Monday, January 13, 2020

The Development of Second-generation & Transformational Advanced CO2 Capture Technologies

Chapter 1: Solvent-Based CO2 Capture


The removal of CO2 from industrial gas streams is not a new process. Gas absorption processes using chemical solvents to separate CO2 from other gases have been in use since the 1930s in the natural gas industry.

However, concentrations of CO2 in natural gas are typically much higher than those found in gases resulting from the combustion process, which simplifies separation. The CO2 partial pressures encountered in flue gas applications (≈2 psia [≈13–14 kPa] CO2 in feed flue gas, ≈0.26 psia [≈1.8 kPa] in treated flue gas) necessitated the use of amine-based processes (see Figure 1-1).

Figure 1-1. Chart for Selecting CO2 Removal Technologies Available Commercially
Before the inception of the DOE/NETL Post-Combustion CO2 Capture Program. Source: Kidnay and Parrish.1 Note that the ‘product’ in the figure refers to treated flue gas.

The licensing history of the Econamine FG process provides a good example of past combustion-based applications of CO2 removal technologies.2 Prior to 1999, 25 facilities were built with CO2 capture capacities ranging from 700 to 365,000 tonnes per year using this process (Table 1-1). The captured CO2 from these facilities was used for enhanced oil recovery (EOR), urea production, and in the food and beverage industry. The capture capacities of these facilities reflect the fact that they were built to serve a specific commercial market for CO2. Other amine-based processes (e.g., ABB/Lummus) were implemented at similar capture capacities during this period. By comparison, a single 550 megawatt (MW) net output coal-fired power plant capturing 90% of the emitted CO2 will need to separate approximately 5 million tonnes of CO2 per year. Scaling up these existing processes represents a significant technical challenge.3

1 Kidnay, A. J. and Parrish, W. R. Gas treating in Fundamentals of Natural Gas Processing 91–132 (CRC Press, 2006).
2 Chapel et al. Recovery of CO2 from flue gases: Commercial trends, The Canadian Society of Chemical Engineers Annual Meeting, Saskatoon, Saskatchewan, Canada. (4–6 October, 1999).
3 U.S. Department of Energy. 2010. Coal-Fired Power Plants in the United States: Examination of the Costs of Retrofitting  with CO2 Capture Technology and the Potential
for Improvements in Efficiency, DOE/NETL-402/102309, Revised January 2010.

Table 1-1. CO2 Capture Plants Built Prior to 1999 Using the Econamine Process


The reactive absorption of CO2 at the gas-liquid interface involves the diffusion of CO2 from the bulk-gas phase to the interface, reaction at the interface and the diffusion of aqueous CO2 species away from the interface into the bulk liquid.4 The superscripts i and * in the equations below refer to the concentration of CO2 at the gas-reaction film interface and the bulk liquid respectively. The amount of CO2 is represented by partial pressure or concentration in solution. If the reaction of CO2 at the interface is irreversible and fast, the concentration in the bulk liquid is negligible, and the gas-film resistance controls the mass-transfer rate. For cases where the reaction of CO2 at the gas-liquid interface is slow, CO2 can diffuse into the bulk liquid before reacting with the solvent, and liquid-phase resistance controls the mass-transfer rate. For cases where the rate of reaction is intermediate, the reacting CO2 species is not depleted in the liquid film and the overall mass-transfer rate is controlled by both gas-, and liquid-film resistances.

4 Tsai, R. E. Mass transfer area of structured packing, Masters Thesis. (2010).

The flux of CO2 is given by:
Equation 1-1

where KG is the overall mass-transfer coefficient, Pco2 represent CO2 partial pressures in the bulk gas, gas-liquid interface, and liquid respectively.

The mass-transfer rate can also be expressed in liquid-phase units as:
Equation 1-2

where E represents the enhancement of the liquid-film mass transfer rate due to the reaction, kl0 represents the liquid-film mass-transfer coefficient (physical diffusion), HCO2 is the Henry’s law constant, and the superscripts refer to similar quantities as above, except that the concentration is measured in liquid-phase units.

The overall mass-transfer coefficient is depicted graphically in Figure 1-2 and can be expressed in terms of the gas- and liquid-film mass transfer coefficients as:
Equation 1-3

Figure 1-2. Schematic of CO2 Absorption into a Liquid Solvent with Chemical Reaction
Source: Tsai, 2010.4

The enhancement factor E depends on the rate at which the amine reacts with CO2 at the interface, and also depends on the diffusion of CO2 from the interface to the bulk liquid. Solvents which can react faster with CO2 are beneficial for CO2 capture because the mass-transfer coefficient is higher.5 Similarly, solvents where CO2 can diffuse more readily (i.e., less viscous solvents) also lead to higher E factors and higher mass transfer coefficients. For simple cases of CO2 absorption by highly reactive solvents, the enhance-ment factor E (and thereby k’g) is directly proportional to the square root of the product of the rate constant and the diffusivity of CO2 in the liquid.6 Therefore, even if the intrinsic reaction kinetics of a particular solvent with CO2 are very fast, if the diffusivity of CO2 in the solvent is slow (e.g., due to high viscosity), the effect of the enhancement factor is limited.


The Carbon Capture R&D program has examined a broad array of solvent-based capture approaches to address the technical chal-lenges described above. These include the use of amines, non-aqueous solvents, carbonate-based solvents, phase-change solvents, and catalyzed processes. In addition, process innovations and improvements have contributed to overcoming the technical issues that serve as barriers to deployment of carbon capture technologies. All of these are described below.


Amines are widely used to separate acid gases such as CO2, H2S and COS from a variety of gas streams. Typically, either a single amine, or a blend of two or more amines in aqueous solution reacts chemically with CO2 in an acid-base reaction, forming a salt. The salts can be converted back into the acid and the base (amine) at high temperatures and reduced acid-gas partial pressures, regenerating the solvent and releasing CO2. Amines are characterized by the degree of substitution (up to 3 atoms) on the nitro-gen atom. If an alkyl group substitutes for hydrogen, the amine is designated as a primary amine (e.g., MEA: HO-CH2-CH2-NH2, aminoethoxyethanol/digylcolamine [AEE/DGA]: H2N-CH2-CH2-O-CH2-CH2-OH). Secondary (e.g., diethanolamine [DEA]: HO-CH2-CH2-NH-CH2-CH2-OH) and tertiary amines (methyldiethanolamine [MDEA]: HO-CH2-CH2-NCH3-CH2-CH2-OH) have two and three groups substituting for hydrogen atoms respectively. Amines can be generally represented as R1R2R3N, where Ri (i = 2, 3) represent hydrogen atoms or organic groups. R2 and R3 are represented by hydrogen atoms in primary amines. Secondary amines have two organic moieties, and one hydrogen atom around the nitrogen atom, whereas tertiary amines have three organic moieties. Examples of primary, secondary, tertiary and hindered amines are shown in Table 1-2.

Table 1-2. Examples of Primary, Secondary, Tertiary, and Hindered Amines

CO2 can react with amines in two different ways: carbamate formation and bicarbonate formation.7

5 More CO2 can be absorbed from the flue gas flow by a given flow rate of solvent if the mass-transfer coefficient is higher.
6 For example, see Bishnoi, S. and Rochelle, G.T. Absorption of carbon dioxide into aqueous piperazine: reaction kinetics, mass transfer and solubility. Chemical Engineering Science 55, 5531–5543 (2000).
7 Wagner, R., Judd, B., Fundamentals–Gas sweetening, Presented at the 2006 Laurance Reid Gas Conditioning Conference, February 26–March 1, 2006, Norman, Oklahoma, USA.

Carbamate Formation

CO2 can react with a primary/secondary amine molecule (R1R2NH) forming a zwitterion which is further deprotonated by a base (amine) to form carbamate (R1R2NCOO-) and a cation (R1R2NH2+). The overall reaction consumes 2 moles of amine per mole of CO2.
Equation 1-4
Carbamate Formation

For polyamines that have multiple amine groups, the proton may be donated to the second amine group within the same molecule, resulting in higher molar-CO2 capacity (e.g., piperazine [PZ]):

Equation 1-5
Bicarbonate Formation

CO2 can be slowly hydrated, forming carbonic acid, which forms a bicarbonate ion and an alkylammonium cation by reacting with an amine:

Equation 1-6
Bicarbonate Formation from Tertiary Amines

The formation of bicarbonate may occur with any amine (primary, secondary, tertiary) but is slow because the dissociation of car-bonic acid to bicarbonate is slow.

The formation of carbamate requires the amine to have labile hydrogen atom(s), and thus is limited to primary and secondary amines. The rate of carbamate formation is considerably faster than the rate of bicarbonate formation. Carbamate formation with primary and secondary amines has a stoichiometry of 1:2 (the capture of 1 mole of CO2 requires 2 moles of amine). In some cases, however, the carbamate formed can also be partially hydrolyzed to form bicarbonate, regenerating the amine, and increasing the CO2 loading to values above 0.5.
Equation 1-7

A sterically-hindered amine is defined as a primary amine where the amino group is attached to a tertiary carbon atom (e.g., 2-ami-no-2-methyl-1-propanol [AMP]), or a secondary amine where the amino group attached to a secondary or tertiary carbon atom.8, 9 Several proprietary amine technologies for CO2 capture incorporate hindered amines. The presence of bulky organic groups around the amine destabilizes the carbamate, forming the bicarbonate (Equation 1-7), resulting in increased CO2 capacity and lowering the heat of absorption while maintaining a relatively high-rate of CO2 absorption compared to tertiary amines. Equation 1-8 represents the reaction of AMP with CO2 forming the bicarbonate:


The primary characteristics of a solvent that make it a feasible candidate for post-combustion flue gas CO2 capture are:
  • Fast reaction with CO2: The cost of CO2 capture is dictated to a large extent by the size of the absorber needed to separate a given quantity of CO2 from the incoming flue gas. CO2 can be separated from the flue gas using a smaller absorber volume if the solvent can absorb CO2 at a faster rate, which lowers the capture costs. As noted in section 1.3, faster sol-vent kinetics results in better mass-transfer performance provided other factors are not limiting CO2 transfer across the gas-liquid interface.
  • Large net-CO2 carrying capacity: The quantity of solvent needed to separate CO2 from the flue gas determines the re-generation load, auxiliary unit costs, and energy requirements. Solvents need to have a high-net CO2 carrying capacity.
  • Low-enthalpy of reaction with CO2: Because the amine chemically reacts with CO2, energy is required to break the amine-CO2 bond to desorb CO2. Primary amines and secondary amines forming carbamates have higher enthalpy of reaction with CO2 compared to tertiary amines which form carbonates.
  • Low energy lost to heat and vaporize water: Typically amines are used in concentrations varying from 30–50 w/w% in aqueous solutions. Because of its high latent heat of vaporization, water evaporation increases the steam require-ments, parasitic load, and the energy penalty. Both CO2 carrying capacity and the energy lost to evaporate water can be increased by reducing the water content in solution; however this increases the solution viscosity and may increase the corrosion rates. Solvent processes are typically operated with the highest feasible concentration of amine where viscos-ity, corrosion, carbamate precipitation, and column temperature bulge issues are minimal or well controlled.
In addition to these desirable characteristics, other favorable features for solvents include low CO2 equilibrium backpressures at ab-sorption conditions, easily reversible reactions at regeneration temperatures, low volatility, and resistance to oxidative and thermal degradation.

Primary and secondary amines typically have higher rates of reaction with CO2 compared to tertiary amines. Among various prima-ry amines, MEA has the highest reaction rate, and blends of MEA and other tertiary or hindered amines are typically used to exploit this feature while maintaining relatively-low reboiler loads. Primary/secondary (mono)amines with a 1:2 stoichiometry have lower CO2 carrying capacity compared to tertiary amines which bind 1:1 with CO2. Further, polyamines such as piperazine have a higher carrying capacity because they have two amine groups per molecule. Tertiary amines have higher CO2 capacities but the reaction kinetics with CO2 are significantly slower than primary and secondary amines. Because the CO2 carrying capacity is expressed in wt% CO2 in the solvent, or the quantity of solvent circulated to capture a unit quantity of CO2, the molecular weight and density of the solvent also play a role in determining its volumetric or weight-based CO2 carrying capacity.

8 Kohl, A. L. and Nielsen, R., Alkanolamines for hydrogen sulfide and carbon dioxide removal, in Gas Purification (Gulf Professional Publishing, 1997).
9 Sartori, G. and Savage, D. W. Sterically hindered amines for carbon dioxide removal from gases. Ind. Eng. Chem. Fund. 22, 239–249 (1983).

From a health and environmental safety perspective, MEA is highly biodegradable, and has no direct adverse effects on human health, animals, and vegetation. Other amine solvents such as AMP, MDEA and PZ are toxic and are not easily biodegraded com-pared with MEA.10 The reaction of amines with NOx in the flue gas leads to the formation of nitrosamines, which are carcinogenic. The reactivity with NOx varies with the amine structure. MEA

MEA scrubbing is the oldest of the amine-based CO2 capture technologies. A schematic representation of the process is presented in Figure 1-3.

Figure 1-3. Schematic of Amine-Based Post-Combustion CO2 Capture Process

After conventional air pollutant cleanup (SOx, NOx, PM), the combustion flue gas enters an absorber reactor and flows counter-currently to a CO2-lean solvent where CO2 is absorbed into, and reacts with, the MEA to form water-soluble compounds. The treated flue gas (mostly N2) is discharged to the atmosphere and the CO2-rich solution is pumped to a stripper reactor for regeneration. In the stripper, the CO2-rich solution is heated in order to break down the salt and regenerate the MEA solvent. A reboiler, supplied with steam extracted from the turbine cycle, provides the heat for regeneration of the MEA solvent in the stripper. Consequently, CO2 is released, producing a concentrated stream which exits the stripper and is then cooled and dehumidified in preparation for compres-sion, transport, and storage. From the stripper, the CO2-lean solution is cooled and returned to the absorber for reuse.

Usable MEA solution concentrations are typically limited by viscosity and corrosion. Therefore, current systems use only between 20% and 30% MEA, with the remaining being water. Although the water present in the solution helps control the solvent tempera-ture during absorption, which is an exothermic reaction, the water also requires significant amounts of sensible heating and stripping energy upon CO2 regeneration. Not every MEA system is the same and various vendors offer different designs. In general, depend-ing on the amount of heat integration, anywhere from 1,550 to more than 3,000 British thermal units (Btu) per pound of CO2 (3.6–7 GJ/T CO2) in the form of low-pressure steam (approximately 45 psia [3.1 bar]) is required to regenerate the solvent to produce a concentrated CO2 stream at a pressure of approximately 25 psia (1.7 bar). PROPRIETARY AMINE BLENDS

10 Shao, R., Stangeland, A., Amines used in CO2 capture—Health and environmental impacts, The Bellona Foundation, September 2009. uploads/media/fil_Bellona_report_September__2009_-_Amines_used_in_CO2_capture.pdf.

Because of its roots in refinery-gas processing and natural gas processing, many proprietary solvents have been developed to treat acid gases containing H2S and CO2, mostly focused on removing acid gases to very low levels meeting natural gas pipeline speci-fications. Historically, these solvent formulations have not been optimized exclusively for CO2 capture alone, and further progress is feasible. Proprietary-amine blends might be comprised of sterically-hindered amines and patented amine blends. Activators (rate enhancers), and chemicals which retard oxidative/thermal degradation of amines are also considered proprietary additives. The reactions involved in the promotion or activation of an amine are complex. Astarita et al. note that several promoters are Lewis bases—molecules with lone pairs of electrons.11 Further, they also note that the difference between the various explanations for amine promotion, e.g., “shuttle mechanism,”12 and “homogeneous catalysis” simply lies in difference in the rate of the reaction of the CO2-amine intermediate with a second base (OH- in the case of carbonate solutions).10 By modeling the CO 2 mass transfer in promoted amines in falling thin films, Puxty and Rowland suggested the basicity of the ‘promoted’ amine to be a primary factor controlling synergistic amine interactions.13 For example, although AMP and PZ have similar acid strengths (pKa values 9.29 and 9.46 at 313 K), because AMP is present at a higher concentration (≈2.49 M vs. 0.94 M for one case presented in the article), it acts as a pH buffer maintaining high pH, allowing more free-PZ to react with CO2, thereby enhancing mass transfer within the gas-liquid reaction film compared to both PZ and AMP considered separately. Similarly, for PZ-MDEA mixtures, MDEA acts as a buffer, tak-ing up the protons released due to the PZ-carbamate formation. However, it is a weaker base than AMP (pKa of 8.22 vs. 9.29 at 313 K), and as a result accepts a lower fraction of the protons released upon PZ-carbamate formation.12

MDEA-Based Mixed Amine Blends

Amine mixtures of MDEA and DEA, MEA or piperazine are used to enhance CO2 removal compared to pure MDEA. The added primary/secondary amine typically does not exceed 20% of the total amine on a molar basis. Total amine concentrations in liquid can approach 55 w/w%.14 The use of PZ as an additive to MDEA was patented by BASF in 1982 (patent expired in 2002).15 BASF’s aMDEA® solvent has been used where very low residual CO2 levels in the treated gas were desired. Unpromoted MDEA is suit-able for bulk-CO2 removal (at intermediate-to-high pressures), but is unsuited for deep-CO2 removal at intermediate pressures, or bulk-CO2 removal at low pressures. Flue-gas CO2 capture (90%) does not require very low levels of CO2 in the treated gas (≈13.5% CO2 in the feed, ≈1.8% CO2 in the treated flue gas) and therefore the solvent circulation rates can be reduced. Additives such as monomethanolamine (MMEA) and DEA could also be used in addition to PZ.16

BASF OASE® Blue Solvent

The amine-based BASF OASE® blue solvent, was developed for post-combustion CO2 capture and has significantly higher stability compared to MEA, lowering operational costs. The composition of the BASF OASE® blue solvent is proprietary, and might consist of combinations of primary/secondary/tertiary/hindered amines and activators (heterocycles, primary or secondary alkanolamines, alkylenediamines or polyamines).17, 18 Aminocarboxylic19 or aminosulfonic acids, which become significantly more acidic with tem-perature may also be used to increase the cycle capacity (lower lean loading), and decrease the solvent circulation rate and steam load while only slightly affecting the rich loading.20 DOE/NETL is funding a 1-MW slipstream demonstration of the BASF OASE® blue solvent in a CO2 capture process developed by Linde. Technoeconomic analyses conducted by Linde indicate that the BASF OASE® blue solvent resulted in significantly lower steam requirement to strip CO2 compared to MEA (2.4 to 2.6 GJ/T CO2, vs. 3.55 GJ/T CO2 for MEA), leading to higher net-plant efficiency, and lower fuel costs. Compared to 8 m piperazine (PZ), the BASF OASE® blue solvent has almost the same level of steam consumption (≈2.4 to 2.6 GJ/T CO2 without heat integration).

11 Astarita, G., Savage, D. W. and Longo, J. M. Promotion of CO2 mass transfer in carbonate solutions. Chemical Engineering Science 36, 581–588 (1981).
12 The shuttle mechanism involves the fast reaction of the promoter with CO2, diffusion of the CO2-amine intermediate product to the bulk liquid, and reaction of the intermediate with a base to form the carbonate.
13 Puxty, G. and Rowland, R. Modeling CO2 mass transfer in amine mixtures: PZ-AMP and PZ-MDEA. Environmental Science and Technology 45, 2398–2405 (2011).
14 Polasek, J., Bullin, J.A., Process considerations in selecting amine, Proceedings GPA Regional Meeting, Sept. 1994, Tulsa, OK.
15 Appl, M. et al. United States Patent: 4336233—Removal of CO2 and/or H2S and/or COS from gases containing these constituents. (1982).
16 Piperazine—Why it’s used and how it works, The Contactor. 2(4), (2008).
17 Asprion, N., Clausen, I. and Lichtfers, U. United States Patent: 7887620—Removal of carbon dioxide from flue gases. (2011).
18 Asprion, N., Clausen, I., Lichtfers, U. and Wagner, R. United States Patent: 8034166—Carbon dioxide absorbent requiring less regeneration energy. (2011).
19 Aminocarboxylic acids exist in the zwitterionic form from pH ≈2.2–9.4, but can exist as anions at high-pH values and cations at low-pH values, accelerating CO2 absorption and desorption respectively.
20 Asprion, N., Sieder, G., Lichtfers, U. and Andarcia, H. R. G. United States Patent: 8388738—Method for removing carbon dioxide from fluid flows, in particular combustion exhaust gases. (2013).


Kansai Electric Power Company (KEPCO) and Mitsubishi Heavy Industries, Ltd. (MHI) has developed the KM-CDR® process which uses the KS-1™ solvent.

The composition of KS-1™ is proprietary, and may consist of activated hindered amines21, 22, 23 (e.g., hindered amine compound selected from the group consisting of 2-amino-2-methyl-1,3-propanediol, 2-amino-2-methyl-1-propanol, 2-(ethylamino)-ethanol, 2-(methylamino) ethanol, 2-(diethylamino)-ethanol, 2-amino-2-ethyl-1,3-propanediol, t-butyldiethanolamine and 2-amino 2-hy-droxymethyl-1,3-propanediol; an activator selected from the group consisting of piperazine, piperidine, morpholine, glycine, 2-me-thylaminoethanol, 2-piperidineethanol, and 2-ethylaminoethanol). Additives such as CuCO3 may also be used as corrosion inhibi-tors to permit the use of carbon steel instead of stainless steel.21

Ten CO2 capture plants using the KM-CDR process, with a maximum capacity of 450 T/d have been built to date. In the majority of these plants, CO2 is recovered from natural gas-fired reformer flue gas and used for urea synthesis.24 A 500 T/d CO2 plant capturing CO2 from coal-fired flue gas started operating at Southern Company’s (Alabama Power) Plant Barry in June 2011.25 The capture CO2 is injected into a saline formation for storage. The plant capacity is equivalent to a 25 MWe power plant with a design flue gas flow rate of 1,16,840 Nm3/h (90% capture with a design CO2 concentration of 10.1 mol% [wet basis]).24 The KM-CDR® process consists of a direct-contact flue gas cooler cooling the flue gas to 95–113 °F, a packed tower absorber for countercurrent flow of gas and solvent, a water-wash section for removing volatiles in treated flue gas, and a stripper/reboiler section. Saturated steam at 3 bar gauge fed to the reboiler, and the temperature at the bottom of the stripper is ≈250 °F.24 In operation, the capture plant removed CO2 at the design rate of 90%, and also exceeded the specific steam consumption measure of 0.98 kg steam/kg CO2 (3 bar g saturated). Further reduction in energy consumption and the cost of electricity can be achieved through heat integration, e.g., using the heat content of the stripper overhead gas and flue gas to heat the boiler condensate, replacing the preheaters.26

The KS-1™ solvent was also tested for three months from January to March 2012 at the Brindisi power station in Italy capturing 60 T/d CO2 (approximately equivalent to ≈2.9 MWenet) from one boiler (flue gas flow rate 12,000 Nm3/h at ≈13% CO2).27 The results indicate that KS-1™ could achieve a lower thermal regeneration energy (1.2 kg-steam/kg CO2) vs. 30 w/w% MEA (1.5 kg-steam/ kg CO2), with a lean-solvent flow rate 25% lower than that of MEA. Further, ammonia emissions in the treated flue gas were also significantly lower than that with MEA.

Amino-Acid Salts (AAS)

Amino-acid salts are a family of solvents which contain both the amine and carboxylic acid functionalities, e.g.,
where M is a alkali metal and R1 and R2 are organic moieties.

The metal carboxylate (-COO-M+) group lowers the solvent vapor pressure and increases water solubility, while the amine group is active for CO2 absorption. AASs exhibit low volatility, high cyclic-molar-CO2 loadings, and high resistance to oxidative degra-dation. They are typically used in their metal salt form because of the increased solubility in aqueous solutions. The amine group reacts with CO2 similar to primary/secondary amine, forming carbamates with a stoichiometry of 1:2. Amino-acid carbamates can further react to form bicarbonate and carbonate ions (especially if the amine group contains a hindered amine or a tertiary amine). The Alkazid process formerly licensed by BASF used sodium or potassium salts of amino acids (alanine, or diethyl- or dimethyl glycine [KDiMGly]) to absorb CO2 and H2S from gas streams.28, 29 The sodium alanine solution (marketed as Alkazid solution “M”) was used to separate either H2S, or CO 2 when present alone, or for absorbing both gases simultaneously. Alkazid solution “di-K” contained potassium diethyl- or dimethyl glycine (KDiMGly) and was used for the selective removal of H2S from gases containing CO2 and also from gases containing trace quantities of CS2 and HCN.27 Weiland and Hatcher estimate that piperazine-promoted KDiMGly would have 20% lower regeneration energy and 20% lower solvent circulation rates compared to 30 w/w% MEA.30, 31 Wetted-wall column tests show that potassium taurate AAS solutions exhibit faster mass-transfer kinetics (at 323 K to 353 K) com-pared to 30 w/w% (5-M) MEA (at 313 K).32

21 Iijima, M. et al. United States Patent: 6036931—Method for removing carbon dioxide from combustion exhaust gas. (2000).
22 Yoshida, K. et al. United States Patent: 6689332—Process for removing carbon dioxide from combustion gases. (2004).
23 Yoshida, K. et al. United States Patent: 6500397—Method for removing carbon dioxide from combustion exhaust gas. (2002).
24 At a plant currently under construction in Qatar, 500 T/d CO2 would be captured using the KM-CDR™ process from steam methane reformer flue gas to increase methanol production from an existing facility.
25 Ivie II, M. A. et al. Project Status and Research Plans of 500 TPD CO2 capture and sequestration demonstration at Alabama Power’s Plant Barry. Energy Procedia 37, 6335–6347 (2013).
26 Wall, T. Development and demonstration of waste heat integration with solvent process for more efficient CO2 removal from coal-fired flue gas. Presented at the 2013 CO2 Capture Technology Meeting, Pittsburgh, PA. (2013).
27 Kamijo, T. et al. Result of the 60 tpd CO2 capture pilot plant in European coal power plant with KS-1™ solvent. Energy Procedia 37, 813–816 (2013).

