Membrane-Mediated Delivery of Carbon Dioxide for Consumption by

CC.C. EEE. III.... JJJ............. LL. MM. NN. OO.. loading. To add this web app to the home screen open the browser option menu and tap on Add to ho...
0 downloads 0 Views 615KB Size
Article pubs.acs.org/IECR

Membrane-Mediated Delivery of Carbon Dioxide for Consumption by Photoautotrophs: Eliminating Thermal Regeneration in Carbon Capture James D. Noel,† William J. Koros,† Benjamin A. McCool,‡ and Ronald R. Chance*,‡ †

School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332, United States Algenol Biofuels Inc., 28100 Bonita Grande Drive, Bonita Springs, Florida 34135, United States



ABSTRACT: Photoautotrophs, such as microalgae and cyanobacteria, use sunlight to convert inorganic carbon, especially dissolved CO2 and bicarbonates, to organic materials, commonly biofuels or biofuel precursors. For large-scale biofuel production, the CO2 abundance in ambient air is not sufficient and CO2 must be delivered in a concentrated form, ideally captured from anthropogenic sources such as flue gas streams from power plants. Carbon dioxide capture and delivery to the photoautotrophic culture can add significantly to the cost of biofuels. Much of that cost can be attributed to the regeneration of the CO2 sorbent, a generally energy intensive process. Herein, a liquid-based CO2 delivery methodology is described and demonstrated. The system involves a conventional CO2 sorption system (exemplified by an amine-functionalized sorbent), and a carbonate stripping solution that produces a bicarbonate-rich solution after contacting the absorber. The latter solution serves as the feed stream for a CO2 selective membrane with the permeate side of the membrane being swept by a seawater solution, designed as a model for a photoautotrophic culture medium. Changes in pH and carbon content of the solutions, as well as the lack of any ionic transport across the membrane, are consistent with molecular transport of CO2. The CO2 flux is shown to be consistent with the chemical potential differential provided by the carbon-rich stripping solution and the carbon-depleted seawater solution. In application to biofuel production, the driving force would be sustained by carbon consumption in the photosynthetic process. Thus, photosynthesis provides much of the energy for the CO2 sorbent regeneration in the form of the regenerated carbonate stripping solution. The overall liquid-based CO2 delivery process described here provides the potential for a more cost-effective biofuel production process.

1. INTRODUCTION AND BACKGROUND The development of clean and renewable energy sources is one of the most challenging technological issues facing society today. A pressing reason to promote the use of carbon neutral fuels to replace fossil fuels and to capture anthropogenic CO2 is mitigating the risk of global climate change. The International Panel on Climate Change (IPCC) reports that global greenhouse gas (GHG) emissions have increased 70% between 1970 and 2004.1 The IPCC found that CO2 is the most important anthropogenic GHG and the annual global CO2 emissions grew by 80% between 1970 and 2004. The potential impact of anthropogenic CO2 production on the environment has led to research and development of methods for capturing it from flue gases from combustion sources.2 Flue gases from U.S. power plants are responsible for approximately 30% of the total U.S. CO2 emissions and can contain up to 15% CO2.1−3 Traditional CO2 capture technologies include absorption by aqueous amines but this technology is energy intensive and expensive when used to capture relatively dilute mixtures of gases. This is primarily due to the high energy use associated with the temperature swings required for CO2 desorption from solution. Many amine-modified silica materials have been investigated for capture of CO2 from gaseous streams;4 there also, recovery is the major energy consumer. Hollow fiber sorbents have been shown to provide an overall energy reduction compared to conventional technologies,5 but even there recovery is the major contributor to energy use and parasitic load.4 Though the removal of CO2 through adsorption processes and subsequent © 2012 American Chemical Society

sequestration can help mitigate greenhouse gas emissions, the photosynthetic conversion of CO2 to biofuel has the potential for comparable mitigation benefits while lessening the demand for nonrenewable energy resources.6 The production of biofuels (ethanol and biodiesel) offers an attractive opportunity to reduce carbon dioxide (CO2) emissions and lower the dependence of the United States on oil imports.6 Microalgae and cyanobacteria (both referred to as algae in this paper) are photoautotrophs and many have been shown to produce ethanol.7 Biofuel can be produced from algae and because they are grown in water rather than soil, their production can be sited on land that is not suitable for agricultural use.6,8 The most commonly discussed approach to algal biofuel is to grow and harvest the algae through a dewatering process to produce biodiesel fuel.9 However the harvesting of algae is both water and energy intensive. A different type of biofuel system has been proposed in which ethanol synthesizing cyanobacteria (blue−green algae) cultures are not harvested but are maintained for continuous ethanol production.6,10 In this project we are developing a novel operational system to capture and deliver CO2 making use of the continuous production of ethanol and the nonharvest aspect of the cyanobacteria-based process. Received: Revised: Accepted: Published: 4673

November 22, 2011 January 16, 2012 February 20, 2012 March 16, 2012 dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681

Industrial & Engineering Chemistry Research

Article

Methods for delivering CO2 to the photobioreactors have been discussed elsewhere,9 including the use of membranes in conjunction with CO2 gas streams.11−14 Our process integrates the most energy intensive part of sorption-based carbon capture (stripping the CO2 from the adsorbate) into the CO2 delivery system. In our system the CO2 is desorbed from the sorbent and the sorbent is regenerated by using sodium carbonatebased aqueous solutions. The CO2 enriched sodium carbonate aqueous stream is exposed to a membrane contactor with a CO2 selective dense film membrane to transmit CO2 to the CO2 depleted, seawater-based algal broth. Once in the algal broth the CO2 is consumed by the algae to produce ethanol (eq 1). In the photosynthetic process of main interest here,6,10 the ethanol is then produced without harvesting the algae. The photosynthetic process provides a sink for CO2 as long as ethanol or biomass is being produced and thus creates a CO2 driving force across the membrane. 2CO2 + 3H2O ↔ C2H5OH + 3O2

