Comparative study of solvent properties for carbon dioxide ... · PDF fileComparative study of solvent properties for carbon dioxide absorption ... solubility of carbon dioxide

PAPER www.rsc.org/ees | Energy & Environmental Science

Comparative study of solvent properties for carbon dioxide absorption

Ortrud Aschenbrenner and Peter Styring*

Received 10th February 2010, Accepted 9th June 2010

DOI: 10.1039/c002915g

Several inexpensive and non-toxic solvents with low vapour pressures were investigated for their

suitability as alternative solvents for the absorption of carbon dioxide from flue gas. The solvents

include poly(ethylene glycol)s, poly(ethylene glycol) ethers, poly(ethylenimine) and glycerol-based

substances. Solvent properties such as thermal stability, solubility of carbon dioxide and selectivity over

nitrogen were investigated in a systematic study using a thermogravimetric analyser. Absorption results

are reported for pure carbon dioxide and nitrogen as well as a mixture of both gases. Desorption and

long-term sorption behaviour are also discussed. Glycerol and poly(ethylene glycol)s show a high

solubility of carbon dioxide. Due to the high viscosity of the solvent, carbon dioxide absorption in

poly(ethylenimine) is very slow in spite of the presence of favourable amine groups. PEG 300 was found

to be the best solvent in this study and shows a high carbon dioxide solubility as well as good selectivity

over nitrogen. The advantages of high stability, low solvent loss and low desorption energy of PEG 300

may outweigh its lower absorption capacity compared to the state-of-the-art solvent

monoethanolamine, making it a potentially advantageous solvent for industrial carbon dioxide

absorption processes.

Introduction

Due to international efforts for the reduction of greenhouse gas

emissions, the capture of carbon dioxide (CO2) from flue gas has

gained increasing interest in recent years.1 Carbon capture

technologies are still far from being the industrial standard, and

processes vary from pre-combustion to post-combustion tech-

nologies. Current approaches are targeted at carbon capture and

storage (CCS) where the gas is transported, post-capture, and

stored in underground reservoirs or aquifers. However, this is

a net loss of carbon from the economy and so the need has been

identified to not only capture carbon dioxide but also to re-cycle

it by using it in reactions to synthesise value-added chemicals.

Various alternative methods have been developed for the fixation

of carbon dioxide in commercially useful products. Recent and

ongoing research includes the chemical conversion of carbon

dioxide to a wide range of potential products, photochemical

reduction or artificial photosynthesis and biological trans-

formation using bacteria or algae.2 Although carbon dioxide

separation processes have been widely used in industry, for

Department of Chemical & Process Engineering, The University ofSheffield, Sir Robert Hadfield Building, Sheffield, UK S1 3JD. E-mail:[email protected]

Broader context

The absorption capacity and selectivity for carbon dioxide over ni

polymers. Thermogravimetric analysis was used in order to mini

screening procedure. While the best of these (PEG 300) shows arou

desorption and regeneration costs are significantly lower and less en

low vapour pressures and so solvent loss by evaporation is significan

stable for prolonged periods over the expected range of operation.

1106 | Energy Environ. Sci., 2010, 3, 1106–1113

example in the purification of natural gas, capturing carbon

dioxide from flue gas presents an additional challenge.1 Flue gas

usually discharges at atmospheric pressure and at a partial

pressure of approximately 0.15 bar in nitrogen. Membrane

processes, which have been used for the effective separation of

carbon dioxide from natural gas, are therefore not viable due to

the high pressures required.3 One of the most widely used

methods for carbon dioxide capture to date is the absorption in

a solvent at near-ambient temperature and pressure and subse-

quent solvent regeneration at elevated temperature and/or

reduced pressure.4

The selection of a suitable solvent is crucial for the economic

viability of the process.5 The main selection criteria are high

solubility of carbon dioxide and, equally important, high

absorption selectivity of carbon dioxide over nitrogen (N2).

Furthermore, easy desorption is highly desirable, as it reduces

the necessary regeneration temperature and pressure difference.

In order to prevent the loss of solvent, a low vapour pressure and

high thermal stability as well as long-term stability are beneficial.

Additionally, the cost and environmental toxicity of the solvents

have to be taken into account, especially when evaporative loss

and chemical degradation are taken into account.

Currently, the most widely used solvents for carbon dioxide

separation are amine solutions.5,6 Amines selectively absorb

trogen have been investigated for a number of low cost liquid

mise the quantity of adsorbent required and provide a rapid

nd 30% efficiency compared to the industry standard MEA, the

ergy intensive. Furthermore, the liquids investigated have very

tly reduced. The liquids are also shown by TGA to be thermally

Therefore, the overall environmental impact may be reduced.

