feasibility study of using brine for carbon dioxide ... ﬂuid in two separate, ... (bicarbonate or...

Feasibility Study of Using Brine for Carbon Dioxide Captureand Storage from Fixed Sources

Daniel Dziedzic, Kenneth B. Gross, Robert A. Gorski, and John T. JohnsonGeneral Motors Research and Development Center, Warren, MI

ABSTRACTA laboratory-scale reactor was developed to evaluate thecapture of carbon dioxide (CO2) from a gas into a liquid asan approach to control greenhouse gases emitted fromfixed sources. CO2 at 5–50% concentrations was passedthrough a gas-exchange membrane and transferred intoliquid media—tap water or simulated brine. When usingwater, capture efficiencies exceeded 50% and could beenhanced by adding base (e.g., sodium hydroxide) or thecombination of base and carbonic anhydrase, a catalystthat speeds the conversion of CO2 to carbonic acid. Thetransferred CO2 formed ions, such as bicarbonate or car-bonate, depending on the amount of base present. Add-ing precipitating cations, like Ca��, produced insolublecarbonate salts. Simulated brine proved nearly as efficientas water in absorbing CO2, with less than a 6% reductionin CO2 transferred. The CO2 either dissolved into thebrine or formed a mixture of gas and ions. If the chemistrywas favorable, carbonate precipitate spontaneouslyformed. Energy expenditure of pumping brine up anddown from subterranean depths was modeled. We con-clude that using brine in a gas-exchange membrane sys-tem for capturing CO2 from a gas stream to liquid istechnically feasible and can be accomplished at a reason-able expenditure of energy.

INTRODUCTIONCarbon dioxide (CO2) emissions can be divided into twomain types: emissions from fixed sources (e.g., industry,heating, and power generation) and emissions from mo-bile sources (e.g., in-use cars and trucks). Fixed-sourceemissions compose about two thirds of total U.S. CO2

emissions, whereas mobile sources compose the remain-der. Globally, in 2000, emissions from fossil fuels totaled�23.5 Gt of CO2/yr, with most from North America,Europe, East Asia (i.e., China), and South Asia (the Indiansubcontinent).1 The Intergovernmental Panel on ClimateChange Special Report on Emission Scenarios estimatesCO2 emissions at 29–44 Gt of CO2/yr in 2020 and 23–84Gt of CO2/yr in 2050.1 Sixty percent of fixed-source emis-sions are large-scale (e.g., �0.1 Mt of CO2/yr) and occur at

specific locations. Most are amenable to CO2 control strat-egies, such as CO2 capture and storage, albeit at an eco-nomic cost.1 One form of CO2 capture and storage (alsoreferred to as CO2 sequestration) is the process of captur-ing CO2 from emission streams and placing it in geologicsites, such as depleted oil and gas wells, coal bed forma-tions, or saline formations for indefinite periods. Thisapproach is attractive, because fossil fuels can still be usedwhile addressing concomitant concerns about global cli-mate change. Indeed, part of the U.S. Department ofEnergy long-term strategy for a hydrogen economy isbased on coal (of which the United States has 1 billion t,about a 200- to 250-yr supply) with concomitant CO2

capture and storage. Because fossil fuels provide 85% ofworld energy, the transition toward more environmen-tally attractive alternatives, like a hydrogen-based econ-omy, will take significant time. Initially, large amounts ofhydrogen fuel will most likely be generated by reformingfossil fuels, which generates large amounts of CO2 as aby-product.2 Thus, a strategy for dealing with fixed-sourceCO2 emissions is essential for technologic approaches togreenhouse gas reduction.

A number of analyses suggest the opportunities andproblems posed by using geologic capture and storage.These studies are based on the traditional approach ofusing alkanolamine scrubbing. Our approach to CO2 cap-ture and storage is based on a different capture method(i.e., using membranes). However, the data available fortraditional scrubbing provide useful reference bench-marks for understanding the potential for geologic se-questration. For example, Hendriks and Blok3 provided anearly estimate of underground storage. They indicatedthat 500–1100 Gt of CO2 could be stored in saline forma-tions and permeable beds at varying depths. The thick-ness of the formation varies, and its porosity, permeabil-ity, and cap rock formation determine its suitability as apermanent reservoir. Hoffert et al.4 reviewed options forstabilizing atmospheric CO2 levels and noted that a majorsequestration effort would be required unless major emis-sion-free primary sources come online by 2050. White etal.5 reviewed CO2 capture with deposition in coal bedsand deep saline formations. They noted the vast potentialof saline storage reservoirs, reviewed approaches to cap-turing CO2, discussed the depositing and fate of CO2, andreviewed the health, safety, and environmental impactsof such a strategy. Shafeen et al.6 reviewed the techno-logic challenges of capturing CO2 and storing it as asupercritical fluid in two separate, deep saline formationsin southwestern Ontario with a capacity of 289 and 442

IMPLICATIONSCO2 capture and storage is one possible solution to green-house gas control. This work supports the concept thatcapture and storage options from fixed sources of CO2 maybe achieved in certain instances using direct transfer ofCO2 to brine without the need for alkanolamine scrubbing.

TECHNICAL PAPER ISSN 1047-3289 J. Air & Waste Manage. Assoc. 56:1631–1641

Copyright 2006 Air & Waste Management Association

Volume 56 December 2006 Journal of the Air & Waste Management Association 1631

million t each. They found that 14,000 t/day could bestored economically.7

In the present work, we assess the feasibility of se-questering CO2 from fixed sources using a strategy basedon membranes and gas-to-liquid transfer with the goal oftransferring CO2 gas directly to brine. We developed alaboratory-scale reactor that captures and chemically con-verts CO2 to other chemical species that can be stored invarious ways. This approach could expand options forCO2 capture and storage by providing an alternative totraditional monoethanolamine scrubbing and by broad-ening the possible storage options for the CO2 (e.g., gasvs. mineralization).