One potentially beneficial property of AASs is that they can form solids (bicarbonates, or the neutral form of the amino acid) as the CO2 loading is increased. This results in a higher driving force at the higher CO2 loading, resulting in faster mass-transfer kinetics.33 The proprietary DECAB process is based on forming slurries in the CO2 absorber using amino-acid salts and thermally regenerating the amino-acid salt.34 The limitations of such processes include reduced interfacial contact area, and complications arising out of handling slurries. This may be offset by increasing solvent (molar) CO 2 loading,35 and lower solvent circulation rates, and the higher driving force for absorption. Brouwer et al. note that a heat-integrated stripper is required to fully leverage the advantages of slurry formation.32 Such processes may also use a two-stage absorber containing packed column-stage and a spray absorber stage to handle the solids formed at high-CO2 loadings.36

An AAS-based solvent is the basis for the Siemens POSTCAP™ CO2 capture technology which was tested at the 0.1 MWe-scale at the Staudinger coal-fired power plant in Germany starting in 2009.37 Results indicate that compared to amine-based technologies, the AAS-based POSTCAP™ process did not require water wash, had ≈25% lower thermal regeneration energy (2.7 GJ/T CO2 vs. 3.6 GJ/T CO2 for 30 w/w% MEA), low solvent makeup requirements, and very low solvent emissions.36 CO2 carbonate or carba-mates do not precipitate under the operating conditions of the process.38


Cansolv absorbents have relatively fast kinetics, very low degradation rates, high resistance against oxidation, and low thermal re-generation energy. In brief, the Cansolv process is similar to a typical amine process with the difference that it can separate both CO2 and SO2 from gas streams. Cansolv solvents are stable in SO2, whereas MEA forms heat-stable salts with SO 2. The need for kinetic separation between CO2 and SO2 indicates that tertiary amines may be relevant solvent candidates for this process. For example, DC-103® solvent is reported to consist of a tertiary amine.39 Activators and free-radical scavengers are also added for improving kinetics and for reducing solvent degradation.40, 41 It is reported that DC-103® solvent requires lower regeneration energy but higher absorber surface area compared to the faster-acting DC-103B® solvent.42

28 Kohl, A. L. and Nielsen, R. Alkaline salt solutions for acid gas removal, in Gas Purification, 330–410 (Gulf Professional Publishing, 1997).
29 BASF currently licenses the amine-based OASE®-family of solvents of gas processing applications instead of the amino-acid salt-based Alkazid solvents.
30 The selectivity of the unpromoted Alkazid “dik” or KDiMGly solution toward H2S over CO2 is the result of its slow CO2 mass-transfer kinetics. The addition of pipera-zine as a promoter increases the rate of CO2 absorption but also increases the thermal regeneration energy.
31 Weiland, R., H. and Hatcher, N., A. Post-combustion CO2 capture with amino-acid salts. Digital Refining (2011).
32 Wei, S. C.-C., Puxty, G. and Feron, P. Amino acid salts for CO2 capture at flue gas temperatures. Energy Procedia 37, 485–493 (2013).
33 Brouwer, J., Feron, P., Ten Asbroek, N. Amino-acid salts for CO2 capture from flue gases, in Proceedings of the 4th Annual Conference on Carbon Capture and Seques-tration, Pittsburgh, PA (2009).
34 Versteeg, G., et al. International Patent Number WO 03/095071 Al: Method for absorption of acid gases. (2003).
35 Lower molar CO2 loading (i.e., mol CO2/mol amine or mol AAS) does not translate to lower weight-based or volumetric CO2 loading. The molecular weight and density of the amine/AAS/ionic liquid are important parameters affecting these measures.
36 Goetheer, E. L. V., Fernandez, E. S. and Roelands, C. P. M. United States Patent Application: US20120282158 A1: Method for absorption of acid gases using amino acids. (2012).
37 Schneider, R., Schramm, H. Environmentally friendly and economic carbon capture from power plant flue gases: The Siemens POSTCAP™ technology. Presented at the 1st Post-Combustion Capture Conference, Abu Dhabi, Dubai. (2011). 18_%20Siemens%20PostcapTM_PCCC1_ext_2.pdf.
38 Winkler, J.L. Slipstream development and testing of post-combustion CO2 capture and separation technology for existing coal-fired plants. Presented at the 2011 CO2 Capture Technology Meeting, Pittsburgh, PA. (2011).
39 Figueroa, J. D., Fout, T., Plasynski, S., McIlvried, H. and Srivastava, R. D. Advances in CO2 capture technology—The U.S. Department of Energy’s Carbon Sequestra-tion Program. International Journal of Greenhouse Gas Control 2, 9–20 (2008).
40 IEA Greenhouse Gas R&D Programme (IEAGHG). Evaluation of novel post-combustion CO2 capture solvent concepts. 2009/14. (2009).
41 Hakka, L. E. and Ouimet, M. A. United States Patent: 7056482—Method for recovery of CO2 from gas streams. (2006).
42 Rameshni, M. Carbon capture overview. (2010).

The unique features of the Cansolv integrated CO2/SO2 capture process are the possibilities for process integration between the CO2 and SO2 capture steps. For example, the SO2 stripper overhead stream contains considerable water vapor, whose latent energy can be recovered using mechanical vapor recompressors (MVRs) to reduce the primary steam requirement in the CO2 stripper by about 15%.43 The exothermic heat of reaction in sulfuric acid synthesis from SO2 can be used to raise steam for the CO2 capture process, and the latent heat of the lean amine exiting the CO2 stripper may also be recovered. Further, a rectangular integrated CO2/ SO2 absorber tower with a built-in prescrubber section is constructed with concrete and lined with ceramic/carbon tiles, reducing the construction and materials costs.44

DOE/NETL-funded tests at EERC indicate that the Cansolv next generation solvents have ≈38% lower solvent circulation rates (at 90% CO2 capture) and 21% lower regeneration energy compared to 30 w/w% MEA.44 In 0.5 MWe slipstream tests at the National Carbon Capture Center (NCCC), the Shell Cansolv DC-201® solvent had 50% lower solvent circulation rate compared to MEA and a thermal regeneration energy of 2.1 GJ/T CO2 (35% lower than that of MEA).45 It was found that the optimum solvent concentration for minimizing the thermal regeneration energy was 45 w/w% to 55 w/w%.46

Cansolv solvents will be used to capture CO2 and SO2 from SaskPower’s 150 MWe lignite-fired Boundary Dam (Unit 3) power plant where the flue gas contains ≈12% CO2 and ≈1,000 ppm SO2.44 Approximately 1.2 million T/y CO2 would be captured and sold for enhanced-oil recovery, while the captured SO2 would be converted to H2SO4 (60 T/d) and sold to the fertilizer market. Two different solvents, Cansolv DS® and Cansolv DC-103® would be used to capture SO2 and CO2 at the Boundary Dam project.47, 48 The project construction came online late 2014.49

Hitachi H3-1

Babcock Hitachi developed an amine-based solvent (H3-1) containing additives for reducing the regeneration energy and solvent degradation rates. It is also claimed that the H3-1 solvent has low degradation, low amine loss and low corrosivity. H3-1 was tested for 1,300 hours at NCCC, is being used at LG&E and KU Energy Service Company (LG&E-KU)’s EW Brown Generating Station as part of a DOE-funded University of Kentucky 0.7 MWe pilot-scale heat integrated absorption capture system test and is going to be used at the 20 MWth-scale at SaskPower’s coal-fired Shand Power Station in Canada, starting in 2014.50 Tests at EERC indicate that 2.88 GJ of thermal energy (21.3% lower than that of MEA) is required to regenerate a tonne of CO2 using the H3-1 solvent.51 The solvent flow rate needed for 90% CO2 capture is ≈46% lower than that required for 30 w/w% MEA. Results from tests at the NCCC show that the thermal regeneration energy was reduced by 33% compared to MEA (2.4 GJ/T CO2 vs. 3.6 GJ/T CO2) while capturing a higher fraction of the incoming CO2 (96% capture vs. 92% capture with MEA), with 38% lower liquid flow rates.52 Wetted-wall column testing at University of Kentucky showed that H3-1 has a lower overall mass-transfer coefficient compared to MEA at lean CO2 loadings but higher overall mass-transfer coefficient at rich CO2 loading conditions.

KoSol Solvents

A 10-MW post-combustion CO2 capture plant operating on a proprietary solvent KoSol-4 began operating in May 2013.53 The pro-cess captures ≈70,000 T/y CO2 from 14% CO2 coal-fired flue gas, some of which would be used for precision welding products and agricultural products. The Korea Electric Power Company (KEPCO) developed the KoSol family of solvents. Thermal energy to regenerate the KoSol-4 solvent is estimated to be 30% lower than that of MEA (≈2.5 to 3 GJ/T CO2).54

43 Shaw, D. Cansolv CO2 capture: The value of integration. Energy Procedia 1, 237–246 (2009).
44 Holmes, M.J. Fossil generation and the Clean Air Act. Presented at the 125th Annual Meeting of NARUC, Bonnet Creek, FL (2013). Presentations/MJHolmes-NARUC%20Nov2013.pdf.
45 Just, P.-E. Shell Cansolv deploying CCS worldwide. Presented at the 2nd Post-Combustion Capture Conference, Bergen, Norway. (2013).
46 Southern Company Services Inc. 2012. The National Carbon Capture Center at the Power Systems Development Facility. Topical Report: Budget Period 4. http://
47 Sarlis, J. Providing the capture process. Presented at the SaskPower CCS Information and Planning Symposium, Regina, Canada. (May 21–23, 2013). http://www.
48 Global CCS Institute. CO2 capture technologies. Post-combustion capture (PCC).
50 Honoki, M. et al. Hitachi’s Carbon dioxide scrubbing technology with H3-1 absorbent for coal-fired power plants. Energy Procedia 37, 2188–2195 (2013).
51 Pavlish, B. Partnership for CO2 capture: Results of the pilot-scale solvent evaluations. Presented at the 2010 NETL CO2 Capture Technology Meeting, Pittsburgh, PA. (September 13–17, 2010).
52 Eswaran, S. et al. Recent developments of Hitachi’s advanced solvent technology for post-combustion CO2 capture.’s%20Advanced%20Solvent%20Technology%20for%20Post-Combus-tion%20CO2%20Capture.pdf.

In summary, many of the proprietary amine blends discussed in this section have significantly better resistance against oxidative and thermal degradation, tolerance to other acid gases, lower vapor pressures, higher carrying capacities and lower energy consumption (30–40% lower) compared to 30 w/w% MEA. Higher carrying capacity also results in lower operating (fuel) costs and also lower capital costs (smaller pumps, heat exchangers, reboilers, strippers). Further, some of these proprietary amines can be regenerated at higher temperatures compared to MEA, resulting in higher CO2 partial pressure in the stripper overhead and lowering the compres-sion load. Notwithstanding such advances, considerable improvements are further required to attain the DOE capture cost goals. Novel gas-liquid contactors, new, more cost-effective routes to solvent synthesis, and solvents with considerably higher mass-transfer coefficients, and lower thermal regeneration energies represent some of the means to attain DOE goals. PIPERAZINE AND OTHER NON-PROPRIETARY AMINES

Piperazine (PZ) is a diamine. Concentrated eight molal (8-m) PZ has been used in several DOE-funded capture projects as a CO2 capture solvent because it has a higher cyclic capacity, faster kinetics, is thermally stable at higher temperatures (150 °C) compared to MEA, and requires less energy to strip CO2, resulting in lower regeneration energy. Benefits include lower capital costs and lower operating costs. However, PZ forms a solid phase with water (PZ.6H2O) and also with CO2 (H+PZCOO-.H2O), and solvent develop-ers need further experience in operating PZ-based capture plants at low-temperature and large scale.


As discussed in section, amines used commercially for gas treatment contain 70–50% w/w water. The water content nega-tively affects solvent working capacity, thermal regeneration energy, and solvent circulation rates. To overcome these issues, several developers are in the process of testing solvents containing low quantities of water (as-formulated). The working solution might nevertheless pick up water from the flue gas, a part of which would be stripped off in solvent regeneration. Common features of this class of solvent technologies are their comparatively higher weight-percent loading of CO2, higher solution viscosity, potentially slower mass transfer rates, and potential issues with solid solvent-CO2 products precipitating in the absorber. NAS development efforts currently underway are focused on the use of hydrophobic amines, CO2-binding organic liquids, aminosilicone solvents, functionalized-ionic liquids, and amine/organic solvent mixtures. These efforts are described below.  HYDROPHOBIC AMINES

As part of a recently-initiated, DOE/NETL-funded effort, RTI and BASF are developing hydrophobic amine solvents with signifi-cantly lower thermal regeneration energy requirements, capable of being regenerated at low temperatures (70–90 °C) compared to MEA.55 The process uses a hydrophobic primary or secondary amine to form a carbamate in the non-aqueous phase. Water con-densed from flue gas or otherwise present in the working solution is separated from the CO2-loaded amine solution in a decanter. The separated aqueous phase is used as water for amine wash at the top of the absorber. The hydrophobic amine phase containing the carbamate is fed to a stripper/reboiler thereby regenerating the hydrophobic amine. This process combines the relative fast kinetics of using primary or secondary amines with the lower quantity of regeneration energy made possible by the absence of water in the stripper. The technology is at the laboratory/bench scale of development, and preliminary analyses indicate a 30–50% reduction in thermal energy consumption compared to state-of-the-art solvents. The use of low regeneration temperatures permits low-pressure steam to be used for regeneration, reducing parasitic load. Preliminary analyses also suggest that the solvent circulation rates with the RTI/BASF NASs would be higher compared with conventional amines, which might increase the capital cost for solvent pumps, heat exchangers and regenerators.55 CO2-BINDING ORGANIC LIQUIDS

CO2BOLs are a class of switchable ionic liquids (i.e., molecular liquids which become ionic in the presence of CO2) having lower specific heat capacities and higher CO2 working capacities compared to aqueous amines, resulting in potential savings in the sensible heat required to strip CO2.
Equation 1-9

55 Kail, M., et al. Non-aqueous solvents for post-combustion CO2 capture. Presented at the 2012 NETL CO2 Capture Technology Meeting, Pittsburgh, PA. http://www.

Similar to aqueous amines, CO2BOLs are basic, but the base (e.g., guanidine, amidine) does not directly react with CO2. Instead, the alcohol component, reacts with CO2 forming alkyl carbonic acid, and subsequently transfers a proton to the base, forming liquid alkylcarbonate.

The current generation of CO2BOLs in DOE/NETL funded research at the Pacific Northwest National Laboratory (PNNL) combine the base and the alcohol moieties in a single molecule, lowering volatility (e.g., Equation 1-9). The addition of non-polar solvent (anti-solvent) to CO2BOLs and other switchable solvents during the solvent regeneration destabilizes bound CO2, thus potentially lowering the temperature at which the stripper can be operated (polarity-swing-assisted regeneration [PSAR]). Preliminary results indicate that PSAR could reduce the regeneration temperatures of CO2BOLs by more than 20˚C. This allows novel possibilities for heat integration such as transferring heat from the absorber to the stripper using heat pumps, thereby lowering steam demand for solvent regeneration. The anti-solvent can be separated out from the CO2BOL by cooling, and liquid-liquid phase separation.

CO2BOLs offer the potential to lower the parasitic energy requirements of CO2 capture by reducing the specific heat of the solvent, and lowering the amount of water present in the solvent. The lower regeneration steam requirements also result in higher power plant output. The polarity swing represents a means to lower the regeneration temperature further. One significant challenge to the CO2BOL (or other switchable ionic liquid) technology is the high viscosity of the solvent after CO2 absorption. Test results indicate that the viscosity of the CO2-loaded solvent needs to be reduced by one-to-two orders of magnitude for the process to be technically feasible.56 AMINOSILICONE SOLVENTS

Aminosilicones are amine-functionalized siloxane compounds having a CO2-philic backbone (physisorbing CO2) and an amine group chemically reacting with CO2. They exhibit lower vapor pressure, higher thermal stability, and lower thermal energy require-ment for CO2 capture, leading to up to 27% lower thermal energy consumption (2,954 kJ/kg CO2) compared to MEA. In DOE/ NETL-funded projects, General Electric (GE) has developed a family of aminosilicones starting with GAP-0 (bis(aminopropyl)tet-ramethyldisiloxane). Although GAP-0 resulted in a high CO2 loading, it was found that the GAP-0 carbamate precipitated as a solid, even when triethylene glycol (TEG) was used as a solvent. Therefore, a mixture of GAP compounds (40% GAP-0. 33% GAP-1, 19% GAP-2, 8% GAP-3)57 with an average molecular weight similar to GAP-1 (designated as GAP-1m) and TEG in a 60/40 w/w% mixture is being tested as a non-aqueous solvent for CO2 absorption.

Equation 1-10

56 With the current alkanolguanidine CO2BOL, the viscosity of the hydrated CO2BOL increased from 22 cP at 0 w/w% CO2 to 117 cP at 5 w/w% CO2. The viscosity would exceed 550 cP at 10 w/w% CO2, a typical value for rich-solvent loading.
57 The number in GAP-0, GAP-1, GAP-2, etc. refers to the number of repeating [SiO(CH3)2] units between the end siloxane members.

The GAP-1/TEG solvent reacts slower than MEA,58 and may require larger reaction vessels. The solvent is regenerated in a stirred tank reactor with solvent recirculation, enabling CO2 recovery at higher pressures (4.3 bar), lowering parasitic load and CO 2 com-pression costs. Bench-scale test results indicate that the solvent has a lower CO2 carrying capacity compared to MEA (≈4.1 w/w% vs. 5% to 7%), and ≈40% higher liquid-to-gas (L/G) ratio (kg lean solvent/Nm3 flue gas) compared with MEA. However, the en-ergy consumption is lower than that for MEA due to the low water-content of the solution and the low specific heat capacity of the solvent. FUNCTIONALIZED-IONIC LIQUIDS

Researchers at the University of Notre Dame, funded by DOE/NETL, have synthesized, tested, and evaluated several candidate ionic liquids (ILs) for post-combustion CO2 capture. A best-candidate solvent IL was identified, NDIL0157, and techno-economic analysis were performed. Extensive experimental studies could not be performed on this IL because of the high cost of making its anion component on a large scale. However detailed data is available on another IL (NDIL0046).59 Laboratory-scale results suggest that a viscous solvent (diluted NDIL0046, 100 cP) can be operated in a column with structured packing (possibly with significant modifications). The corrosion rates of materials in contact with NDIL0046 and CO2 is also very low, which has the potential to lower costs. Techno-economic calculations using NDIL0157 as a solvent to capture CO2 from a supercritical coal power plant indicate an ≈80% increase in the cost of electricity (COE), similar to the DOE Base Case (Fluor Econamine process) using MEA. NDIL0157 has significantly lower parasitic electricity loss compared to MEA, but slightly higher CO2 capture and compression equipment costs.

The Notre Dame IL technology is unique in that the viscosity does not change after CO 2 absorption. NDIL0046 solvent is viscous, and reacts about 2.5 times slower than MEA (liquid film mass transfer coefficient [kg’] at 5 kPa rich-solvent loading and 40 °C for NDIL0046: 1.4e-7 mol/s/m2-Pa vs. 3.5e-7 mol/s/m2-Pa for MEA). Because absorber volume is inversely proportional to the mass transfer coefficient, the size of the absorber needed for NDIL0046 or similar solvents would be larger (approximately at least twice) than that needed for MEA assuming that the driving force is the same. The IL used for techno-economic study, NDIL0157 has a lower viscosity (156 cP vs. 126 cP at 40 °C) than NDIL0046, and could potentially react faster than NDIL0046.60

The net-CO2 carrying capacity of NDIL0157 was estimated to be ≈0.0656 kg CO2/kg IL, or 0.059 kg CO2 for a 90:10 diluted solu-tion. This is marginally higher than that of MEA, and is advantageous. The use of diluents such as tetraglyme to lower viscosity may also lead to increased water absorption and higher parasitic energy requirements.

58 ≈70% lower mass-transfer coefficient, may require 42% larger surface area in absorber for the same driving force.
59 Apparently, NDIL0046 and NDIL0157 do not share any common components (cation/anion), and so it is not apparent if the trends in physical properties for NDIL0046 could be relevant for NDIL0157.
60 The Notre Dame wetted-wall column (WWC) experiments compared the height of a transfer unit (HTU) (or absorber volume needed for a single stage of separa-tion) for a diluted 90 wt% NDIL0046+10% tetraglyme solution with 10–20 wt% MEA, whereas the commercial Fluor Econamine process uses ≈30 wt% MEA (quarterly progress report, Nov 6, 2012). It is significant that the HTUs measured with diluted NDIL0046 were comparable to that of 10–20 wt% MEA, and that a 100 cP solvent could be operated in a column. However, the HTUs for 30 wt% MEA would be lower than that of the diluted NDIL0046 due to higher amine concen-tration and higher reactivity (lower absorber volume or packing height needed to remove the same amount of CO2).  AMINE-ORGANIC SOLVENT MIXTURES

DOE/NETL and ION Engineering have been developing a solvent with relatively low heat of regeneration (≈1.6 to 2.5 GJ/T CO2) which has low water-content and is resistant to thermal degradation with low heat-stable salt buildup on exposure to SOx and NOx. The solvent consists of a non-aqueous amine with ‘organic liquid’ as the solvent for the amine instead of water, resulting in low thermal regeneration energy due to the lower sensible and latent heats (see for example, Bara et al.61). The ION solvent (C) also has ≈30% lower solvent circulation vs. MEA for 90% CO2 capture, which may lower the capital costs of the pumps, heat exchangers, and the flash vessels. The solvent is regenerated in dual-stage flash vessels at two different temperatures (<120 °C). The use of flash vessels instead of conventional packed bed stripper columns and reboilers has the potential to lower capital costs.


This category consists of alkali or ammonium salts with weak acids (e.g., carbonic acid) used to absorb CO2. A key feature is the reaction of CO2 to form carbonate or bicarbonate ions in solution. The heat of desorption of CO2 from carbonate-based solvents is lower than that from amine-based solvents (e.g., 600 kJ/kg CO2, or 26.4 kJ/g-mol CO2 vs. ≈80 kJ/g-mol CO2 for MEA), and can po-tentially lower the overall thermal regeneration energy. Historically, these solvents were primarily used in natural gas and synthesis gas processing. Previous and current technologies include:
  • Processes based on hot potasssium carbonate solutions used for CO2 (and H2S) removal from high-pressure gas streams (with promoters to enhance CO2 absorption,27 e.g., Benfield process, Giammarco-Vetrocoke process),
  • Processes based on ambient-temperature absorption of CO2 with vacuum regeneration (e.g., vacuum carbonate process,27 aqueous ammonia-based processes [CSIRO, ECO2®]62),
  • Processes based on low-temperature absorption (e.g., chilled ammonia63).
The reaction of CO2 with basic carbonate solutions takes place by two mechanisms: reaction with hydroxide ion (OH-) (Equation 1-11), and reaction of dissolved CO2 with water followed by dissociation of carbonic acid (Equation 1-12).10, 27
The reaction shown in Equation 1-12 occurs in the absence of strong bases. It involves the reaction of CO2 with water, forming carbonic acid (H2CO3), which dissociates into bicarbonate and a proton. This is a slower reaction compared to the reaction of CO2 with hydroxide ions (Equation 1-11). The latter is the dominant reaction in the pH ranges of interest for commercial solvents (pH >8). The reaction rate of CO2 in carbonate-bicarbonate solution is slow at room temperature, and high temperatures are required to increase the reaction rate. Rate-increasing additives, or promoters improving the rate of CO2 absorption, include formaldehyde, methanol, phenol, ethanolamines, arsenious acid, glycine, and the enzyme carbonic anhydrase.27 Promoters previously used com-mercially include diethanolamine (DEA), hindered amines, glycine, and arsenious oxide.27

One unique feature of the carbonate-based processes is that if the stripper and absorber are operated at similar temperatures (either both hot or otherwise), there is no need for an intermediate lean-rich heat exchanger, lowering capital costs. In general, carbonate-based solvents cannot absorb CO2 faster than MEA from coal-fired power plant flue gas, resulting in higher absorber capital costs.

61 Bara, J. E. et al. WO2012158609 A1: Compositions and methods for gas capture processes. (2012).
62 Yang, N. et al. Aqueous ammonia (NH3) based post-combustion CO2 capture: A review. Oil & Gas Science and Technology—Revue d’IFP Energies nouvelles (2013). doi:10.2516/ogst/2013160
63 Telikapalli, V. et al. CCS with the Alstom chilled ammonia process development program–Field pilot results. Energy Procedia 4, 273–281 (2011). AMMONIA-BASED PROCESSES

Ammonium carbonate or aqueous ammonia solvents have been used for flue gas treatment. Although ammonium carbonate solutions are good CO2 solvents, the loss of ammonia to flue gas is a serious drawback. This is partly overcome by the use of low temperatures or extensive wash sections, or the use of a second salt to lower the vapor pressure of ammonia. Processes under development include the Alstom chilled-ammonia process (CAP), ambient temperature (Powerspan ECO2®, CSIRO) processes, and the ammonium carbonate-potassium carbonate mixed-salt process.

Alstom chilled ammonia process (CAP)

A detailed description of the process is provided in reference 39 and Cerimele, 2011.64 In brief, the use of a high concentration of aqueous ammonium carbonate in the absorber results in the formation of ammonium bicarbonate, which precipitates under the absorber operating conditions. The rich solvent consists of a slurry of ammonium bicarbonate (NH4HCO3) and ammonium carbon-ate ((NH4)2CO3). The formation of solid NH4HCO3 phase permits higher CO2 loading in solution compared to MEA, with carrying capacities up to 20 kg CO2/100 kg solution, with a lean loading of 24–28 w/w% CO2 versus ≈5 kg CO2/kg solution working capacity for MEA.39 The absorber operates at 0 ° to 10 °C, and requires a separate refrigeration system. The rich solution exiting the absorber can be concentrated to 30–50 w/w% solids and heated by heat exchangers and the stripper/reboiler to separate CO2. Ammonium car-bonate can be regenerated at high temperatures (and pressures, ≈300 psig, or 21 bar65) without solvent degradation, unlike amines. However, the partial pressure of ammonia (and reflux to the top of the stripper) would increase with increased stripper temperatures (and pressures). It is reported that the mass transfer kinetics in the CAP are slower than those with aqueous MEA, and are primarily affected by the free ammonia content in solution.39 With DOE/NETL support, the chilled-ammonia process was demonstrated at AEP’s Mountaineer coal-fired power plant and was designed to capture 100,000 T CO2/y. Third-party assessments indicate that the chilled ammonia process has a parasitic load comparable to 30 wt% MEA, ≈385 kWh/T CO2.66 ENZYME-CATALYZED PROCESSES

In living cells, the reaction of CO2 with water is accelerated by the enzyme carbonic anhydrase (CA). The in vivo reaction rate of CA is very high, with a turnover rate of 106 reactions per second. CA helps to maintain the acid-base balance in living cells (e.g., by converting produced CO 2 into bicarbonate for transport in the blood to the lungs where it is converted back into CO2 and expelled). In CO2 capture systems, the reactions involved are:
where M is a monovalent cation
Equation 1-13

64 Cerimele, G., L. Mountaineer Commercial Scale Carbon Capture and Storage Project. Topical Report: Preliminary Public Design Report. DOE Award No.: DE-FE0002673,
Reporting Period: 2/01/10–9/30/11 (American Electric Power Service Corporation, Columbus, OH, 2011). Coal/major%20demonstrations/ccpi/MTCCS-II---Preliminary-Public-Design-Report-Rev-12_14_2011-b.pdf.
65 Telikapalli, V. et al. CCS with the Alstom chilled ammonia process development program–Field pilot results. Energy Procedia 4, 273–281 (2011).
66 Calculated from values provided in Versteeg, P. and Rubin, E. S. Technical and economic assessment of ammonia-based post-combustion CO2 capture. Energy Procedia 4, 1957–1964 (2011).