This journal is ª The Royal Society of Chemistry 2010

carbon dioxide over nitrogen with a high absorption capacity, as

they react with CO2 to form carbamates. The most commonly

used amine-based absorbent is monoethanolamine (MEA).7

However, the amine-based absorption process is still at the

research stage,8,9 albeit at the pilot plant scale, as the use of amine

solutions for CO2 absorption has some disadvantages. The main

problems are the high energy requirement for solvent regenera-

tion, their high vapour pressure and subsequent mass loss

through evaporation, degradation of the solvent and associated

plant corrosion.4,7,10

If physical absorption is used rather than chemical absorption,

solvent regeneration is easier and less energy intensive. Solvents

currently used for physical absorption of carbon dioxide are, for

example, methanol, sulfolane and poly(ethylene glycol) ethers.11

The best-known example is Selexol, a commercial mixture of

poly(ethylene glycol) dimethyl ethers (poly ¼ 1 to 11) with

optimised properties.12,13 The main disadvantage of these phys-

ical solvents is the energy needed for gas adsorption and solvent

regeneration via pressure or temperature swing and potential

solvent loss due to volatility.14 When carbon dioxide capture and

activation pathways are considered,2 physical adsorption

solvents have an additional benefit by potentially directly

combining the absorption process with a subsequent or simul-

taneous catalytic conversion of carbon dioxide to useful

products.

Ionic liquids have been suggested as alternative physical

solvents for carbon dioxide absorption due to their extremely low

vapour pressures. Various ionic liquids were found to absorb

CO2 with high selectivity over N2.15,16 Polymers of ionic liquids

have also been reported to have high CO2 absorption capacity

and selectivity over N2, with fast and completely reversible

absorption.17,18 However, ionic liquids are at present considered

to be too expensive for large-scale industrial applications.

In this study, more commonly available solvents with low

vapour pressures for the absorption of CO2 from flue gas are

proposed: poly(ethylene glycol), for example, has been reported

to exhibit high CO2 solubility selectivity over N2.19 Liquid

poly(ethylene glycol) enhances CO2 adsorption over amine-

based solids.20 Poly(ethylene glycol)s have low vapour pressures

and are stable and non-toxic. Glycerol is another stable and non-

toxic liquid with low vapour pressure that is available in vast

quantities as a by-product of bio-diesel production, that can be

used as solvent for CO2. Kovvali and Sirkar reported21 a low

selectivity of CO2 over N2 for glycerol, whereas a much higher

selectivity was found for glycerol carbonate, which is also stable,

non-volatile and non-toxic. It is, however, difficult to compare

the suitability of all these substances for CO2 absorption, because

the reported data for solubility and selectivity were obtained with

very different experimental methods. A systematic study of the

relevant properties of these substances is required in order to

evaluate their performance as solvents for CO2 capture from flue

gas.

The aim of this study was to investigate and compare the

suitability of several commercially available alternative solvents

for the capture of carbon dioxide from flue gas. The study

includes glycerol-based substances (glycerol and glycerol

carbonate) as well as polymeric liquids, namely poly(ethylene

glycol)s, poly(ethylene glycol) ethers and poly(ethylenimine).

These solvents have low vapour pressures, low toxicity and low


or moderate cost. Solubilities of carbon dioxide and nitrogen,

measured using a thermogravimetric method, are reported along

with further properties such as thermal stability and vapour

pressure. Liquids were selected rather than solid adsorbents due

to their ease of transportation around processes through

pumping through pipes. The performance of the solvents was

evaluated and compared to the state-of-the-art solvent mono-

ethanolamine, which shows a CO2 solubility of 43.8 g l�1. To our

knowledge, this is the first reported systematic study of carbon

dioxide and nitrogen solubility for these solvents.

Materials and methods

Materials

Glycerol (98%) was supplied by Prolabo. 4-Hydroxymethyl-1,3-

dioxolan-2-one (glycerol carbonate), tetra(ethylene glycol)

dimethyl ether (tetraglyme, 99%), poly(ethylene glycol) 150

dimethyl ether (PEGDME 150), branched poly(ethylenimine),

poly(ethylene glycol) 200 (PEG 200) and poly(ethylene glycol)

300 (PEG 300) were obtained from Sigma Aldrich. Poly(ethylene

glycol) 600 (PEG 600) was obtained from Acros. All solvents

were used as received. The structures of the solvents are shown in

Fig. 1. Dry carbon dioxide (99.8%) and nitrogen were supplied

by BOC.

Method

A thermogravimetric analyser (Perkin Elmer Pyris 1 TGA) was

used for the experiments. All experiments were performed at

atmospheric pressure with carbon dioxide, nitrogen or a 50% v/v

mixture of both gases, the ratio being maintained by flow

controllers. The gases were obtained dry and were also passed

through in-line silica gel drying tubes. The flow rate was 50 cm3

min�1 for each gas. A small amount of solvent (20–40 mg) was

placed in a ceramic sample pan and suspended in the furnace of

the thermogravimetric analyser.

For the thermal stability experiments, the sample was heated

up to 700 �C at a heating rate of 10 K min�1 with a flow of

nitrogen or carbon dioxide. For some experiments, a heating rate

of 20 K min�1 was used. There was no significant effect of the

heating rate on the results.