EXPERIMENTAL WORKLaboratory Reactor

Design. A laboratory-scale reactor was constructed to de-termine critical parameters for efficient transfer of CO2

gas from a gas stream to a capture liquid (Figure 1). A gasstream and a liquid stream simultaneously passedthrough a gas exchange module containing a micro-porous gas-exchange membrane. The CO2 gas diffusedacross the membrane to the liquid side and was sweptaway as dissolved CO2, acid (carbonic acid), or anions(bicarbonate or carbonate).

Gas Exchange Module. The membrane module consistedof a commercially available Medtronic Affinity NT 511hollow-fiber gas exchanger that is used for efficient O2

and CO2 exchange in heart-lung machines. The mem-brane is composed of microporous, polypropylene hollowfibers with a 2.5-m2 gas-exchange surface area. The mod-ule was modified on the gas side using epoxy plugs to

occlude gas vents on the module. This allowed gas pres-sure to be controlled at 5–25 psi, the approximate rangespecified by the manufacturer. In these experiments, tapwater, tap water with added substances (see below), waterpurified by reverse osmosis, or simulated brine was fedinto the liquid side.

Gas Flow. CO2 (BOC Gases) in varying concentrations(5%, 10%, 30%, and 50% nominal concentration in air)was metered through an Alicat mass flow controller at 1–9L/min into the gas inlet of the gas exchange module.Corrections were made to the mass flow controller setpoint to account for changes in gas viscosity at varyingconcentrations of CO2 gas using established procedures.8

A back pressure of 5–25 psi was maintained by using a gatevalve and a 0–50 psi pressure gauge (McDaniels Controls).The effluent gas from the module then passed through aModel 100 Waterless Chiller (California Analytical) hav-ing a 10-L/min flow capacity to remove water. An Alicatmass flow meter was used to determine the effect of losingCO2 on the mass flow of gas after the exchange process.Then, the dried gas flowed into a California AnalyticalModel 200 nondispersive IR analyzer to determine CO2

concentration to �0.1%.

Liquid Flow. Because seasonal fluctuations in water tem-perature affected measurements significantly, tap waterwas tempered to 22 °C using a Lawler 9700 Type TP ther-mostatic mixing valve. The liquid was pumped from areservoir using a Micropump Model 2211562 L19301. Thewater passed through an Omnifilter U25 water filter, andthe flow was monitored using a Blue-White Industriesliquid flow meter at 1–9 L/min. A back pressure of 10 psi

Figure 1. Schematic of CO2 capture and storage reactor. This basic design was used in various configurations to study the transfer of CO2

from a gaseous feed stream to a capture liquid and its conversion to storable forms. Pres � pressure gauge.

Dziedzic, Gross, Gorski, and Johnson

1632 Journal of the Air & Waste Management Association Volume 56 December 2006

was maintained on the water side of the module by usinga gate valve and a McDaniels Controls pressure gauge. Forsome experiments, a Micropump O/C GA Gear PumpModel 75211–22 drew off 100 mL/min from a T fittingconnected to the aqueous effluent stream and fed it intoan Orbisphere Model 3610 CO2 sensor that measureddissolved CO2. A sidestream from the main effluent flowwas passed through a flow cell fitted with a Cole ParmerSI499 FF03 pH probe and a Thermo Orion Model 410A�

pH meter for pH determination.

Simulated Michigan Brine. CO2 at a concentration of 30%in air was used for experiments employing simulatedMichigan brine. Simulated brine was made using a for-mula derived from the U.S. Bureau of Mines for “typical”Michigan brine9 and contained the following substances:8.2 g/L of MgCl2, 13.64 g/L of CaCl2, 5.45 g/L of NaCl,and 0.48 g/L of KCl (all J.T. Baker). These componentswere dissolved into tempered tap water (TTW) at 22 °C or,in certain experiments where we performed a mass bal-ance, into water purified by reverse osmosis.

Baseline Experiments Using WaterCO2 Transfer. In the first set of experiments, we evaluatedthe transfer efficiency of CO2 from gas streams with con-centrations of 5%, 10%, 30%, and 50%. We used filteredTTW as the capture liquid while maintaining back pres-sure of 5 psi on the gas side and 10 psi on the liquid side.The mass fraction, the mass transfer, the concentration ofCO2 in the capture liquid, and the pH of the effluent weredetermined. In the next experiment, we assessed the ef-fect of simultaneously varying gas and water pressure onextraction efficiency. We compared a 30% gas stream at 5psi, tap water at 10 psi, to a 30% gas stream at 20 psi, tapwater at 25 psi. We then evaluated the effect of the inter-action of pH and dissolved sodium bicarbonate on extrac-tion efficiency. First, we adjusted 100 L of TTW to pH 9,8.3, and 5.9 (all �0.1) with 50% NaOH and measured CO2

extraction efficiency using the TTW at the three pHs. Wethen made up a solution of TTW with 2000 mg/L ofsodium bicarbonate and readjusted each pH to the samethree starting pH values. We measured CO2 extractionefficiency for each combination of sodium bicarbonateand pH and compared the effect of adding sodium bicar-bonate with the effect of pH alone. In the next experi-ment, we assessed the effect of varying pH and/or addingthe catalyst carbonic anhydrase (EC.4.2.1.1, C-3934, Lot111K1102, from bovine erythrocytes, 4860 W-A units/mgprotein; Sigma) on capture efficiency. In this experiment,we used 30% CO2 at 7 L/min and compared CO2 extrac-tion using TTW with extraction using 40 L of water thathad been alkalinized by adding 19.2 mL of 50% NaOH(19.4 N). We then added 10 mg/mL of carbonic anhydraseand compared the CO2 extraction again. In a final set ofexperiments, we measured the yield of carbonate salt pro-duced by adding a continuous flow of reagents to thereactor effluent. We used a 10% CO2 feed stream, TTW,TTW with 0.02 M KOH, or TTW with 0.02 M KOH, towhich 10 mg/L of carbonic anhydrase had been added.These solutions were run at 1 L/min through the mem-brane module, and the effluents were treated with 1 MKOH at 17 mL/min and 1 M CaCl2 at 25 mL/min. The

precipitate that formed was collected on filter paper,washed, dried, and weighed.