The pH and ionic strength of the solution play a crucial role in the total amount of inorganic carbon that can be dissolved into solution. For example at a pH of 8.25 and a partial pressure of CO2 at 380 ppm (the average pH of seawater under ambient conditions and the concentration of CO2 in the air), deionized water has 3.1 times less total dissolved inorganic carbon (DIC) capacity than the total DIC capacity of seawater (12.53 and 38.95 mg/L, respectively). Dissolved inorganic carbon is the sum of the inorganic carbon species (CO2, H2CO3, HCO3−, and CO32−) written in units of mg of carbon per liter. Also the dominate carbon species differs at different pH values between deionized water and seawater. The dominant aqueous carbon species of deionized water is CO2(aq) below a pH 6.35, HCO3− between a pH of 6.35 and 10.33, and CO32− above a pH of 10.33, while the dominant aqueous carbon species in seawater is CO2(aq) below a pH 5.85, HCO3− between a pH of 5.85 and 8.92, and CO32− above a pH of 8.92. Carbon concentrations are calculated from solubility products defined by Zeebe and WolfGladrow.19 By controlling the pH, ionic strength, sodium, and initial carbon concentrations, one can create a solution that can both desorb CO2 from a solid sorbent to become carbon enriched and be able to create a CO2 concentration gradient that can be used to deliver CO2 to a receiving reservoir. The CO2 enriched aqueous stream is then passed over a CO2 selective membrane which is swept on the permeate side with a receiving solution stream. A typical CO2 selective silicone membrane with a high CO2 solubility and permeability is polydimethylsiloxane (PDMS) which has a permeability of 3800 barrer (1.27 × 10−12 mol/m-s-Pa) at 35 °C.20 The source of the receiving solution stream can be seawater broth depleted in CO2 by photoautotrophs found in photobioreactors (PBR). To prevent the formation of biofilm onto the dense film membrane (PDMS), the broth removed from the reactor should be first passed through micro- or ultrafiltration membrane to remove the microorganisms. The carbonate chemistry in water allows sea algae to deplete the CO2 concentration through the consumption of CO2 and HCO3−.21 The consumption of CO2 by the algae to form ethanol or other products provides for a CO2 driving force across the PDMS membrane. A conceptual diagram of the overall setup is illustrated in Figure 1. The process is similar to the Benfield process in which hot carbonate solutions are used to scrub CO2 from acid gases. However, in this case a photobioreactor is used in place of a regenerator.22 The CO2 adsorption column will work in two separate modes. The first mode (1) CO2 from the feed stream (a), in this case a CO2 rich flue gas, is adsorbed by the sorbent. The exit stream in mode 1 (b) is a CO2 stripped flue gas. In the second mode (2) the CO2 is desorbed from the solid sorbent by a water−carbonate stream (d) in which the sodium carbonate (Na2CO3) reacts with the surface and removes complexed CO2 held as RNHCOO− to form sodium bicarbonate consistent with eq 2. The sodium bicarbonate reacts to a dissolved inorganic form and becomes enriched in dissolved inorganic carbon. This enriched carbon stream (c) is delivered to the water−carbonate reservoir. After some time the water− carbonate reservoir exit stream (e) is pumped in contact with the membrane contactor in which aqueous CO2 dissolves into the dense film membrane. The membrane retentate (f) now has less CO2 corresponding to a lower DIC concentration than stream e. The membrane permeate (g) is now enriched in CO2 corresponding to a higher DIC concentration than the feed stream (h) from the receiving solution reservoir. After some time, the receiving solution reservoir reaches a DIC concentration

(1)

The CO2 delivery concept is based on the chemical reaction of carbonate with CO2 and water to form bicarbonate. Aqueous based carbonate scrubbing has been investigated for capture of CO2 from flue gas and it compares favorably with liquid amines. As with liquid amines, carbonate scrubbing of air for removal of CO2 is not feasible due to solubility limits of carbonates in water and the low capacities at dilute CO2 concentrations found in air.15 The reaction of sodium carbonate to form sodium bicarbonate is illustrated in eq 2. The heat of reaction (ΔHrxn) of reaction 2 was calculated to be −61.6 kJ/mol using heats of formation of all species.16 Na2CO3 + H2O + CO2 ↔ 2NaHCO3

(2)

− 2RNH2 + CO2 ↔ RNH+ 3 + RNHCOO

(3)

The regeneration of an amine-based sorption system involves the removal of CO2 from the carbamate complex that is formed during CO2 capture. The carbamate formation reaction is shown in eq 3. The heat of reaction of reaction 3 for primary amines is reported to be approximately −80 kJ/mol.17,18 − Na2CO3 + H2O + RNH+ 3 + RNHCOO

↔ 2RNH2 + 2NaHCO3

(4)