For the absorption experiments, the sample was heated to

100 �C and held at this temperature for at least 30 min prior to

the experiment in order to remove any absorbed gas and water.

The sample was then cooled at 200 K min�1 to the absorption

temperature 25 �C and held at this temperature until the weight

remained constant. This took up to 5 hours. Each experiment

was performed at least three times and the average value

determined.

Results and discussion

Stability

Thermal stability of the solvents for carbon dioxide absorption is

important with respect to solvent regeneration at elevated

temperatures. All the substances were therefore examined using

thermogravimetric analysis.

Energy Environ. Sci., 2010, 3, 1106–1113 | 1107

Fig. 1 Structures of the selected solvents for absorption studies: (a) glycerol, (b) glycerol carbonate, (c) poly(ethylene glycol), (d) poly(ethylene glycol)

dimethyl ether and tetraglyme (n ¼ 4), and (e) poly(ethylenimine).

Table 1 Thermogravimetrically obtained onset temperatures for totalevaporation/decomposition and vapour pressure at 95 �C

SubstanceMolar massin g mol�1

Onsettemperature/�C

Vapourpressure/Pa

Glycerol 92.09 237 15.8a

Glycerolcarbonate

118.09 237 7.6b

Tetraglyme 222.28 177 60.5b

PEGDME 150 150 205 78.8b

PEG 200 200 202 9.9b

PEG 300 300 272 1.5b

PEG 600 600 404 0.2b

Poly(ethylenimine) ca. 10 000 370 0.2b

a Data from: Cammenga, et al.,22 Ross and Heideger;23 b Data from:Aschenbrenner, et al.24

For all the substances used in this study, the temperature scan

graphs show only one step where the weight is rapidly reduced to

approximately zero, as shown in Fig. 2 for PEG 300. Generally

no residue was found in the sample pan. This behaviour was

independent of the carrier gas used and can be attributed to total

evaporation or decomposition of the sample. The onset

temperature of the step was determined for the various

substances and is listed in Table 1 along with the vapour pressure

at 95 �C. The type of carrier gas used (carbon dioxide or

nitrogen) had no influence on the onset temperature. As the onset

temperature marks the beginning of a weight loss of 100%, this

temperature can be regarded as the approximate boiling point of

the substances. In case of the polymeric substances that do not

exist as a vapour, it represents the approximate decomposition

temperature. As seen from Table 1, the obtained boiling or

decomposition temperatures of the substances are consistent

with the vapour pressure data, since in most cases the substances

with higher vapour pressure at 95 �C show the lower boiling or

decomposition temperature.

As can be seen from Table 1, PEG 600 and poly(ethylenimine)

have the highest boiling/decomposition temperatures and lowest

Fig. 2 Thermogravimetric data for PEG 300 under nitrogen at

a temperature scan rate of 10 K min�1.


vapour pressures of the substances used. Both have a very high

viscosity and are paste-like at room temperature, whereas the

other solvents are liquids. The high viscosity of PEG 600 and

poly(ethylenimine) is due to their high molar masses compared to

the other substances in this study.

Solubility of carbon dioxide

The solubility of carbon dioxide at 25 �C in the various

substances was measured using the thermogravimetric method

described in the Experimental section. The results are shown in

Table 2 together with the solvent densities. The solubility data

are shown in mg g�1 as obtained directly from the experiments.

The listed densities were then used to calculate the solubility in g

l�1. Data for the common solvents methanol and water are also

included in the table for reference.

For most of the solvents, equilibrium adsorption was

reached in less than four hours. However, in the case of poly-

(ethylenimine) the absorption of carbon dioxide was so slow that

equilibrium was not reached even after six hours. This may be

explained by the high viscosity of poly(ethylenimine) compared


Table 2 Solubility of carbon dioxide at 25 �C. Solvent density data asprovided by supplier

SubstanceSolventdensity in g l�1

Solubilityin mg g�1

Solubilityin g l�1

Glycerol 1250 13.8 17.2Glycerol carbonate 1400 7.9 11.0Tetraglyme 1011 4.8 4.9PEGDME 150 1089 6.4 6.9PEG 200 1124 13.4 15.1PEG 300 1124 13.5 15.1PEG 600 1124 7.7 8.7Poly(ethylenimine) 1030 >3.0 >3.1Methanol25,26 788 7.7 6.1Water27,28 997 1.5 1.5

Fig. 3 Polar interaction of a terminal hydroxyl group with a free carbon

dioxide molecule.

to most of the other substances in this study, making diffusion of

carbon dioxide into the solvent more difficult. The data for

poly(ethylenimine) in Table 2 represent the average concentra-

tion after six hours when the weight was still increasing, albeit

much more slowly than at the start of the absorption process.

From the shape of the uptake curve, the equilibrium solubility

value can be estimated to lie between 4.5 and 9.0 mg g�1 and

therefore in the same range as the other substances in this study.