Chemical Characterization of Precipitate. To chemicallycharacterize the white precipitate that formed during ex-perimentation, the reactor was operated at the followingconditions: 30% CO2 feed stream at 5 or 7 L/min andTTW at 3 L/min. An excess of CaCl2 (20 g/L as CaCl2�2H2O) was then added to the collected effluent, and the pHwas adjusted to pH 8 or 9 using 0.02 M KOH. Calcium andsodium were determined quantitatively by inductivelycoupled plasma/atomic emission spectroscopy. Carbonand sulfur mass determinations were performed quanti-tatively by combustion IR analysis. Chloride was deter-mined qualitatively by silver nitrate precipitation and wascorrelated with X-ray fluorescence spectroscopy compar-ing k-� Cl and Ca peaks using quantitative Ca data fromabove.

Experimental Protocols Using Simulated BrineLong-term membrane performance was evaluated by run-ning simulated Michigan brine intermittently for 4-hrperiods at 10- to 50-day intervals over 190 days. After each4-hr run, membrane performance was determined by run-ning 5 L/min of 30% CO2 with 5 L/min of TTW alone, andthe CO2 transfer was compared with baseline values. Wenext determined the ability to capture and convert CO2

into carbonate salts using brine in the reactor. In contrastto previous work, no components were added to the re-actor effluent to facilitate precipitation. In these experi-ments we prepared brine at neutral pH and brine that wasalkalinized using 3630 mg/L of tris(hydroxymethyl)ami-nomethane (TRIS) base (Sigma). To perform an accuratemass balance, we used water purified by reverse osmosis toremove dissolved CO2. Effluent was collected and allowedto react overnight. The precipitate was collected, and theyield was determined as described previously.10

Energy Use ModelingGeneral Considerations. We created a model, Energy Modelfor Carbon Dioxide Sequestration Strategies (EMSS), toassess energy requirements for using a strategy based onmembranes and brine to capture and store CO2. A copy ofEMSS is available at http://www.awma.org/journal/pdfs/2006/12/dziedzicsupplementalmaterial.xls. The energyrequired to pump liquid from subterranean reservoirs andto return the liquid to the reservoir was calculated. Resultsare reported as megawatt hours per ton of net CO2 cap-tured (i.e., the amount captured less that amount pro-duced in supplying the power for the operation) and asthe ratio of the rate of net CO2 captured to the rate of CO2

captured unadjusted for CO2 from energy use, which is ameasure of efficiency of the sequestration operation. Theenergy calculations were divided into three parts: (1)power to pump the brine, (2) power to compress the inputgas stream, and (3) energy penalty for generating electric-ity to run the operations. The last step in the modeloptimizes the liquid flow rate based on the theoreticalCO2 input concentration.

The model was used to iterate solutions for energyconsumption and efficiency from a given set of inputvariables (e.g., CO2 concentration, flow, etc.). A given



energy expenditure and/or storage efficiency was targetedfor a set of input parameters that described a process (e.g.,hydrogen production) and determined solutions in termsof the requirements (e.g., depth, base concentration, etc.)to attain those targets. The target was chosen to providean energy input for sequestration that was consistent withthe Department of Energy target of a small (�10%) costadded to energy generation expenditure. Many solutionsare possible for the same output and depend on thechoice of depth, amount of added base, and so forth.

Brine Pumping Power Calculation. The power required topump brine up from and back to an underground reser-voir was calculated using the reservoir depth, liquid flowrate desired, sp gr of the brine, and the motor/pumpefficiency (eq 1). This equation was modified from Turtonet al.11 by adding a sp gr term (�) to account for increasedweight of brine salts. The equation by Turton et al.11 wasadapted from Walas,12 where ε is the motor/pump effi-ciency:

kW � 1.67 � �Flow m3/min�� P bar�/ε (1)

After the resultant power was calculated, the additionalpower required for any desired gas compression (see fol-lowing section) was added.

Gas Compression Power Calculation. The power requiredfor compressing a mixture of air and CO2 was calculatedbased on eq 2, also from Turton et al.11

kW � m ZRTP2/P1a � 1/1000 a ε (2)

where m is the molar flow rate, Z is the compressibility(assumed to be 1), R is the gas constant (8.314 J permol-K), T is temperature (K), P1 and P2 are initial and finalpressure (atmospheres), and a is 1 � k �1, where k is(Cp/Cv), the ratio of the heat capacity of the gas at con-stant pressure to the heat capacity of the gas at constantvolume. The value of k was assumed to be 1.41, the valuefor an ideal diatomic gas, and, therefore, a remains con-stant at 0.2908. Although Z and a are different for CO2,the modeled values result in a conservative estimate ofcompression power for mixtures of CO2 and air and hadno impact on the energy use for net CO2 captured.

CO2 Penalty and Energy Use Calculations. The rate of CO2

produced as a by-product of the energy used in the pump-ing operation was calculated based on the input value of1.35 lb of CO2 per kilowatt-hour of electrical energyused.13 Energy consumed per year was calculated frominputs of power in kilowatts and duration of use.

CO2 Concentration and Capture Calculations. In the finalstep, the rate of CO2 capture, the highest theoretical con-centration of CO2 in the liquid, and the lowest theoreticalliquid flow rate were calculated. The mass flow of CO2 inthe gas stream was calculated using inputs of gas streamflow, pressure, temperature, and concentration of CO2.The input value of gas stream pressure was added to the

total head pressure requirement for the submerged pumpin the pumping power section. After the mass flow of CO2

was calculated, the concentration of CO2 (100% transferefficiency) was found by dividing the mass flow of CO2 inthe gas stream by the liquid stream flow. The theoreticalmolarity of the CO2 was calculated, and if base waspresent, the theoretical concentration of CO2 in the liq-uid was reduced by the concentration of the base, assum-ing the reaction proceeded in a one-to-one manner. Thecalculated theoretical concentration of CO2 remaining inthe liquid was corrected by the fraction that was trans-ferred through the membrane and into the liquid. It wasassumed that a value of 0.9 could be achieved based onlaboratory-scale experimental results (e.g., see Figure 2).