Normal variants of CA are unstable at the conditions involved in CO2 absorption and desorption. There has been work on develop-ing versions of CA which are more stable at absorber and desorber conditions, and immobilizing the CA so that it is not exposed to high temperatures in the CO2 stripper. Technologies under development include:
  • Regenerating potassium carbonate (K2CO3) solution containing carbonic anhydrase (CA) at milder conditions compared to the use of high-temperature steam (Novozymes),
  • The use of sol-gel chemistry to immobilize the enzyme within micelles and coat it over commercial high-surface area packing (Akermin).
Bench-scale testing results and techno-economic analyses indicate that both absorption and regeneration of CO2 need to be enzyme catalyzed to fully leverage the low-heat input advantage using K2CO3. The development of encapsulated enzymes stable at tempera-tures from 70 ° to 80 °C is a challenge being addressed by currently funded NETL/DOE research. Low-temperature regeneration also results in low partial pressures of CO2 in the stripper overhead, requiring additional power for a compressor or vacuum blower. AMINE-PROMOTED CARBONATE

The Benfield hot potassium carbonate process, developed for acid-gas removal from natural gas streams used a hot potassium car-bonate solvent with diethanolamine (DEA) to promote CO2 absorption. This process is not directly relevant to coal-fired power plant CO2 capture because the absorber is operated at higher pressures (≈40 psig) and temperatures (≈75 °C) than those encountered in post-combustion CO2 capture. However, certain process design elements and heat integration features might be of interest. The ag-gressive process conditions in the hot-potassium carbonate process require the use of corrosion inhibitors (typically based on vana-dium ions). Recently, new amine promoters such as LRS-10 (London Research Station-10, a 3 w/w% solution of several formulated secondary amines67) have been used to increase the percentage of CO2 captured, and to reduce the reboiler load.

NETL-funded research has also examined piperazine as an activator in K2CO3 solutions.68 It was found that 4-m K+/4-m PZ solu-tion had faster absorption kinetics, higher solvent cyclic capacity, and lower parasitic load compared to 7-m MEA. One limitation encountered was that potassium sulfate (K2SO4) formed by the reaction of K2CO3 with flue gas SO2 could precipitate in 4-m K+/4-m PZ solvent, necessitating slurry operation of the CO2 absorber.

NETL-funded research resulted in the proof-of-concept testing of an integrated vacuum carbonate absorption process (IVCAP), where CO2 would be stripped under vacuum using a combination of direct and indirect steam stripping.69 The low-quality steam (2 to 9 psia) would be withdrawn from the low-pressure (LP) turbine,70 minimizing the derating loss due to the extraction of high-pressure (temperature) steam. A related process, but operating at higher temperature and pressure (Hot-CAP) is also being devel-oped at the bench scale.


The use of solvents which can undergo phase change is advantageous because:
  • It increases driving force for CO2 absorption by lowering the equilibrium CO2 pressure in the rich solvent,
  • It lowers regeneration energy by reducing the quantity of material that has to be heated to liberate CO2 in the regenerator.
In spite of these thermodynamic and kinetic advantages, development of phase change solvent technologies is challenging because:
  • Novel contactors for optimal gas-liquid-solid contact need to be designed and tested,
  • There is limited operational experience in operating conventional packed column absorbers with slurries at power plant scales,
  • Liquid-solid separation, or slurry concentration require additional equipment, which may increase capital costs and auxiliary load.
68 Rochelle, G., et al. CO2 capture by absorption with potassium carbonate. Final Report: DE-FC26-02NT41440. 2007.
69 Lu, Y, et al. Development and evaluation of a novel integrated vacuum carbonate absorption process. Presented at the 2011 DOE/NETL CO2 Capture Technology Meeting, Pittsburgh, PA. (August 22–26, 2011).
70 Chen, S., Lu, Y. and Rostam-Abadi, M. WO2007133595 A2: Integrated vacuum absorption steam cycle gas separation. (2007).

The ammonia-based CO2 absorption process described in section, and the CO2-binding organic liquids summarized in sec-tion are two examples of phase-change solvent technologies. Another example of this category of processes is the GE phase-changing aminosilicone absorbent process. In the DOE/NETL-funded solvent project described in section, it was observed that the GAP-0 aminosiloxane reacts with CO2 forming a solid carbamate. It was further observed that the GAP-0 carbamate was a chunky solid in the wet-salt form and a powder in its dry state. The proposed process, which is at the bench scale of development, consists of a spray absorber to contact the flue gas with the liquid lean GAP-0 solvent thereby forming a solid carbamate. A cyclone downstream of the absorber separates the fines from the flue gas, and an extruder is used to pressurize and heat the solid carbamate. The heated, pressurized solid carbamate is fed to stirred tank pressure vessels to dissolve the solid carbamate, liberate CO2 at two pressures, and the lean liquid GAP-0 is recycled back to the spray absorber.


In addition to the advancements made in solvent technologies described above, significant advancements have been made in cap-ture processes that have moved technologies toward achievement of DOE program goals. These can be generally categorized as advanced solvent regeneration (stripping) technologies, advanced gas-liquid contactors, and advanced heat integration. These are described below. ADVANCED SOLVENT REGENERATION TECHNOLOGIES

NETL/DOE is supporting the development of innovative means to regenerate CO2-rich solvents such as:
  • Gas-pressurized stripping (GPS) to improve the efficiency of the stripping process and regenerate solvent at high pres-sure, reducing CO2 compression cost (Carbon Capture Scientific [CCS] LLC). Salient features of the GPS process are:
    • The use of a stripping gas at high pressures to recover CO2 from the CO2-rich solvent, with a GPS column heated throughout its height, producing high-pressure CO2 along with the stripping gas.
    • Absorption of CO2 from the GPS column overhead gas at high pressures using the rich solvent leaving the flue-gas CO2 absorber, and subsequent recovery of CO2 at high pressure by flashing (pressure reduction).
    • The GPS process has a higher heat of absorption (due to the second absorber tower), but lower thermal energy needed to heat the solvent and evaporate water compared with the conventional MEA process.
    • The energy required to compress the CO2 is also reduced because CO2 is stripped at pressures of 15–90 atm.
    • In addition to the conventional lean-rich solvent heat exchanger, the GPS process also requires a lean-stripping/gas-rich-stripping gas heat exchanger, which may lead to higher equipment costs. Modeling results indicate that the GPS process could lower the parasitic (steam+electric) load of CO2 capture as much as 40% compared to conventional MEA CO2 capture process, resulting in a load of 180 kWh/T CO2.
  • Air stripping with heat integration (University of Kentucky Center for Advanced Energy Research [UKy-CAER]). The UKy-CAER process integrates three elements:
    • A secondary air stripper reduces the lean loading of the solvent subsequent to primary steam stripping, with the sepa-rated CO2 being recycled to the boiler, increasing the driving force for absorption by increasing the incoming CO2 concentration in the flue gas and reducing the amount of CO2 in the lean solvent,
    • Utilizing the waste heat from the primary stripper overhead and the CO2 compression intercooling to reduce the water content of the air provided to the cooling tower, in turn decreasing the cooling water temperature and lowering the condenser temperature, improving the turbine (and plant) efficiency,
    • Advanced H3-1solvent with lower thermal regeneration energy and comparable or faster kinetics compared to MEA. ADVANCED GAS-LIQUID CONTACTORS

The capital cost of the flue-gas CO2 absorber is a major (≈40%) component of overall capture plant costs. The development of equipment capable of contacting the capture solvent and the flue gas effectively has also been a focus of NETL/DOE efforts. Examples include:
  • Flat-jet absorbers for high interfacial mass transfer areas: (Neumann Systems Group [NSG]): The size of absorber needed is proportional to amount of CO2 absorbed/(mass transfer coeffficient * effective surface area * log-mean driving force). NSG’s novel jet-based absorber lowers absorber volume by having high effective surface area (≈930 m2/m3 pro-jected for the full-scale system, and 400 m2/m3 for the 0.5 MWe pilot) and relatively high superficial gas velocity (≈4–7 m/s). The theoretical maximum gas velocity in the Neustream™-C absorber is ≈15 m/s. In typical packed bed absorbers, overall surface areas (which are higher than the ‘effective’ surface area) vary from 100 to 250 m2/m3. In conventional packed-bed absorbers, flue gas velocities are often limited to 1–2 m/s to avoid flooding and high pressure drops. The low gas velocities lead to larger absorber dimensions and higher costs. In contrast, the NSG flat-jet design permits the use of high gas velocities, which could decrease the size of the equipment needed to treat a given volumetric flow rate of flue gas.
  • Membrane liquid contactors for high interfacial areas: Gas Technology Insitute (GTI) is developing a hybrid membrane/ absorption process in which a polyether ether ketone (PEEK)-based membrane contactor is used to separate CO2 from flue gas. The membrane device has high interfacial gas-liquid contact area. CO2 permeates through the membrane, react-ing with the solvent. Relatively high mass transfer coefficients (≈1.8 s-1) and low gas-side pressure drops were demonstrated during bench-scale testing.71 ADVANCED HEAT INTEGRATION

The use of intermediate-to-low pressure steam from the steam cycle for CO2 solvent stripping results in significant energy penalties because of the loss of electric power which would otherwise be generated by the steam. Thermal integration of the capture process with the overall power plant cycle can minimize the steam withdrawn from the power plant steam cycle, improving the efficiency. For example, CO2 strippers and CO2 compressors produce low-quality heat which can be used elsewhere in the power plant. Similarly, heat recovered from the flue gases before SOx removal can be used to offset part of the heat requirement for solvent regeneration. Another example is the use of mechanical vapor compression in the Cansolv process (discussed in section It is to be noted that increased heat integration requires additional capital expenditure, and often gains in energy efficiency due to advanced heat integration may be offset by increased capital costs, with the result that the overall capture cost is not reduced proportionally.
  • Heat integration of the CO2 stripper overhead and the flue gas for condensate feedwater preheating: Southern Company, with NETL/DOE support, will design and operate a 25 MWe pilot using the Mitsubishi high-efficiency system (HES) waste heat recovery technology integrated with an amine (H3-1)-based CO2 capture process unit. Waste heat in flue gas and CO2 stripper overhead streams will be used to preheat a slipstream of boiler condensate in the power plant steam cycle. Ancillary advantages include reduction in the gaseous SO3 emissions from the flue gas desulfurization (FGD) unit and savings in FGD make-up water consumption, while simplifying the boiler steam cycle.
  • Use of heat recovered from flue gas for interstage heating of solvent in the CO2 stripper: Linde proposed the use of a heat recovery unit, where energy recovered from flue gas exiting the power plant baghouse (375 °F) would be used to partly offset the steam requirement in the CO2 stripper. Low-pressure steam (<4 bar) produced in the heat recovery unit would be used to heat solvent in the CO2 stripper, akin to absorber intercooler. It is estimated that this use of low-grade heat and the use of a let-down turbine to expand the IP-LP crossover steam to <6 bar would lower the energy penalty of CO2 capture from 28–25%.

The technologies under development as part of the Carbon Capture R&D program are focused on overcoming barriers to deploy-ment of capture technologies. Table 1-3 summarizes the barriers that the projects in the Capture Program portfolio are addressing as part of their technology development efforts.

71 Zhou, S.J. et al. Hybrid membrane/absorption process for post-combustion CO2 capture. Presented at the NETL CO2 Capture Technology Meeting, July 10, 2013.

Table 1-3. Key Challenges Addressed by Solvents-Based Post-Combustion CO2 Capture Projects in Carbon Capture R&D Program Portfolio


Significant reductions in the quantity of electricity lost per unit quantity of CO2 have been achieved. Currently funded solvent technologies have lower thermal regeneration energy, lower equivalent work loss due to steam withdrawal for stripping, and require smaller quantities of solvent to capture a given amount of CO2.


Figure 1-4. Progress in Reducing Parasitic Load for Solvent-Based Post-Combustion CO2 Capture in DOE/NETL Funded Projects

Figure 1-4 and Figure 1-5 represent the progress made by the DOE/NETL post-combustion CO2 capture program in reducing the parasitic load due to CO2 capture (kWh lost/T CO 2 captured),72 and the concomitant reduction in energy penalty due to CO2 capture. Technologies being developed currently have parasitic loads from less than 200–250 kWh/T CO2 (including CO 2 compression). Reduction in parasitic load is significant because it also reduces the amount of coal to be burnt to produce a given quantity of net power. In general, the reduction of parasitic load was achieved due to a combination of more thermally stable solvents with higher cyclic capacities and lower thermal regeneration energies. Regeneration at high temperatures is advantageous because it increases the partial pressure of CO2, lowering the compression load. However, this needs to be weighed against the loss of electricity due to the use of higher quality steam for regeneration. In practice, stripping steam is typically withdrawn from the intermediate pressure-low pressure (IP-LP) crossover steam and is desuperheated to reboiler conditions.

72 Parasitic load is estimated as the equivalent work lost due to steam used for CCS + capture auxiliaries (e.g., pumps, blowers) + energy required for CO2 compres-sion from the stripper overhead conditions to 152.7 bara (2,214.5 psia). There are several means to account for the loss in steam cycle output due to the withdraw-al of steam for the capture process. IEAGHG (Retrofitting CO2 capture to existing power plants, Report 2011/02, May 2011) used a coefficient of performance for steam extraction (COPx) defined as the ratio of the heat requirement to the drop in steam cycle output. It is claimed that typical COPx values will be ≈5. Trimeric uses a derating factor depending on the quality of steam extracted for stripping. A derating factor of 0.0902 hp-h/lb steam was used in Trimeric techno-economic analyses for Case 12, and is being used here. Note that a COPx value of 5 is equivalent to a derating factor of 0.08144 hp-h/lb steam. The actual specific steam consumption estimate from the bituminous coal baseline revision 2 is 0.1120 hp-h/lb steam, which reflects a more efficient steam cycle, and higher parasitic load for steam used for CO2 stripping.

Figure 1-5. Progress in Reducing Overall Energy Penalty for Solvent-Based Post-Combustion CO2 Capture in DOE/NETL Funded Projects

For a greenfield CCS project, it may be possible to match this crossover steam pressure to the pressure required for regeneration in the reboiler (e.g., ≈3 to 4 bara pressure for amine systems73), minimizing LP turbine power losses. Other options to minimize the impact of the loss of high-grade steam include a letdown turbine which produces additional power while desuperheating the steam to match the requirements in the reboiler, which also has high capital costs. Examples of similar tradeoffs encountered in brownfield retrofits are discussed by Grol, 201274 and IEAGHG, 2007.71


Power plant efficiency and energy penalty have been considerably improved. However, significant reduction of capital costs and operating costs are required to achieve DOE goals. Key parameters to be evaluated in the future include:
  • Solvent-CO2 mass-transfer kinetics: Some of the solvent technologies with low energy penalty and parasitic load also re-act slowly with CO2 compared with aqueous MEA solutions, resulting in higher capital costs. The absorber, intercoolers, solvent circulation pumps, heat exchangers, reboiler and stripper are some of the capital-intensive pieces of equipment in a CO2 capture plant. Although solvents with improved carrying capacities can lower the capital costs of the pumps, stripper, reboiler, and cross heat exchangers, a solvent with slower kinetics adversely impacts the size of the absorber. Fundamental wetted-wall column studies, or qualitative CO2 loading studies using different solvents in the same piece of equipment are required to understand the capital cost implications of selecting a solvent with slow mass transfer kinetics.
  • Costs of second-generation solvents: The solvent technologies being developed in the DOE/NETL portfolio are novel and some of the solvent candidates have considerably higher cost compared to MEA, partly because they are first-of-a-kind technologies. However, significant advances in reducing the costs of solvent production are also needed to drive down the cost of solvents and thereby lower the cost of CO2 capture.
  • Operability concerns: These include: formation of solids in the absorber, solvent aerosol emissions, and emissions of amine degradation products.
73 IEA Greenhouse Gas R&D Programme (IEA GHG). CO2 capture ready plants. 2007/4, May 2007.
74 Grol, E., Techno-economic analysis of CO2 capture-ready coal-fired power plants. DOE/NETL-2012/1581. August 1, 2012.


Alternate means to regenerate solvents: In conventional solvent regeneration processes, the entire solvent (with the water) has to be heated to liberate CO2. Processes which could selectively supply energy to the solvent molecules so that they would liberate CO2 without energy input to the rest of the solution would reduce the parasitic energy penalty associated with CO2 capture. Examples of Capture such processes include the use of electrochemistry to capture and/or regenerate CO2.

Phase change: There has been a drive toward minimizing the energy penalty by reducing water content of solvents, and the devel opment of novel phase-change technologies is a logical outgrowth of that approach. It is relatively easy to identify solvents which undergo phase change in the presence of CO2. However, the development of contactors and ancillary equipment to handle slurries  or solids represents a potential focus area.

Hybrid approaches to CO2 capture: Technologies offering potential value for the future also include those which combine multiple mass-transfer phenomena. Such processes aim to combine the advantageous feature of each mass-transfer process while attempting to limit the drawbacks of each approach. Technologies include but are not limited to:
  • Approaches to immobilize viscous solvents in membranes or adsorbents to overcome viscosity limitations,
  • Membrane contactors with solvents used to selectively permeate CO2,
  • The use of additional mass-transfer operations (e.g., membranes, adsorbents) to reduce the water content of the rich solvent and thereby lower solvent regeneration energy.
In general, process equipment for hybrid approaches needs to be developed and tested at pilot- to slipstream-scales. Approaches combining several mass-transfer unit operations would also need additional equipment to carry out the separation processes.

Chapter 2: Sorbent-Based CO2 Capture


The removal of CO2 from gas streams using solid sorbents is not a new concept. In fact, lithium hydroxide (LiOH) sorbent was historically used for removal of CO2 at low concentrations (<1%) from air in closed environments such as spacecraft. Demand for a practically regenerable alternative to LiOH needed for extended space missions led to development of regenerable sorbents such as the solid amine beads called HSC+, developed by Hamilton Sunstrand Space Systems International and used since the early 1990s, most notably on the space shuttle. This proprietary material consists of a liquid amine, polyethyleneimine (PEI), bonded to a high-surface-area, solid polymethyl methacrylate polymeric support. The material also consists of a second liquid phase coating, poly(ethylene glycol) (PEG), to enhance CO2 adsorption and desorption rates.1 For the International Space Station, CO2 removal is accomplished with common, commercially available sorbents: there are four packed beds, two containing silica gel and zeolite 13X desiccant media, the other two containing zeolite 5A CO2 sorbent media.2

Cryogenic air separation requires removal of trace species including water vapor and CO2 from incoming feed air to avoid freezing of these in process. In large-scale commercial plants, this is typically performed in molecular sieve units containing sorbents; a very commonly used sorbent for this application is zeolite 13X3 which has high selectivity for CO2, and high affinity for water common to zeolites in general.

One of the most widespread and important applications of sorbents in large-scale gas separations is for the production of hydrogen at refineries. In this case, the objective is separation of hydrogen from various gas mixtures, such as the syngas produced in a steam methane reformer consisting of H2, CO, CO2, unconverted CH4, and N2. Generally the objective is not to isolate CO 2, but rather to separate relatively pure H2 for refinery process use, with the leftover gas consisting of unrecovered H 2 and most of the rest of the other species, which is used as fuel gas in the refinery. The most commonly used hydrogen separation technology used at refineries is pressure swing adsorption (PSA) (see section 2.2.4 for further discussion). The sorbents utilized are commercially available types including molecular sieve (zeolites), activated carbon, activated alumina or silica gel.4 In many refinery applications the adsorption process is not specific to CO2 capture, but because CO2 is a major gas species which is always targeted for capture in such PSA processes, these applications qualify in that sense as sorbent-based CO2 capture.

On the other hand, in at least one instance CO2 is targeted specifically for capture in a refinery process, this example being the currently operating carbon dioxide capture and storage project at Valero Energy’s Port Arthur Texas refinery. In this project, 90% of the CO2 from the existing steam methane reformers located within the refinery is captured in new vacuum swing adsorption (VSA) units at a rate of 1 million tonnes per year, upstream from the existing PSA process for capturing hydrogen. The process concentrates the initial gas stream from the steam methane reformers (containing 10–20% CO2 prior to water gas shift, and ≈40% after WGS) to greater than 97% CO2 purity. A common, commercially available sorbent is used in the VSA units. The CO2 is transported to the West Hastings oil field in Brazoria County south of Houston, Texas, where it is injected for enhanced oil recovery (EOR).5

Notwithstanding certain instances of established uses of sorbents for CO2 capture described above, sorbent-based CO2 capture is usually not chosen over alternatives such as solvent-based and membrane-based technology for industrial-scale carbon capture applications. For example, out of the Global CCS Institute’s database of 55 large-scale CCS projects in various stages of development to operation, 13 projects are classified as operating, with only one of those 13 projects using sorbent-based CO2 capture (the Air Products Valero Refinery project described in preceding section). Most of the 12 remaining projects totaling 25.6 million tonnes per year CO2 captured are natural gas processing, synthetic natural gas, or fertilizer production, and 11 of these plant use solvent-based processes (amine, Selexol, Rectisol) and one uses membrane separation.6

1 Satyapal, S.; Filburn, T.; Trela, J.; Strange, J., Energy Fuels 2001, 15 (2), 250–255.
2 Bernadette Luna, George Somi, J. Parker Winchester, Jeffrey Grose, Lila Mulloth, Jay L. Perry, “Evaluation of Commercial Off-the-Shelf Sorbents and Catalysts for Control of Ammonia and Carbon Monoxide,” American Institute of Aeronautics and Astronautics.
4 Zahra Rabiei, “Hydrogen Management in Refineries,” Petroleum and Coal 54(4) 357–368, 2012; pc_4_2012_rabeie_188.pdf.
5 Global CCS Institute database, information on the Air Products Steam Methane Reformer EOR Project. Accessed 15 Jan 2015.
6 Global CCS Institute database.


Notwithstanding the existence of commercial CO 2 sorbents such as various zeolites, silica gels, activated carbons, supported amine sorbents, etc., which are suitable for utilization in CO2 capture applications, the previously noted trend to opt for solvent and membrane technology alternatives to sorbents for large-scale CO2 capture implies that the sorbent based technology is simply not as cost effective as the solvent or membrane separation alternatives in many or most cases. This can probably be attributed to the following challenges facing sorbents:
  • Cost of many sorbents is high (i.e., particularly zeolites, some other types)
  • CO2 adsorption capacity limitations
  • Operation at higher flue gas/exhaust gas temperatures
  • Sorbent attrition
  • Sorbent regeneration (energy demands)

A solid sorbent post-combustion CO2 capture technology potentially offers several advantages over a conventional aqueous solvent-based process such as monoethanolamine (MEA):

Energy penalty: Solid sorbents have the potential to significantly reduce the regeneration energy penalty compared to solvents. This energy penalty reduction can be attributed to: (1) the heat capacity of solids is significantly lower than that of water (i.e., by approximately a factor of four), which dramatically reduces the sensible heat input required to accomplish the temperature swing; and (2) the moisture content during regeneration will be significantly lower for solids compared to solvents, so less evaporation will occur.

Range of operating temperature: Aqueous solvent-based processes are limited in their operating temperature range. In contrast, a range of solid sorbent types exist allowing operation from ambient temperatures to as high as 700 °C.

Miscellaneous benefits: Solid sorbents yield less waste during cycling, and spent solid sorbents have lower environmental impact than solvents easing their disposal.7



Adsorbents being considered for carbon capture employ one of the two possible adsorption mechanisms, namely physical adsorption (physisorption) and chemical adsorption (chemisorption). In physisorption, target molecules are attracted to the surface of pore walls within a high surface-area sorbent by van der Waals forces and have a low heat of adsorption that is only slightly greater than heat of sublimation of the adsorbate. The physisorption is reversible and fast. In chemisorption, the target gas undergoes a covalent chemical reaction to bind to certain sites on the sorbent with a much greater heat of adsorption, roughly equal to the heat of reaction.

Figure 2-1. Physisorption vs. Chemisorption

In physisorption the CO2 is adsorbed weakly by the substrate itself, in chemisorption, the CO2 is adsorbed more strongly by specific binding sites.8

7 Junya Wang, Liang Huang, Ruoyan Yang, Zhang Zhang, Jingwen Wu, Yanshan Gao, Qiang Wang, Dermot O’Hareb and Ziyi Zhong, “Recent advances in solid sorbents for CO2 capture and new development trends,” Energy Environ. Sci., 2014, 7, 3478.
8 Adam Hughmanick Berger and Abhoyjit S. Bhown. “Comparing Physisorption and Chemisorption Solid Sorbents for use Separating CO2 from Flue Gas using Temperature Swing Adsorption,” Energy Procedia 4 (2011) 562–567.

An example of a sorbent exhibiting the physisorption mechanism is activated carbon; CO2 adsorbs onto its high surface area pore walls by weak dipole interactions. The heat of adsorption of carbon dioxide on activated carbon sorbent ranges from -25 to -40 kJ/ mole, which is close to the heat of sublimation. This low heat of adsorption reduces the amount of energy needed to desorb a given quantity of carbon dioxide. However, it also means that carbon dioxide is less likely to adsorb onto the porous substrate.

On the other hand, solid sorbents that capture carbon dioxide through a chemical process, such as bonding with an amine that is grafted or coated onto the surface of the sorbent, have a much higher heat of adsorption. These heats of adsorption can range be-tween -60 and -100 kJ/mole depending on the amine used. While this increases the amount of energy needed to separate the carbon dioxide from the chemisorbent, it also increases the affinity of the amine for carbon dioxide which allows the amine-based solid sorbent to have a higher capacity at low concentrations of carbon dioxide.

While physisorbents such as activated carbon can to be stable even past 200 °C, grafted amines tend to volatilize and degrade above 120 °C which restricts their regeneration temperatures. Another difference is that physisorbents tend to attract various other molecules to their surface lowering the selectivity of the sorbent, which leads to more nitrogen contamination and lower carbon dioxide purity than with chemisorption for the same process configuration. On the other hand, chemisorbents can permanently bind to species such as SO2 and poison binding sites, decreasing the capacity of the sorbent.


The energy necessary to be added during the regeneration step to reverse the CO 2 absorption reaction is found in Equation 2-1. The energy balance assumes that state 1 is the system after CO2 absorption, and state 2 is the system after CO 2 regeneration. The energy balance also assumes a change in temperature occurs between states 1 and 2 (i.e., thermal regeneration is employed). The subscripts represent the equipment (e), the sorbent (s), and the CO2 (c), and with the indices 1 and 2 added to reference the states. The variable m is the mass, C is the specific heat, Cp is the constant pressure specific heat for the gaseous CO 2. Q is the heat input, and Qr is the heat of reaction (represented as a positive number) for breaking the chemical bond between the active species and the CO2. B is a constant of proportionality for dimensional units, and L is the CO2 loading capacity expressed in gmole CO2/kg.

Equation 2-1

The equation shows that for fixed swing in temperature, regeneration energy can be reduced by increased loading L or decreased specific heat C. The energy balance also assumes that the only gas adsorption reaction occurring is CO2. If there are additional reactions taking place, such as adsorption/desorption of H2O, then additional analogous terms would need to be included in Equation 2-1.

The energy balance equation highlights the need for several sorbent parameters to be determined in value. They include the heat of adsorption of the gas (CO2 reaction), the specific heat (heat capacity) of the sorbent, and the CO2 loading capacity of the sorbent. A fourth but very important consideration is the role of moisture, whether as a participant (reactant) in the CO2 reaction or as a competitor in adsorption/desorption.9

9 “Factors in Reactor Design for Carbon Dioxide Capture with Solid, Regenerable Sorbents,” James S. Hoffman, George A. R ichards, Henry W. Pennline, Daniel Fischer, and George Keller.