Poly(ethylenimine) contains –NH2 and –NH groups that can be

expected to increase the absorption of carbon dioxide due to

strong polar interactions and reaction between the functional

groups and the carbon dioxide molecules. However, due to the

branched network structure of poly(ethylenimine) shown in

Fig. 1e, many of these groups may be located in the interior and

may therefore not be accessible to carbon dioxide molecules. The

presence of these functional groups therefore does not present an

advantage for carbon dioxide absorption in polymeric liquids.

However, a lower degree of polymerisation and a higher number

of amine groups may lead to increased carbon dioxide absorp-

tion, especially if porous structures can be engineered.

Rolker et al. obtained high carbon dioxide absorption29 for

a hyperbranched poly(ethylenimine) with a molar mass of just

615 g mol�1 as opposed to a molar mass of ca. 10 000 g mol�1 for

the poly(ethylenimine) used in this study. The hyperbranched

poly(ethylenimine) investigated by Rolker et al. showed far

higher absorption of carbon dioxide than poly(ether)s and

poly(ester)s of similar molar mass.29 A recent study by

Ismael et al. indicates that the presence of water is essential for

the reaction of amine groups with carbon dioxide.9 The total

absence of water in our experiments may therefore be another

reason for the poor performance of poly(ethylenimine) for

carbon dioxide absorption. Carbamates are notoriously insol-

uble in non-protic solvents, often leading to the formation of

a precipitate film at the liquid surface. In these studies, no

precipitate was observed on carbon dioxide adsorption with

poly(ethylenimine).

According to Table 2, all of the solvents in this study exhibit

far better carbon dioxide solubility than water, and the majority

of the solvents perform significantly better than methanol. The

best carbon dioxide solubility is obtained for glycerol, PEG 200

and PEG 300. These are the solvents with the highest density of

free –OH groups. Although carbon dioxide is a non-polar gas,

the polarity of the individual C–O bonds in the molecule allows

for interaction with polar groups. Thus, carbon dioxide can act


as Lewis acid or Lewis base and participate in hydrogen

bonding.30 Previously, the high solubility of carbon dioxide in

liquid poly(ethylene glycol) membranes found in several studies

was attributed to acid–base reactions of the acidic carbon dioxide

with the electron-rich ether oxygen in the PEG molecules.31

However, this cannot be the only explanation, as solvents such as

poly(ethylene glycol) dimethyl ethers show lower solubility of

carbon dioxide in spite of their ether groups, as seen in Table 2. It

is therefore likely that the terminal –OH groups have a higher

affinity for carbon dioxide molecules and increase absorption

compared to substances which contain only ether groups. The

high polarity of the bonds in the carbon dioxide molecule allows

a strong electron interaction with the highly polar –OH groups in

the solvent molecules as shown in Fig. 3.

It is interesting to note the dependence of the carbon dioxide

solubility on the average molar mass of the different poly-

(ethylene glycol)s. The solubility of CO2 in PEG 200 and PEG

300 is almost identical. Obviously, these two solvents have

a similar number of –OH and –O– groups, as well as a similar

average chain length of 8 and 12 carbon atoms, respectively, and

therefore similar affinity to carbon dioxide. PEG 600 has much

larger molecules with an average chain length of 26 carbon

atoms. This drastically reduces the density of available –OH

groups with high affinity to carbon dioxide, compared to

the number of less favourable ether –O– groups. Furthermore,

the high number of carbon atoms per molecule increases the

viscosity of PEG 600 compared to PEG 200 and PEG 300,

making diffusion of carbon dioxide to the functional groups

more difficult. Both effects result in lower carbon dioxide

absorption for PEG 600. PEGs with higher molar mass will most

likely show similarly low carbon dioxide solubility.

Some literature data were found for the solubility of carbon

dioxide in similar solvents at 25 �C. Kovvali and Sirkar found

a value of 2.8 g l�1 for the solubility of carbon dioxide in glycerol

carbonate,21 which is a low value compared to the 11.0 g l�1

found in this study. However, the value given by Kovvali and

Sirkar was calculated from membrane permeability rather than

direct absorption and can therefore not be compared directly.

The same authors present a carbon dioxide solubility of 6.5 g l�1

in PEG dimethyl ether, which is close to the values of 4.9 g l�1 and

6.9 g l�1 found in this study for tetraglyme and PEG dimethyl

ether 150, respectively. Unfortunately, Kovvali and Sirkar do

not specify the molar mass of the PEG dimethyl ether. A solu-

bility value of 4.4 g l�1 carbon dioxide in PEG given by Saha and

Chakma does not specify the molar mass of PEG,31 making it

also difficult to compare with the values obtained in this study for

PEG 200, 300 and 600.