To find the lowest theoretical liquid flow, we firstdetermined the maximum concentration of CO2 theoret-ically possible in the liquid, given a set of model condi-tions. The baseline concentration, which was adjusted bythe model, was 0.033 mol/L, the solubility of CO2 at room

Figure 2. CO2 transfer for a 30% gas stream to tap water (initialpH � 6.78). The fraction of CO2 transferred (top); the CO2 trans-ferred in mmol/min (middle), and the CO2 concentration (mmol/L;bottom).



temperature and pressure. This baseline value was multi-plied by the gas stream pressure, the percentage of CO2 inthe gas stream divided by 100, and the transfer efficiency(fraction) to obtain the maximum concentration of CO2

theoretically possible in the liquid. No correction wasmade for the temperature of the liquid. After the maxi-mum theoretical CO2 concentration was calculated forthe model conditions, liquid flow rates were automati-cally iterated until the modeled CO2 concentration con-verged on the maximum theoretical CO2 concentration.This value was the lowest theoretical liquid flow rate. Thisflow rate was then input into the pumping power calcu-lation and the energy values adjusted accordingly.

The rate of CO2 capture was found by adjusting theexpected theoretical capture rate. Finally, the energy useper ton of net CO2 captured was determined from theenergy use for the year divided by the yearly rate of netCO2 captured. The overall efficiency of the sequestrationwas represented by the ratio of the rate of net CO2 cap-tured to the rate of CO2 captured, unadjusted for CO2

from energy expenditure. The relationship of the compo-nents of the model is shown in the appendix.

Limitations. EMSS uses simplifying assumptions so thatan approximation of operating energy expenditure couldbe made. It does not include energy for drilling or main-tenance, which would be one-time or small expendituresof energy. Energy needed to overcome friction from liquidmovement through pipe is not included.

RESULTSCO2 Transfer and Chemical Conversion

We determined the efficiency of CO2 transfer over a broadrange of conditions by varying CO2 concentration (5–50%), water flow (1–7 L/min), and gas flow (1–9 L/min)while maintaining fixed back pressures of 5 psi on the gasside and 10 psi on the water side. A representative figureusing 30% CO2 is provided (Figure 2). In general, theextraction efficiency depended on the CO2 concentrationand the ratio of gas-to-liquid flow rate. The fraction ofCO2 extracted from the gas stream varied as a function ofwater and gas flows, ranging from 10% to �100% (Figure2, top). At the highest CO2 concentration (50% CO2),

varying gas flow and water flow produced extraction effi-ciencies from 100% to �20%. With 5% CO2, somewhatless (80–90%) of the CO2 was maximally extracted underthe most favorable gas-to-liquid flow ratio used. The rateof CO2 transferred to the water for 30% gas (Figure 2,middle) was dependent on the gas and water flow rates. At30% CO2, the maximal rate was 63 mmol/min at optimalgas and water flow. Overall, the rate of CO2 transferredvaried from �115 mmol/min (50% CO2 flow at 9 L/min;liquid flow at 7 L/min) to �2 mmol/min (5% CO2 flow at1 L/min; liquid flow at 1–7 L/min). We calculated theconcentration of carbon species (i.e., dissolved CO2 andionic forms, expressed as dissolved CO2; Figure 2, bot-tom). The values correlated with dissolved CO2 (at neutralpH) as measured using a CO2 sensor using an Orbispheremonitor as described in the methods (data not shown). At30% CO2, the maximal concentration occurred at lowwater flows and high gas flows. The maximal concentra-tion of CO2 was 27 mmol/L when 50% CO2 was used.

CO2 reacts with water according to the followingreaction:

CO2 � H2O3 H2CO3 (3)

The carbonic acid then rapidly dissociates to form HCO3�

and H�. The effect of this reaction could be readily seen ineffluent pH changes as a function of gas and liquid flow.For example, at 50% CO2 concentrations, a maximalchange of �2 pH units was observed in our system at gasand water flows of 9 L/min and 7 L/min, respectively.Lesser pH changes were observed at lower CO2 concen-trations and depended on gas and water flow rates.

We next assessed the effect of varying the pressure ofthe gas stream on extraction efficiency. We changed thegas and water back pressures from gas (30% CO2) at 5 psiwith water at 10 psi to gas at 20 psi with water at 25 psi(Figure 3). The extraction efficiency increased signifi-cantly at elevated pressure for comparable gas and waterflows. In the next experiment, we determined the effectsof adding base (NaOH) alone or with added carbonicanhydrase catalyst on transfer efficiency (Figure 4). Add-ing 19.2 mL of 50% NaOH to 40 L of tap water increased

Figure 3. Effect of increasing gas and water-side pressures on CO2 transfer efficiency. [CO2] � 30%.



the capture efficiency by �1.2-fold over water alone.When we used the same alkalinized liquid and addedcatalyst at 10 mg/L, the capture efficiency increased to1.4-fold over water alone.

The ability of liquid to extract CO2 is affected bythe pre-existing “carbon load” of the liquid. To under-stand this effect on extraction efficiency, we added acarbon load in the form of sodium bicarbonate to TTWand adjusted the liquids to varying pHs. The pH affectsthe extraction efficiency by determining whether theCO2 load exists predominately as dissolved gas andcarbonic acid (pH � 6); predominately as dissolvedbicarbonate with some dissolved CO2, carbonic acidand carbonate ions (pH 6–10); or as a mixture of bicar-bonate and carbonate ions with a minor component ofdissolved CO2 and carbonic acid (pH 8–12). We as-sessed the combined effect of adding 2000 mg/L ofsodium bicarbonate and varying pH on extraction effi-ciency using 30% CO2 at gas flows of 1–7 L/min and aTTW flow of 3 L/min (Figure 5). At alkaline and inter-mediate pH, the effect of bicarbonate and pH was neg-ligible. The effect of acid pH was negligible on extrac-tion efficiency when using TTW but was significantwhen bicarbonate was present, where efficiency wasreduced by �90%.