The temperature swing adsorption (TSA) process is based on the tendency of sorbents such as activated alumina, silica gel, and zeolites to have higher adsorption capacities for gases at lower temperatures than at elevated temperatures. The sorbent adsorbs relatively large quantities of the target gas such as CO2 at the lower temperatures of flue gas (≈100 °F or ≈40 °C). The sorbent loaded with CO2 can then be regenerated simply by increasing its temperature to ≈100 °C or greater (typically with heated purge gas) in a regeneration step which drives off the CO2, after which the regenerated sorbent can be cooled back down to operating temperature and returned to CO2 capture. The simplest configuration possible is a two-bed system, with one bed undergoing adsorption of CO2 from the feed (flue gas), while the other bed undergoes regeneration by heated purge gas (often steam). When breakthrough is about to occur in the bed capturing CO2 from flue gas, the beds are switched. Also, gases exiting the adsorption bed may be recycled through it if necessary to increase uptake of CO2. See Figure 2-2 below which depicts the basic idea of TSA. TSA takes its name from the essential process of modulating or “swinging” temperature to drive off the adsorbed gas, in direct contrast to pressure swing adsorption (PSA) which uses changes in pressure, not temperature, to release adsorbed gas. TSA technology began commercially in the 1960s, and has current application in drying compressed air and natural gas, as well as other purification applications such as carbon dioxide removal from air. Large-scale use to capture CO2 from flue gas is contemplated.

Figure 2-2. Two-Bed Temperature Swing Adsorption Process for Post-Combustion CO2 Capture10


  • Suitable for operation on low pressure feed streams (<60 psig) where PSA or VSA may be less practicable.
  • Typically less expensive to operate than PSA.
  • A practical problem in TSA processes is the reduction in the capacity or life of the adsorbent when it is subjected to repeated thermal cycling.
  • Because beds of adsorbent cannot normally be heated and cooled quickly, the cycle time of a typical TSA process may range from several hours to several days. Long cycle times inevitably mean large bed lengths resulting in high adsorbent inventories, and high capital cost.
Effects of Sorbent Heat of Adsorption in TSA11

Increasing the heat of adsorption also increases the temperature dependence of the isotherms (which describe sorption capacity as a function of pressure for fixed temperatures). Sorbents with small heats of adsorption will require a much larger change in tempera-ture to affect the same change in adsorbed capacity as a sorbent with a large heat of adsorption would experience from a small temperature swing. Figure 2-3 shows this difference in the isotherms for two sorbents that are identical expect for the heat of adsorption. Because the TSA cycle involves both adsorption at low temperatures with a low partial pressure of carbon dioxide and desorption at a high temperature with a high partial pressure, the difference between the low temperature low purity and high temperature high purity determines the working capacity or quantity of carbon dioxide that is produced per cycle. Therefore, having a high absolute capacity on its own is not enough to ensure a high working capacity.

Figure 2-3. Isotherm for CO2 Sorbents with Heat of Adsorption of -25kJ/kg and -60 kJ/kg

The Langmuir isotherm
Equation 2-2

Equation 2-3

11 Adam Hughmanick Berger and Abhoyjit S. Bhown. “Comparing Physisorption and Chemisorption Solid Sorbents for use Separating CO2 from Flue Gas using Temperature Swing Adsorption,” Energy Procedia 4 (2011) 562–567.

assumes that there are a fixed number of sites qs to adsorb onto, with each site having the same heat of adsorption ∆h. The number of sites that are occupied, q, is determined by the partial pressure of the adsorbed species PCO2 and the Langmuir parameter bCO2. In this equation R is the universal gas constant and bo is a constant that is fit to the adsorption data. The Langmuir model is a reasonable assumption for both chemisorption and physisorption. In chemisorption, there are a fixed number of amine sites, each of which will bind to the carbon dioxide with roughly the same energy. In physisorption, the fixed number of sites is equivalent to the monolayer coating of carbon dioxide that would correspond to a saturated sorbent.12 Further, from Equation 2-2, the higher the Langmuir value bCO2 is, the faster the isotherm approaches saturation as PCO2 increases. In Equation 2-3, bCO2 depends exponentially on temperature and heat of adsorption, changes in temperature impact the isotherm more for higher heat of adsorption.

DOE Research

DOE NETL funded research in sorbent capture using TSA in the Georgia Institute of Technology project addresses some of the key issues noted above. The project is developing a supported amine sorbent-based CO2 capture module at bench scale. The sorbent incorporates supported amines in composite polymer/silica hollow fibers, with high loadings of the sorbent to facilitate large CO2 adsorption capacities but without the high pressure drops normally associated with traditional packed and fluidized beds. Also the inclusion of an impermeable layer lining the fiber’s interior bore allows for rapid cooling during adsorption and heating during desorption by passing water through the bore of the fibers, essentially turning the fiber module into a shell and tube heat exchanger. This will allow rapid temperature swing adsorption with the cycles running on the time scale of seconds/minutes instead of hours/ days. High capital costs typical of TSA would be significantly reduced by the fast cycling; also, operating costs and parasitic loads will be reduced through heat integration of the adsorption step with preheating the boiler feed water. See Figure 2-4 for illustration of these concepts.

The project is also addressing issues of oxidation, SOx exposure, and NOx exposure in deactivation of the sorbent fibers, and developing techniques for effective field regeneration.

Figure 2-4. Sorption (top) and Desorption (bottom) Modes in Hollow Fiber Sorbents

Many of the other sorbent-based CO2 capture projects in the DOE NETL portfolio assume conventional TSA process operation (adsorption at 40 °C, regeneration at 100 °C).



In PSA a pressurized gas mixture containing CO2 flows through a bed of adsorbent, ideally with high selectivity and capacity for CO2 adsorption, until the adsorption of the CO2 approaches equilibrium with the solid (after which significant breakthrough of CO2 would occur). The bed is then regenerated by stopping the feed mixture and reducing the pressure, which releases the CO2 gas which can be removed. Figure 2-5 depicts the general steps of PSA, on the left with adsorption occurring and on the right with regeneration occurring and capture of the CO2-rich gas stream.

12 Figueroa J.D., Fout T., Plasynski S., Mcilvried H., Srivastava R.D. “Advances in CO2 capture technology—The U.S. Department of Energy’s Carbon Sequestration Program,” Intl. J. of Green House Gas Control 2008; 2:9-20.

Figure 2-5. Pressure Swing or Vacuum Swing Adsorption Process for CO2 Capture13

A variation of PSA is Vacuum Swing Adsorption (VSA) in which the process stream containing CO2 is not pressurized. CO2 is re-moved from the sorbent by establishing a vacuum on the regenerating sorbent bed which draws off the CO2. The mechanism is the same in either case: CO2 adsorbs at higher pressure and desorbs at lower pressure, and the pressure differential can range above or below ambient pressure.

  • Well-established process technology, used on very large scale for certain applications
  • Needs no steam or water; only electricity required to operate compressors/pumps; energy intensity generally better than conventional solvent-based CO2 removal technology (amines)
  • Tolerant to trace contaminants; possibly with use of guard or layered beds
  • Wide range of sorbent materials are commercial and widely available
  • Increase in COE potentially lower than that of other capture technologies
  • Sorbent beds require limited hands-on maintenance, translating into flexibility in locating them
  • Energy intensity still too high to meet goals: improved sorbent and novel process designs needed to overcome
  • Large sorbent beds required for high-throughput applications such as flue gas treatment
    • High capital cost for sorbent and associated process equipment
    • Implies large pressure drop and associated power penalty: could possibly be overcome by structured adsorbents and faster cycling
    • Large plant footprint for beds: possibly overcome by underground installation and faster cycling or smaller beds


The Carbon Capture Program has supported research and development in a diverse set of solid sorbent-based CO2 capture approach-es to address the technical challenges described above. A major part of the R&D sorbents portfolio consists of post-combustion capture technology projects, in which the process context is capture of CO2 from relatively low-temperature exhaust and flue gases. The remainder of the portfolio consists of pre-combustion capture technology, in which the CO2 would be captured from higher-temperature syngas. Recent projects are referenced in Table 2-1.

Sorbent Types

Types of sorbents for CO2 capture are highly varied, and as a rule the types are distinctly suited to the temperature regimes in which they are intended to operate for adsorption and regeneration. In their extensive surveys of the literature for sorbent-based CO2 capture, Wang and fellow researchers14 have adopted a general classification of adsorbents based on three sorption and desorption temperatures ranges at which they operate, which are as follows:
  • Low temperature (<200 °C)
  • Intermediate temperature (200–400 °C)
  • High temperature (>400 °C)
Applicable sorbent types fall into these categories as follows:

Low Temperature Sorbents
  • Solid amine-based adsorbents (with variations in supports for these sorbents including silica; carbon in various forms including activated carbon, carbon nanotubes, graphite/graphene; polymers; low-cost materials such as clays, etc.)
  • Carbon-based adsorbents
  • Zeolite-based adsorbents
  • Metal-organic framework (MOF)-based adsorbents
  • Alkali metal carbonate-based adsorbents
  • Others such as immobilized ionic liquid-based sorbents, etc.
Intermediate Temperature Sorbents
  • Layered double hydroxide (LDH)-based sorbents
High Temperature Sorbents
  • CaO-based sorbents
  • Alkali ceramic-based sorbents
This classification of sorbent types is useful, because it captures essentially all of the types currently being investigated in the field of sorbent-based carbon capture, differentiates among the types according to the major temperature regimes characteristic of their application for either pre- or post-combustion capture, and provides a framework for showing how the DOE/NETL Carbon Capture Programs’ R&D projects cover areas of this field of R&D.

In addition, process innovations and improvements have contributed to overcoming the technical issues that serve as barriers to deployment of carbon capture technologies. All of these are described below.

14 Junya Wang, Liang Huang, Ruoyan Yang, Zhang Zhang, Jingwen Wu, Yanshan Gao, Qiang Wang, Dermot O’Hare and Ziyi Zhong, “Recent advances in solid sorbents for CO2 capture and new development trends,” Energy Environ. Sci., 2014, 7, 3478. Qiang Wang, Jizhong Luo, Ziyi Zhong and Armando Borgna, “CO2 capture by solid adsorbents and their applications: current status and new trends,” Energy Environ. Sci., 2011, 4, 42.

Table 2-1. DOE/NETL Sorbent-Based CO2 Capture Projects Portfolio

Sorbent tradeoff

Table 2-2. Comparison of Sorbent Types for CO2 Capture

15 UOP’s work is not continued in current portfolio; UOP has not pursued funding in recent FOAs.


Aqueous amine-based solvent processes are often chosen for current, commercial applications of CO2 capture. However, given the drawbacks of such processes, including the high regeneration energy, fouling of process equipment, solvent foaming problems, emissions from volatile solvents, and corrosion caused by amine solutions, effort has been made to prepare amine-based solid ad-sorbents for CO2 capture by immobilizing organic amines on certain solid support materials. Solid amine adsorbents promise lower capital cost, lower pressure for gas recovery, and lower energy consumption for regeneration compared to baseline amine solvent processes.16

Generally the solid supports should have a good affinity for the amine molecules, high surface area, proper porosity, good mechanical strength and hydrothermal stability. Non-porous and microporous materials are generally avoided, while mesoporous silica, carbon, and certain polymers and resins are preferred due to their proper combination of high surface areas and large pore size and volume.17

Mechanism of Capture

Amine-base sorbents undergo the same types of reactions with CO2 which are known to occur with amines in aqueous solutions. In the case of amine-based solid adsorbents, the reactions of airborne CO2 and water vapor occur with amine functional groups chemically bonded to and located on the sorbent surface, or with liquid amines immobilized within a porous support. The reactions involving primary, secondary and tertiary amines are as follows:

Intermediate zwitterion formation

CO2 forms intermediate zwitterions by reaction with primary or secondary amines:
Equation 2-4

Carbamate formation

Intermediate zwitterions further react with more amine to form stable carbamates:
Equation 2-5

The zwitterion from reaction of a secondary amine with CO2 is further deprotonated by a base (amine) to form carbamate (R1R2N-COO-) and a cation (R1R2NH2+) as shown here:

(secondary amine)
Equation 2-6

16 R. A. Khatri, S. S. C. Chuang, Y. Soong and M. Gray, Energy Fuels, 2006, 20, 1514.
17 Qiang Wang, Jizhong Luo, Ziyi Zhong and Armando Borgna, “CO2 capture by solid adsorbents and their applications: current status and new trends,” Energy Environ. Sci., 2011, 4, 42.

Without water vapor present, carbamate formation is the mechanism for CO2 removal. Stoichiometrically, this allows capture of 1 mole of CO2 for every 2 moles of amine (CO2 loading of 0.5). However, in practice water is usually present in capture of CO2 from flue or exhaust gases, which allows subsequent reaction of the carbamate to form carbonate, regenerating the amine, and increasing the CO2 loading to values above 0.5 and approaching a theoretical loading of 1:
Equation 2-7

Protonation of tertiary amine

Tertiary amines can react with carbonic acid, formed from slow hydration of CO2. Bicarbonate ion and an alkylammonium cation result:

Equation 2-8

Recent Developments and DOE Contribution

An emphasis of current work in solid amine-based sorbents development has been in trying to improve the CO2 sorption capacity by:
  • Improved approaches for supporting amines on porous materials. Different synthesis methods include (1) impregnating amines into supports (generally resulting in high capture capacity but issues of transport limitations of CO2 to active sites and the leaching of amines over multiple regeneration cycles which tend to limit their performance and long-term viability), (2) post synthesis grafting (covalently attaching amines onto solid surfaces, e.g., hyperbranched amines covalently attached to silica, amine groups functionalized onto silica or alumina supports, etc.), and (3) direct condensation (an example which did away with the amine support altogether is polyamine porous particles prepared by a precipitation–polymerization method).18
  • Selecting appropriate amines. Tradeoffs exist among the choices of primary amines, secondary amines, tertiary amines, and polyamines. Primary amines can form stable complexes with CO2, capture CO2 efficiently but are difficult to regenerate. Tertiary amines do not capture CO2 as efficiently as primary amines, but can be regenerated readily at relatively low temperatures. Secondary amines may offer a compromise between primary and tertiary amines.
  • Enhancing CO2 diffusion. For example, surfactant-promoted adsorbents in which a CO2-neutral surfactant is introduced in the support might create extra CO2 transfer pathways to facilitate CO2 diffusion into the deeper films.
For a survey of recent work in the field, the reader is referred to Wang et al. 2014 pp. 3479–348019 for much more detailed descrip-tion than can be included here.

18 H. B. Wang, P. G. Jessop and G. Liu, ACS Macro Lett., 2012, 1, 944.
19 Junya Wang, Liang Huang, Ruoyan Yang, Zhang Zhang, Jingwen Wu, Yanshan Gao, Qiang Wang, Dermot O’Hareb and Ziyi Zhong, “Recent advances in solid sorbents for CO2 capture and new development trends,” Energy Environ. Sci., 2014, 7, 3478.

The DOE NETL-funded research portfolio includes a number of projects investigating solid amine-based adsorbents:
  • The sorbent selected for the ADA-ES project is a commercial sorbent consisting of an ion exchange resin (polystyrene) with a primary benzyl amine that removes CO2 in a TSA process cycle (40 °C adsorption, 120 °C regeneration).
  • Aspen Aerogels, Inc. is conducting research on amine-functionalized aerogel pellet sorbents, which have high surface area and porosity, tailored pore size distribution, highly-stable functionality, and excellent hydrophobicity for resisting degradation from flue gas and its contaminants over long-term use. Amines are grafted onto hydrophobic aerogel.
  • RTI is working on polyethyleneimine (PEI) supported on commercial silica, based on Pennsylvania State University’s work in “molecular basket sorbents” which had looked to replace expensive mesoporous molecular sieve supports with more cost-effective alternatives. They investigated mesocellular silica foam, hexagonal mesoporous silica, silica gel, fumed silica, activated carbon, and carbon black as supports.
  • Georgia Tech Research Corporation is investigating amines loaded into composite polymer/silica hollow fibers in innovative module configurations. Both impregnated PEI and class 2 (grafted 3-aminopropylsilane) fiber sorbents utilizing composite commercial mesoporous silica and low-cost commercial polymers (e.g., cellulose acetate) were investigated.
  • University of Akron is investigating sorbents based on porous carbon and silica support structures with immobilized aliphatic amines distributed inside the material’s pores and immobilized aromatic amines placed on the external surface and pore mouth of the material. The idea is that the aromatic amines, while having negligible CO2 capture capacity, are capable of adsorbing and desorbing SO2 in the same temperature range as aliphatic amines do for CO2. The expectation is that the aromatic amines on the pore mouths and particle surface will prevent sorbent deactivation by selectively capturing SO2, thus allowing the aliphatic amines within the pores to capture CO2 instead of irreversibly adsorbing residual SO2 from the flue gas desulfurization-treated flue gas.
  • Advanced Fuel Research was investigating amines supported on low-cost activated carbon, with the carbon produced from scrap tires. This directly addressed the issue of sometimes high capital costs of sorbent based on expensive support materials such as molecular sieves, etc.  CARBON-BASED ADSORBENTS


Carbon-based materials have advantages of low cost, high surface area, high amenability to pore structure modification and surface functionalization, and relative ease of regeneration. However, the CO2 adsorption mechanism on carbon materials is physisorption and therefore weak, resulting in relatively poor CO2 selectivity, and significant drop in CO2 sorption capacity at temperatures associated with power plant flue gas (50–120 °C). Accordingly, research has been focused mainly on methods to increase CO2 capacity by increasing surface area and tuning the pore structure of carbon sorbents, increasing alkalinity by surface modifications (such as nitrogen doping, amine modification, oxidation, fluorination, modification with metal oxides), and synthesis of carbon-based hybrid composites. The reader is referred to the reviews of R&D work in this field prepared by Wang et al.20, 21for much more detailed discussion.

DOE Research

The current research portfolio’s coverage of carbon-based CO2 sorbents is dominated by TDA’s projects in which mesoporous car-bon modified by functional groups for improved CO2 capture capacity and performance is being evaluated for both pre-combustion and post-combustion applications. Further detail on the makeup of the mesoporous carbon sorbent is regarded as proprietary. How-ever, it is noted that the functional modification allows relatively strong adsorption but without the CO2 forming a true covalent bond with the surface sites. Therefore, regeneration can be carried out with only a small energy input of 4.9 kcal per mol of CO2, which is much lower than that for chemical absorbents or amine based solvents.

The process implementation of the work is innovative; in recent work the approach is to integrate a membrane module and a sorbent capture system in a two-step CO2 separation process.

20 Junya Wang, Liang Huang, Ruoyan Yang, Zhang Zhang, Jingwen Wu, Yanshan Gao, Qiang Wang, Dermot O’Hare and Ziyi Zhong, “Recent advances in solid sorbents for CO2 capture and new development trends,” Energy Environ. Sci., 2014, 7, 3478.
21 Qiang Wang, Jizhong Luo, Ziyi Zhong and Armando Borgna, “CO2 capture by solid adsorbents and their applications: current status and new trends,” Energy Environ. Sci., 2011, 4, 42. ZEOLITE-BASED ADSORBENTS


Zeolites are porous crystalline aluminosilicates, whose structure consists of interlocking tetrahedrons of SiO4 and AlO4 joined together in various regular arrangements through shared oxygen atoms. They have open crystal lattices containing pores with molecular dimensions, into which molecules can penetrate. The negative charge created by the substitution of an AlO4 tetrahedron for a SiO4 tetrahedron is balanced by exchangeable positively charged cations (e.g., Na+, K+, Ca2+, Mg2+), which are located in the channels and cavities throughout the structure.22 Such zeolites have ion-exchange, catalytic and adsorptive properties, which are highly dependent upon the size, charge density, and distribution of these cations in the porous structure. The structure is depicted in Figure 2-6 for a synthetic faujasite X zeolite (faujasite zeolites are divided into two types X and Y; for X zeolites the silica to alumina ratio is between 2 and 3, while in Y zeolites it is 3 or higher).

Figure 2-6. Framework Structure of Low-Silica X Zeolite (cations not shown)23

Mechanism of Capture

Zeolites’ defined crystalline structures result in uniform pore sizes in the interval of 3 to 10 Å, a property that allows them to sepa-rate molecules by means of the molecular sieve effect. Separation of gases in zeolites can also take place through the mechanism of selective adsorption of those molecules that have relatively large energetic non-saturations (π bonds, dipoles and quadrupoles). Generally speaking, separation of gases by these adsorbents then depends on three factors: structure and composition of the frame-work, cationic form, and zeolitic purity (both natural zeolites and synthetic zeolites are utilized, the former often contaminated with other minerals which drives the issue of purity).

Different cationic forms of a given zeolite may lead to significant differences in the selective adsorption of a given gas such as CO2, due to both the location and size of the interchangeable cations which affect the local electrostatic field, and the polarization of the adsorbates.24

22 Qiang Wang, Jizhong Luo, Ziyi Zhong and Armando Borgna, “CO2 capture by solid adsorbents and their applications: current status and new trends,” Energy Environ. Sci., 2011, 4, 42.
23 Yasuo Nozue, “Novel Properties of Correlated Electrons in Alkali-Metal Clusters Incorporated in Regular Nanospace of Zeolite Crystals,” Croatian-Japanese Workshop on Materials Science, Zagreb, 29–30 June 2009.
24 Rosario Hernández-Huesca, Lourdes Díaz, Gelacio Aguilar-Armenta, “Adsorption equilibria and kinetics of CO2, CH4 and N2 in natural zeolites,” Separation and Purification Technology, Volume 15, Issue 2, 8 March 1999, pp. 163–173.

Specifically, the CO2 adsorption mechanism on zeolites includes physisorption of CO2 via an ion–dipole interaction, with the CO2 in a linear orientation:
In addition to physical adsorption, more strongly bound carbonate species are also observed. These adsorbed CO2 sites are bent and associated with bi-coordination, as shown in the following two types of carbonate species, as shown in this case involving sodium cations:

Equation 2-9

Suitability for CO2 capture

In recent years extensive studies have examined zeolites for their suitability as CO2 capture sorbents. For example, one study ex-amined a number of common zeolites for CO2 capture from nitrogen in flue gas: 5A, 13X, NaY, NaY-10, H-Y-5, H-Y-30, H-Y-80, HiSiv 1,000, H-ZSM-5-30, H-ZSM-5-50, H-ZSM-5-80, H-ZSM-5-280, and HiSiv 3,000. The CO2 pure component adsorption isotherms and expected working capacity curves for an assumed pressure swing adsorption (PSA) application were determined for a selected promising subgroup of these adsorbents (these included 13X, NaY, H-Y-5, HiSiv 1,000, ZSM-5-30, HiSiv 3,000). The results showed that promising adsorbent characteristics are a near linear CO 2 isotherm and a low SiO2/Al2O3 ratio with cations in the zeolite structure which exhibit strong electrostatic interactions with carbon dioxide.25

The CO2 capacity of zeolites is affected by temperature, with capacities strongly decreasing with increasing temperature. Zeolites 13X and UOP WEG-592 were studied in the context of CO2 removal from flue gas utilizing a combination of PSA and TSA (PTSA) in an NETL systems study.26 Both of the zeolites were found to have good CO 2 adsorption capacity at 120 °C. However, the capaci-ties at 120 °C were considerably lower than that at ambient temperature; and in fact zeolites cannot be used above 200 °C for CO2 capture since their capture capacities are extremely low at these temperatures. The marked temperature sensitivity of zeolites must be considered in process context in terms of column design: since the adsorption process is exothermic, the heat released during adsorption will tend to increase the column temperature as the heat transient moves through the column, driving down capacities. Therefore, the importance of the heat effects and operating temperature on the adsorption capacities cannot be overlooked.

So, although adsorbent capacity is an important parameter for zeolite performance in CO2 capture, the optimal design of a PSA separation unit is a complex task as the separation and recovery effectiveness are highly sensitive to the chosen operating conditions as well as adsorbent type. By way of illustration, Harlick and Tezel concluded that if a low-pressure CO2 feed and very low regeneration pressure are used then the NaY and 13X adsorbents would be preferable, but that as CO2 feed and regeneration pressure increases, adsorbent with a linear isotherm would be in order.27

Zeolites’ affinity for water represents a particular challenge in CO2 capture from water vapor-containing gas streams, since water molecules directly complete with CO2 molecules for the adsorption sites on zeolites, particularly X zeolites favored for CO2 capture. Although CO2 adsorption capacity can be regained by increasing the temperature during regeneration, the loss of adsorption capacity during the adsorption cycle has negative impacts on overall cost-effectiveness of zeolite sorbent-based processes. Also, water may have a detrimental effect on the stability of zeolite frameworks; the combination of CO 2 and water forming weakly acidic conditions may cause dealumination of zeolite structures, leading to a partial or total destruction of the framework.28

25 Peter J.E. Harlick, F. Handan Tezel, “An experimental adsorbent screening study for CO2 removal from N2,” Microporous and Mesoporous Materials, Volume 76, Issues 1–3, 1 December 2004, pp 71–79.
26 “CO2 Capture Utilizing Solid Sorbents,” Ranjani Siriwardane, Ming Shen, Edward Fisher, and James Losch. U.S. Department of Energy, National Energy Technology Laboratory.
27 Harlick and Tezel 2004.
28 Wang et al. 2011.


Current Trends

Current research in improving zeolites for CO2 capture is mainly focused on:
  • Changing the composition and structure of the zeolitic framework.
  • Cationic exchange (the cations influence the electric field inside the pores as well as the available pore volume, and provide a convenient mean for tuning the adsorptive properties of these porous materials). For example, for certain zeolites, Li+ provides the highest CO2 capture capacity among all the univalent cations; in other situations certain exchanged cations such as Ca2+, Mg2+, and Sr2+ result in much higher capacities in given zeolites.
  • Modifying the existing zeolites with various amine groups (by co-condensation, impregnation or grafting) to further increase the CO2 capture capacity. Initial work tended toward simpler approaches of impregnation; however, adsorbents impregnated with amines experience low thermal stability. To overcome this problem, recent proposals have included grafting of aminosilanes through silylation onto the intrachannel surface of mesoporous silica. In general, amine-grafted adsorbents have exhibited a comparatively higher adsorption rate and higher stability in cyclic runs than the amine-impregnated adsorbents.
  • Preparing zeolite-based hybrid materials. For example, zeolite-based hybrid materials such as SBA-3/cotton fiber composites have also been studied for enhanced CO2 capture.
The reader is referred to Wang et al.29 for a thorough review of recent work in the field.