The solubility of carbon dioxide in various PEG dimethyl

ethers was investigated by Henni et al.13 as well as Sciamanna and


Table 3 Solubility of carbon dioxide in mg g�1 at 25 �C. The substancesare listed in order of increasing molar mass

Substance Henni13 Sciamanna32 Present study

Di(ethylene glycol)dimethyl ether

6.8 8.9 —

PEGDME 150 — — 6.4Tri(ethylene glycol)

dimethyl ether5.6 7.3 —

Tetraglyme 6.6 6.6 4.8PEGDME 250 5.5 — —

Table 4 Solubility of nitrogen at 25 �C

Substance Solubility in mg g�1 Solubility in g l�1

Glycerol 7.5 9.3Glycerol carbonate 4.4 6.2Tetraglyme 1.5 1.5PEGDME 150 2.3 2.5PEG 200 8.9 10.1PEG 300 4.7 5.3PEG 600 4.7 5.3Poly(ethylenimine) 6.3 6.5

Table 5 Comparative molar solubilities of carbon dioxide and nitrogenat 25 �C for the pure gases in the test adsorbents under identical condi-tions. Data for methanol and water are included for reference

SubstanceSolubility ofCO2/mmol l�1

Solubility ofN2/mmol l�1

Glycerol 391 333Glycerol carbonate 251 220Tetraglyme 111 52PEGDME 150 158 91PEG 200 343 359PEG 300 344 189PEG 600 197 190Poly(ethylenimine) >71 233Methanol25,26 138 7Water27,28 34 1

Lynn.32 Table 3 shows the solubility data at 25 �C together with

the corresponding values found in this study. The substances are

listed in order of increasing molar mass. PEGDME 150 is

a mixture of poly(ethylene glycol) ethers which has a molar mass

higher than di(ethylene glycol) dimethyl ether but lower than

tri(ethylene glycol) dimethyl ether.

The literature values lie in the range of 5 to 9 mg g�1 for the

substances in Table 3. This is in agreement with the results found

in this study. The deviation between the values in this study and

the literature values is of the same magnitude as the deviation

between the values of the two literature sources. The literature

values show a general trend of a slight decrease in carbon dioxide

solubility with increasing molar mass of the PEG dimethyl ether.

This trend is confirmed in the results from this study.

It is important to compare the carbon dioxide solubility of the

solvents in this study with state-of-the-art solvents for carbon

dioxide absorption. At the present time, the most widely used

solvent for carbon dioxide absorption is an aqueous solution of

monoethanolamine (MEA).4 In a state-of-the-art carbon dioxide

absorption process described in the literature using MEA as the

solvent the concentration of MEA in solution is 0.3 g/g.5,33 This

gives a typical concentration of carbon dioxide in the MEA

solution was calculated to 43.8 g l�1 using a solution density of

1013 g l�1 (ref. 34) and showing a typical carbon dioxide uptake

of 0.2 mol mol�1.33,35 The calculated adsorption value of 43.8 g l�1

is approximately three-times higher than the 15 to 17 g l�1

obtained in this study for the best absorbents PEG 200, PEG 300

and glycerol. It also has to be taken into account that the typical

uptake for MEA solutions is still relatively inefficient and below

the maximum uptake concentration which can theoretically be

five times as high, corresponding to 1 mol carbon dioxide per mol

MEA as obtained in a stoichiometric reaction. This would result

in a maximum concentration of 219 g l�1, which is more than 10

times the maximum concentration reached in glycerol. In reality,

such high uptake values are not achieved and in addition the

regeneration energy is high as the carbamates need to be broken

down. However, it is possible that the lower energy requirement

for solvent regeneration and the reduced loss of solvent in the

case of glycerol, PEG 200 or PEG 300 can compensate the

disadvantage of lower carbon dioxide solubility.

Selectivity over nitrogen

In order to gain knowledge about the selectivity of carbon

dioxide absorption over nitrogen absorption, the solubility of

nitrogen was measured for comparison. The results for pure, dry

nitrogen are shown in Table 4.


Glycerol and PEG 200, which are among the substances with

the highest molar solubility for carbon dioxide, also have the

highest solubility for nitrogen. Low nitrogen solubility is found

for tetraglyme and PEGDME 150, but these also have low

solubility for carbon dioxide. From the solubility data in mg per

solvent unit, however, it is not possible to assess the different

affinities of the carbon dioxide and nitrogen molecules for the

solvents. Table 5 shows the solubilities of carbon dioxide and

nitrogen in mmol per litre of solvent. Again a general trend can

be observed that solvents with high carbon dioxide solubility also

exhibit comparatively high nitrogen solubility.

However, the data for pure gases give no indication of selec-

tivity as adsorption in a mixed gas system may be competitive. In

order to examine the possibility of competitive adsorption,

studies were carried out on a 1 : 1 by volume mixture of the two

gases over the different adsorbents. The results are presented in

Table 6 as the mass of gas adsorbed per mol of adsorbent for

pure carbon dioxide, pure nitrogen and the 1 : 1 mixture for the

different adsorbents under the same conditions. The value for

poly(ethylenimine) is an estimated value based on a relative

molecular mass of 10 000. The final entry is the mass adsorption

per monomer unit of the polymer in order to relate the perfor-

mance back to the low molecular mass materials. An experi-

mental error of 10% based on uncertainties in the data has also

been included. It is recognised that the partial pressure of CO2 in

this mixed gas system is high (0.5 bar) in comparison with

a typical flue gas, however this is a model system. Other partial

pressure compositions are currently under investigation, both on

a bench-scale and in a scaled-up system.

