °C from the lowest to highest values). All of the ionic liquids tested had decomposition temperatures higher than 300 °C (Table 1). These values are higher than those for the smaller [P4444]+ cation.7,18 The [P66614]+ cation is a very stable cation with a decomposition temperature as high as 420 °C, achieved when paired with a more stable anion.22 We suspect asymmetry of the [P66614]+ cation increases its stability. Although multiple groups have demonstrated that the introduction of asymmetry to the phosphonium cation affects the viscosity and glass transition temperature,22,23 to our knowledge the effect on decomposition temperature has not yet been reported and should be the subject of further research. CO2 Absorption. [P66614][Ala], [P66614][Gly], [P66614][Ile], and [P66614][Sar] were selected to experimentally measure the CO2 uptake at room temperature. Their isotherms are plotted in Figure 4 with previously reported data for [P66614][Met] and [P66614][Pro].13 The size of the symbols in the graphs are a reasonable estimate of the uncertainty in the data. There are two distinct portions of each isotherm: a large uptake at low pressures due to chemical reaction and a further increase in capacity at higher pressures due to physical absorption. All isotherms rise significantly above 0.5 mol of CO2 per mole of IL, which would be the expected capacity if the 1:2 mechanism was dominant. Initial experiments with [P66614][Ala] showed it absorbed significantly less than the other ILs, and therefore it was not considered for further investigation. Based on preliminary FTIR measurements, it is possible that the [P66614][Ala] sample was already partially complexed with CO2 from short exposure to the atmosphere, which could explain its lower apparent capacity. [P66614][Ile], [P66614][Sar], [P66614][Met], and [P66614][Pro] reached saturation at values slightly less than 1:1 stoichiometry. This deviation is consistent for different batches

114

Ind. Eng. Chem. Res., Vol. 50, No. 1, 2011

Figure 4. CO2 capacity in CO2:IL molar ratio at room temperature (22 °C) for [P66614][Gly] (filled circles), [P66614][Ile] (empty squares), [P66614][Sar] (filled triangles), [P66614][Ala] (filled diamonds), [P66614][Pro] (gray dashes),13 and [P66614][Met] (gray stars).13

Figure 5. CO2 capacity in CO2:IL molar ratio at room temperature (22 °C) for [P66614][Ile] (empty squares) fitted with the two-reaction model when all variables are free (dashed line) and with the Henry’s law constant fixed at 100 bar (dotted line).

Table 2. Calculated Parameters for the Two-Reaction Modela compound

H (bar)

k1 (bar-1)

k2 (unitless)

[P66614][Ile] [P66614][Sar] [P66614][Gly] [P66614][Pro] [P66614][Met]

26 1000 5.5 1000 18

130 9400 350 350 850

0.72 87 -1.3 0.45 47

a

Calculated parameters for the CO2 absorption model in eqs 5-7.

and cannot be easily explained by impurities. The NMR spectra do not show any significant impurities, and the residual halide content (