The net heat of reaction for the removal of CO2 from loaded sorbent into the aqueous solution is the difference in reaction energies of reaction 2 minus reaction 3, producing a low regeneration energy requirement of about 18.4 kJ/mol (reaction 4). This approach also utilizes the chemistry of bicarbonate conversion to drive CO2 into the aqueous phase. The sodium bicarbonate (NaHCO3) will disassociate and react to form a mixture of aqueous CO2, H2CO3(aq), HCO3−, CO32−, Na+, NaHCO3(aq), and Na2CO3(aq) (reactions 5 and 6) dependent on the ionic strength of solution, total concentration of sodium, total concentration of inorganic carbon, and pH. NaHCO3 ↔ Na+ + HCO− 3

(5)

H2O + CO2 ↔ H2CO3 ↔ H+ + HCO− 3 ↔ 2H+ + CO32 −

(6) 4674

dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681

Industrial & Engineering Chemistry Research

Article

Figure 1. Conceptual diagram and carbon streams of the overall process. The first operational mode of the adsorption column, labeled (1), is sorption of CO2 from a flue gas feed stream. The second operational mode, labeled (2), is the desorption of CO2 from the sorbent. A detailed description of the process is found in the text (Section 1).

Ocean salts with deionized water to simulate photobioreactor broths. 2.2. Membrane Synthesis. A polydimethylsiloxane (PDMS) of typically 90 μm thickness was used as a CO2 selective membrane to transfer aqueous CO2 between a bicarbonate (CO2 rich) aqueous solution to a CO2 lean aqueous solution. The PDMS membrane was synthesized by first weighing out a 9:1 wt/wt ratio of silicone elastomer base to curing agent and dissolving in heptane to make a 20 wt %/wt solution of PDMS in heptane. The solution of 20 wt %/wt PDMS in heptane was then pipetted onto a Teflon plate and placed in a vacuum oven at 80 °C for 2 hours. After evaporating the heptane, the PDMS film was left to cool at room temperature and subsequently placed into a permeation cell to determine gas permeability and selectivity. The permeability of the PDMS film was determined from gas transport measurements with pure gases at 22 °C (CO2 = 2170, O2 = 398, N2 = 206 barrer) [Barrer = 10−10 cm3 (STP) cm/cm2-s-cmHg)]. The gas selectivities were then calculated from the permeabilities (αO2/N2 = 1.9, αCO2/O2 = 5.5, αCO2/N2 = 10.5). The selectivities are consistent with the formation of a defect-free PDMS membrane.23 2.3. CO2 Delivery. A membrane contactor with fluids on each side was used to measure CO2 transfer across a selective membrane. The PDMS membrane was supported inside the membrane contactor by two highly porous (2 mm pores) metal supports (0.7 mm thick) (Figure 2A). The two fluids (counter current flow) were a bicarbonate (CO2 rich) aqueous solution transferring CO2 to a receiving (CO2-poor) downstream solution. A conceptual diagram of the experimental setup is given in Figure 2B. A 35 mg/L DIC solution was used to simulate the amount of DIC that would be found in the algal broth in a closed seawater based photobioreactor. Each solution reservoir was contained in a 1 L Tedlar bag with two polypropylene 2in-1 valves with a septum valve and a 1/8 in. fitting. The Tedlar bags were used to eliminate CO2 transfer from the atmosphere to solution and from the solution to the atmosphere. Each solution was prepared by weighing the appropriate amount of NaHCO3 and placing the sample inside the bag. A vacuum was then applied to the bag to remove access air. Then 500 mL of

usable by microorganisms in the photobioreactor (PBR). The PBR feed (i) contains carbon that is consumed to make ethanol or other biofuels. The biofuel is removed (l) from the PBR and processed further. The PBR algal broth (j) is then pumped through a filter (a micro- or ultrafiltration membrane) in which a return stream (k) is enriched in biosolids and returned back to the reactor while a biosolids free stream (m) is returned to the receiving solution reservoir to be enriched in carbon via contact with the membrane contactor. In this project we focus on the transfer of CO2 across the dense film membrane from the water− carbonate solution reservoir to the receiving solution reservoir, and the ability to regenerate the solid CO2 sorbent using viable stripping solutions. In future work, we will address the operation and analysis of the other process components in Figure 1. We will address this system in two parts: (i) CO2 delivery from one aqueous solution to another using a CO2 selective membrane and (ii) CO2 capture onto a solid adsorbent and desorption using a sodium carbonate aqueous solution.

2. MATERIALS AND METHODOLOGY 2.1. Instrumentation and Materials. A total carbon (TC) analyzer (Shimadzu TOC-VCPH) was used to determine dissolved inorganic carbon concentrations in aqueous solutions. An inorganic carbon calibration was performed daily before each measurement using carbon standard solutions. A quadrupole mass spectrometer (Pfeiffer Vacuum OminStar Gas Analysis System) was used to determine the gas concentrations during the carbon dioxide sorption studies. A Mettler Toledo pH meter was used to measure the pH of the aqueous solutions in real time, while peristaltic pumps were used to circulate the aqueous solutions. All chemicals used in experiments were ACS grade or better. Water used in the experiments was deionized (>15 MΩ•cm) using a Millipore Elix. Sylgard 184 Silicone Elastomer Base and Curing Agent were used to synthesize the polydimethylsiloxane (PDMS) membranes. The 3-aminopropyltriethoxysilane (APTES) functionalized glass beads (30−50 μm) were purchased from Polyscience, Inc. (Cat. no. 23584) and used as the CO2 sorbent. Synthetic seawater was made by mixing Instant 4675

dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681

Industrial & Engineering Chemistry Research

Article

Figure 2. (A) Membrane contactor design in which the membrane is supported by two porous metal supports. (B) Experimental setup of a liquid/ liquid membrane contactor.

