Carbon Capture Using Amine-modified Carbon Nanotubes

The increasing rise in the rate of both anthropogenic CO2 emissions and atmospheric CO2 concentration, combined with an evolving understanding of what this means for global climate change both now, and in the future, has increased the need for novel technologies for CO2 capture from large point sources. Current industry standards involve the use of amine solvents, which have a large parasitic load, primarily due to the heating and evaporation of water. A robust, inexpensive sorbent material that can operate under flue gas conditions (8 to 12% CO2 with high humidity) could have a large impact on the economics of CO2 capture. For this purpose, multi-walled carbon nanotubes (MWNTs) were functionalized with the aminosilane compounds aminopropyltriethoxysilane (APTES) and N-dimethylaminopropyltrimethoxy-silane (DMAPS) and the resulting materials were characterized and tested for CO2 capture under relevant conditions. Amine loadings were determined to be larger by comparison of silane grafting groups using X-ray Photelectron Spectroscopy (XPS), when primary amines (APTES) were used. The morphology of the MWNTs with functionalization did not seem to be significantly altered. Porosity measurements indicate insignificant change in both the inner micropore space of the MWNTs and the larger mesopore space due to MWNT bundling. Experiments of CO2 breakthrough indicated that APTES functionalized MWNTs captured 1.2 mmol CO2 g sorbent under dry conditions, with an expected decrease in the presence of water. Although the DMAPS functionalized MWNTs were shown to not effectively capture CO2 under dry or humid conditions, important lessons were learned regarding the need for a novel functionalization procedure, an increase in the loading of the tertiary amine DMAPS and a need for a breakthrough experiment with higher humidity. Introduction The purpose of this research project is to develop a robust, inexpensive solid sorbent for CO2 capture from large point sources. Multi-walled carbon nanotubes (MWNTs) have been chosen for functionalization using both primary and tertiary aminosilane compounds. The use of carbon nanotubes, which are essentially an activated carbon material that can be tuned to achieve desired properties, and aminosilane compounds creates a combination of materials that are well understood, and have been used widely in industry as individual components. Carbon nanotubes offer an advantage over other engineered materials, such as zeolites and metal-organic frameworks, which may preferably adsorb H2O, making them difficult to work with in humid environments. The research objectives of this work include the: 1) development of a method for successful functionalization of MWNTs; 2) characterization and testing of sorbents for CO2 capture and release under relevant conditions; and 3) use of knowledge gained from these experiments to better tune the sorbents for enhanced CO2 capture. Background The increase in atmospheric CO2 concentrations since the industrial revolution is thought to be the leading contributor to global climate change. Natural sources and sinks for atmospheric CO2 can be expected to maintain a rough equilibrium according to the carbon cycle, implying the change in atmospheric concentration is the result of anthropogenic emissions. In 2007, global energy-related CO2 emissions reached 28.8 gigatons (Gts) and are projected to increase to 40.2 Gts by 2030.[1] Carbon-based fossil fuels account for 86% of global energy usage and 75% of anthropogenic CO2 emissions.[2] Given the world’s current energy portfolio, increasing energy demand due to the rise of large and emerging economies, such as China and India and the global abundance of accessible and relatively cheap carbon-based energy sources, such as coal and natural gas, energy-related CO2 emissions are expected to continue increasing. Carbon capture at fixed point stations, such as coal-fired power plants (CFPPs) combined with sequestration (CCS) has the potential to mitigate Gts of anthropogenic CO2 emissions, and is regarded as a key method for global-scale CO2 emissions reduction. A major hurdle in the implementation of CCS is that current industrial carbon capture technologies are energy intensive and not cost-effective.[2, 3] To be viable, a capture technology must achieve 90% CO2 capture with a maximum energy penalty of 10%.[4] The current state-of-the-art capture technology for CFPPs, i.e., aqueous amine solutions, involves countercurrent gas-liquid stripping of CO2 in a packed column. This method requires the addition of new capital equipment that must be retrofitted to existing CFPPs and may carry a prohibitively high parasitic load due to the energy costs of regeneration [5]. Moreover, the cost of electricity from a CFPP with 90% CO2 capture using aqueous amine solutions is expected to increase by 81%.[5, 6] As an alternative to gas-liquid absorption, gas-solid adsorption processes have been proposed as promising technologies for carbon capture[7-9]. Solid sorbents have the potential to reduce the energy demand of capture processes due to potentially higher loading capacities, absence of solvent heating and vaporization during regeneration, lower material heat capacities (i.e., carbon compared to water), and lower heats of sorption.[8, 10-12] Solid sorbents require a large surface area-to-mass ratio and a preferential interaction with CO2 to be efficient and effective. While a range of solid sorbents have been considered, including activated carbon[13], zeolites[14], and metalorganic frameworks[15], amine-modified carbon nanotubes (amine-CNTs) have recently been identified as promising sorbents for CO2 capture.[10, 12, 16-21] CNTs are attractive for adsorbing gases because they have high surface areas and relatively controllable porosity. The surfaces of pristine CNTs are relatively inert to covalent modification, but have highly polarizable delocalized π electrons that enable physisorption. CNT growth has been extensively studied[22] and CNT materials can be grown controllably to a range of diameters and number of walls.