Tables 5 and 6 show that those adsorbents possessing terminal

hydroxyl groups have the highest affinity for carbon dioxide with


Table 6 Solubility by mass of carbon dioxide, nitrogen and a 1 : 1volumetric mixture of the gases per mole of adsorbent at 25 �C. All datahave an accuracy of �10%

Substance CO2/g mol�1 N2/g mol�1 CO2/N2/g mol�1

Glycerol 1.27 0.69 1.13Glycerol carbonate 0.93 0.52 0.72Tetraglyme 1.07 0.32 0.81PEGDME 150 0.96 0.35 0.61PEG 200 2.68 1.79 2.25PEG 300 4.04 1.42 4.70PEG 600 4.63 2.84 6.56Poly(ethylenimine) 63.44 30.27 19.13in g/momomer 1.92 4.02 1.21

Table 7 Absorption of carbon dioxide, nitrogen and a 1 : 1 mixture ofboth gases at 25 �C. The theoretical values were calculated from thesolubility of the pure gas with the assumption of independent absorptionand the validity of Henry’s law for both gases

SubstanceSolubility ofCO2 in g l�1

Solubility ofN2 in g l�1

Solubility of CO2/N2

mixture in g l�1

Experimental Theoretical

Glycerol 17.2 9.3 15.3 13.3Glycerol

carbonate11.0 6.2 8.5 8.6

Tetraglyme 4.9 1.5 3.7 3.2PEGDME 150 6.9 2.5 4.4 4.7PEG 200 15.1 10.1 12.6 12.6PEG 300 15.1 5.3 17.6 10.2PEG 600 8.7 5.3 8.5 7.0Poly(ethylenimine) >3.1 6.5 >2.0 >4.8

glycerol having the highest volumetric capacity than PEG 300

and PEG 200. However, the highest molar capacity is shown by

PEG 600 then PEG 300. The adsorption of nitrogen is highest for

PEG 200 and glycerol volumetrically and the methyl ethers in

terms of moles of adsorbent. Table 6 shows the mass of gas

adsorbed per mol of adsorbent for the mixed gas system as TGA

does not permit the analysis of gas composition so that molar

values cannot be determined. The absorbed mass is therefore the

sum of both gases absorbed. The highest gas capacity is observed

for PEG 600 and PEG 300. Due to the high molecular weight of

poly(ethylenimine) the mass of adsorbed gas per monomer unit is

given as this relates more closely to the other solvents, however

these values are very low and surprisingly suggest a greater

selectivity for nitrogen.

A theoretical value for the solubility of the gas mixture,

calculated from the solubility of the pure gases, is also included in

Table 7. This theoretical solubility is based on the assumption

that both gases obey Henry’s law, where the solubility is

proportional to the partial pressure of the gas, and that the

absorption occurs independently for each gas. Henry’s law has

generally been used to describe the solubility of carbon dioxide in

a wide range of liquids.13,32,36

Most of the substances show good agreement of the experi-

mental value with the theoretical value calculated from Henry’s

law. However, in the cases of glycerol, PEG 300 and PEG 600 the

experimental value is higher than the value found from theory.

This can be due to competitive absorption, one of the gases

absorbing faster than the other and so inhibiting the absorption

of the second gas. If the experimental solubility value is higher

than the predicted one, as is the case here, the gas with higher

solubility (carbon dioxide) will then be the compound with faster

absorption. This behaviour can be explained by a stronger

interaction of carbon dioxide with the solvent molecules due to

the strong polarity of the molecular bonds in carbon dioxide and

the functional groups of the solvent. Unfortunately, an experi-

mental confirmation of this kinetic behaviour was not possible as

the thermogravimetric method used was not accurate enough for

kinetic studies.

Another possible reason for the high experimental values

compared to the theoretical calculation is a deviation of the

behaviour of the gas mixture from Henry’s law. This is particu-

larly obvious for PEG 300 where the solubility of the gas mixture

is higher than the solubility of pure carbon dioxide. This can only

be explained by the assumption that both gases absorb to


a higher extent than the 50% of the value for the pure gas

expected from Henry’s law. This also means that each of the two

gases interact with different parts of the solvent molecules, so

that the sum of the dissolved gas molecules is more than the

number of molecules corresponding to saturated solution for

each individual gas. This is in agreement with the different

mechanisms for molecular interaction as discussed in the

previous section, with strong electron interaction between the

carbon dioxide and the –OH and –O– groups of the solvent, and

far weaker interaction between the nitrogen molecules and the

entire surface of the solvent molecules.