deionized water was pumped into the bag and thoroughly mixed with an embedded stir bar. Afterward 100 mL of the stripping solution was pumped out of the bag for a final volume of 400 mL and used for initial measurements. Each aqueous solution was thoroughly mixed using stir bars in each bag. During the experiment each solution was pumped at a flow rate of 2, 10, or 20 mL/min to investigate the effect of flow rate and examine the impact of external mass transfer resistances within the membrane contactor. For all experiments, the flow rates investigated were countercurrent and numerically the same on either side of the membrane. A pH probe was placed in the receiving solution to monitor the change in pH. The change in pH is due to the uptake of CO2 into solution. At periodic intervals the aqueous solutions were sampled though the septum valves using a syringe to determine the dissolved inorganic carbon concentration using a total organic carbon (TC) analyzer (Shimadzu TOC-VCPH). Sodium concentration was measured with a sodium ion probe in each grab sample. No sodium was observed to transfer across the PDMS membrane. 2.4. CO2 Adsorption and Desorption. A packed bed column with a 3-aminopropyltriethoxysilane (APTES) sorbent was used to remove CO2 from a mixed CO2/N2 gas stream. A conceptual diagram of the experimental setup is illustrated in Figure 3. A 7 mL column was first filled with 1.3 g of 3 mm of glass beads, followed by 1 g of the amine-grafted glass bead sorbent purchased from Polyscience, Inc. and an additional 1.7 g of 3 mm glass beads on top of the sorbent. Column frits (10 μm filters) at the top and bottom of the column prevented the movement of glass beads and sorbent. CO2 sorption was not observed to occur on the unfunctionalized glass beads. A 12.5% mol/mol CO2 (the rest N2) stream was fed into the column at a rate of 0.05 SCFH (standard cubic feet per hour) (22.39 mL/min). The amount of CO2 in the exit stream was measured as a function of time with a quadrupole mass spectrometer. After CO2 adsorption, valves were opened to allow an aqueous solution to pass over the sorbent to desorb the CO2. Each aqueous solution was prepared in a Tedlar bag by weighing out the necessary combination of NaOH and NaHCO3 to obtain the desired pH and DIC. Deionized water (115 mL) was then pumped into the bag reactor and mixed to eliminate CO2 transfer from the solution, similar to the preparation in the liquid/liquid membrane experiment. The solution was thoroughly mixed with a stir bar in the bag reactor, and 15 mL of

Figure 3. Adsorption column experimental setup.

solution was used to determine the initial pH and DIC concentrations resulting in a final water reservoir volume of 100 mL. The aqueous solution was pumped from the bottom of the column to top at a rate of 2 mL/min and returned back to the water reservoir. Dissolved inorganic carbon concentrations were determined using the TC analyzer and measured as a function of time. 2.5. Integrated CO2 Adsorption and Desorption with CO2 Delivery. An integrated experiment was used to demonstrate CO2 adsorption and desorption coupled with CO2 delivery via a membrane contactor. A packed bed column with APTES sorbent was used to capture CO2 from a gaseous stream. The 17.4 mL column was first filled with 0.5 g of glass wool then 1 g of 3 mm glass beads. Then 10 g of the APTES sorbent was placed on top of the glass beads and another 1 g of 3 mm glass beads was placed on top of the APTES sorbent. A 12.0% mol/mol CO2 (the rest N2) stream was fed into the column at a rate of 0.05 SCFH (22.39 mL/min). The amount of CO2 in the exit stream was measured using a quadrupole mass spectrometer and the amount of CO2 adsorbed was calculated using a breakthrough curve. The adsorption of CO2 by the amine sorbent was very fast, occurring within 20 min. After an hour of flowing gaseous mix of CO2/N2, the gas valves were closed and the NaHCO3 stripping solution was flowed over the column to remove CO2 from the APTES sorbent. A NaHCO3 stripping solution was made similarly to that outlined in Section 4676

dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681

Industrial & Engineering Chemistry Research

Article

2.3. The NaHCO3 stripping solution was in contact with the sorbent for 24 h while 1.5 mL grab samples were used to determine the DIC concentration over time. After 24 h the flow of the NaHCO3 stripping solution was then redirected to the membrane contactor. The receiving water solution, on the permeate side of the membrane contactor, was a CO2-poor seawater solution used to simulate the algal broth in a photobioreactor. The receiving aqueous solution was made by placing 17.97 g of Instant Ocean inside a Tedlar bag reactor similar to Section 2.3. Then 500 mL of boiling deionized water was pumped into the bag to dissolve the Instant Ocean salts. The DI water was boiled to minimize uptake of ambient CO2. Afterward, solution was pumped out of the bag reactor and used for initial measurements leaving a final volume of 400 mL. The synthetic seawater and the NaHCO3 stripping solution were in contact with the PDMS membrane for about 24 h. Grab samples were removed from the synthetic seawater solution to determine the DIC concentration over time.