Finally, we ran the reactor and added a continuousflow of base and calcium chloride to the reactor effluent.

In this case, CO2 was converted into carbonate salts (Fig-ure 6). Chemical analysis of the precipitate formed undersimilar conditions indicated that the precipitate consistedof nearly all CaCO3 and that other salts (e.g., NaHCO3

and NaCl) were absent (Table 1).

Simulated Michigan BrineWhen using simulated Michigan brine, the capture effi-ciency reduced by a small amount compared with TTW.The largest effect seen over a range of gas flows (1–7L/min) with water flow fixed at 3 L/min was a 6% reduc-tion. The magnitude of reduction depended on gas flow(i.e., the amount of CO2 transferring into the liquid). Thisreduction was readily reversed by using TTW, indicatingthat the observed effects were transient changes in CO2

loading capacity because of the salt content of the simu-lated brine.

We determined the effect of using brine on mem-brane integrity over time (Figure 7). Our experimentalsystem was used intermittently over weeks with in-terspaced periods of “down time” to determine the poten-tial impact of repeated runs and prolonged periods offluid stagnancy in the system. We initially conductedthree runs with TTW to establish a baseline and thenswitched to using simulated brine. The switch showed aninitial reduction in efficiency of �5%. Over the next sev-eral weeks, we assessed membrane performance between

Figure 4. The effect of adding base alone or base � catalyst on thecapture efficiency of tap water. [CO2] � 30%; gas flow � 7 L/min.

Figure 5. The effect of the interaction of sodium bicarbonate addition (2000 mg/L) and pH on CO2 extraction efficiency. [CO2] � 30%; TTWFlow � 3 L/min.

Figure 6. The effect of base alone or base with catalyst on theefficiency of converting 10% CO2 from a gas feed stream (gas-to-liquid flow rate ratio � 5:1) into a carbonate precipitate.



brine experiments, using TTW to compare the efficiencywith our initial TTW runs. Gas and liquid flows were keptconstant. The extraction efficiency as measured with TTWruns remained within 6% of initial values over the testingperiod indicating that membrane efficiency was well pre-served. In addition, a wash with strong acid could be used torestore the efficiency of the membrane to within 2% of theinitial value (data not shown).

In a final experiment, we assessed the ability of ourreactor to form carbonate salt precipitate in the presenceof a nonbicarbonate base, TRIS base. We alkalinized thesimulated brine by adding TRIS at 3630 mg/L. In thisexperiment, we used water purified by reverse osmosis toperform a mass balance. A key issue was to determinewhether the membrane would function properly when allof the chemical components that could produce carbon-ate precipitate were present in the brine. Another keyquestion was to assess the efficiency of precipitate forma-tion using the simulated brine. The extraction efficiencyof the gas was consistent with previous work using simu-lated brine (Figure 8, top). The precipitation efficiency ofthe gas dissolved into the liquid was nearly 100% underthe conditions chosen. Thus, the calculated net efficiencyof transfer and conversion of CO2 gas to precipitatedproduct could be superimposed on the gas transfer effi-ciency curve. To fully assess the impact of base on the

transfer and mineralization process, we ran the same ex-periment at neutral pH (Figure 8, bottom). The extractionefficiency was somewhat lower than that observed withalkalinized brine, as expected, when comparing a neutralpH liquid with an alkaline liquid. However, this effect wassmall (i.e., �6%). Therefore, the transfer of CO2 to liquidremained quite efficient. However, even after allowing theeffluent to react overnight, no precipitate formed. Theseresults indicate that one can choose different chemicalforms (i.e., mineral vs. soluble forms) as storage optionsfor the CO2 depending on the chemical properties of thebrine.

Energy Model of CO2 Capture Using BrineTo better understand the feasibility of using this labora-tory chemistry on a larger scale, we developed the energyuse model entitled “EMSS.” A flow diagram of the modelis shown in the appendix. There are 14 input variables,each of which is independently adjustable. The four mainsections of EMSS are brine pumping power, gas compres-sion power, CO2 penalty as result of energy consumed,and CO2 capture. The CO2 concentration and capturesection contains an input that addresses the presence ofbase, also shown in the diagram.

Two scenarios were modeled using EMSS. A summaryof energy use for CO2 capture and storage for an industrialgas burner that might be used in a typical manufacturingoperation and that produced a dilute CO2 stream wascompared with hydrocarbon reforming from methane forhydrogen production (Table 2). For a natural gas emis-sions control system, the concentration of CO2 in the gasstream is 2% at 1.1 atm pressure. We found that it wasnecessary to use brine having a base content of 0.135 Mfrom a reservoir 500 ft below the earth to achieve anenergy expenditure of 0.34 MWh/t of net CO2 captured,with an overall operational efficiency of 0.81. When mod-eling a hydrocarbon reforming operation, the CO2 con-centration is 60% at high pressure (25 atm), and thisresults in energy use of 0.37 MWh/t of net CO2 capturedand an overall operational efficiency of 0.80 at 2000-ftdepth, with no base required for capture.

Table 1. Mass percent of substances in precipitate.

Sample Carbon CalciumChloride

(%)Sodium

(%)RatioCa/C

1 10.6 38 �1 �0.1 3.582 11.0 39 �1 �0.1 3.543 10.8 38 �1 �0.1 3.514 10.7 38 �1 �0.1 3.555 10.6 37 �1 �0.1 3.496 10.7 37 �1 �0.1 3.457 10.8 37 �1 �0.1 3.42

Notes: The theoretical ratio of Ca to C for CaCO3 (40.0% Ca and 12.0% C) is3.33.