Recent NETL research related to zeolite type sorbents for CO2 capture has tended to emphasize the overall CO2 capture cycle, and improved zeolite structuring in the sorbent beds to address process concerns. The key idea of the current W.R. Grace project is to use high sorbent packing density in the fastest possible cycling rate in a PSA cycle or rapid PSA, which would result in multiple benefits including reduced capital costs (smallest possible sorbent beds and correspondingly smaller columns, and lower footprint occupied), reduced energy consumption (mainly associated with improved pressure drop, shorter sorbent regeneration times, and improving mass transfer issues), and reduced environmental burdens (less attrition of sorbent, etc.) Work demonstrated that zeolite crystals can be coated onto a crushproof metal foil structure (Catacell core structure) which has lower density, lower pressure drop and energy penalty for the same CO2 removal capacity, and better plug flow behavior (corrugated structure not subject to premature breakthrough) compared to a conventional packed bed of zeolite pellets. InnoSepra LLC’s project’s key innovation is sorbent in structured monolithic form, which again has significantly lower pressure drops, faster kinetics, virtually no attrition losses, requires less parasitic power, and has a lower overall capital cost compared to conventional sorbent forms of pellets and spheres. METAL-ORGANIC FRAMEWORK-BASED ADSORBENTS


Metal-organic frameworks (MOFs) are novel hybrid materials that combine bridging organic ligands and metal ions or metal-containing clusters. MOF materials have usually been created by crystallization from hot solutions; upon removal of “guest” solvent molecules from the resultant crystalline network by evaporation and pressure reduction, a large internal surface area results typi-cally with positively charged metal sites lining the cavities or channels, providing extensive active sites for other guest molecules (such as gases like hydrogen, methane and CO2) to adsorb. MOFs’ robust 3D structures are crystallographically well-defined, and many of them possess superior surface areas relative to those of traditional adsorbents such as activated carbon and zeolites.30 One of the archetypical examples is MOF-5, constructed from zinc atoms as the metal centers and terephthalic acid as the organic linker depicted in Figures 2-7 and 2-8 below. In this MOF, each [Zn4O] unit is bridged by six benzene-1,4-dicarboxylates. MOF-5 forms stable crystals (extended crystalline lattice of MOF-5 is depicted in the second figure) having remarkably low density for a crystal-line material (0.59 g/cm3), pore volumes of 0.61±0.54 cm3/cm3 and very high estimated Langmuir surface area of 2,900m2/g,31 even exceeding the values of pore volumes and surface areas typical of zeolites.

29 Junya Wang, Liang Huang, Ruoyan Yang, Zhang Zhang, Jingwen Wu, Yanshan Gao, Qiang Wang, Dermot O’Hare and Ziyi Zhong, “Recent advances in solid sorbents for CO2 capture and new development trends,” Energy Environ. Sci., 2014, 7, 3478.
30 Yangyang Liu, Zhiyong U. Wang and Hong-Cai Zhou, “Recent advances in carbon dioxide capture with metal-organic frameworks,” Greenhouse Gas Sci Technol. 2:239–259 (2012).
31 Li H, Eddaoudi M, O’Keeffe M and Yaghi OM, Design and synthesis of an exceptionally stable and highly porous metal-organic framework. Nature 402:276–279 (1999).

Figure 2-7. MOF-5 Framework Designed and Synthesized by Li and Co-Researchers
Top three figures show the Zn4(O)O12C6 cluster, as ball and stick model (Zn, blue; O, green; C, grey) left, and with the Zn4(O) tetrahedron indicated in green in middle and ZnO4 tetrahedra indicated in blue on right. Bottom, one of the cavities in the Zn4(O)(BDC)3, MOF-5, framework. Eight clusters (only seven visible) constitute a unit cell and enclose a large cavity, indicated by a yellow sphere of diameter 18.5Å in contact with 72 C atoms (grey).32

Figure 2-8. 3D Structure of Crystalline MOF-533

32 Li H, Eddaoudi M, O’Keeffe M and Yaghi OM, Design and synthesis of an exceptionally stable and highly porous metal-organic framework. Nature 402:276–279 (1999).
33 Jarad A. Mason, Mike Veenstra and Jeffrey R. Long, “Evaluating metal–organic frameworks for natural gas storage,”Chem. Sci., 2014, 5, 32–51.

Another well-known MOF is HKUST-1 (depicted in Figure 2-9) having the formula Cu₃(TMA)₂(H2O)₃]n (where TMA is benzene-1,3,5-tricarboxylate); structurally, it consists of copper(II) paddlewheel dimers bridged by the TMA.34

Figure 2-9. Basic Structure of HKUST-1
Turquoise: Cu; grey: C; red: O (hydrogen atoms not depicted for clarity).35

Tailoring MOFs for CO2 Capture

Through judicious selection of metal ions and organic linkers, the structure and properties of MOFs can be systematically tuned and used for specific applications—hydrogen storage, methane storage, catalysis, and CO2 capture figure prominently in the literature as applications of interest for MOFs. Beyond the tailoring of the MOF through selection of metal ions and organic bridging ligands, it is also possible to chemically modify MOFs after the crystalline materials have already been formed in a process known as post-synthetic modification (PSM) of MOFs, typically involving chemical reactions involving covalent bond formation with the framework. This is expected to open up new opportunities for functionalizing MOFs for specific applications: (1) ability to include a more diverse range of functional groups in MOFs, freed of the restrictions posed by MOF synthetic conditions; (2) easy purification and isolation of modified products because the chemical modification is performed directly on crystalline solids using reagents in non-solid phases; (3) a single given MOF structure can be modified with different reagents thereby generating a large number of topologically identical, but functionally diverse MOFs; (4) control over both the type of substituent and the degree of modification allows introduction of multiple functional units into a single framework in a combinatorial manner, enabling an effective way to systematically fine-tune and optimize MOF properties for a given intended application.36

The combined favorable properties of large surface area, permanent porosity, tunable pore size/functionality, and potential for fine tailoring possible by post-synthetic modification, have suggested great promise of MOFs as superior sorbents for CO2 capture.

DOE Research

DOE has supported research in the use of metal organic frameworks to remove CO2 from flue gas. In the period of 2007–2010, a team led by UOP investigated microporous MOFs and an associated vacuum-pressure swing adsorption (vPSA) process for the re-moval of CO2 from coal-fired power plant flue gas.37 Out of an array of over 35 MOF materials evaluated for favorable CO2 adsorption at more realistic flue gas conditions, they found that seven MOFs (MIL-101, MIL-53, MIL-96, Zn, Co, and Ni/DOBDC, and HKUST-1) exceeded initial performance targets. Notably, the particular MOF structure type, M/DOBDC (M designates Zn, Co, Ni, or Mg and DOBDC refers to the form of the organic linker in the resultant MOF structure, dioxybenzenedicarboxylate) performed well, and they found that one particular type, Mg/DOBDC outperformed all MOF and zeolite materials they had evaluated, with about 25 wt% CO2 captured by this MOF at flue gas conditions (≈0.13 atm CO2 pressure, 311K), as opposed to Ni/DOBDC with under 10 wt% CO2 captured. On the other hand, they noted that MOFs fully equilibrated with water do not pick up appreciable CO2, and they ventured the conclusion that the flue gas stream would need to be dried before being passed over an adsorbent bed of Ni/ DOBDC or any other MOF. Process-wise this may not be unreasonable, though flue gas drying/dehumidification would represent added process complication and cost.

34 Stephen S.-Y. Chui, Samuel M.-F. Lo, Jonathan P. H. Charmant, A. Guy Orpen, Ian D. Williams, “A Chemically Functionalizable Nanoporous Material [Cu3(TMA)2(H2O)3] n,” Science 19 February 1999: Vol. 283, no. 5405, pp. 1148–1150.
35 Pure Appl. Chem., Vol. 85, No. 8, pp. 1715–1724, 2013.
36 Wang Z and Cohen SM, “Post-synthetic modification of metal-organic frameworks,” Chem Soc Rev 38:1315–1329 (2009).
37 Richard Willis, “Carbon Dioxide Removal from Flue Gas Using Microporous Metal Organic Frameworks,” Final Technical Report for DOE Award Number: DE-FC26-07NT43092, October 2010.

Kizzie et al. examined effects of water on the same series of M/DOBDC MOFs identified by UOP as outperformers, quantifying the CO2 adsorption capacities of the regenerated sorbents after exposure to 70% relative humidity. For Mg/DOBDC, Zn/DOBDC, Ni/ DOBDC, and Co/DOBDC, 16%, 22%, 61% and 85% of the initial CO2 capacities were recovered, respectively.38 Liu et al. note that the different degree of capacity retention for these likely reflects the different stability of the MOFs toward hydrolysis. Even though Mg/DOBDC has excellent CO2 adsorption in dry conditions, the results show that Co/DOBDC, having much better response and recovery from exposure to moisture, might be more suitable to CO2 capture from flue gas considering the high relative humidity of flue gases in practice, and the added expense and process complication of engineering solutions such as flue gas dehumidification to accommodate moisture sensitive sorbents.39

Increasing performance of MOFs for CO2 capture

Recent work in the field of improving/developing MOFs for CO2 capture is reviewed by Wang et al.40 In this review they note recent trends in schemes and new materials; the reader is referred to this source for a detailed description of the literature review they completed in 2014. By way of summary, the schemes/areas of investigation being pursued which they noted are as follows:
  • Modification of metal sites: coordinately unsaturated metal centers are initial sites of interaction with CO2 and act as physisorptive sites for CO2 molecules enhanced by ion induced dipole interactions, so the metal identity within a series of MOFs will have effect on heat of adsorption and other parameters. Research in this area continues to better understand these effects.
  • Selecting appropriate organic linker: choice of the appropriate organic linker, which can either increase the porosity and specific surface area or provide extra adsorbing sites, has in the last few years resulted in multiple new MOFs targeted for CO2 capture.
  • •Novel structures: synthesis of MOFs with different microstructures and morphologies such as 1D tubular, 2D, 3D, core– shell MOFs, etc.
  • Metal ion doping; for example, an ionic rho-ZMOF’s CO2 capture capacity was shown to be improved by doping a certain amount of Li+.
  • Functionalization: CO2 adsorption capacities of MOF-based adsorbents are usually not fully utilized at low pressures (<0.15 bar) characteristic of flue gas. Modifying MOFs with amines, carboxyl groups, or some other polar groups has been tried to solve this problem; various basic amine groups including EDA, 4-picolylamine, 3-picolylamine, dimethyl-acetamide, acylamide, PEI, etc. have been investigated with varying results.
  • Hybrid MOFs: In order to utilize the MOF pore space effectively and improve the gas adsorption capacity, incorporation of some other materials such as graphite oxide or carbon has also been recently explored.
Challenges of MOF-based sorbents for CO2 capture

MOFs have potential to significantly outperform traditional zeolites and other sorbents. Some examples have already been shown to have high CO2 capacities and adsorption selectivity, and MOF performance may be further improved by tailoring the pore structures and the chemical compositions of MOFs, taking advantage of the unique opportunities afforded by synthesis and post-synthesis techniques discussed above.

However, MOF synthesis cost and material stability in the presence of water vapor remain as significant challenges to be over-come.41 The latter issue has already been mentioned in the context of discussion of previous DOE work in this area.

38 Kizzie AC, Wong-Foy AG and Matzger AJ, Effect of humidity on the performance of microporous coordination polymers as adsorbents for CO2 capture. Langmuir 27:6368–6373 (2011).
39 Yangyang Liu, Zhiyong U. Wang and Hong-Cai Zhou, ‘Recent advances in carbon dioxide capture with metal-organic frameworks,” Greenhouse Gas Sci Technol. 2:239–259 (2012).
40 Junya Wang, Liang Huang, Ruoyan Yang, Zhang Zhang, Jingwen Wu, Yanshan Gao, Qiang Wang, Dermot O’Hare and Ziyi Zhong, “Recent advances in solid sorbents for CO2 capture and new development trends,” Energy Environ. Sci., 2014, 7, 3478.
41 Jian Liu, Praveen K. Thallapally, B. Peter McGrail, Daryl R. Brown and Jun Liu, “Progress in adsorption-based CO2 capture by metal–organic frameworks,” Chem. Soc. Rev., 2012, 41, 2308–2322.

Hydrothermal stability

CO2 capture from flue gas is distinct from other applications, because it involves capture of CO2 mainly diluted by nitrogen, i.e., at partial pressures much less than atmospheric pressure (0.15 atm partial CO 2 pressure is common) and in humid conditions consider-ing the saturated condition of flue gas following sulfur removal processes which results in cooled flue gas with 5–7% water vapor on a volume or molar basis.

The researchers at UOP funded by NETL for the MOF-based CO2 capture study investigated hydrothermal stability (combination of thermal stability and resistance to irreversible reaction with water via hydrolysis) of MOFs in detail. This is a critical property for sorbents, because damage caused to a sorbent by moisture at process conditions or during regeneration would ultimately limit or prevent its commercial use.

The study focused on development of a model to predict energies of ligand displacement and hydrolysis and activation energies for ligand displacement for cluster models for different MOFs, tested by comparison of the modeling predictions with experimental measurements of sorbent hydrothermal stability for representative sorbent samples. The model utilized a simplified quantum mechanical approach in which interactions between metal oxide clusters and water could be manageably calculated. The calculations were made tractable by assuming that the often complex linking ligand is approximated by a simplified functional equivalent and that the flexible framework of MOFs may not impose significant restraints on metal oxide clusters during reactions. This approach was verified to be valid for estimating hydrothermal stability of MOFs, given excellent observed agreement between model predictions and experimental results. The key conclusion is that the strength of the bond between the metal oxide cluster and the bridging linker is most important in determining the hydrothermal stability of the MOF in question.

Using this method, a steam stability map was developed for a number of known MOF types, as illustrated in Figure 2-10. The method should provide a useful tool to help select and invent hydrothermally stable MOFs for desired applications such as CO2 capture from flue gas.

Figure 2-10. Steam Stability Map for MOFs
The position of the structure for a given MOF represents its maximum structural stability by X-ray diffraction measurement, while the energy of activation for ligand displacement by a water molecule determined by molecular modeling is represented by the magenta number (in kcal/mol).42

42 Low, J.J., Benin, A.I., Jakubczak, P., Abrahamian, J.F., Willis, R.R. J. Am. Chem. Soc. 2009, 131, 15834–15842.

Costs of MOFs

Costs for synthesis of MOFs made in small quantities in laboratory settings are relatively high; ultimately, unit costs for manufacture of MOFs for large-scale CO2 capture must be relatively low for them to be a cost-effective alternative to conventional CO 2 capture technologies. As far as manufacturing techniques are concerned, cost of reactors and utilities to manufacture MOFs can be assumed to be comparable to those for synthetic zeolites, and manufacturing methods are adaptable for conventionally available precipitation and crystallization manufacturing methods. However, cost of reagents which usually include metal source, organic linkers, sol-vents for reactions, and solvents for exchange processes may be high, especially the organic linkers which are specialized chemicals not available in large-scale commercial quantities. Costs would be greatly reduced if these were derived on large-scale from cheap sources such as petroleum. Assuming economies of scale in manufacturing can be realized (tonne quantities), starting costs of MOF manufacture, consisting of raw material costs, may be tractable; Liu et al. estimate these ranging from a little as $1.19/kg to under $10/kg for several MOFs, with only a couple of examples exceeding $20/kg, compared to about $1/kg for common silica gel.43 ALKALI METAL CARBONATE-BASED ADSORBENTS

Because of their high CO2 sorption capacity and low cost, alkaline metal carbonate solids such as K2CO3 and Na2CO3 have received attention for CO2 capture applications. Their CO2 capture takes place within the temperature range of 50–100 °C, while regeneration occurs in the range of 120–200 °C, which afford a good potential match to conditions for post-combustion capture of CO2 from flue gas.44

Important reactions involved in the capture of CO2 using Na2CO3 are as follows:45

Both reactions are reversible and highly exothermic, so energy management is an important process issue. Conversion decreases with an increase in the reaction temperature and pressure, and reaction rate increases with an increase in temperature and H2O concentration, and with a decrease in pressure, though maximum carbonation temperature is limited by the reaction thermodynamics. The resulting problem is that the overall carbonation reaction rate for Na2CO3/K2CO3 is rather slow, and reactivity decreases with increases in sorption/regeneration operations due to sorbent degradation and attrition. Also, the carbonates are highly and irreversibly reactive with flue gas contaminants HCl and SO2, so stringent flue gas cleaning is required.

DOE Research

TDA had past projects in sodium carbonate adsorbents for DOE NETL; ultimately funding for that work was discontinued mainly because the regeneration energy demand for their sorbent was ≈1,350 BTU/lb, similar to that for amine-based systems and therefore regarded as non-competitive.


Layered double hydroxides (LDHs), also known as anionic clays or hydrotalcites, are layered basic solids. The structure consists of two types of metallic cations accommodated with the aid of a close-packed configuration of OH− groups in a positively-charged brucite-like layer, with anions and water located in the interlayer space for charge compensation.46 The metal cations occupy the centers of the octahedral structures, whose vertices contain hydroxide ions, and the octahedrons are connected by edge sharing to form an infinite sheet, as depicted in Figure 2-11.47

Figure 2-11. Layered Double Hydroxide Structure

43 Jian Liu, Praveen K. Thallapally, B. Peter McGrail, Daryl R. Brown and Jun Liu, “Progress in adsorption-based CO2 capture by metal–organic frameworks,” Chem. Soc. Rev., 2012, 41, 2308–2322.
44 Qiang Wang, Jizhong Luo, Ziyi Zhong and Armando Borgna, “CO2 capture by solid adsorbents and their applications: current status and new trends,” Energy Environ. Sci., 2011, 4, 42.
45 Y. Liang, D. P. Harrison, R. P. Gupta, D. A. Green and W. J. McMichael, Energy Fuels, 2004, 18, 569.
46 W. Yang, Y. Kim, P. K. T. Liu, M. Sahimi and T. T. Tsotsis, Chem. Eng. Sci., 2002, 57, 2945.

In general, LDH derived mixed oxides possess high surface area, abundant basic sites, and higher temperature stability favorable for absorbing acidic CO2 at 200–400 °C. Potentially, these characteristics and working temperature range should allow LDH-based CO2 sorbents to be used in the sorption-enhanced water gas shift reaction and biomass reforming processes.48 In terms of carbon capture, LDHs can be operated and regenerated at higher temperatures and pressures than the low temperature sorbents, potentially allowing more thermally efficient production of higher pressure CO2. MGO SORBENTS

Like LDH sorbents just discussed, MgO sorbents have potential as CO2 sorbents in conjunction with the water gas shift reaction. MgO has catalytic activity for the water gas shift reaction, while simultaneously taking up CO2:49

Equation 2-12

The regeneration reaction is simply the reverse of the above, with ∆Hrxn ≈+100kJ/mol, effected by temperature swing and/or pres-sure swing. MgO sorbent is capable of achieving over 95% CO2 capture and 40% conversion in the water gas shift (WGS) reaction. The relevant operating temperature range is 350–450 °C, which is approximately the required temperature of the first stage of the typical water gas shift reactor; it may be possible to replace the conventional two-stage water gas shift with one stage using the MgO sorbent, by benefit of the increased driving force for CO conversion provided by simultaneous removal of product CO2.50



Calcium-based materials are good adsorbent candidates for capturing CO2 due to their high reactivity with CO2, high capacity and low material cost. The reversible reaction between CaO and CO2, occurs at 600–700 °C for carbonation, and the regeneration temperature is 850–950 °C at which the CO2 release reaction occurs rapidly and to completion, producing a nearly pure stream of CO2. Capability of operation at elevated temperature offers great potential for reducing CO2 from various clean energy systems, particularly high-temperature pre-combustion carbon capture from gasification processes.

Equation 2-13

47 Wang et al. 2011.
48 Wang et al. 2014.
49 Emadoddin Abbasi, Armin Hassanzadeh, Shahin Zarghami, Hamid Arastoopour, Javad Abbasian, “Regenerable MgO-based sorbent for high temperature CO2 removal from syngas: 3. CO2 capture and sorbent enhanced water gas shift reaction,” Fuel, Volume 137, 1 December 2014, pp. 260–268

The carbonation reaction is highly exothermic and it is possible to efficiently recover the large amount of energy released during the CO2 capture.51 Abundant and inexpensive limestone (CaCO3) is the raw material for the sorbent. Taken together, these factors tend to support favorable process economics.

Challenges in utilization of CaO-based sorbents arise principally from loss of reversibility for the carbonation reaction due to the sintering of the adsorbent particles in repeated use-regeneration cycles. To a lesser extent, attrition and irreversible reaction of calcium with sulfur species are also involved in CaO-based sorbent degradation. In order to combat this problem, a variety of ap-proaches have been taken:
  • Utilizing different sorbent synthesis methods or pretreating limestone with acids, yielding pure CaO particles with higher specific surface areas and smaller particle size, or special microstructures (e.g., hollow CaO, mesoporous CaO, etc.) that have more favorable performance.
  • Preparation of CaO-based mixed oxide type CO2 sorbents
  • Incorporating CaO particles into inert materials that act as structural supports or matrices, in order to significantly improve sorbent durability.
DOE Research

DOE funded URS’s “Evaluation of Dry Sorbent Technology for Pre-Combustion CO2 Capture” project, which investigated engineered sorbents for a process combining CO2 capture with the water gas-shift (WGS) reaction. The sorbents fall into the CaO-based mixed oxide type CO2 sorbents mentioned above, and the synthesis methods used yielded more surface area and hollow structure to improve sorbent stability and performance. Ultrasonic spray pyrolysis (USP) and flame spray pyrolysis (FSP) were used to prepare the sorbents, which included zirconia (ZrO2)-doped CaCO3, yttrium oxide (Y2O3)-doped CaCO3, and MgO/CaO sorbents.53 The engineered sorbents were found to perform much better than natural limestone in evaluation testing. ALKALI CERAMIC-BASED SORBENTS

Alkali zirconate-based (e.g., Li2ZrO3) and alkali silicate-based sorbents (e.g., Li4SiO4) are also high-temperature sorbents, which are of interest because of their excellent CO2 sorption capacities as well as small volume change during the CO2 sorption/desorption cycles. In recent years the latter alkali-silicate sorbents have been most studied, given their advantages of using cheaper raw materials (as opposed to expensive zirconia) and their considerably lower regeneration temperatures (<750 °C) compared to CaO-based sorbents. The relatively low regeneration temperatures are a beneficial result of the easily reversible reaction between Li4SiO4 and CO2:54

Li4SiO4 + CO2 ↔ Li2SiO3 + Li2CO3

Equation 2-14


Significant reductions in the quantity of electricity lost per unit quantity of CO2 have been achieved. Currently funded sorbent technologies have improved performance in terms of better sorbents (higher CO2 capacity, selectivity), decreasing thermal regeneration energy or regeneration at milder temperatures, better process configurations optimizing sorbents with more efficient application of pressure swing adsorption and temperature swing adsorption cycles, and utilization of increasingly cost-effective sorbents.

51 Wang et al 2011.
52 Wang et al 2014.
53“Evaluation of a Dry Sorbent Technology for Pre-Combustion CO2 Capture” [PDF-1.64MB] (July 2012) Presented by Bill Steen, URS Group, 2012 NETL CO2 Capture Technology Meeting, Pittsburgh, PA.
54 Wang et al 2014.


When considering sorbent-based CO2 capture technologies, increasing reduction of the energy penalty incurred per ton of CO2 captured is usually a useful parameter to directly gauge the advance of technological capability. This is reflected in Figure 2-12, which depicts energy penalty on this basis for a number of sorbent-based CO2 capture projects in the NETL portfolio in recent years. With the exception of a stray outlier, the trend from higher to lower energy penalty is generally apparent from earlier to latter projects, showing significant progress in NETL-supported R&D sorbent-based capture technology. Projects recently underway, which have not progressed yet to calculating process performance and techno-economic parameters, cannot yet be represented in this figure.

Figure 2-12. Progress in Reducing Parasitic Load of Sorbent-Based CO2 Capture by DOE/NETL Funded Projects

Certain NETL sorbent-based projects of past years should not be represented in the above figure because in some cases energy penalty per ton CO2 is not a meaningful basis of comparison. For example, DE-FE0000465, “Evaluation of Dry Sorbent Technology for Pre-Combustion CO2 Capture” considered a dry calcium-based sorbent integrated into a water-gas shift reactor, which necessitated a radical revision of the overall plant process compared to the base case IGCC process configuration, most notably in terms of an added sorbent regenerating boiler that exceeds the gas turbine in equivalent electrical output. Techno-economic analyses showed that there was potential to better the baseline cost of electricity in such a configuration, even though the energy penalty for CO2 capture is significantly higher than that of the baseline in these cases. In such instances, the recourse is to compare on basis of COE only.


Power plant efficiency and energy penalty have been considerably improved. However, significant reduction of capital costs and operating costs are still required to achieve DOE goals. It is anticipated that the following issues will be increasingly emphasized as work goes forward in sorbents technology development:
  • Sorbent performance under realistic conditions: Prior work has already begun to address the problems of sorbent stability in low CO2 partial pressure, humid flue gas streams, but ongoing work must thoroughly lead to viable operation under realistic operating conditions. More stable sorbents may have lower capacities and reaction kinetics in both CO2 uptake and regeneration, spelling higher capital costs in process equipment. Adsorption columns, compression equipment to establish pressure differentials or vacuums in PSA or VSA process schemes, heat exchangers for TSA schemes, are some of the capital-intensive pieces of equipment in a CO2 capture plant; these capital costs need to be optimized in relation to sorbent costs and performance in overall process cycles.
  • Costs of advanced sorbents: Advanced sorbents are novel and materials costs tend to be significantly higher than those associated with conventional CO2 capture technologies (amine solvent-based), partly because they are first-of-a-kind technologies. Advances in reducing the costs of feedstocks are needed, along with lower production process costs through improved synthesis, economies of scale, etc.
  • Operability concerns: These include: sorbent degradation, solids handling issues particular to sorbent based-processes which are not experienced in conventional solvent-based technologies, and understanding and managing emissions associated with novel sorbent-based processes.


Membrane separation of gases such as CO2 takes advantage of differences in the relative transfer rates or permeation of various gases through a membrane barrier to effect a greater or lesser degree of gas separation, which is influenced by both the relative diffusivity and surface adsorption of the various gases present. The exact mechanisms for transport and adsorption vary for different membrane materials, such as polymers, metals, and ceramics. To illustrate, the relative permeabilities1 of various gas species in the polymer polydimethylsiloxane are shown in Figure 3-1.

Figure 3-1. Relative Permeabilities of Gases in Polydimethylsiloxane2

Qualitatively, this distribution is typical of most polymeric membrane materials, nitrogen having the lowest permeability and carbon dioxide having higher permeability. The permeability of CO2 is approximately one order of magnitude higher than that of N2 in this particular polymer, meaning that the ratio of permeabilities or selectivity of CO2 to N2 is about 10. The latter selectivity would not be high enough for CO2 separation from N2 in envisioned commercial application, however. For that, selectivities would need to be higher, comparable to those typical of commercially available membranes for CO2 separation from natural gas, which have selectivities of about 30.

Higher selectivity values are desirable because they result in increased purity of the captured CO2, combined with high membrane permeance to minimize required membrane area. This is a significant challenge, because permeance and selectivity tend to be in-versely proportional, i.e., the higher the selectivity of membrane for any pair of gases, the lower the permeance of the more selec-tive gas. For high permeance membranes less membrane area is needed but the desired gas in the permeate is dilute, and for high selectivity membranes the desired gas in the permeate is relatively concentrated but much more membrane area is needed. However, even a membrane with both high selectivity and high permeance may be too expensive to fabricate.