Figure 5. Real time pH measurements as a function of time in the receiving solution as a function of flow rate.

of the DIC and pH values for a 2 mL/min flow rate to those for a 20 mL/min flow rate). The observed change in pH decreases to an approximate value of 0.002 pH units per hour and steadily decreases due to the decrease in the driving force. The increase in DIC concentration as indicated by the change in pH agree well with the grab samples measured by the TC analyzer (39.6, 41.4, and 46.1 mg/L for 2, 10, and 20 mL/min, respectively). A full comparison of the two measures is given in Figure 6 and shows reasonably good agreement. The lingering differences in

3. RESULTS AND DISCUSSION 3.1. CO2 Delivery. A liquid/liquid membrane contactor experiment was used to determine the viability of a CO2 selective membrane to transfer aqueous CO2 between a bicarbonate (CO2 rich) aqueous solution to a receiving downstream solution which was CO2 poor. A conceptual diagram of the experimental setup is illustrated in Figure 2. To determine the viability of this CO2 delivery method, a solution of 1 M NaHCO3 was pumped into contact with membrane contactor. The downstream solution was an aqueous solution with a low dissolved inorganic carbon (DIC) concentration, ≈ 35 mg/L solution made from DI water and the appropriate amount of NaHCO3. Solutions were prepared as described in the experimental section. The initial DIC concentration of each of the receiving solutions was essentially the same: 34.9, 34.6, and 35.1 mg/L for the flow rates of 2, 10, 20 mL/min, respectively (Figure 4).

Figure 6. Comparison of TOC analyzer measured dissolved inorganic carbon (DIC) concentration (open symbols) and pH derived DIC (closed symbols) at the three flow rates.

DIC concentrations at long times are an indication of significant external mass transfer resistances in the membrane contactor. In fact, in the absence of external mass transfer resistance, all three curves in Figure 5 (after flow rate normalization) should superimpose at all time scales, the changes with time reflecting the decrease in driving force across the membrane in these “batch” experiments. Future contactor designs will minimize these mass transfer resistances. We will return to this issue later in the discussion. In each case studied, the DIC concentration rapidly increases but levels off approaching an asymptotic value. The rate of change continually decreases due to the decrease in driving force. This indicates that even though there still is a significant difference in the amount of DIC in each reservoir (≈ 1 to 3.5 × 10−3 M in the upstream and downstream reservoirs, respectively) the CO2 driving force across the membrane decreases substantially over time. We investigated the change in the CO2 chemical potential

Figure 4. Dissolved inorganic carbon (DIC) concentration in the receiving solution as a function of time at different flow rates.

The final DIC concentration for each flow rate condition was 40.8, 42.9, and 45.5 mg/L for 2, 10, and 20 mL/min, respectively, after 30 h of operation. The pH decreased in the receiving solution due to the uptake of CO2 (Figure 5). The DIC and pH for all three flow rates will eventually approach the same value. The difference observed at the 2000 min mark is due to the time response which is a function of the flow rate and the mass transfer resistances in the membrane contactor. (Experiments run to 80 h, not shown, reveal a close approach 4677

dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681

Industrial & Engineering Chemistry Research

Article

on both sides of the membrane contactor. The effective concentrations or activities (αi) of dissolved species are defined as o αi = γm i i /m

The difference between the upstream and downstream chemical potentials is D ΔμCO2 = μU CO2 − μCO2

(7)

o o = μCO 2 + RT ln αCO2, U − μCO2

where γi is the activity coefficient and mi is the molal concentration of species i. The standard state of molality (mo) is 1 mol/kg so that eq 6 can be rewritten as αi = γimi. The chemical potential (μi) of individual dissolved species is μi = μio + RT ln αi

− RT ln αCO2, D

ΔμCO2 = RT ln

(8)

1 Σ(miz i) 2

(9)

where Zi is the charge of species i. The activity coefficient (γi) is strongly dependent on ionic strength (I), which is dependent on the concentration of anions and cations. Other factors that affect the activity coefficient include temperature, density of water, dielectric constant of water, and effective size of hydrated ions. The Debye−Huckel equation (eq 10) was used to solve for the activity coefficients of charged dissolved aquatic species.24 log γi =

−Az i2 I 1 + Bai I

(10)

where A = 1.824928x106ρo1/2(εT)−3/2 and B = 50.3(εT)−1/2, ρo is the density of water, ε is the dielectric constant of water, T is temperature in Kelvin, Zi is the ionic charge of species i, and ai is the ion size parameter of species i. Activity coefficients for uncharged, molecular species obey the empirical Setchenow equation up to high ionic strengths and generally include dissolved gases, weak acids, and molecular organic species. The Setchenow equation (eq 11) is24 log γi = K iI

(11)

A 20 mL/min the downstream solution initially has an ionic strength of 2.94 × 10−3 mol/kg which leads to the activity coefficient of CO2 (γCO2,D) of 0.9984 and a molar concentration of CO2 equal to 3.12 × 10−5 M and a CO2 activity (αCO2,D) of 3.115 × 10−5. For the downstream solution at the final sample point (DIC = 46 mg/L), the ionic strength is 2.917 × 10−3 mol/kg which leads to the CO2 activity coefficient (γCO2,D) of 0.9984 and a molar concentration of CO2 equal to 9.11 × 10−4 M or a CO2 activity (αCO2,D) of 9.095 × 10−4. Similarly, for the upstream solution the initial (t = 0) ionic strength is 0.707 mol/kg, the γCO2,U = 0.6866, γCO2,D = 1.69 × 10−2, and initial CO2 concentration =2.46 × 10−2 M. For the upstream solution for the final data point the following are calculated for a CO2 concentration of 2.41 × 10−2 M: γCO2,U = 0.6861 and the γCO2,D = 1.47 × 10−2. From the calculated CO2 activities, the chemical potential of CO2 in both the upstream and downstream can be described as o μU CO2 = μCO2 + RT ln αCO2, U

(12)

D o μCO 2 = μCO2 + RT ln αCO2, D

(13)

αCO2, U αCO2, D

(15)