Figure 7. Extraction efficiency of CO2 gas by polypropylene gas exchange membrane as a function of time. [CO2] � 30%; Gas flow � 5 L/min;Liquid flow � 5 L/min. See Experimental Work section.



DISCUSSIONIn previous work, Xu et al.14 used a polypropylene mem-brane contactor with aqueous absorbent solutions, suchas sodium hydroxide, monoethanolamine, and dietha-nolamine to capture CO2 from a CO2/N2 mixture. Theyused a self-made polypropylene hollow fiber membrane,CO2 concentrations of 5–50%, and gas and liquid flows of�8.3–16.6 and 0.6 L/min, respectively. They found thattheir membrane efficiently transferred the CO2. The pro-cess was characterized by a mass transport coefficient thatincreased with increased liquid flow and increased gasflow and concentration of base added. The transport also

varied with CO2 concentration and temperature of theabsorbent liquid. Trachtenberg and colleagues15–17 devel-oped an approach to CO2 scrubbing based on using car-bonic anhydrase for air purification applications in space-craft. The use of enzyme-facilitated transport membraneswas effective at low concentrations of CO2 and couldcapture CO2 to near-ambient levels. The carbonic anhy-drase selectively enhanced the flux of CO2 over O2. Themembranes they used included immobilized liquid, alu-minum oxide, and polyester types, and varying immobi-lization schemes for both native and mutated forms ofcarbonic anhydrase were used. Bond et al.18 evaluatedcarbonic anhydrase for accelerated CO2 transport for ef-fective CO2 capture into a liquid using a scrubber ap-proach. Using catalyzed hydration, they proposed thatCO2 could be captured efficiently in a traditional scrubberthat contained a catalyst, thereby avoiding the process ofconcentrating CO2. They developed various forms of thecatalysts19 and then used a suitable form with simulationsof liquids that are produced from oil and natural gasexploration (so-called “produced water”) to determine theefficiency of CO2 capture and conversion to carbonateminerals. An appeal of this approach is that these“waters” are brine and brine-like liquids containingprecipitating counterions (e.g., calcium and magnesium)that are already being generated, transported, and rein-jected into the earth by the oil and gas industry duringthe process of energy exploitation. Based on this work, Liuet al.20 estimated that 3.49 Mt of CO2/yr could be con-verted to carbonate from liquid from the San Juan and

Figure 8. Efficiency of CO2 capture and mineralization using brine. [CO2] � 30%; brine flow � 3 L/min. Brine was alkalinized with 3630 mg/Lof TRIS base (top). pH was neutral (bottom). “CO2 absorbed” is the percent of CO2 gas from the feed stream transferred to the capture liquid;“CO2 Mineralized Based on CO2 in” refers to the amount of CO2 captured and transformed into mineral; and “CO2 Mineralized Based on CO2

Dissolved” refers to the percentage of dissolved CO2 that was mineralized.

Table 2. Summary of one example of critical variables that determinesenergy use/net ton of CO2 captured, operation efficiency, amount of baserequired, and net CO2 captured and stored for a small gas burner and forhydrogen production.

VariableNatural Gas

BurnerHydrogen

Production

Depth of reservoir (ft) 500 2000Volume of CO2 in gas stream (%) 2 60Temperature of gas stream (C) 25 400Pressure of gas stream (atm) 1.1 25Base content of liquid stream (mol/L) 0.135 0MWh/t of net CO2 captured 0.34 0.37Ratio of rate of net CO2 captured to

rate of CO2 captured, unadjustedfor energy use

0.81 0.80

Notes: Other solutions are possible.



Permian basins based on the known calcium and magne-sium in the liquids.

In the present work, we combined and extended ele-ments of these approaches. We confirmed that a mem-brane can efficiently capture CO2 in a liquid medium overa range of conditions, including changes in CO2 concen-tration, gas flow, liquid flow, and gas-side pressure. Wedemonstrated additive effects of base and dissolved car-bonic anhydrase on the membrane efficiency and showedthat simulated brine could be used intermittently as thecapture liquid over a period of weeks without clogging orunacceptable degradation of the membrane. Simulatedbrine captured CO2 from a gas stream at nearly the sameefficiency (90–95%) as that of TTW, the benchmark usedin our early reactor experiments. The decline in efficiencyusing simulated brine at neutral pH is �6% for the de-fined conditions. The reductions in CO2 capture effi-ciency are due to the ionic effects in the brine, which tendto mildly impede the transfer of CO2 gas and its conver-sion to acid and ions. The capture efficiency in brine isquite good at all pH levels (i.e., �98%), except when highbicarbonate loads are present with low pH, which im-peded CO2 transfer significantly. A membrane-based ap-proach is flexible in that it is consistent with strategies tocollect or convert the CO2 in solution as a gas, acid orions, or as precipitated carbonate salts. The reactor canwork continuously with the fluids passing though in asingle pass, with or without added chemicals, such asbase. If base is present or added, the process of convertingCO2 to a stable, nontoxic salt is possible. These findingssupport the concept that direct transfer of CO2 to brinefor storage using a gas-to-liquid exchange membrane inone of several chemical forms is feasible.