For poly(ethylenimine), the absorption of the gas mixture was

very slow and equilibrium was not reached during the experi-

mental time. This is similar to the result with pure carbon dioxide

and due to the very high viscosity of poly(ethylenimine), as

mentioned earlier. However, one would expect to see a fast

absorption of the nitrogen present in the gas mixture. This was

not the case, indicating that nitrogen absorption is inhibited by

the presence of carbon dioxide, which may react selectively but

very slowly with the amine groups in the polymer. It is possible

that carbon dioxide is selectively adsorbed at the surface of the

polymer and that this inhibits mass transfer into the interior.

Desorption and absorption cycles

For industrial applications, it is necessary to know the desorp-

tion and long-term absorption behaviour of the solvents.

Therefore, experiments were performed with two subsequent

absorption cycles using pure carbon dioxide. After equilibrium

absorption was reached, the sample was heated to a desorption

temperature of 100 �C for 30 min and then cooled for a second

absorption cycle. The gas was not changed during the procedure.

This means desorption was conducted in the same carbon

dioxide atmosphere, the driving force for desorption being only

the higher temperature. The experiments were only performed

with tetraglyme and PEGDME 150 because of their relatively

short absorption time (1 h compared to up to 5 h for the other

solvents in this study). Fig. 4 shows the two absorption cycles of

carbon dioxide in PEGDME 150.

As seen in Fig. 4, a sharp decrease in weight is observed during

the desorption period. The weight loss is far higher than the


Fig. 4 Thermogravimetric data for PEGDME 150 under carbon dioxide

in two subsequent absorption cycles. The temperatures are the set values.

Table 8 Absorption of carbon dioxide at 25 �C in two subsequentabsorption cycles

Substance

Solubility of CO2 in mg g�1

1st cycle 2nd cycle

Tetraglyme 4.8 3.7PEGDME 150 6.4 5.0

absorbed amount of carbon dioxide. This effect arises from

evaporation of the solvent. It is supposed that most of the

absorbed carbon dioxide is desorbed at 100 �C. However, this is

not evidently clear due to the underlying weight loss caused by

evaporation. It is possible that some carbon dioxide remains in

the solvent, especially as there was still a constant flow of carbon

dioxide during desorption in these experiments.

Table 8 shows the average absorption of carbon dioxide for

PEGDME 150 and tetraglyme in the first and second absorption

cycle. The absorbed amount was always lower in the second

cycle. This may indicate that desorption at 100 �C was not

complete. It might also indicate degradation of the solvent over

time. It is possible that PEG 300 as the best solvent in this study

regarding solubility and selectivity shows similar problems in

long-run absorption and desorption experiments. A thorough

further investigation of the desorption and long-term absorption

behaviour will be necessary in order to draw conclusions about

the solvent’s potential for industrial absorption of carbon

dioxide from flue gas. The TGA method is not suitable for this

investigation, as the experiments should be performed under real

desorption conditions without presence of carbon dioxide and

possibly at reduced pressure. It is obvious that such conditions

will make the desorption of carbon dioxide much easier and will

probably lead to relatively fast and complete desorption for all

the solvents in this study, with exception of poly(ethylenimine)

due to its high viscosity as well as chemical reactions between its

amine groups and carbon dioxide.

Conclusions

In this study, a range of alternative solvents for the absorption of

carbon dioxide from flue gas was investigated. All the solvents

have high thermal stabilities and low vapour pressures, reducing

the loss of solvent in the process to a minimum.


In spite of its potential to react with carbon dioxide, poly

(ethylenimine) as the only solvent with amine groups in this study

showed very slow absorption due to its high viscosity. It per-

formed particularly badly in a mixed gas system when compared

to adsorption of pure gases.

The best substance in this study was shown to be PEG 300 with

a high carbon dioxide solubility and good selectivity over

nitrogen. The affinity for pure carbon dioxide was higher than

for pure nitrogen, and the mixed gas system showed a positive

deviation from Henry’s Law with a higher mass of gas being

adsorbed than for either of the pure gas streams. The enhanced

performance has been attributed to a combination of internal

ether oxygen groups together with terminal hydroxyl groups. It is

proposed that the latter show enhanced interactions with the

carbon dioxide molecules leading to stronger adsorptive inter-

actions. The long-term sorption behaviour of PEG 300 still needs

to be investigated in order to be able to compare the overall

performance and costs with state-of-the-art solvents such as

monoethanolamine solutions. Although monoethanolamine has

a higher carbon dioxide absorption capacity than PEG 300, it is

possible that the advantages of PEG 300 such as the high

stability, reduced solvent loss and lower desorption energy will

outweigh this disadvantage in comparison with monoethanol-

amine, making PEG 300 a suitable and favourable alternative

solvent for industrial processes.

Acknowledgements

We gratefully acknowledge support from the EPSRC for OA

under the ‘C-Cycle’ project (EP/E010318/1).

References

1 IPCC Special Report: Carbon Dioxide Capture and Storage, ed. B.Metz, O. Davidson, H. de Coninck, M. Loos and L. Meyer,Cambridge University Press, 2005, www.ipcc.ch/publications_and_data_reports.html#1, accessed 02.02.2010.