Equation 15 describes the driving force at initial conditions assuming a perfectly mixed solution. Initially the ratio of the upstream to downstream activity is (αCO2,U)/(αCO2,D) = 540 while after 24 h the ratio of the upstream to downstream activity decreases to (αCO2,U)/(αCO2,D) = 16. As can be seen the initial activity of CO2 in the upstream is 540 times greater than the activity of CO2 in the downstream. As the experiment progresses the initial activity of CO2 in the upstream decreases to be 16 times greater than the activity of CO2 in the downstream. This is a 97% decrease in driving force which explains, to a large degree, the decrease in CO2 flux across the membrane and indicates that only CO2 crosses the membrane. As noted earlier, the later conclusion was confirmed by the absence of sodium ion transport during the experiment. An important design parameter, relevant to potential for process scale up and commercial deployment, is the CO2 permeance of the membrane. In the case of the fastest flow rate (20 mL/min) there is a change in DIC concentration of 0.25 mg/L within the first 15 s, which results in an initial CO2 flux of 0.134 mmol/m2 sec and a permeance of about 5 GPU [GPU = 10−6 cm3 (STP)/(cm2-s-cmHg)]. In comparison to the expected permeance of 24 GPU determined from gas transport measurements on the same PDMS films (Section 2.2), i.e., without liquid mass transfer resistances, the impact of the external mass transfer resistance is about a factor of 5 reduction in CO2 flux. The magnitude of this effect is at least qualitatively consistent with the observations and discussion in Section 3.1. After 1 h under the same conditions, the flux drops to 0.017 mmol CO2/m2-s and to about 6 × 10−5 mmol CO2/m2-s after 24 h due primarily to changes in the CO2 driving force across the membrane. In the envisioned application of this system, the decreasing flux is not an issue as the photosynthetic activity of the photoautotrophs will maintain a strong driving force for CO2 transport. However, significant reduction in external mass transfer resistance will be necessary for practical application of this system. Algenol biofuels is looking to produce a target of 6000 gallons of ethanol per acre per year which corresponds to about 2% solar energy conversion.6 This CO2 demand would require about 2100 mol CO2 per acre per day. Assuming a CO2 permeance consistent with elimination of external mass transfer resistance (i.e., 24 GPU) and continuous delivery for 12 h of daylight each day, the system would require about 80 m2 of membrane surface per acre or about 0.013 m2 per gal of ethanol produced per year. At $100 per m2 for the full membrane unit cost, the capital investment would be about $1.3 on this per gal basis. Amortized over 15 years, the contribution to the cost of a gallon of ethanol would be about $0.09, a cost that would be offset, at least partially, by the cost savings in the CO2 capture system. To set the scale, a $50/tonne cost for CO2 capture1 translates, stoichiometrically, to about $0.30 per gallon of ethanol. Though this is a very rough estimate and does not

where μio is the chemical potential of the species at a reference condition. The ionic strength (I) of the solution is defined as

I=

(14)

4678

dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681

Industrial & Engineering Chemistry Research

Article

exit stream was measured as a function of time using a quadrupole mass spectrometer and the amount of CO2 adsorbed was calculated from the breakthrough curve. The amount of CO2 adsorbed was 1.54 mg as C. After an hour of flowing gaseous mix of CO2/N2 the gas valves were closed and the NaHCO3 stripping solution was flowed over the column to remove CO2 from the APTES sorbent. The initial DIC concentration in the stripping solution was measured to be 33.76 mg/L. The stripping solution was in contact with the sorbent for 24 h while 1.5 mL grab samples were used to determine the DIC concentration over time (Figure 7). The final DIC

include network piping costs for gas and liquid transfers, the results are encouraging. 3.2. CO2 Adsorption and Desorption. A packed bed column with an amine-grafted sorbent (3-aminopropyltriethoxysilane, APTES) was used to remove CO2 from a gaseous stream. The CO2 loaded sorbent was then stripped with an aqueous solution to assess the viability of desorbing CO2 from the sorbent using aqueous solutions. Different pH values and DIC concentrations of the stripping solution were used to probe what would be the best condition for removing CO2 from the sorbent. A conceptual diagram of the experimental setup is illustrated in Figure 3. The amount of CO2 in the exit stream was measured as a function of time and the amount of CO2 adsorbed was calculated using a breakthrough curve. The adsorption of CO2 to the amine is very fast and occurs within 10 min. The average sorption capacity of the sorbent was determined to be 1.024 mg CO2 as C/g of sorbent based on a single adsorption/regeneration cycle. The working adsorption capacity will be less than 1.0 mg CO2 as C/g sorbent and should be taken into consideration when scaling up this process. Valves were then opened to allow an aqueous solution to pass over the sorbent to desorb the sorbed CO2. Each aqueous solution was prepared as stated in the experimental setup. The aqueous solution was pumped from the bottom of the column to top at a rate of 2 mL/min and returned back to the water reservoir. Both external mass transfer in bulk aqueous solution and the reaction kinetics at the surface are important factors in determining the overall rate of CO2 elutriated, and in the future we will evaluate their relative importance. However, our focus was probing the effect of pH on the effective equilibrium of the elutriation of CO2 to begin system design feasibility. A mass balance on the amount of carbon uptake in the solution is used to calculate the total amount desorbed from the APTES sorbent (Table 1). The results indicate that at higher pH values

Figure 7. Concentration of DIC as a function of time in the NaHCO3 stripping solution in contact with the APTES sorbent. The initial concentration was 33.76 mg/L and the final concentration was 37.18 mg/L.