There are several issues related to using membranesfor industrial-scale CO2 capture. They must be cost-effec-tive and must stand up to the rigors of industrial-scaleoperations. Developing new and robust membranes is anactive area of research.21 In the present work, we used ahigh quality but relatively fragile polypropylene mem-brane system because of its high probability of providinggood data on CO2 transfer. As more options are devel-oped, more industrially suitable membranes will emerge.Another possible issue is that, in this system, net transferefficiencies of CO2 are lower in streams with a low CO2

concentration compared with those with a high concen-tration. This is attributable, in part, to the partial pressureof the CO2 in the initial gas stream. Henry’s law predictsthat gas with a higher partial pressure of CO2 will transfermore CO2 to the liquid than gas with a lower partialpressure of CO2. The concentrations of CO2 from typicalgas, oil, and coal-burning generators and hydrogen re-formers are 2–5%, 5–10%, 10–15%, and 15–60%, respec-tively. At the lower CO2 concentrations, modificationmay be needed to increase the mass of CO2 captured perunit of energy used for capture and storage. One way to dothis is to develop new ways of combusting fossil fuels toproduce high CO2 concentrations. An example of oneapproach is to burn oxygenated fuels, which producehigh CO2 concentrations per unit of energy. Anotherexample is to eliminate N2 before combustion using pres-sure swing absorption or adsorbents. This effectively con-centrates the combustion reactants and produces a highly

concentrated CO2 stream.22 Another way, as was shownin our current model, is to choose brines with high basecontent and/or add very inexpensive base. These modifi-cations would incur costs that we have not addressed inthe current model. However, the high CO2 concentrationcase (hydrogen reforming) would not require these mod-ifications or their added expense indicating potential usefor such applications.

Storage of CO2 is a critical part of any CO2 captureand storage strategy. Although brine formations are aleading candidate for terrestrial storage, the best chemicalform in which CO2 should be stored (gas, ions, and/ormineral) is less clear. Storing the CO2 as compressed gas isnow a leading approach. This is the result of extensiveexperience using CO2 flooding of oil and gas fields toenhance tertiary recovery of oil and gas. There are a num-ber of issues that are being investigated when appliedto CO2 sequestration on a large scale. First, the amount ofCO2 to be sequestered is enormous: tens of gigatons ofCO2 are generated per year. Maintaining such huge vol-umes of CO2 as a gas without concomitant leakage andpossible risk to life and property is being intensively stud-ied.23 In addition to the gas storage option, we evaluatedthe possibility of using brine chemistry to form carbonatesalts, which are stable, nontoxic forms of carbon. Salts ofthe carbonates could be formed into “mountains” aboveground. Alternately, the CO2-containing brine could bepumped into the original brine formation and the precip-itation could be allowed to occur underground (Figure 9,top). These experiments show that the chemistry of thebrine can be chosen in a membrane-based system to effi-ciently produce carbonate salts. Consistent with the workof Bond and colleagues,18,20 we found that, under theproper conditions, alkaline brine works quite effectivelyin converting CO2 to carbonate ions that will react withappropriate counterions to produce solids. Base is neededto drive this reaction. Being able to augment any naturallyoccurring base with added base would increase the feasi-bility of this storage strategy. Although mineralization ispreferable to storing CO2 as a gas, we do not rule out thetechnical feasibility of storing CO2 as gases or ions inbrine formations (Figure 9, bottom) if this proves accept-able from leakage and safety perspectives. However, theadvantages of a mineralization strategy would make it thepreferred approach.

To understand the energy penalty of brine-based CO2

sequestration, we created a model to calculate the energyrequirement to pump liquid from subterranean reservoirsand return the liquid to the reservoir. The model reportsthe results as the megawatt hour per ton of net CO2

captured and the ratio of the rate of net CO2 captureddivided by the rate of CO2 captured unadjusted for CO2

from energy expenditure, a measure of efficiency of thesequestration operation. The user can compare operatingconditions and required resources (e.g., use of pressure,base, CO2 content of the gas steam, or changes in theefficiency of gas-to-liquid transfer) of a CO2 sequestrationscheme to achieve an acceptable operating energy use.This information makes it easy to compare the energyoperating expenditure of a proposed sequestrationscheme to other literature estimates.



The major energy use during brine-based CO2 captureand storage is pumping brine against pressure up fromand down into the earth. We simplified the calculationsby adding the energy required for each pass. This is sim-plistic but appropriate for an initial estimate of energyneeded. Once the liquid is brought to the earth’s surface,it needs to be moved horizontally through the mem-brane-equipped exchanger units. This additional powerrequirement was calculated according to an examplegiven by Chopey,24 but it was not made part of EMSS,because the power requirement is low. As an example, for1000 ft of 8-in., horizontal pipe, the additional powerrequirement is equivalent to power required to pumpfrom about an additional 50 ft of depth. The power re-quirement can be reduced by increasing pipe diameter.

Our analysis of the operating energy supports theidea of using brine to capture CO2. This is especially likelyfor hydrogen reforming when high CO2 concentrationsare produced and is likely true for fossil fuel combustionwith modifications. In the latter case, costs of the modi-fications would determine viability. In summary, mem-brane-based gas-to-liquid transfer can serve as a possibleapproach to CO2 capture and storage. Using naturallyoccurring brines and brine formations appears technicallyfeasible and practical based on net energy expended.

REFERENCES1. Intergovernmental Panel on Climate Change. Carbon Dioxide Capture

and Storage; Mertz, B.; Davidson, O.; de Coninck, H.; Loos, M.; Meyer,L. Eds.; Cambridge: New York, NY, 2005; pp 19-21.

2. Muralov, H. Hydrogen from Fossil Fuels without CO2 Emissions. InAdvances in Hydrogen Energy; Padro, C.E.G.; Lau, F. Eds.; Kluwer Aca-demic/Plenum: New York, NY, 2000; pp 1-16.

3. Hendriks, C.A.; Blok, K. Underground Storage of Carbon Dioxide;Energy Convers. Mgmt. 1993, 34, 947-957.

4. Hoffert, M.I.; Caldeira, K.; Benford, G.; Criswell, D.R.; Green, C.; Her-zog, H.; Jain, A.K.; Khesghi, H.S.; Lackner, K.S.; Lewis, J.S.; Lightfoot,H.D.; Manheimer, W.; Mankins, J.C.; Mauel, M.E.; Perkins, L.J.;Schlesinger, M.E.; Volk, T.; Wigley, T.M.L. Advanced TechnologyPaths to Global Climate Stability: Energy for a Greenhouse Planet;Science 2002, 298, 981-987.