2 M. Mikkelsen, M. Jorgensen and F. C. Krebs, Energy Environ. Sci.,2010, 3, 43–81.

3 A. Meisen and X. Shuai, Energy Convers. Manage., 1997, 38, S37.4 C. M. White, B. R. Strazisar, E. J. Granite, J. S. Hoffman and

H. W. Pennline, J. Air Waste Manage. Assoc., 2003, 53, 645–715.5 R. Notz, N. Asprion, I. Clausen and H. Hasse, Chem. Eng. Res. Des.,

2007, 85(A4), 510–515.6 S. Ma’mun, H. F. Svendsen, K. A. Hoff and O. Juliussen, Energy

Convers. Manage., 2007, 48, 251–258.7 J. C. Abanades, E. S. Rubin and E. J. Anthony, Ind. Eng. Chem. Res.,

2004, 43, 3462.8 M. Mikkelsen, M. Jorgensen and F. C. Krebs, Int. J. Greenhouse Gas

Control, 2010, 4, 452–458.9 M. Ismael, R. Sahnoun, A. Suzuki, M. Koyama, H. Tsuboi,

N. Hatakeyama, A. Endou, H. Takaba, M. Kubo, S. Shimizu,C. A. Del Carpio and A. Miyamoto, Int. J. Greenhouse GasControl, 2009, 3, 612–616.

10 M. Wilson, P. Tontiwachwuthikul, A. Chakma, R. Idem, A. Veawab,A. Aroonwilas, D. Gelowitz, J. Barrie and C. Mariz, Energy, 2004, 29,1259.

11 Y. J. Heintz, L. Sehabiague, B. I. Morsi, K. L. Jones andH. W. Pennline, Energy Fuels, 2008, 22, 3824–3837.

12 Selexol Patents, US Pat., 3 594 985, 1971; Selexol Patents, UK Pat.,1 277 139, 1985.

13 A. Henni, P. Tontiwachwuthikul and A. Chakma, Can. J. Chem.Eng., 2005, 83, 358.

14 M. Kanniche and C. Bouallou, Appl. Therm. Eng., 2007, 27, 2693–2702.


15 J. L. Anthony, J. L. Anderson, E. J. Maginn and J. F. Brennecke, J.Phys. Chem. B, 2005, 109, 6366.

16 J. Jacquemin, P. Husson, V. Majer and M. F. Costa Gomes, FluidPhase Equilib., 2006, 240, 87.

17 J. Tang, H. Tang, W. Sun, M. Radosz and Y. Shen, Polymer, 2005, 46,12460.

18 P. Styring, O. Aschenbrenner and S. Supasitmongkol, Prepr. Pap. -Am. Chem. Soc., Div. Fuel Chem., 2009, 54(1), 002.

19 H. Lin and B. D. Freeman, J. Mol. Struct., 2005, 739, 57.20 S. Satyapal, T. Filburn, J. Trela and J. Strange, Energy Fuels, 2001,

15, 250.21 A. S. Kovvali and K. K. Sirkar, Ind. Eng. Chem. Res., 2002, 41, 2287.22 H. K. Cammenga, F. W. Schulze and W. Theuerl, J. Chem. Eng. Data,

1977, 22, 131.23 G. R. Ross and W. J. Heideger, J. Chem. Eng. Data, 1962, 7, 505.24 O. Aschenbrenner, S. Supasitmongkol, M. Taylor and P. Styring,

Green Chem., 2009, 11, 1217–1221.25 E. Wilhelm and R. Battino, Chem. Rev., 1973, 73, 1–9.


26 P. L€uhring and A. Schumpe, J. Chem. Eng. Data, 1989, 34, 250.27 F. Daniels and R. A. Alberty, Physical Chemistry, Wiley, New York,

1980.28 S.-W. Park, J.-W. Lee, B.-S. Choi and J.-W. Lee, J. Ind. Eng. Chem.,

2005, 11, 202.29 J. Rolker, M. Seiler, L. Mokrushina and W. Arlt, Ind. Eng. Chem.

Res., 2007, 46, 6572–6583.30 P. Raveendran, Y. Ikushima and S. L. Wallen, Acc. Chem. Res., 2005,

38, 478–485.31 S. Saha and A. Chakma, J. Membr. Sci., 1995, 98, 157–171.32 S. F. Sciamanna and S. Lynn, Ind. Eng. Chem. Res., 1988, 27, 492–

499.33 A. B. Rao and E. S. Rubin, Environ. Sci. Technol., 2002, 36, 4467–

4475.34 R. H. Weiland, J. C. Dingman, D. B. Cronin and G. J. Browning,

J. Chem. Eng. Data, 1998, 43, 378–382.35 A. Chakma, Energy Convers. Manage., 1997, 38, S51–S56.36 Y. Miyano and I. Fujihara, Fluid Phase Equilib., 2004, 221, 57–62.


Documents

Comparative study of solvent properties for carbon dioxide ... · PDF fileComparative study of solvent properties for carbon dioxide absorption ... solubility of carbon dioxide