concentration was measured to be 37.18 mg/L. The resulting 3.42 mg/L increase in DIC concentration corresponds to stripping 1.37 mg of CO2 from the sorbent or 88.9% removal. After 24 h the flow of the NaHCO3 stripping solution was then redirected to the membrane contactor. The receiving water solution on the permeate side of the membrane contactor was a CO2-poor synthetic seawater solution used to simulate the seawater in a photobioreactor (algal broth). The receiving aqueous solution was synthetic seawater solution with an initial DIC concentration of 24.28 mg/L and a pH of 8.28. The synthetic seawater and the NaHCO3 stripping solution were in contact with the PDMS membrane for 24 h. Grab samples were removed from the synthetic seawater solution to determine the DIC concentration over time (Figure 8). The final DIC concentration in

Table 1. Percentage of Carbon Dioxide Desorbed from the Sorbent under Different Initial Conditions DIC (mg/L) pH

1.0

35

7 8.25 seawater 8.25 10 12

19.4%

20.9% 52.1% 57.4% 50.9% 59.4%

55.3% 80.8%

more of the CO2 can be desorbed from the APTES sorbent. However, at higher pH values the stripping solution has less of a CO2 driving force to promote the transfer of CO2 across the PDMS membrane. 3.3. Integrated Experiment (CO2 Adsorption/Desorption and CO2 Delivery). An experiment was used to demonstrate CO2 adsorption and desorption coupled with CO2 delivery via a membrane contactor. This experiment integrated CO2 capture from flue gases using an APTES sorbent, the CO2 desorption from the sorbent using a NaHCO3 solution, and used the membrane contactor to transfer CO2 from the stripping solution to a CO2 poor aqueous solution in a sequential fashion. In this case, the solution used to strip CO2 from the APTES sorbent is used to transfer CO2 to a CO2 poor synthetic seawater solution to demonstrate the technical feasibility of this system. The packed bed column with APTES sorbent was used to capture CO2 from a gaseous stream. The amount of CO2 in the

Figure 8. Concentration of DIC as a function of time in the synthetic seawater solution. The initial concentration was 24.28 mg/L and the final concentration was 25.71 mg/L.

the seawater was 25.71 mg/L resulting in a 1.43 mg/L increase in DIC concentration. The increase in DIC concentration corresponds to 0.57 mg of CO2 delivered to the synthetic seawater solution from the captured CO2 APTES sorbent. 4679

dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681

Industrial & Engineering Chemistry Research

Article

nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

In practice the CO2 in the seawater would be consumed by the photoautotrophs to produce a biofuel. The continuous production and removal of the biofuel provides a nearly continuous CO2 driving force. This experiment demonstrates a method of delivering CO2 from gaseous phase to aqueous phase in a form that can be consumed by photoautotrophs during photosynthetic processes using a solid sorbent and a gas permeable membrane contactor.

4. CONCLUSIONS In this work we have demonstrated a method for delivering carbon which lowers the energy required in comparison to steam stripping regeneration of the CO2 sorbent. The system described uses a conventional CO2 sorption system, and a carbonate stripping solution that produces a bicarbonate-rich solution after contacting the absorber. The stripping solution then serves as the feed stream for a CO2 selective membrane with the permeate side of the membrane being swept by a seawater solution, designed as a model for a photoautotrophic (algal broth) culture medium. The changes in pH and carbon content of the solutions, the lack of any sodium ion transport across the membrane, and the demonstrated change in CO2 chemical potential are consistent with molecular transport of CO2 across the membrane. The percent delivery of captured CO2 to the photoautotrophic culture medium solution was 37%. This capture percentage can be improved by optimizing the carbon flux across the membrane via membrane optimization and minimization of liquid mass transfer resistances. Also, this method can be adjusted to use traditional amine based solutions (for example methylethanolamine, MEA or triethanolamine, TEA) for CO2 capture from flue gas as long as the CO2 selective membrane prevents the amine from passing through. Currently we are investigating this technique for CO2 capture and delivery to photobioreactors through dense film membranes. In application to biofuel production, the CO2 driving force across the membrane would be sustained by carbon consumption in the photosynthetic process providing a nearly continuous CO2 driving force in contrast to the batch mode experiments herein. Photosynthesis provides much of the energy needed for the CO2 sorbent regeneration in the form of the regenerated carbonate stripping solution. Thus, the overall liquid-based CO2 delivery process described here provides the potential for a more cost-effective biofuel production process. In remains to be seen if capital investments are low enough for economic development of such a system and we are currently exploring that question.