5. White, C.M.; Strazisar, B.R.; Granite, E.J.; Hoffamn, J.S.; Pennline,H.W. Separation and Capture of CO2 from Large Stationary Sourcesand Sequestration in Geological Formations-Coalbeds and Deep SalineAquifers; J. Air & Waste Manage. Assoc. 2003, 53, 645-715.

6. Shafeen, A.; Croiset, E.; Douglas, P.L.; Chatzis, I. CO2 Sequestration inOntario, Canada. Part 1: Storage of Potential Reservoirs; Energy Con-vers. Mgmt. 2004, 45, 2645-2659.

7. Shafeen, A.; Croiset, E.; Douglas, P.L.; Chatzis, I. CO2 Sequestration inOntario, Canada. Part II: Cost Estimation; Energy Convers. Mgmt. 2004,45, 3207-3217.

8. Bird, R.B.; Stewart, W.E.; Lightfoot, E.N. Transport Phenomena; JohnWiley and Sons: New York, NY, 1960; pp 25-27.

9. Kramer, D.A. Magnesium, Its Alloys and Compounds. In U.S. Geologi-cal Survey Open-File Report; 01-341; U.S. Geological Survey: Reston, VA,2001.

10. Abtahi, M.; Kaasa, B.; Vinstead, J.E.; Ostvold, T. Calcium CarbonatePrecipitation and pH Variations in Oil Field Waters. A Comparisonbetween Experimental Data and Model Calculations; Acta Chem.Scand. 1996, 50, 114-121.

11. Turton, R.; Baile, R.C.; Whiting, W.B.; Wallace B.; Shaeiwitz, J.A.Analysis, Synthesis, and Design of Chemical Processes, 2nd ed.; PrenticeHall: Upper Saddle River, NJ, 2003; pp 343-344.

12. Walas, S.M. Chemical Process Equipment: Selection and Design; Butter-worths: Boston, MA, 1988; p 98.

13. U.S. Department of Energy and Environmental Protection Agency.Carbon Dioxide Emissions from the Generation of Electric Power in theUnited States; PB2001-100635; National Technical Information Service:Springfield, VA, 2000.

14. Xu, Z.; Wang, J.; Chen, W.; Xu, Y. Separation and Fixation of CarbonDioxide Using Polymeric Membrane. In First National Conference onCarbon Sequestration; U.S. Department of Energy, National EnergyTechnology Laboratory: Morgantown, WV, 2001.

15. Trachtenberg, M.C.; McGregor, M.L.; Tu, C.; Laipis, P.J.; Willson,R.C.; Kennedy, J.F.; Paterson, M.; Rudolph, F.B. Enzyme-Enhanced

Figure 9. Two possible strategic options for capturing and storing CO2 based on an approach of storing the CO2 as carbonate mineral (top)or storing it as gas and ions (bottom).



Membranes for Gas Separation. In SAE Technical Series Paper; Society ofAutomotive Engineers: Warrendale, PA, 1999.

16. Ge, J.J.; Trachtenberg, M.C.; McGregor, M.L.; Cowan, R.W. Enzyme-Based Facilitated Transport of Vacuum Induced Sweep for EnhancedCO2 Capture. In SAE Technical Series Paper; Society of AutomotiveEngineers: Warrendale, PA, 2001.

17. Trachtenberg, M.C.; Bao, L.; Goldman, S.L.; Smith, D.A. DynamicMaintenance of CO2 Levels in Closed Environments. In SAE TechnicalSeries Paper; Society of Automotive Engineers: Warrendale, PA, 2004.

18. Bond, G.M.; Stringer, J.; Brandvolt, D.K.; Simsek, F.A.; Medina, M.G.;Egeland, G. Development of Integrated System for Biomimetic CO2Sequestration Using the Enzyme Carbonic Anhydrase; Energ. Fuel2001, 15, 309-316.

19. Simsek-Ege, F.A.; Bond, G.M.; Stringer, J. Polyelectrolyte Cages for aNovel Biomimetic CO2 Sequestration System. In EnvironmentalChallenges and Greenhouse Gas Control for Fossil Fuel Utilization in the21st Century; Maroto-Valer, M.M.; Song, C.; Soong, Y. Eds.; KluwerAcademic: Amsterdam, The Netherlands, 2002; pp 133-146.

20. Liu, N.; Bond, G.M.; Abel, A.; McPherson, B.J.; Stringer, J. BiomimeticSequestration of CO2 in Carbonate Form: Role of Produced Waters andOther Brines; Fuel Process. Technol. 2005, 86, 1615-1625.

21. Lee, A. CO2 Capture Project Update. In MIT Sequestration Forum III—Moving toward a Regulatory Regime; 2002; pp 13-14.

22. Herzog, H.; Golomb, D. Carbon Capture from Fossil Fuel Use. In Encyclo-pedia of Energy; Elsevier Academic: Boston, MA, 2004; pp 277-287.

23. Holloway, S.; Chadwick, A.; Pearce, J. Geological Sequestration—HowSerious a Problem Are Potential Leaks? In MIT Sequestration ForumIII—Moving toward a Regulatory Regime; 2002; pp 13-14.

24. Chopey, N.P. Handbook of Chemical Engineering Calculations; McGraw-Hill: New York, NY, 1984; pp 6-5–6-11.

About the AuthorsDaniel Dziedzic and Kenneth Gross are Technical Fellows inthe Chemical and Environmental Sciences Laboratory ofthe General Motors Research and Development Center inWarren, MI. Robert Gorski and John T. Johnson are staffscientist and staff research scientist, respectively, at theGeneral Motors Research & Development Center. Addresscorrespondence to: Daniel Dziedzic, GM R&D Center,30500 Mound Rd., Mail Code 480-106-269, P.O. Box 9055,Warren, MI 48090-9055; phone: �1-586-986–1656; fax:�1-586-986-1910; e-mail: [email protected].

APPENDIX

Appendix Figure 1. Flow diagram for EMSS (note 14 input fields and solver button).



feasibility study of using brine for carbon dioxide ... ﬂuid in two separate, ... (bicarbonate or...

Documents