The earliest experimental studies of gas separation were conducted in the 1850s by Thomas Graham, who measured permeation rates for a large number of gases through a wide assortment of materials. Based on the results of his experiments, Graham devel-oped the solution-diffusion model for permeation and Graham’s law of diffusion. The solution-diffusion model is still used today to describe permeation through dense materials such as polymers. Graham’s work was first exploited in the early 1940s as part of the Manhattan project as a means of separating isotopes of uranium. Also in the 1940s and 1950s, gas permeation theory was further developed by Barrer, van Amerongen, Stern, Meares, and others leading to improvements to the solution-diffusion model. Com-mercial application, however, would not emerge for another twenty years.3

The development of high flux membranes and large surface area membrane modules for reverse osmosis applications in the late 1960s and early 1970s catalyzed the development of membrane-based gas separation technology. Monsanto was first to commercialize polymer-based gas separation membranes with the introduction of the PRISM™ membrane in 1980. The main application targeted was hydrogen separation from refinery waste gases, but the potential for other gas separations involving helium, CO2 sepa-ration from fuel gases, and air separation were also recognized from the start. The success of PRISM resulted in the introduction of other gas separation membrane technologies by companies such as Cynara, Separex, Grace, Dow, Ube, and DuPont / Air Liquide.

1 See section on nomenclature for definition.
2 Metz, D.S., 2003. “Water Vapor and Gas Transport through Polymeric Membranes, “Thesis. Enschede, The Netherlands: University of Twente. http://doc.utwente. nl/40766/.
3 Baker, R.W., Membrane Technology and Applications, Second Edition, Wiley, 2004.

Following the successful application of H2 separation membranes for industrial processes, membranes were introduced for the removal of carbon dioxide from natural gas and for nitrogen separation from air. Since the 1980s, continuous improvements have been made in flux and selectivity through the introduction of advanced polymer materials and improved membrane fabrication technologies, and the cost of gas separation membrane systems have steadily declined. In addition to PRISM membranes currently produced by Air Products, other examples of commercial membranes are MEDAL™ membranes produced by Air Liquide and PolySep™ membranes produced by UOP.

Notwithstanding these established instances of membrane-based gas separations, commercial CO2 separation/capture using membranes competes against the alternatives (dominated by solvent-based technologies) and is fairly limited to applications such as smaller-scale natural gas purification. To illustrate, baselines for large-scale pre-combustion CO2 capture from syngas and post-combustion CO2 capture from flue gas typically specify conventional solvent-based technologies (amine, Selexol, Rectisol, etc.) and not membrane-based capture. For example, out of the Global CCS Institute’s database of 55 large-scale CCS projects in various stages of development to operation, 13 projects are classified as operating totaling 25.6 million tonnes per year CO2 captured; most of these plants are for natural gas processing, synthetic natural gas, or fertilizer production. Eleven of these plant use solvent-based processes (amine, Selexol, Rectisol), one uses sorbents, and only one (the Petrobras Lula Oil Field CCS Project, 0.7 million tonnes per year) uses membrane separation.4 In fact, there are no commercial membranes that can effectively separate H2 from CO2 (be-cause of low selectivity).


Notwithstanding the limited extent of membrane-based separation technology in the commercial marketplace, membrane-based CO2 capture technology potentially offers several advantages over a conventional aqueous solvent-based process such as monoetha-nolamine (MEA) in post-combustion CO2 capture:
  • Simple operation; no chemical reactions, no moving parts, no heating/temperature swing required to recover CO2, no use of hazardous chemicals
  • Tolerance to high levels of wet acid gases (for certain membrane types); inert to oxygen
  • Compact and modular with a small footprint (at least at lower capacities); easily scalable
  • Potential for inherent energy efficiency (≈20% plant energy at 90% capture)
  • No additional water used (and in most configurations recovers water from flue gas)
In most configurations of membrane separation no steam use is required,5 unlike solvent and sorbent based processes which typically demand steam for solvent or sorbent regeneration. Therefore, no modifications to existing boiler and steam turbine are required.


The limited extent of existing capacity for commercial large-scale CO2 capture by membranes suggests that competing technologies are usually more cost favorable. This is the consequence of several significant challenges facing currently membrane based CO2 capture technology:
  • Permeability and permeance (see next section for further discussion) of available membranes are lower than desirable. This translates into large membrane areas, requiring both large unwieldy footprint for membrane modules, and result-ing in high capital costs (the membrane itself and module housings). High CO2 permeance (or H2 permeance in certain configurations of pre-combustion capture) is essential to reducing required membrane areas.
  • Although utility steam is not usually needed for membrane separations, significant compression or vacuum capacity to generate the driving force for separation is often demanded. Operation of compressors or vacuum pumps requires expensive auxiliary power, decreasing the net electrical generation of the power plant. Reducing cost through more efficient process arrangements and approaches is essential.
  • Potential harmful contaminants (fly ash, SO2, NOx, water, and trace metals in flue gases; H2S, ammonia, fly ash and trace metals in syngas) may reduce effectiveness and lifetime of membranes/modules in post-combustion and pre-combustion membrane-based capture configurations, respectively.
  • The lack of experience operating membrane-based CO2 capture systems, particularly at the scales associated with large coal-fueled plants, contributes to technological and financial risk that impedes deployment of these systems.
4 Global CCS Institute database.
5 In certain process configurations of pre-combustion membrane-based CO2 capture, steam sweep might be utilized.


Understanding of membrane separations demands definition of some of the more important technological parameters involved. In membrane-based gas separations, the feed gas is introduced into a membrane unit or module, and comes in contact with the membrane surface. The penetrant or permeant is the molecular species in contact with the membrane surface that passes through the membrane. The permeate is the stream containing penetrants that leaves the membrane module. The non-permeate or retentate is the stream that has been depleted of one or more penetrants that leaves the membrane module without passing through the membrane. Other important definitions include:7
  • Permeability: Transport flux per unit trans-membrane driving force per unit membrane thickness. SI units: [kmol-m/m2-s-kPa]
  • Permeance: Transport flux per unit trans-membrane driving force; permeance equals permeability divided by membrane thickness. SI units: [kmol/m2-s-kPa]
  • Pressure Ratio: Ratio of the feed pressure to the permeate pressure
  • Selectivity: Ratio of the permeability or permeance of one component to another for a given membrane material at stated conditions
  • Separation Factor: Ratio of the compositions of two components in the permeate relative to the composition ratio of these components in the retentate
  • Stage Cut: the fractional amount of the total feed entering a membrane module that passes through the membrane as permeate
It should be noted that many investigators employ the above terms haphazardly; for example, selectivity as defined above is some-times referred to as separation factor (see discussion below). Care must be taken in interpreting reported data for permeability, permeance and selectivity.

The following units of measure are frequently employed for permeability:

and for permeance:

where standard pressure and temperature (STP) are 0 °C and 1 atmosphere, and GPU stands for Gas Permeation Unit. Barrer and GPU are compound units based on the units of measure selected for quantity transported, membrane thickness, membrane surface area and partial pressure (i.e., component fugacity), respectively:

6 Content of this section primarily from “Integration of H2 Separation Membranes with CO2 Capture and Compression,” Prepared for U.S. Department of Energy, National Energy Technology Laboratory, under DOE TAMS Contract No. DE-AC26-05NT41816, John Marano, Technology and Management Services, Inc.
7 IUPAC Working Party on Membrane Nomenclature. MEMBRANE SEPARATION FACTOR

For most perm-selective membrane materials, the flux Ji of component i through the membrane will be proportional to the difference in partial pressures on either side of the membrane barrier:
Equation 3-1

where Pi is the permeability of gas component i, l is the thickness of the membrane barrier, and Δpi is the partial pressure driving force across membrane. Equation 3-1 is the basis for the definition of permeability given above. Also, the quantity P i / l can be viewed as the pressure normalized flux; and is the basis for the definition of membrane permeance:
Equation 3-2

Permeability is a property of the membrane material; whereas, permeance also depends on the thickness of the membrane. Obviously, it is desirable to minimize the membrane thickness to achieve the highest permeance and flux. The minimum thickness is limited by material properties and membrane fabrication technique, and also depends on mechanical strength requirements for any gas separation application.

Separation Factor is defined as:
Equation 3-3

where Xi and Xj are mole fractions of components i and j on either side of the membrane.

It can be shown for a differential membrane element that:8
Equation 3-4

Clearly, the separation factor is a function of material properties, membrane thickness and partial pressure driving force. It is useful because it can be directly calculated from experimental data; but is not a good indicator of membrane’s ability to separate gas pairs because it is a function of operating conditions (i.e., Δp/p).

8 Derivation of equation A4 involves the substitution of A1 for the terms in the numerator, and the ideal gas law for the terms in the denominator. A4 is only applicable for a differential membrane. The true relationship between selectivity and separation factor is more complex and depends on actual geometry of the membrane contactor. For more details see: Koros, W.J.; Fleming, G.K., “Membrane-based gas separation,” Journal of Membrane Science, 83, pp.1–80, 1993.

Equation 3-4 can be used as the starting point for considering a number of different definitions for the term selectivity. The ratio (Pi/Pj) in Equation 3-4 can be defined as the true selectivity of components i and j:
Equation 3-5

It is solely a property of the membrane material, and includes any effects of pressure, temperature and interactions between the molecular species present.

An ideal selectivity may be defined as:
Equation 3-6

where the superscript ’o’ indicates the permeabilities are for transport of pure gas components through membrane. Unlike the true selectivity, the ideal selectivity is not a function of gas composition. Ideal selectivity is also referred to as permselectivity.

It is the true selectivity that is required for the accurate design of commercial gas separation membrane systems. P i does not necessarily equal P io nor does P j equal P jo; however, it is much easier to design experiments to measure pure gas permeabilities.

Thus, the experimental determination of permselectivity can be a reasonable, cost-effective starting point for screening membrane materials at bench scale, but must be supplemented later on in the development process with testing under realistic operating conditions using the mixed gases of interest.

Finally, if measurements of the pure gas permeabilities, P io and P jo, are made at the same conditions of temperature, feed and permeate pressure, then based on Equation 3-4, the ideal selectivity Sijo is equivalent to an ideal separation factor, SFijo. THEORY OF MEMBRANE-BASED GAS SEPARATION

Conceptually, the behavior of gases permeating a membrane can be considered from the viewpoint of a thin barrier containing a single large pore, which is in contact with a gas containing molecules of two components A and B.

For pore diameters 0.1 μm or larger, flow of the gas through the pore will be by viscous Poiseuille flow and no separation will occur.

As the pore diameter is decreased, a point is reached where the radius of the pore is of the same magnitude as the mean free path of the gas molecules (λ/r ≈ 1).

For λ/r less than one, a gas molecule will experience more collisions with the pore walls than with other gas molecules. Since molecule-to-molecule collisions are rare each molecule moves independently through the micropore. Therefore, for gas mixtures containing components with different average velocities, a gas Viscous Flow separation is possible.

This pore flow regime is known as Knudsen diffusion and for an ideal cylindrical pore; the mo-lecular flux of each component is governed by the equation:

Knudsen Diffusion

Viscous Flow

where Ji is the flux of component i in units of kg-mol/m2-s, r is the pore radius in meters, Ԑ is the porosity of the membrane, R is the ideal gas constant in kg-m2/s2-kg-mol-K, T is absolute temperature in Kelvin, Mi is the molecular weight of the gas in kg/kg-mol, l is the pore length in meters, pHi is the partial pressure at the entrance of the pore in Pascals, and pLi is the partial pressure at the exit of the pore in Pascals. The subscripts H and L designate the high and low partial pressure sides of the pore, respectively.
Equation 3-7

Based on Equation 3-7, the gas permeance is defined as:
Equation 3-8

Equation 3-8 illustrates that for Knudsen diffusion, the permeance is inversely proportional to the square root of the molecular weight of the gas and also to the square root of the absolute temperature. The former relationship is the basis for Graham’s law for predicting the ideal selectivity of Knudsen diffusion membranes:
Equation 3-9

where i and j designate different molecular species present in the gas.

When the pore diameter decreases to between 5 and 10 Å, the nanoporous membrane begins to separate gas species via a molecular sieving mechanism.

Ideally, for molecular sieving, the flux is related to the ratio of the area of the pore available for transport to the total area of the pore. This is given by:

Molecular Sieving

Equation 3-10

where dp is the diameter of the pore and dki is the kinetic diameter of the gas molecule. The ideal selectivity is then given by:
Equation 3-11

In addition to the phenomena described above, Knudsen diffusion and molecular sieving, surface adsorption and diffusion can con-tribute to molecular transport through membrane pores.

Table 3-1 lists molecular parameters for gases typically used in evaluating membrane performance. Also presented are ideal selectivities calculated from these parameters for both molecular sieving and Knudsen diffusion.

Table 3-1. Molecular Parameters of Membrane Diffusion for Various Gases9

As the pore diameter is decreased, selectivities can become quite large, approaching infinity when the molecule becomes larger than the pore opening. However, the rate of transport/permeance will also decrease for the smaller gas species as the pore opening decreases. It can also be seen that H2 and CO2 are nearly the same size, making them difficult to separate from each other based on molecular size alone. The same holds true for CO2 and N2 which are relatively close in size. Most porous materials other than zeolites do not possess precise pore sizes, making their use impractical with small gas molecules. Molecular weight-based separation based on Knudsen diffusion is also an inefficient means of separating H2 and CO2.

Adsorption may become significant when the pore diameter drops below about 100 Å. In particular, if a gas species is condensable, the amount of gas adsorbed can be significantly greater than that occupying the free pore volume, and surface diffusion can be the dominant transport mechanism. When the diameter of the adsorbed species is of the same magnitude as the pore diameter, it is also possible for the adsorbed species to effectively block the pores, significantly decreasing or entirely preventing the transport of other gaseous species.

Surface Adsorption
9 Compiled from: Baker, R.W., Membrane Technology and Applications, Second Edition, Wiley, 2004; Handbook of Chemistry and Physics, 56th Edition 1975–1976, CRC Press, 1975; Skelland, A.H.P., Diffusional  Mass Transfer, Krieger Publishing Company, 1985.

Transport by surface diffusion is complex, since it is also likely to occur in parallel with either Knudsen diffusion or molecular sieving. The combined phenomena cannot be easily modeled using a single equation for predicting the total molecular flux.

At pore diameters less than about 5 Å, the membrane transport mechanism may again change from molecular sieving to one dominated by molecular diffusion. This is the mechanism for transport in most polymer membranes, where the molecular species first adsorbs onto the membrane surface, dissolves in the membrane material, and then diffuses across the membrane (i.e., solution-diffusion).
Surface Adsorption

For molecular diffusion the flux of each component is governed by the equation:
Equation 3-12

where Si is the surface solubility of component i, in units of kg-mol/m3-Pa, and Di is the diffusivity of i in bulk membrane material, in units of m2/s. Solubility is primarily a function of the chemical composition of the membrane and diffusion of the structure of the membrane. Gases can have high permeation rates due to high solubility, high diffusivity, or both.

Based on Equation 3-12, the gas permeance is defined as:
Equation 3-13

The ideal selectivity for molecular diffusion in dense membranes is then given by:
Equation 3-14

The ratio of Di/Dj can be viewed as mobility selectivity, reflecting the different sizes of the molecules, and the ratio of Si/Sj can be viewed as solubility selectivity, reflecting the relative condensabilities of the gases. In polymers, the diffusion coefficient always de-creases with increasing molecular size, because big molecules interact with more segments of the polymer chain. Thus, the mobility selectivity always favors the passage of the smaller molecule. The solubility selectivity favors larger, more condensable molecules. For molecules similar in size, molecular diffusion may not provide effective gas separation; e.g, for CO2 and N2 which are relatively close in size, diffusivity selectivity in polyimide is an inefficient 2–4 (the ideal separation factor would be 15–30).10

In addition to molecular diffusion, transport may also occur in dense materials via either an atomic or ionic diffusion mechanism.

Atomic diffusion of hydrogen occurs in palladium and related transition metals and alloys. Molecular hydrogen is adsorbed on the metal surface and disassociates into atomic hydrogen, which diffuses through the metal atom matrix. The hydrogen atoms then recombine to form molecular hydrogen and desorb from the surface. When diffusion is the controlling step, the flux is given by Sievert’s law:
Equation 3-15

10 Materials Science of Membranes for Gas and Vapor Separation, edited by Benny Freeman, Yuri Yampolskii, Ingo Pinnau, John Wiley and Sons, England, (2006).

Note that the partial pressure driving force is no longer linear. The partial pressure term raised to the 1/2-power results from the disassociation of diatomic hydrogen. The equilibrium constant for disassociation of H2 on the membrane surface K i has units of kg-mol/m3-Pa1/2.

Based on Equation 3-15, a gas permeance can be defined as:
Equation 3-16

However, permeance as defined above for atomic diffusion is not directly comparable to that for all the other transport mechanisms previously discussed, which take the form of the general flux equation given in Equation 3-1.

The selectivity of metallic membranes for hydrogen separations is essentially infinite due to the disassociation on the membrane surface of diatomic hydrogen, with a molecular diameter of ≈2.9 Å, into atomic hydrogen, with an atomic diameter of less than 1 Å. While size loses much of its meaning on these scales as the wave nature of matter begins to noticeably manifest itself, atomic hydrogen is so small that it can easily diffuse through the dense metal phase.

Ionic diffusion can occur in high-temperature dense ceramics, such as doped rare-earth oxides. In this case, hydrogen adsorbs on the ceramic surface, disassociates, and diffuses as hydrogen ions (protons) through the membrane. The hydrogen ions then recombine to form molecular hydrogen and desorb from the surface. The flow of protons across the membrane must be balanced by an equivalent flow of negatively charged electrons in the same direction to maintain neutrality. The hydrogen flux for this transport mechanism at elevated temperature is given by the equation:
Equation 3-17

where R is the ideal gas constant in units of J/kg-mol-K, T is the absolute temperature in Kelvin, F is the Faraday constant in Coulomb/kg-mole, and σamb is the ambipolar conductivity of the material in units of 1/Ohm-m, defined as:
Equation 3-18

where σamb and σamb are the protonic and electronic conductivities.

Based on Equation 3-17, a gas permeance can be defined as:
Equation 3-19

Again, permeance as defined above for ionic diffusion is not directly comparable to that for all the other transport mechanisms previously discussed.

Based on Equation 3-18, the flux will be high when σamb is large. This requires both the protonic and electronic conductivities to be large. Conductivity can be raised by operating the membrane at high temperatures and by doping the membrane material. Ceramic materials normally do not possess high electronic conductivity. Therefore, ceramic/metallic composites, referred to as cermets, are being researched to enhance this conductivity.

In general, permeability, and thus permeance, is a function of temperature; however, this dependence will vary based on the exact mechanism of membrane transport. Knudsen diffusion is proportional to the square root of temperature. Molecular sieving in principle does not possess a temperature dependence; but, in real systems an effect is observed due to other interacting phenomena. All mechanisms involving surface adsorption and/or diffusion will show an exponential-type temperature dependence, which can be modeled using an Arrhenius Law expression:
Equation 3-20

where Ki and Ei are constants for any given component, and Ei is the activation energy. The quantity R is the ideal gas constant expressed in energy units and the temperature T is in absolute units. In general, gas adsorption decreases with increasing temperatures; whereas, diffusion increases. These interactions can result in quite complex temperature behavior in systems involving both adsorption and diffusion. Based upon Equation 3-19, it might appear that permeance is only proportional to temperature. However, ionic conductivity also exhibits an exponential temperature dependence, increasing with increased temperatures.

For membrane systems involving molecular diffusion, most notably polymers, an inverse relationship exists between selectivity and permeability; as the selectivity is increased, the permeability decreases, and vice versa. It has also been shown experimentally that a log-log plot of the permselectivity of a binary gas pair versus the permeability of the “faster” gas molecule, results in a clearly defined upper limit on selectivity for any given value of permeability. Such a plot is referred to a Robeson plot after the membrane researcher to first describe this phenomenon in polymer-based membranes. This relationship is depicted in Figure 3-2 for H2 and CO2 for a wide range of polymeric membrane materials.

Figure 3-2. Robeson Plot for H2/CO2 Permselectivity in Polymers11

11 Robeson, L.M., “Correlation of Separation Factor Versus Permeability for Polymeric Membranes,” Journal of Membrane Science, 62, 1991.

No completely theoretical explanation has been identified for the Robeson limit, and many researchers have investigated modified materials with the hope of pushing this limit. However, these efforts have so far only had limited success.

Figure 3-3 shows the Robeson limits for H2/CO2, H2/N2, and H 2/CH4. This plot clearly demonstrate the difficulty of separating H2 from CO2 using purely a solution-diffusion mechanism, relative to H2/N2 and H2/CH4 membrane-based separations which have been commercialized. As discussed above, both solubility and diffusion are molecular size dependent phenomena, and as can be seen from Table 3-1, H2 and CO2 molecules have very similar sizes.

Figure 3-3. Robeson Limits for H2 Permselective Membranes12 MEMBRANE MODULE DESIGNS

Schematics of the four most common membrane module designs are shown in Figure 3-4. Tubular and plate-and-frame designs are common in filtration and water treatment applications. Multiple tubes are often packed into a single cylindrical vessel. This is referred to as a shell-and-tube design. Currently, all industrial, gas separation membrane applications employ polymer membrane technology based on hollow fibers or flat sheets, and these designs will be discussed in more detail below.

12 Based on parameters from: Robeson, L.M., Polymer, 35, p. 4970, 1994.

Figure 3-4. Common Membrane Module Designs13

In hollow fiber (HF) based modules, the outside layer of the fiber is composed of a dense polymer on the order of 0.05–010 μm thick. The separation occurs across this thin layer. The porous sub-layer provides structural support to the fiber so that it can withstand large pressure gradients. Individual fibers are about three times the diameter of a human hair (≈125 μm). In a typical membrane module, thousands of fibers are bundled together in a cylindrical arrangement around a central core consisting of perforated tube known as the bore. An epoxy resin seals each end of the cylindrical fiber bundle to a tubesheet, thus separating the feed gas from the permeate. The fiber bundle and bore are enclosed in a shell similar to that used with shell-and-tube heat exchange equipment. Feed gas flows into the shell through the bundle and around the tubes. The retentate, that is the depleted feed gas, exits the membrane module through the bore. If a sweep gas is employed, it is routed through a shell head and distributed to one end of the hollow fiber bundle, exiting at the other end of the module enriched with permeate. If no sweep is used, one end of the fiber bundle may be completely sealed. Flow in this arrangement is predominantly cross flow, though co-current and counter-current arrangements are possible.

An alternative design for polymer membrane equipment uses thin sheets of polymer arranged in spiral-wound (SW) modules. The feed gas enters the module and flows between the membrane leaves. Components of the feed gas permeate the membrane and spi-ral inward toward a central collection pipe. The depleted feed flows across the membrane surface and exits at the other end of the module. The leaves of the membrane alternate between porous membrane sheets and non-porous spacer sheets providing passage for both the retentate and permeate. This arrangement results in cross flow. The membrane industry standard spiral-wound module is an 8-inch diameter module containing 15–30 membrane envelopes with a total membrane area of 20–50 m2 per module. The spiral-wound module design is robust, fouling resistant, and very economical.14 Spiral-wound modules have captured more than 90% of the reverse osmosis market, more than 70% of the ultrafiltration market, and perhaps 30% of the gas separation market.15

Membrane Technology and Research (MTR), in the context of CO2 capture projects funded by DOE NETL, has devised a modification of the standard spiral-wound module that allows introduction of sweep gas on the permeate side to establish driving force without additional pressurization. It involves closing the permeate collection pipe in the middle with a plug, forming two separate compartments; also, during module fabrication, additional glue lines can be applied to direct gas flow in the permeate channel. As shown in the Figure 3-5, the modifications allow the permeate channel to be swept with a sweep gas and the module to operate in a partial countercurrent mode. The permeate gas, together with the sweep gas, flows countercurrent to the feed gas flow.16

13 Torzewski, K. (Ed.),“Facts at your Fingertips—Membrane Configuration,” Chem. Eng., Mar.1, 2009
14 “Membrane Process to Sequester CO2 from Power Plant Flue Gas,” Final Technical Report, Tim Merkel, July 2009, DE-FC26-07NT43085.
15 W. Baker, Membrane Technology and Applications, 2nd ed., John Wiley and Sons Ltd., Chichester, England pp. 544 (2004).
16 “Membrane Process to Sequester CO2 from Power Plant Flue Gas,” Final Technical Report, Tim Merkel, July 2009, DE-FC26-07NT43085.

Figure 3-5. Spiral-Wound Membrane Module Flow Patterns

A less commonly used membrane module arrangement involves the use of flat sheets in a plate-and-frame (PF) design. This configuration has been explored in recent DOE-NETL work in MTR’s project Low-Pressure Membrane Contactors for CO2Capture. The project focus was a compact large membrane area (>500 m2), low pressure drop plate-frame sweep module for CO2 capture applica-tion. As in MTR’s modified spiral-wound modules, an essential aspect of this PF module is use of sweep combustion air to establish the separation driving force (CO2 partial pressure difference) instead of relying on expensive compressors or vacuum pumps (which have both high capital cost and have high energy cost to operate). Additionally, PF-type membrane module designs have the advantage of lower pressure drop compared to typical hollow fiber modules or spiral-wound modules, also saving on energy.

HF and SW membrane-based systems are typically smaller than other types of gas separation plants, such as absorption and adsorption systems, and have high surface area-to-volume ratios (packing density). Achievable packing densities are much greater for hollow fibers (HF; ≈6,000 m2/m3 or ≈1,800 ft2/ft3) versus either spiral-wound sheets (SW; ≈650 m2/m3 or ≈200 ft2/ft3) or plate-and-frame sheet arrangements (PF; ≈325 m2/m3 or ≈100 ft2/ft3). All of these are greater than what is achievable in for example an absorption column (≈260 m2/m3 or ≈80 ft2/ft3). Since membrane units are modular, there is no limit to maximum capacity and plants consisting of over a 100 individual modules, with H2 delivery rates of greater than 50 MM scfd have been built. However, modular-ity has the drawback that plant costs essentially scale linearly with plant capacity and the economic advantages of membranes versus other processes can be lost at high throughputs. Almost 400 hydrogen gas-separation membrane systems are operating worldwide.

The cost of H2 recovery systems is proprietary, but may be estimated. Typical costs for hollow fiber and spiral wound sheet membranes are $45–55 and $375 per m2 ($4–5 and $35 per ft2), respectively.17 Housing can be an additional 20% of the surface area cost and final assembly an additional 1%. Installation of the system at the plant site can be assumed to be 20% of the equipment cost due to the modular nature the equipment. Installed costs of plate-and-frame or shell-and-tube modules utilizing metallic plates or tubes to support the active membrane layer would typically be expected to be higher, given higher materials and fabrication costs.18

In addition to permeance and selectivity, the useful life of the membrane and module is a critical parameter in evaluating the economic performance of a membrane-based gas separation process. A shortened lifetime results in higher maintenance and replacement costs. Membrane life for H2 recovery systems with proper operation and maintenance is reported to be high, exceeding 15 years. However, operation in harsh chemical environments can shorten this lifetime to as little as 3–5 years. Reliability can also be extremely high, with on-stream factors exceeding 99%. Modularity coupled with proper valve placement allows individual mod-ules to be taken off-line without shutting down the entire system. This feature also provides large swings in throughput. Individual module turn-down/up ratios as low as 30% and greater than 100% are possible; however, recovery will be compromised. Additional modules may be brought on-stream if additional capacity is required. PROCESS CONFIGURATIONS FOR MEMBRANE SEPARATIONS

Available membranes for gas separation are usually limited in their combination of permeability characteristics and selectivity for the target gas component to be removed or purified. Accordingly, a single-stage membrane separator often cannot attain required degrees of separation and purity of products (e.g., the DOE NETL goals of 90% CO2 capture with 95%+ purity cannot practically be attained in a single-state process). This principle was documented in a series of parametric studies of CO2 separation from N2 at flue gas concentrations in order to quantify energy penalties compared to the conventional amine process; it was concluded that although energy cost could compare favorably, existing membrane materials did not perform well enough to meet CO2 recovery/purity requirements.19 Other studies have reported on both single and two-stage systems for CO2 separation from N2 at flue gas, including capital and operating (energy) cost considerations in their analyses.20 An important observation is that energy penalty considerations alone are of limited usefulness in membrane-based capture, chiefly given the overwhelming cost penalty that a large membrane area can incur. In a recent series of studies, various arrangements of multistage membrane separators for carbon capture from flue gas were exhaustively analyzed for comparative performance, with complete cost calculations prepared.21

17 Ibid.; adjusted for inflation.
18 MTR has developed a concept for lowering cost of PF modules; they claim costs might be reduced to the range of $50/m2 in large area, compact PF modules. See Low-Pressure Membrane Contactors for CO2 Capture.