REFERENCES

(1) Solomon, S., Qin, D.; Manning, M.; Chen, Z.; Marquis, M.; Averyt, K. B.; Tignor, M.; Miller, H. L. IPCC 2007: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change; Cambridge University Press: Cambridge, United Kingdom and New York, 2007. (2) Aaron, D.; Tsouris, C. Separation of CO2 from Flue Gas: A Review. Sep. Sci. Technol. 2005, 40 (1−3), 321−348. (3) Choi, S.; Drese, J. H.; Jones, C. W. Adsorbent Materials for Carbon Dioxide Capture from Large Anthropogenic Point Sources. Chemsuschem 2009, 2 (9), 796−854. (4) Lively, R. P.; Chance, R. R.; Koros, W. J. Enabling Low-Cost CO2 Capture via Heat Integration. Ind. Eng. Chem. Res. 2010, 49 (16), 7550−7562. (5) Lively, R. P.; Chance, R. R.; Kelley, B. T.; Deckman, H. W.; Drese, J. H.; Jones, C. W.; Koros, W. J. Hollow Fiber Adsorbents for CO2 Removal from Flue Gas. Ind. Eng. Chem. Res. 2009, 48 (15), 7314−7324. (6) Luo, D. X.; Hu, Z. S.; Choi, D. G.; Thomas, V. M.; Realff, M. J.; Chance, R. R. Life Cycle Energy and Greenhouse Gas Emissions for an Ethanol Production Process Based on Blue-Green Algae. Environ. Sci. Technol. 2010, 44 (22), 8670−8677. (7) Deng, M. D.; Coleman, J. R. Ethanol Synthesis by Genetic Engineering in Cyanobacteria. Appl. Environ. Microbiol. 1999, 65 (2), 523−528. (8) Weyer, K. M.; Bush, D. R.; Darzins, A.; Willson, B. D. Theoretical Maximum Algal Oil Production. Bioenergy Res. 2010, 3 (2), 204−213. (9) Greenwell, H. C.; Laurens, L. M. L.; Shields, R. J.; Lovitt, R. W.; Flynn, K. J. Placing Microalgae on the Biofuels Priority List: A Review of the Technological Challenges. J. R. Soc. Interface 2010, 7 (46), 703− 726. (10) Woods, R. P.; Legere, E.; Moll, B.; Unamunzaga, C.; Mantecon, E. Closed Photobioreactor System for Continued Daily in situ Production, Separation, Collection, and Removal of Ethanol from Genetically Enhanced Photosynthetic Organisms. U.S. Patent #7682821, 2007. (11) Ai, W.; Guo, S.; Qin, L.; Tang, Y. Development of a groundbased space micro-algae photo-bioreactor. Adv. Space Res. 2008, 41 (5), 742−747. (12) Cheng, L. H.; Zhang, L.; Chen, H. L.; Gao, C. J. Carbon dioxide removal from air by microalgae cultured in a membrane-photobioreactor. Sep. Purif. Technol. 2006, 50 (3), 324−329. (13) Kim, H. W.; Marcus, A. K.; Shin, J. H.; Rittmann, B. E. Advanced Control for Photoautotrophic Growth and CO2-Utilization Efficiency Using a Membrane Carbonation Photobioreactor (MCPBR). Environ. Sci. Technol. 2011, 45 (11), 5032−5038. (14) Kumar, A.; Yuan, X.; Sahu, A. K.; Ergas, S. J.; Van Langenhove, H.; Dewulf, J., A hollow fiber membrane photo-bioreactor for CO2 sequestration from combustion gas coupled with wastewater treatment: A process engineering approach. J. Chem. Technol. Biotechnol. 85, (3), 387-394. (15) Knuutila, H.; Svendsen, H. F.; Anttila, M. CO2 Capture from Coal-Fired Power Plants based on Sodium Carbonate Slurry; A Systems Feasibility and Sensitivity Study. Int. J. Greenhouse Gas Control 2009, 3 (2), 143−151.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We acknowledge Christopher Jones, Cheng Chen, Ke Zhang, and Ryan Lively for their helpful discussions and technical suggestions. We acknowledge Jason Ward and Cyrus Rashedi for helpful discussions regarding the Algenol process. This material is based upon work supported by the Department of Energy under Award DE-FOA-0000096 and also supported by Algenol Biofuels. This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, 4680

dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681

Industrial & Engineering Chemistry Research

Article

(16) Felder, R. M.; Rousseau, R. W. Elementary Principles of Chemical Processes, 3rd ed.; Wiley: New York, 2005; pp 751−787. (17) Kim, I.; Svendsen, H. F. Heat of Absorption of Carbon Dioxide (CO 2 ) in Monoethanolamine (MEA) and 2-(Aminoethyl)ethanolamine (AEEA) Solutions. Ind. Eng. Chem. Res. 2007, 46 (17), 5803−5809. (18) Kim, I.; Hoff, K. A.; Hessen, E. T.; Haug-Warberg, T.; Svendsen, H. F. Enthalpy of Absorption of CO2 with Alkanolamine Solutions Predicted from Reaction Equilibrium Constants. Chem. Eng. Sci. 2009, 64 (9), 2027−2038. (19) Zeebe, R. E.; Wolf-Gladrow, D. A. CO2 in Seawater: Equilibrium, Kinetics, Isotopes; Elsevier: Amsterdam, New York, 2001; pp 251−255. (20) Merkel, T. C.; Bondar, V. I.; Nagai, K.; Freeman, B. D.; Pinnau, I. Gas Sorption, Diffusion, and Permeation in Poly(dimethylsiloxane). J. Polym. Sci., Part B: Polym. Phys. 2000, 38 (3), 415−434. (21) Coleman, J. R.; Colman, B. Inorganic Carbon Accumulation and Photosynthesis in a Blue-Green-Alga as a Function of External pH. Plant Physiol. 1981, 67 (5), 917−921. (22) Bartoo, R. K. Removing Acid Gas by the Benfield Process. Chem. Eng. Prog. 1984, 80 (10), 35−39. (23) Baker, R. W. Membrane Technology and Applications, 2nd ed.; McGraw-Hill: New York, 2004; pp 301−309. (24) Langmuir, D. Aqueous Environmental Geochemistry; Prentice Hall: Upper Saddle River, NJ, 1997; pp 140−144.

4681

dx.doi.org/10.1021/ie2027124 | Ind. Eng. Chem. Res. 2012, 51, 4673−4681