Conventional Two-Stage Processes

When considering two-stage membrane separation systems, these studies have noted that two main system circuitries are possible that improve separation performance, classified as follows: (1) serial “stripper” circuits, in which retentate from the first stage be-comes the feed for a second membrane separation stage, and (2) serial “enricher” circuits, in which the permeate from the first stage becomes the feed for the second stage. Figure 3-6 provides examples of the first class of stripper process arrangement, while Figure 3-7 provides examples of the second class.

Figure 3-6. Two-Stage Membrane Circuitries: Serial Stripper Concept

19 Roda Bounaceur, Nancy Lape, Denis Roizard, Cecile Vallieres, Eric Favre, “Membrane processes for post-combustion carbon dioxide capture: A parametric study,” Energy, Volume 31, Issue 14, November 2006, pp. 2556–2570.
20 Dongxiao Yang, Zhi Wang, Jixiao Wang, and Shichang Wang, “Potential of Two-Stage Membrane System with Recycle Stream for CO2 Capture from Post-Combustion Gas,” Energy Fuels, 2009, 23 (10), pp. 4755–4762; Minh T. Ho, Guy W. Allinson, Dianne E. Wiley, “Reducing the Cost of CO2 Capture from Flue Gases Using Membrane Technology,” Industrial and Engineering Chemistry Research 2008 47 (5), pp. 1562–1568.
21 Li Zhao, Reinhard Menzer, Ernst Riensche, Ludger Blum, Detlef Stolten, “Concepts and investment cost analyses of multi-stage membrane systems used in post-combustion processes,” Energy Procedia, Volume 1, Issue 1, Greenhouse Gas Control Technologies 9, Proceedings of the 9th International Conference on Greenhouse Gas Control Technologies (GHGT-9), 16–20 November 2008, Washington DC, USA, February 2009, pp. 269–278; Li Zhao, Ernst Riensche, Reinhard Menzer, Ludger Blum, Detlef Stolten, “A parametric study of CO2/N2 gas separation membrane processes for post-combustion capture,” Journal of Membrane Science, Volume 325, Issue 1, 15 November 2008, pp. 284–294; Li Zhao, Ernst Riensche, Ludger Blum, Detlef Stolten, “Multi-stage gas separation membrane processes used in post-combustion capture: Energetic and economic analyses,” Journal of Membrane Science, Volume 359, Issues 1–2, Membranes and CO2 Separation, 1 September 2010, pp. 160–172.

Figure 3-7. Two-Stage Membrane Circuitries: Serial Enricher Concept

In each class, three variations are depicted, which vary only in how the pressure gradient across the membrane stages is applied, whether through upstream compression by a compressor and exhaust to ambient pressure, or by downstream vacuum established by vacuum pump with the upstream at ambient pressure. Usually, the preferred method is to use vacuum pumping on the first stage, in order to avoid the necessity of compressing the whole volume of flue gas at considerable energy penalty, with compression ap-plied to the second stage, which is economical because of the much reduced quantity of gas flow in the second stage. Obviously, a trade-off is present in that using compressors to establish the pressure gradient requires less membrane surface area to accomplish the same degree of gas recovery, and because membranes represent the majority of capital cost of a membrane separation system, higher pressures tend lead to a significant lowering of capital cost.

Calculations show that the serial enricher arrangement is more energy efficient than parallel schemes or other serial schemes. More-over, the retentate from the second stage of the enricher system is often similar in CO 2 concentration to the original feed to the system, which suggests that the second-stage retentate might be recycled to the beginning of the cycle to maximize system performance and efficiency still further. This “serial enricher” with recycle circuit is depicted in Figure 3-8.

Figure 3-8. Two-Stage Membrane Circuitries: Enricher with Recycle

The latter arrangement has the best performance to cost ratio of all of the above.

Sweep Gas Utilization

A persistent drawback of the foregoing process arrangements remains in that they all require energy-intensive compression or vacuum pumping to provide the driving force for all of the membrane separation stages. In the case of post-combustion CO2 capture, MTR proposes a process solution that utilizes combustion air as the sweep gas to generate separation driving force at the sweep module, thereby reducing the overall need for energy-intensive compressors or vacuum pumps. This is depicted in Figure 3-9. Stream 4 is the combustion air sweep; the air enriched in CO2 (i.e., ≈8% CO2, ≈18% O2) is utilized as combustion air at the boiler.

Figure 3-9. MTR Proposed Membrane Process to Capture and Sequester CO2 in Flue Gas from a Coal-Fired Power Plant22

Here, the combustion air sweep generates a CO2 partial pressure gradient in the second membrane stage, eliminating the need for compressors or vacuum pumps to drive this stage. In this way, the sweep module avoids the energy penalty of compression or vacuum treatment and provides an essentially “free” separation. Additionally, by recycling CO2 to the boiler via the combustion air sweep loop, the CO2 concentration in the flue gas exiting the boiler increases from about 13% to approximately 18%. This increases the CO2 partial pressure driving force for transport in the first membrane stage. Consequently, the membrane area and system cost is reduced.23 Optimization of process cycles in this way and improved membrane performance in selectivity and permeability should, in combination, result in increasingly efficient and cost-effective membrane-based gas separation technology that will be able to meet challenging DOE goals for carbon capture.

Ohio State University is assuming the same type of two stage configuration with combustion air sweep in their work on novel inorganic/polymer-supported composite membranes (see Table 3-5, fourth row).

Hybrid Cycles

The possibility of combining solvent or sorbent-based separations with membrane-based processes to further improve energy efficiency and performance has been increasingly evident in R&D in the carbon capture field. The DOE-NETL portfolio reflects this, with a number of concepts involving hybrid process cycles emerging in recent work. For example, MTR is exploring a hybrid cycle which includes the same combustion air sweep-driven membrane stage discussed in the previous section, but with the initial CO2 separation stage consisting of an absorption column utilizing 5m piperazine solvent (see Table 3-5 first row in post-combustion listings). The CO2 enriched combustion air results in higher than normal CO2 concentration in the flue gas, and the amine solvent stage performance can be increased/optimized for it accordingly. MRT is evaluating both hybrid-parallel and hybrid-series process arrangements to explore the opportunities of the concept, as shown in Figure 3-10.

Figure 3-10. MTR Hybrid Post-Combustion CO2 Capture Process Cycles

22 “Membrane Process to Sequester CO2 from Power Plant Flue Gas,” Final Technical Report, Tim Merkel, July 2009, DE-FC26-07NT43085.
23 Ibid.


GTI is investigating a membrane contactor which utilizes polyether ether ketone (PEEK) hollow fiber membranes in a contactor having flue gas on one side and an amine solvent (MDEA) on the other side (see Table 3-5, second row of post-combustion listings). The solvent efficiently takes up the CO2 permeate, lowering the CO2 partial pressure on the permeate side to almost zero, thereby establishing the driving force for CO2 separation without need of feed side compression or permeate side vacuum. This arrangement should be able to achieve post-combustion capture with >90% CO2 removal and high CO2 purity in just one stage, as opposed to most processes which demand at least two stages to attain the required removal and purity. Of course, it does require the regeneration of rich solvent as any solvent-based process (performed in this case in another PEEK contactor operated in reverse mode). This approach combines the advantages of the compactness of a membrane contactor with the high selectivity of a solvent-based absorption process, and is expected to significantly reduce the cost of CO2 capture relative to the baseline amine-based process.

The considerable diversity of the DOE NETL portfolio in membrane separations is illustrated by the Fuel Cell Energy project, which is examining a concept of CO2 separation from flue gas using electrochemically driven separation membranes. As shown in Figure 3-11, the process concept is markedly different from conventional membrane process arrangements, consisting of electro-chemical membranes within fuel cell modules driven by supplemental fuel (presumably natural gas); the fuel cell modules generate additional electricity in addition to performing CO2 capture.

Figure 3-11. Fuel Cell Energy Combined Electric Power and CO2 Separation System Concept24


Currently, gas separation membranes are used industrially for hydrogen separation within ammonia plants (H2/N2 separation) and petrochemical plants (H2/hydrocarbons); and for separating nitrogen from air, removing CO2 and water from natural gas, and recovering organic vapors from air or nitrogen. The most widely used membrane materials for gas separation are polymers. Polymers are attractive because they can be processed into hollow fibers or thin sheets with high surface areas per unit volume. Each membrane unit can contain thousands of fibers. As discussed above, this results in compact, modular membrane units having relatively low costs, even when permeance and selectivity are not extremely high.

24 “Electrochemical Membrane for CO2 Capture and Power Generation,” Hossein Ghezel-Ayagh, NETL-DOE 2014 Transformational Carbon Capture Technology Work-shop, September 23, 2014 Arlington, VA.

The three major providers of membrane technologies for gas separation applications are Air Products and Chemicals, Inc. (APCI), Air Liquide and UOP. The APCI PRISM™ membrane is based on the original Monsanto technology commercialized back in 1980. Air Liquide’s MEDAL™ membrane was developed by a joint venture between DuPont and Air Liquide. The UOP membrane tech-nology is named PolySep™, and also has its roots in technology developed earlier. The relative permeance of these commercial membranes, based on the open literature, is:

Table 3-2. Relative Permeance of Gases in Three Commercial Membranes

Profiles of these three technologies are described in Table 3-3.

Table 3-3. Process Profiles for Commercial Polymer Membrane Technologies

Commercial experience with polymer membranes (hydrogen separation)25

Gases with high permeabilities such as hydrogen enrich the permeate side of the membrane, and gases with lower permeabilities enrich the retentate side of the membrane because of the depletion of components with high permeability. The first fraction of the gas to permeate the membrane consists primarily of the components with the highest permeability. As a larger fraction of the feed gas is allowed to permeate, the relative amount of the components with lower permeability increases in the permeate stream.

Polymer membranes can be used to process gases containing a wide variation in H2 content. Feed pretreatment requirements are typically minimal, and are primarily associated with feeding a dry, clean gas to the unit. Since the water and hydrocarbon content of the feed will be increased as hydrogen is removed, it is necessary that the feed gas be heated to above the dew point of the retentate exiting the module to prevent condensation of a liquid phase within the module. Exposure to liquid water or hydrocarbons, particulates and high concentrations of H2S may permanently damage the polymer material. Polymers in commercial use are limited to temperatures not much greater than the boiling point of water.

For any given membrane, hydrogen recovery is a strong function of the feed H2 content and pressure, and product requirements in regards to H2 purity and delivery pressure. Feed compression or permeate re-compression requirements can add significantly to the cost of a H2 recovery plant. A doubling of the feed pressure can cut H2 delivery costs by as much as a fourth. Higher purity hydrogen is associated with lower recovery, with this effect much more dramatic with membrane systems versus competing PSA technology. For a membrane with a H2 selectivity of 30, a 3% increase in product purity results in a 25% decrease in recovery. Higher H2 recoveries also require more membrane area. For specified feed composition and system pressure levels, the amount of area required increases exponentially at high hydrogen recovery. For a specific membrane system, the recovery versus purity, is primarily dependent on the ratio of the retentate to permeate pressure and is largely independent of absolute pressure levels. However, area requirements are inversely proportional to the feed pressure. Therefore, compressing the feed gas rather than the permeate, even though the permeate flow is smaller, is often preferable when the objective is to achieve a required pressure ratio.

For properly designed and operated membrane systems, pressure drops are typically small. The pressure differential across the membrane is set by the upstream pressure and the desired H2 delivery pressure, unless feed compression or permeate re-compression is included. Feed pressures as high as 2,500 psia (170 bar) have been demonstrated. More significantly, cross membrane differentials as high as 1,650 psia (112 bar) can be achieved. However, very high pressure differentials can cause compression of the membrane, detrimentally affecting performance. This effect is temperature dependent, and the maximum allowable pressure differential will be lower than 1,650 psia as the operating temperature approaches the maximum allowable operating temperature for the membrane.


Conventionally, a water gas shift (WGS) reactor is utilized to maximize the hydrogen content of syngas, following which CO2 may be captured using standard absorption techniques. This conventional approach involves two separate stages of reaction and CO2 separation. Realizing that the conventional approach has inherent drawbacks in efficiency, the DOE NETL R&D portfolio includes several projects developing WGS membrane reactors. The concept is based on increasing conversion of the equilibrium WGS reaction:

CO + H2O ↔ CO2 + H2

Equation 3-21

25 MacLean, D.L.; Stookey, D.J.; Metzger, T.R., “Fundamentals of gas permeation,” Hydrocarbon Processing, pp.47–51, Aug. 1983. Mazur, W.H.; Chan, M.C., “Membranes for Natural Gas Sweetening and CO2 Enrichment,” Chem. Eng. Prog., pp.38–43, Oct. 1984. Hogsett, J.E.; Mazur, W.H., “Estimate membrane system area,” Hydrocarbon Processing, pp.52–54, Aug., 1983.
Whysall, M.; Picioccio, K.W., “Selection and revamp of Hydrogen Purification Processes,” 1999 AIChE Spring Meeting, Houston, TX, March 13–18, 1999. Air Products Product Brochure, Advanced Prism® Membrane Systems for Cost Effective Gas Separations.
Picioccio, K.; Reyes, E., “Breaking the Hydrogen Barrier with PSA Revamps—Increasing Hydrogen Production of PSA Units,” UOP webpage, 2000. Haggin, J., “Membrane Technology Developments Aim at Variety of Applications,” Chem. and Eng. News, pp.25–26, April 4, 1994.

by withdrawing one of the products (often H2) across membranes in the reactor, causing the reaction to progress further to the right by Le Chatelier’s principle. Figure 3-12 illustrates the concept of the membrane reactor. As illustrated, in this configuration syngas undergoing the WGS reaction flows into membrane modules containing hydrogen-selective membranes tubes. Hydrogen permeates into the tubes and is swept away by nitrogen on the bore side. The resulting H 2/N2 mixture would be suitable for firing in a gas turbine for electricity generation. As hydrogen is withdrawn, the syngas increases greatly in CO2 content. In this way a significant measure of carbon capture is effected within the reactor itself. (Some WGS membrane reactors might utilize CO2 selective membranes in which case the separation still occurs, but with a different process flow configuration that would use steam sweep of the CO2 instead of N2 sweep of the hydrogen). Overall, this concept results in process intensification with its inherent reduction of energy intensity and improved separation efficiency.

The challenges of the WGS membrane reactor come from the requirement for suitable hydrogen or CO2 selectivity as the case may be, permeability and durability in the reactor environment, and reasonable membrane cost. The NETL R&D work has investigated a wide range of membranes including ceramic-carbonate dual phase membranes, carbon molecular sieve membranes, Pd alloy membranes, and exfoliated hydrogen-selective zeolite membranes for the WGS reactor application.

Figure 3-12. Water Gas Shift Membrane Reactor Concept26 TYPES OF MEMBRANES

Membranes possibly suitable for separation of CO2 (or hydrogen separation from CO2-H2 mixtures) fall into several categories:
  • Polymeric membranes (e.g., polyethyleneoxide, polydimethylsiloxane, sulphonated polyimides)
  • Ceramics (aminosilicate structures)
  • Supported liquid membranes (facilitated transport)
  • Metallic membranes (e.g., palladium, palladium alloys for hydrogen separation)
  • Other (zeolites, etc.)
Conventional polymers have often been used for gas separations, though commercially available polymer membranes and module manufacture/sealing technologies are limited to about 150 °C operating temperatures. (Higher operating temperatures of >200 °C typical of syngas represent a challenge for this class of membranes). Many of the projects in the NETL portfolio are investigating processes and membrane configurations using polymer membranes of various types. Some are unspecified/proprietary; others are known such as polyether ether ketone (PEEK)-based hollow fiber membranes in both pre-combustion and post-combustion application.

Ceramic materials are interesting, particularly in the context of gas separation where they might conceivably be employed in high temperature regions (such as 350 °C or higher applicable to syngas or hot flue gas). For example, one recent project in the NETL portfolio is utilizing ceramic-molten carbonate dual phase membranes for CO2 capture within a WGS membrane reactor.

Considerable attention to supported liquid membranes for separation of CO2 can be found in the literature. Here, the chemistry of the membrane is geared toward facilitated transport of CO2, i.e., amine constituents of the membrane can induce transport of CO2 at higher rates than seen in conventional polymeric materials. Some researchers are optimistic about the potential for such membranes in gas separation systems, specifically room temperature ionic liquid supported membranes,27 while others believe that in general, stability problems would limit the use of supported liquid membranes in practical gas separation processes.28 Certain projects in the NETL portfolio have included such membranes: in one a carbonic anhydrase enzyme-catalyzed, contained liquid membrane has been investigated, and others involve use of amine selective layers on polymeric supports.

27 Scovazzo, P., “Determination of the upper limits, benchmarks, and critical properties for gas separations using stabilized room temperature ionic liquid membranes (SILMs) for the purpose of guiding future research” Journal of Membrane Science, 343(1–2), 2009, 199–211; Scovazzo, P. et al., “Long-term, continuous mixed-gas dry fed CO2/CH4 and CO2/N2 separation performance and selectivities for room temperature ionic liquid membranes,” Journal of Membrane Science, 327(1–2), 2009, 41–48.
28 Hong, S.U. et al., “Polymer-ionic liquid gels for enhanced gas transport,” Chemical Communications, (46), 2009, 7227–7229.

Metallic palladium membranes have almost infinite selectivity for hydrogen and high permeance and are suitable for operation at high temperature; they have figured in several WGS applications and syngas cleanup in the NETL portfolio.

Zeolites are microporous crystalline aluminosilicates with defined nanopore sizes, occurring naturally as minerals, and also represented by a much larger number of synthetic zeolites that have been developed for use as sorbents and catalysts. Zeolites can also be deposited as a thin layer on porous substrates, enabling their use as a selective layer in membranes. To illustrate, the zeolite sodalite exhibits pore sizes of 0.29 nm, which lies in the range of the kinetic molecule diameter of hydrogen, while CO2 should be too large to pass through the small pores of the sodalite framework. Therefore sodalite membranes are candidates for molecular sieving to separate hydrogen from CO2. Zeolite membranes are cheaper than palladium or platinum membranes, while still having the latter’s useful property of high specific gas selectivity. On the other hand, they are more expensive than polymeric membranes, yet can have much higher thermal and hydrothermal stability than polymeric membranes, essential for higher temperature applications such as hydrogen-CO2 separations in presence of water vapor at warm syngas temperatures.29


The Carbon Capture Program has supported research and development in membrane-based CO2 capture approaches for both pre-combustion (mainly H2-CO2 separations) and post-combustion (N2-CO2 separations). Pressures, temperatures, and gas composition widely vary between pre-combustion and post-combustion applications, resulting in distinct R&D portfolios for those groupings. Membranes of widely differing materials types have been suggested for development in conjunction with highly varied proposed process configurations. Earlier in the DOE/NETL program, membrane materials were unbounded, but because of excessive materials cost, work on conventional Pd-based membranes has been almost completely discontinued.

Recent pre-combustion projects are referenced in Table 3-4; post-combustion projects are listed in Table 3-5.

Table 3-4. DOE/NETL Membrane-Based Pre-Combustion CO2 Capture Projects Portfolio

Table 3-5. DOE/NETL Membrane-Based Post-Combustion CO2 Capture Projects Portfolio

29 Christiane Günther, Hannes Richter, Ingolf Voigt, “Zeolite Membranes for Hydrogen and Water Separation under Harsh Conditions,” Chemical Engineering Transac-tions, 32 (2013), pp. 1963–1968.


The cost of electricity and cost for capturing unit quantities of CO2 has improved significantly under the DOE NETL Carbon Capture program. Currently DOE-funded membrane technologies have improved performance in terms of higher permeances and selectivities), increasingly compact membrane module configurations, optimized process configurations taking best advantage of pressure differentials via pressurization, vacuum, sweep, etc., and improving manufacturing techniques which result in increasingly cost-effective membranes.


When considering solvent-based CO2 capture technologies, increasing reduction of the energy penalty incurred per ton of CO2 captured is a meaningful parameter in gauging the advance of technological capability.31 However, the same parameter is not useful in assessing membrane-based technologies because multiple factors of (1) membrane cost, (2) membrane performance in terms of separation factor and flux, (3) energy penalty to operate the membrane system, and (4) process configuration all vary significantly from one membrane system implementation to the other, often with interdependencies, and taken together determine the ultimate increase in COE for CO2 capture. Indeed, in some envisioned configurations it is possible that membrane cost would dominate the economics of the CO2 capture process while energy penalty is only a minor contributing factor, considering that some configurations greatly reduce energy penalty by using sweep gas. Also, excellently performing membranes may reduce required membrane areas but may be more expensive to manufacture than poorer performing membranes, so it is not inconceivable that worse-performing membranes may be less expensive overall per unit of capture. Therefore, ranking performance simply on membrane parameters such as selectivity and permeability is also not conclusive. Two parameters that do provide some useful basis of comparison are the COE (typically in units of $/MWh) with capture compared to a baseline capture system (standard amine-based capture) and unit cost of captured CO2 (typically in $/ton captured).

Figure 3-13 depicts these parameters for post-combustion CO2 capture projects in the NETL portfolio in recent years. The trends from higher to lower COE and cost per ton of CO2 are readily apparent from earlier to latter projects, showing significant progress in NETL-supported R&D membrane-based capture technology.

Figure 3-13. Progress in Reducing the Cost of Electricity and Cost of CO2 Capture for Membrane-Based CO2 Capture by DOE/NETL Funded Projects

30 Same as basic MTR scheme.
31 Often the process equipment involved in the solvent-based capture process (columns, regenerators, etc.) is the same no matter what the solvent process used, meaning that energy penalty reductions are the only significant route to increasing performance/cost effectiveness.


Key parameters to be evaluated in the future include:
  • Membrane performance under realistic conditions: Prior work has already begun to include evaluation of membrane stability in actual syngas streams (pre-combustion capture applications) and humid flue gas streams (post-combustion capture), but ongoing work must lead to scaled-up, viable operation under realistic operating conditions. More stable membranes may have lower capacities spelling higher capital costs for required membrane area and associated process equipment. Compression equipment to establish pressure differentials or vacuums are capital-intensive pieces of equipment in a CO2 capture plant, needing to be optimized in relation to membrane costs and performance in overall process cycles.
  • Costs of advanced membranes: Some of the advanced or developmental membranes considered in the DOE NETL port-folio are novel and materials costs tend to be significantly higher than those associated with conventional membranes, partly because they are first-of-a-kind technologies. Advances in reducing the costs of feedstocks are needed, along with lower production process costs through improved synthesis, economies of scale, etc. Prices for material for some membranes (e.g., those incorporating palladium alloys) cannot be controlled, but improved synthesis to minimize required amounts of high-value materials should be sought.
  • Operability concerns: These are fairly limited in the case of membrane-based systems but issues such as membrane degradation remain and must be considered.


CO2 captured from coal flue gas or synthesis gas must be compressed to transport it via pipeline for geologic storage, enhanced oil recovery, or CO2 utilization. Compression represents a significant parasitic load in addition to the energy requirement for separating CO2 from flue gas. For example, compressing CO2 captured from a supercritical pulverized coal plant (capturing 90% CO2 and generating 550 MW net power) consumes 45 MW.

The compression of CO2 is challenging because of the high volumetric flow rate of gas (ca. 109,700 actual cubic foot/minute [acfm]) for a 550 MW supercritical pulverized coal-fired power plant,1 high pressure ratio (≈100:1), large variations in the physical proper-ties of CO2 in the region of its critical temperature and pressure conditions. For example, the data in Figure 4-1 indicates that com-pressing CO2 from a supercritical pulverized coal power plant would require two parallel multi-stage centrifugal compressor trains.

Figure 4-1. CO2 Compressor Coverage Chart for a Range of Discharge Pressure and Inlet Flow
Showing the typical discharge pressure and inlet volumetric flow rate of CO2 captured from supercritical pulverized coal-fired power plant. Source: Gallick et al., 2006.2

In this context, lowering the cost of capture (and the cost of electricity) requires the development of advanced compression technologies which can significantly lower the capital and operational costs.



The reversible work for compressing a fluid in a closed system without flow is given by
, where P is the pressure and V denotes the volume.

For open systems, where fluid flows in and out of a control volume, the reversible work for compression also needs to account for the flow work, and is expressed as 
For a compressible fluid, volume varies significantly with pressure. Adiabatic compression, where the fluid does not lose heat to its surroundings, and isothermal compression, where the fluid is in thermal equilibrium with its surroundings, represent two ends of the spectrum in gas compression. The work involved in isothermal compression is lower because the gas temperature between the inlet and the outlet of the compressor does not change, and the work being performed on the system (i.e., change in enthalpy) is not lost as the increase in the sensible heat of the gas. In a reversible adiabatic (isentropic) compression process, the entropy of the fluid being compressed does not change between the inlet and the outlet. For an ideal gas undergoing an adiabatic compression, PVγ = constant, where γ = Cp/Cv is the ratio of specific heats at constant pressure and constant volume respectively.

1 Black. J. Cost and performance baseline for fossil energy plants—Volume 1: Bituminous coal and natural gas to electricity. Revision 2, November 2010. DOE/NETL-2010/1397.
2 Gallick, P., Phillippi, G., Williams, B.F., What’s correct for my application—A centrifugal or reciprocating compressor? In Proceedings of the Thirty-Fifth Turboma-chinery Symposium, pp. 113–122, (2006). Note that the chart was developed for selecting compressors for natural gas service and is not to be used to gauge the exact boundaries of applicability for CO2 service.
3 A discussion of reversible work and actual work is provided in Appendix 7.1
4 See Appendix 7.1 and Equation 13.

Multi-stage centrifugal compressors more generally employ polytropic processes, which are defined by the following relation: PVn = constant, where n is termed the polytropic exponent. The polytropic exponent n is a function of the polytropic efficiency (ηp) and gas properties. A polytropic process is a path for which the ratio of reversible work input to the enthalpy rise is constant.2 A polytropic process is equivalent to a succession of an infinite number of isentropic (constant entropy) compression steps, each followed by an isobaric (constant pressure) heat addition. This reversible heat addition is such that it would generate the same temperature increase as the irreversible losses in the real process.5 The polytropic efficiency is established by the manufacturer and is a function of the capacity at the inlet of the compressor.6 For an ideal gas, ηp is related to the polytropic exponent and the ratio of specific heats (γ) as:
Equation 4-1

If n is known, the work for compressing an ideal gas in a polytropic process can be estimated by integrating VdP from Ps to Pd and is represented as:

Equation 4-2

where, Hp ideal gas, HR ideal gas, MolWt, and Ts refer to the polytropic head required to compress the ideal gas, actual (effective) head to compress ideal gas, gas molecular weight and temperature at the inlet of the compressor respectively. At high pressures, the ideal gas law, PV = RT (for one mole of gas) is no longer valid because molecules interact strongly with one another. To account for this deviation from ideal gas behavior, the ideal gas polytropic work can be modified using the compressibility factor.

Equation 4-3

In Equation 4-3, Zavg (Z = Pv/RT) represents the average of the gas compressibility at the suction and discharge conditions and requires gas discharge temperature. Note that v represents the volume occupied by a mole of gas (molar volume). Hp real gas represents the polytropic head required to compress the gas, and Hp real gas is the effective head required to compress the real gas.

5 Rasumussen, P.C., Kurz, R. Centrifugal compressor applications—Upstream and downstream. In Proceedings of the Thirty-Eighth Turbomachinery Symposium, pp. 169–186, (2009).
6 Mohitpour, M., Golshan, H. and Murray, A. Gas compression and gas coolers. In Pipeline design and construction: A practical approach, pp. 129–226 (ASME, 2007).

The gas discharge temperature for a polytropic process is related to the pressure ratio as:
Equation 4-4

In general, the polytropic exponent can be calculated from:
Equation 4-5

Because work is a path-dependent function, its value is affected by both the start- and end-points and also the compression path, which is determined by the polytropic exponent n. In a typical compression path, variations in the polytropic exponent n could change the calculated work. Schultz developed a correction factor for isentropic compression, assumed to be directly applicable to polytropic compression processes7 (discussed in Appendix 7.3).

The polytropic exponent varies with the pressure and temperature of the gas and compression work calculations need to account for changes in the value of n. A study on the compression of pure CO2 found that the polytropic exponent was practically unchanged for pressures less than the critical pressure (Pc, 1,069 psia) (P < Pc).8 Further, within the dense-phase region, at P > Pc, T > 1.65Tc (critical temperature) (or T > 444 °F), the exponent is relatively constant. However, at conditions where the pressure is above the critical pressure and the temperature exceeds 1.65 times the critical temperature [P > Pc and T < 1.65Tc (i.e., T < 444 °F for CO2)], the polytropic exponent varied significantly, from approximately 1.8 to 4. Because all three regions [i.e., P < Pc; P > P c, T < 1.65Tc; P > Pc, T > 1.65Tc] are relevant for intercooled CO2 compression (intercooler temperature at 70 °F, or 1.15Tc, Ps = 23.5 psia [0.02Pc], Pd = 2,215 psia [2.07Pc]), it is reasonable to expect that work calculations which assume a constant value of n would result in discrepancies, especially at pressures exceeding the critical pressure (P > Pc).


Figure 4-2 represents the isentropic and isothermal compression paths on a pressure-enthalpy diagram. The initial temperature and pressure and final pressure in Figure 4-2 correspond to those for CO2 compression for Case 12 of the DOE/NETL bitminous coal baseline (2010) report.1

Several methods to calculate stage outlet temperatures and head were evaluated. They included using the polytropic equation for an ideal gas, the Schultz polytropic procedure and an equation-of-state (EOS)-based approach to calculate the area under the V-P curve. Details of the calculations are provided in the Appendix. Results shown in Table 4-1 indicate that the use of an EOS-based integration approach (Prode), or Schultz polytropic procedure with a high-fidelity EOS led to more accurate outlet temperatures and compression work values compared to ideal gas equations for polytropic compression without the Schultz correction factor.9

Because the critical temperature of CO2 is 87.76 °F, isothermal compression at 70 °F would result in a crossover to the vapor-liquid region before compression to the dense-phase region. Figure 4-1 also indicates that the work for isentropic compression would be higher than that for isothermal compression because the volume of the gas changes significantly in the former case. However, it is not realistic to compress a gas isothermally because real compression processes always have losses resulting in an increase of tem-perature (or entropy). Further, a comparison of the enthalpy after the first stage of the polytropic process and the isentropic process indicates that for the inlet pressure, temperature and discharge pressure, the work for the isentropic path is lower.

7 Schultz, J. M. The polytropic analysis of centrifugal compressors. J. Eng. Gas Turbines Power 84, 69–82 (1962).
8 Sandberg, M.R., Colby, G.M. Limitations of ASME PTC 10 in accurately evaluating centrifugal compressor thermodynamic performance. In Proceedings of the Forty-Second Turbomachinery Symposium, October 1–3, 2013, Houston, TX.
9 The PRODE calculations used the Peng-Robinson equation of state (PR-EOS) to calculate the discharge temperature. To ensure compatibility between the Schultz polytropic procedure with the high-fidelity (Span and Wagner) EOS, the head estimated by PRODE was recalculated using the enthalpies estimated by the Span and Wagner EOS in CoolProp.

Table 4-1. Inlet and Outlet Pressures for CO2 Compression Using Conventional Centrifugal Compression with Intercooling Between Stages

Figure 4-2. Comparison of Isentropic, Isothermal, and Intercooled Polytropic CO2 Compression (ηp= 86%) Paths
For compressing CO2 at 23.5 psia, 70 °F to 2,215 psia.

Table 4-2. Stage Outlet Temperature, Head, and Overall Compression Work for Six-Stage Compression of CO2 from 23.5–2,215 psia Using Prode, Schultz, and Ideal Gas Approximation Calculations

Note: Subscripts s and d denote suction and discharge respectively. All calculations assume a polytropic efficiency of 86%. The gas power [hp/(lbm/min)] does not include the mechanical efficiency of the compressor. 1 hp/(lbm/min) = 98.64 kJ/kg.


High-pressure ratio compressors are based on shock compression where a rotor operating at high (sub-sonic) speeds generates shock waves compressing the CO2. Incorporating high-pressure ratio compression in conventional compressor designs results in large forces being exerted on the compressor shaft, leading to complicated and costly designs.6 Therefore, the pressure ratio in conventional compressors is limited to ≈2 to 3 for CO2 compression. Ramgen has been developing a high-pressure ratio shock compression technology with DOE funding, which can increase pressure ratio per stage from 2 to approximately 10.10 This lowers the number of compressor stages needed for CO2 compression to 2,215 psia, and thereby decreases the physical size and lowering the capital cost.11

High-pressure ratio compressors also offer the possibility of integrating heat from the compression fluid with the Rankine cycle of the SCPC plant. A schematic of the high-pressure ratio compression is shown in Figure 4-3, where CO2 is cooled from the compressor exit temperature to 70 °F.

Figure 4-3. Schematic of Two-Stage Compression of CO2 Using Shock Wave Compression
Note: Integration with the SCPC plant Rankine cycle is not shown in the figure.

For the purposes of power plant heat integration, the schematic shows CO2 would be cooled to 150 °F using cooling fluid (e.g., boiler feedwater [BFW]) and further cooled to 70 °F using cooling water. The calculated compressor outlet temperatures, head and compression work for the two stages are presented in Figure 4-4.

Results shown in Table 4-3 are based on a 5 psia pressure drop for cooling 100% pure CO2 (i.e., for both coolers per stage) and assume 86% polytropic efficiency. Both the area integration (Prode) and Schultz methods were used to calculate the exit temperature, head and compression power.


Table 4-3. Stage Outlet Temperature, Head, and Overall Compression Load for the Two-Stage
Compression of CO2 from 23.5–2,215 psia Using Prode and Schultz calculations

Figure 4-4. p-H Diagram for Conventional and High-Pressure Ratio CO2 Compression (ηp = 86%)

The results from Figure 4-4 indicate that the enthalpy at the compressor stage outlet is significantly higher for high-pressure ratio compressors compared to conventional compressors. Therefore, compressor load for high-pressure ratio compression would also be considerably (30.16%) higher because of the higher temperature (446–472 °F vs. 167–192 °F) and higher fluid volume to be compressed.

12 The gas power [hp/(lbm/min)] does not include the mechanical efficiency of the compressor. 1 hp/(lbm/min) = 98.64 kJ/kg.


In isothermal compression, the temperature of the fluid remains constant. Isothermal compression requires that the heat produced during compression be removed as quickly as it is produced. The conventional means to lower the temperature is to cool the gas in between compression stages using external intercoolers. Multi-shaft, integral-gear centrifugal compressors can be used to achieve near-isothermal operation with external intercooling. However, their reliability is low compared to single-shaft in-line centrifugal compressors, which also offer higher operating flexibility.13 The Southwest Research Institute (SwRI) is developing a technology where CO2 is continually cooled in the compressor diaphragm using an internal jacket instead of using interstage cooling. Studies indicate that 6 to 12% reduction in the compression load is achievable using the cooled-diaphragm concept.14 The performance of the cooled-diaphragm technology approaches that of isothermal compression. A large number (16) of semi-isothermal stages was calculated to assess the cooled-diaphragm concept.15

There are two factors of note in the compressing CO2 semi-isothermally. First, it is evident from Figure 4-2 that isothermal com-pression at 70 °F (21.11 °C) would result in a crossover to the vapor-liquid region before compression to the dense-phase region. The physical properties of CO2 fluctuate significantly in the critical region, and compression paths evaluated in this study avoid the vapor-liquid phase transition. Second, in a typical power plant (e.g., Case 12), the temperature to which CO2 can be cooled using cooling water alone (i.e., without external refrigeration) depends on the temperature of the cooling water (Case 12: 60 °F). If the average CO2 temperature (inlet-outlet) in each compressor stage is to remain at 70 °F, external refrigeration is required.16 Two semi-isothermal compression cases were studied (Figure 4-5):
  • Semi-isothermal compression using 16 stages, intercooler exit temperature: 70 °F, average temperature of ≈90 °F (be-tween suction and discharge of each stage) [Table 4-4],
  • Semi-isothermal compression using 16 stages, intercooler exit temperature: 50 °F, average temperature of ≈70 °F (be-tween suction and discharge of each stage) [Table 4-5].
The compression path for conventional compression, isothermal compression at 70 °F, 110 °F (constant temperature lines) and semi-isothermal compression are shown in the CO2 pressure-enthalpy (p-H) diagram in Figure 4-4. CO2 is cooled to the same inter-cooler exit temperature (70 °F) for both the semi-isothermal compression path at an average temperature of 90 °F (blue line) and the conventional compression path (green). However, because only six compression stages were used in conventional compression, the temperature increase per stage is larger than that of the sixteen-stage path, leading to higher power consumption.

Table 4-4. Specific Work and Overall Compression Load for Compressing CO2 from 23.5 to 2,215 psia Using Intercooler Temperature of 70 °F and 16 Compression Stages

13 IEAGHG. Rotating equipment for carbon dioxide capture and storage. 2011/07, September 2011.
14 Moore J. et al. Novel concepts for the compression of large volumes of carbon dioxide—Phase III. Presented at the 2013 CO2 Capture Technology Meeting. http:// of%20CO2.pdf.
15 This case is denoted as semi-isothermal compression because the gas temperature before and after compression (i.e., before intercooler) is almost the same, if not exactly the same. The number of intercooling stages were chosen to provide average constant temperature (Tin+Tout)/2. See Tavg in Table 4-4 and Table 4-5.
16 Assumes a 10 °F approach temperature between the cooling water and CO2 streams.

Table 4-5. Specific Work and Overall Compression Load for Compressing CO2fFrom 23.5 to 2,215 psia Using Intercooler Temperature of 50 °F and 16 Compression Stages16

Figure 4-5. p-H Diagram for Semi-Isothermal CO2 Compression Cases Compared to Conventional CO2 Compression (ηp = 86%)

The compression load and head (specific work) required in both cases indicate that the power consumption is reduced by 12–15% for the semi-isothermal cases compared to the load for conventional compression (Table 4-2). Note that auxiliary cooling load may be needed to cool CO2 to 50 °F, increasing the total power requirement. Therefore, the power estimate for the 70 °F case represents an optimistic estimate for the power required to compress CO2 at 23.5 psia to 2,215 psia. Previous analyses by SwRI suggest that a compression path with low-pressure compression using cooled diaphragm, liquefaction and pumping results in larger energy consumption than a compression path with gas cooling and therefore the liquefaction option was not evaluated.10

17 The intercooler temperature for the stages 13–14 was adjusted to keep the temperature above the phase transition boundary line.


The higher stage outlet temperature for high-pressure ratio compression offers possibilities for recovering part of the lost compression work (lost as sensible heat) for heating boiler feedwater. A previous study by Chen and Matuszewski18 focused on the integration of heat recovered from high-pressure ratio CO2 compression in the context of a subcritical PC plant (Conesville Unit#5). In the Compression current study, a supercritical PC power plant with 90% CO2 capture (Case 12 of the DOE/NETL bituminous coal baseline, revision2) was chosen as the base plant for comparative analyses.

Table 4-6. Comparison of Results from the Excel Model for Case 12 to those from NETL Baseline

To facilitate heat integration studies, the Rankine cycle for Case 12 was modeled in Excel using an Excel add-in to calculate water/ steam properties.19 The parameters used to model the steam cycle were obtained from the bituminous coal baseline and the DOE/ NETL quality guidelines for energy system modeling studies (QGESS).20 These parameters are provided in Appendix 6.4. The power generated by the HP, IP and LP turbines of Case 12 is compared against baseline values, and the results indicate that the Excel model results are consistent with those from the DOE/NETL baseline. Because only the steam cycle and the CO2 compressor were modeled in this study, net power calculation assumed that other loads such as miscellaneous balance-of-plant, steam turbine auxil-iaries, circulating water pumps, ground water pumps, cooling fans, and transformer losses were similar to baseline values. Further, these were assumed not to vary significantly in response to heat integration.21

The compressor stage exit temperatures and the minimum CO2-BFW temperature approach determine the suitability of integrating the recovered heat within the Rankine cycle. The minimum temperature approach used in this study is 20 °F (same as that used in Chen and Matuszewski, (2009)). The feedwater heater (FWH) temperatures for Case 12 are shown in Table 4-7.

18 Chen, S., Matuszewski, M. Benefits of Ramgen shock compression technology for CO2 compression. DOE/NETL-2009/0629. June 2009.
19 The International Association for Properties of Water and Steam Industrial Formulation 1997 (IAPWS IF-97) steam tables from packaged in the ‘All-in-one’ add-in from the University of Alabama ( were used.
20 DOE/NETL Office of Program Planning and Analysis. Quality guidelines for energy system modeling studies: Process modeling design parameters. DOE/NETL-341/081911. (January 2012).
21 The only significant load which could change upon heat integration is the circulating water pump load due to the higher condenser duty for the cases with heat integration. However, because the condenser heat duty is ≈31% of the total cooling tower duty (5,360 MMBTU/h) for case 12, and because information about the other cooling loads for the power plant is not readily available, this value is assumed to remain the same. Integrating heat from CO2 compression would also lower the cooling duty for the CO2 intercoolers where CO2 is cooled to 70 °F, offsetting the increased condenser duty and increased cooling tower load (increased circulating pump and CT fan load), however this study does not take those into account.

Table 4-7. Boiler Feed Water Exit Temperatures for the Feedwater Heaters in Case 12

A comparison of the FWH temperatures from Table 4-7 and CO2 exit temperatures from Table 4-3 shows that a portion of the HP steam for FWH 7 could be potentially offset by heat integration. Decreased steam extraction to FWH 7 would result in increased IP and LP power generation, not an increase the HP turbine output. The actual amount and quality of heat that can be recovered by heat integration can be calculated by temperature-heat transferred (Q) diagrams for the two stages of CO2 cooling (Figure 4-5).

The calculations shown in Figure 4-5 assume that water exiting the condenser pumps can be further pumped to BFW pressure (4,200 psia) before being fed to the CO2 coolers. The temperature of the water stream exiting the HP coolers was determined by iterative calculations using Excel solver with the temperature approach of 20 °F as the minimum criterion. As discussed in Chen and Matuszewski, (2009), the specific heat capacity of CO2 at 2,220 psia changes significantly with temperature, leading to a pinch in the middle of the heat exchanger. However a corresponding pinch for the low-pressure (230.9 psia) CO2 cooler22 was not found.

Although CO2 gas temperature entering the HP cooler is higher (472 °F vs. 447 °F), the existence of the pinch for the HP cooler limits the temperature of BFW to ≈370 °F, lower than that exiting the LP cooler (≈410 °F). The effect of compression load on plant efficiency was analyzed by considering the following scenarios:
  • Case A: High-pressure ratio compression. Condensate pump exit BFW is pumped to 4,200 psia and is used to cool CO2 for both HP and LP coolers and fed to FWH 8 (after auxiliary heating)
  • Case B: High-pressure ratio compression. Condensate pump exit BFW is pumped to 4,200 psia and is used to cool CO2 in LP coolers and fed to FWH 8 (after auxiliary heating). HP CO2 stream (2,220 psia) is cooled from 472 °F to ≈330 °F by BFW from the main BFW feed pumps (i.e., exiting the deaerator ≈306 °F23).
  • Case C: Heat integration similar to Case A, with 90% polytropic compression efficiency24
  • Case D: Heat integration similar to Case B, with 90% polytropic compression efficiency19
  • Case E: Semi-isothermal compression (with an average temperature of 90 °F), with 86% polytropic compression efficiency
  • Case F: Semi-isothermal compression with an average temperature of 70 °F, with 86% polytropic compression efficiency
  • Case G: Semi-isothermal compression with an average temperature of 90 °F, with 90% polytropic compression efficiency
  • Case H: Semi-isothermal compression with an average temperature of 70 °F, with 90% polytropic compression efficiency

The process flow diagram for Case A is shown in Figure 4-7. The T-Q diagrams for Case A are shown in Figure 4-6. The process flow diagram for Case B is shown in Figure 4-8. The T-Q diagrams for Case B are shown in Figure 4-9. The T-Q diagrams for cases C and D are shown in Figure 4-10 and Figure 4-11. The gross steam turbine output, compressor load and efficiency for cases A-H are shown in Table 4-8, Figure 4-12 and Table 4-9.

22 One reason for the difference between the results of this analysis and that of the previous analysis (DOE/NETL-2009/0629) is that the CO2 gas used in that study had ≈31.9 mol% water vapor which would be condensed during CO2 cooling, increasing the specific heat (i.e., decreasing the slope of the T-Q line), leading to the pinch. Because the Case 12 feed gas did not contain an appreciable quantity of water (≈0.0069% H2O—stream 20), this 230.9 psia pinch is not likely to be prominent for the cases in this study.
23 The difference between the deaerator exit (303 °F) and BFW pump exit temperature (306 °F) is because of the pumping work lost to heat.
24 For comparison, a baseline (i.e., conventional compression) Case 12 with 90% polytropic compression efficiency was also modeled.

Figure 4-6. T-Q Diagrams for the 231 psia (Case A, B) and 2,220 psia CO2 Coolers (Case A) (ηp = 86%, high-pressure ratio compression)
Red: CO2, Blue: BFW, ∆Tmin = 20 °F

Table 4-8. Results from CO2 Compression Heat Integration Scenarios

Figure 4-7. Process Flow Diagram for Case A
Yellow: CO2, Light Blue/Violet: BFW, Red: Steam

Figure 4-8. Process Flow Diagram for Case B
Yellow: CO2, Light Blue/Violet: BFW, Red: Steam

Figure 4-9. T-Q Diagrams for HP CO2 Coolers in Case B
The LP cooler T-Q diagram is unchanged from Case A. Red: CO2, blue: HP BFW, purple: LP BFW. ∆Tmin = 20 °F

Figure 4-10. T-Q Diagrams for LP CO2 Coolers in Cases C and D
Red: CO2, blue: HP BFW. ∆Tmin = 20 °F

Figure 4-11. T-Q Diagrams for HP CO2 Coolers (Case C and D) (ηp = 86%, high-pressure ratio compression)
Red: CO2, blue: HP BFW, purple: LP BFW. ∆Tmin = 20 °F

Figure 4-12. Gross Steam Turbine Output and Compressor Load for High-Pressure Ratio Cases (A, B, C, D), Baseline Case 12, and Semi-Isothermal CO2 Compression Cases (E, F, G, H)

The integration of heat of CO2 compression with the Rankine cycle for a SCPC power plant leads to a small (0.2% to 0.5% [percent-age points]) increase in the net efficiency. Similarly, semi-isothermal compression at an average temperature of 70–90 °F would lower the compressor load by 12–15%, resulting in a comparably small increase in net efficiency (0.3%) over that for the base case (Case 12). Note that in the cases F and H, CO2 is cooled to 50 °F, which may require additional parasitic energy load depending on the plant conditions, reducing the net power and efficiency produced beyond that shown in Table 4-9.

The higher compressor efficiency in Cases C and D results in a lower compressor load, lower CO2 stage outlet temperature and lower BFW temperature (386 °F vs. 398 °F, 425 °F vs. 446 °F). However, the increased net efficiency for Cases C and D indicates that the reduction in compressor load outweighs the lower turbine output. Furthermore, because high-pressure ratio is inherently less efficient compared to conventional compression, the percentage reduction in compressor load when the efficiency is increased from 86% to 90% is 4.69% higher for the higher-pressure ratio compression (5.71% reduction vs. 4.6% reduction).

Table 4-9. Comparison of Net Power and Efficiency for Conventional, High-Pressure Ratio, and Semi-Isothermal CO2Compression

The preceding discussion noted that improvements in compression loads do not affect the net efficiency significantly. One feature of high-pressure ratio compression is that it has the potential to lower compressor capital costs significantly (≈50–60%) because only two stages are needed for CO2 compression. The effects of capital costs, and compressor power consumption on lowering the overall cost of CO2 capture were analyzed, using the updated (capital, fixed and variable operating costs) cost estimates (2011$) for Case 12 of the DOE/NETL baseline (Figure 4-12). A reduction in the capital cost of compressors and dryers does not contribute significantly toward lowering the CO2 capture cost. The reduction in CO 2 capture cost (over the baseline) resulting from the several advances discussed in Cases A to H is shown in Figure 4-12. Generally, for both high-pressure ratio compression (when two levels of heat integration are considered) and semi-isothermal compression (at 70 °F and 90 °F), at compressor polytropic efficiencies of 86% and 90%, CO2 capture costs were $2 to $3/T CO2 lower than that estimated for Case 12 of the DOE/NETL bituminous coal baseline. If the compressor costs were zeroed, the potential reduction in capture costs is $4 to $5/T CO2.

The reduction in CO2 capture costs for the high-pressure ratio compression cases does not account for the additional capital costs for the heat exchangers. Therefore the reduction in CO2 capture cost represents an upper bound of the potential improvement using high-pressure ratio compression when the compression heat is integrated within the Rankine cycle.

Figure 4-13. Reduction of CO2 Capture Cost and Cost of Electricity Upon the Integration of Advanced CO2 Compression Technologies25

Bars in light grey represent capture cost reduction with zero compressor capital costs.

The results from Chen and Matuszewski, (2009) indicate a slight decrease in efficiency when heat from high-pressure compression is integrated within the power plant Rankine cycle. This may arise from the higher quantity of water in the CO2 stream modeled in that work, and the differences in the performance of the Rankine cycles (subcritical vs. supercritical). Nevertheless, this slight improvement in performance would also need lower compressor capital costs to lower the CO2 capture costs.


The potential for improving net efficiency and reducing capture costs using advanced compression technologies, high-pressure ratio compression where compression heat is integrated with the Rankine steam cycle, and semi-isothermal compression were examined. Both technologies result in similar improvements in net plant efficiency (≈0.3 to 0.5 percentage points). Further, the CO2 capture costs are reduced $2 to $3/T CO 2. A compressor capital cost of zero would result in a reduction in the cost of capture by $4 to $5/T CO2. Similarly, the COE would be reduced $2 to $5/MWh, for the non-zero cost, and $5 to $6/MWh for the zero-cost scenarios. Although these results indicate that both compression technologies do not lead to significant cost reductions in capturing CO2 from supercritical bituminous-coal fired power plants, reducing the cost of CO2 compression is nevertheless important to achieve DOE CO2 Capture R&D program goals. For example, reduction of the compression load, results in a smaller overall plant, reducing overall plant costs.

25 Note: The compressor capital cost ($/hp) for semi-isothermal cases was assumed to be similar to that for Case 12. Ramgen compressor costs were escalated using the 2007$ estimate for Ramgen and the 2007$ and 2011$ estimates for conventional compression and drying from Case 12. The total variable operating cost, fuel cost and total fixed operating cost were assumed be constant. Capital charge factor: 0.1243, capacity factor: 85%. 


The types of technologies to be employed in the future for compressing CO2 would be partly driven by the pressures at which CO2 is available, either from natural sources such as the Jackson Dome, or from anthropogenic sources such as industrial or power plant gas streams. Several capture technologies being currently developed with DOE/NETL support can release CO2 at pressures from 1 to 10 bar (15 psia to 150 psia), reducing compression load compared to the cases discussed in this report. The pressure at which CO2 would be available may determine how much of an impact CO2 compression would have on the final cost of electricity, cost of capture, or cost of the product. For example, in cases where CO2 is separated at sub-atmospheric pressures, the cost impact will be relatively higher. Similarly, if a technology with low parasitic load is used to capture CO 2, the quantity of CO2 generated from the capture process for the same net power output would be lower, which would lower the costs of CO2 compression because a comparatively lower quantity of CO2 would be captured and compressed for the same net power output.

The current analysis did not consider the capital costs of heat exchange equipment for the case studies. Feed water heater costs could be considered in future techno-economic analyses to better reflect the economics of the high-pressure ratio compression steps.

The heat-integration analysis was based on a supercritical pulverized bituminous coal-fired Rankine cycle. Excess heat from the shock-wave compression is not well integrated within a steam-based Rankine cycle. On the other hand, integrating the heat energy from shock-wave compression with a supercritical CO2-based power cycle may be more effective and could be considered as an avenue for future work.

Industry demand for CO2 compression would also be driven by the requirements of pipeline operators and CO2 producers supplying CO2 for enhanced oil recovery (EOR). CO2 compression and purification is also a critical component of technologies such as oxy-fuel combustion, where the downstream CO2 purification unit (CPU) is a critical component of the overall CO2 capture and com-pression scheme. Although the analysis in the current report indicates that advanced compression technologies may not reduce CO2 capture costs for supercritical power plants considerably, the development of technologies which can improve the efficiency and the capital costs of CO2 compression is nevertheless of primary importance to the use and storage of CO2. For example, advancements in CO2 compression which can lead to smaller, more efficient, less costly compressors are of significance to the upstream and mid-stream oil and gas industry, and are also relevant to the developers of supercritical CO2 power cycles.


The 10 largest coal producers and exporters in Indonesia:

  1. Bumi Resouces
  2. Adaro Energy
  3. Indo Tambangraya Megah
  4. Bukit Asam
  5. Baramulti Sukses Sarana
  6. Harum Energy
  7. Mitrabara Adiperdana 
  8. Samindo Resources
  9. United Tractors
  10. Berau Coal