[23] This control of porosity and structure allows for a material that can be tuned to specific sorption capacities. CNT interactions with gases may be tuned for chemisorption interactions through functionalization of the nanotube surface with various functional groups. Covalent functionalization of SWNTs has also been well-studied[24] and can be performed by facile acid treatment which creates hydroxyl (-OH) and carboxylic acid (COOH) groups on the ends and walls of the SWNTs.[25] These functional groups can be further covalently functionalized by use of commercially available silane grafting agents using condensation chemistry. This results in a material with a large surface area, controllable porosity, and tunable surface chemistry. Amine moieties tethered to high surface area supports have been shown to adsorb carbon dioxide.[26] Depending on the number of moieties, other than hydrogen, attached to the nitrogen atom, an amine is labeled primary (R-NH2), secondary (R-NHR) or tertiary (RNRR). Previous work has shown that the stability of CO2 reaction products, the reaction rate, and heat of reaction depends not only on the type of amine (primary, secondary, or tertiary) but also on the chemical structure of the sorbent molecule, specifically the steric hindrance of the amine group.[27-29] While primary and secondary amines have been applied toward carbon capture more frequently than tertiary amines, the nature of tertiary amine-CO2 reaction chemistry suggests that tertiary amines might be excellent functional groups for CO2 sorption. First, tertiary amines physisorb CO2.[8] As mentioned previously, capture by physisorption, rather than chemisorption, may be ideal for minimizing sorbent regeneration costs. Second, unlike primary and secondary amine-CO2 reactions, where H2O competes with CO2, the tertiary amine-CO2 reaction requires a 1:1 H2O to CO2 ratio to form bicarbonate. Given the ubiquitous presence of H2O in flue gases, the potential for tertiary amines to promote CO2 sorption in the presence of H2O is an important benefit over primary and secondary amines.[8] The proposed reaction for supported primary amines is as follows: CO2 Sorption: 2 R-NH2 + CO2 ➞ R-NH3 + R-NH-COO Regeneration: R-NH-COO + R-NH3 + (Heat) ➞ CO2 + 2 R-NH2 For the tertiary amines the mechanism is believed to require water: CO2 Sorption: R-NRR + CO2 + H2O ➞ R-NRRH + HCO3 Regeneration: R-NRRH + HCO3 + (Heat) ➞ R-NRR + CO2 + H2O In the above schemes, the substituent R is an alkyl moiety connecting the amine to the support material and R, R are different alkyl moieties. The current industrially scalable post-combustion CO2 capture process involves aqueous absorption with alkanolamines. Although adsorption (the focus of the current work) is different from absorption separation processes, a great deal can be learned through the vast number of studies associated with amine-based solvent approaches to CO2 capture based upon absorption. In other words, the mechanism and chemical pathway of CO2 binding in a solvent is similar to that of a porous solid sorbent. Most commonly used are monoethanolamine (MEA) and diethanolamine (DEA).[30] Recently, Chowdhury et al. investigated the CO2 sorption rate, loading capacity, and heat of reaction measurements of twenty-five amine-based absorbents. They were able to correlate their findings to the differences in chemical structure of the amines.[31] The results suggest that the specific amine moieties and structures used may be selected or modified to tune the CO2 capture process. Tertiary amine-CO2 reaction chemistry suggests that the tertiary amine-CNTs may have high CO2 capacity with a relatively low heat of sorption, which makes tertiary amineCNTs promising sorbents for CO2 capture. The primary goal of the proposed work is to synthesize and investigate tertiary amine modified CNTs as CO2 capture sorbents. Initial investigations have been designed to compare the material and CO2 capture properties of tertiary amine-modified CNTs with primary amine modified CNTs. Results Materials and Methods OH-functionalized MWNTs (OH-MWNTs) with outer diameters of 10-20 nm were purchased from Sun Innovations (Item #SN32547). Aminosilanes: 3aminopropyltriethoxysilane (APTES) (99%) and N-dimethylaminopropyltrimethoxysilane (DMAPS) were purchased from Gelest, Inc. Anhydrous toluene was purchased from Arcos Organics. Method 1: Functionalization procedures were based on the methods reported earlier by Su et al.[19] Prior to silane functionalization, the MWNTs were heated in a quartz tube under a nitrogen flow at 300 ̊C for one hour to remove water, amorphous carbon material and other volatile impurities. The silane functionalization was carried out in an inert argon atmosphere to increase batch-to-batch reproducibility by limiting exposure to ambient water. Two grams of heat-treated MWNTs were dispersed in 100 mL of toluene and 10 mL of silane. The reaction was conducted under reflux with continuous stirring for 18 hours in a Schlenk flask in a silicon oil bath at 125 ̊C. The solution was allowed to cool and the dispersion was vacuum filtered through 0.45-μm PTFE filter (Millipore FH) with excess toluene. The product was then dried in a quartz tube under a nitrogen flow at 125 ̊C for 1.5 hours to remove any residual silane and toluene. The synthesis of primary and tertiary amine functionalized CNTs was repeated into two batches. For the second batch of tertiary amine CNT functionalization, the reaction was reduced from 96 hours to 24 hours. No observable difference in material properties was found between the two batches. Method 2: Functionalization procedures were based on the methods reported by Su et al.[32] The CNT support used in this method, CTube 100 MWNT from CNT Co. Ltd., is the same support as that used by Su et al.[32] The silane functionalization was carried out in an ambient atmosphere to facilitate water-induced silane polymerzation. Two grams of as-produced MWNTs were dispersed in 70 mL of toluene and 30 mL of silane. The reaction was conducted under reflux with continuous stirring for 2 hours in a Schlenk flask in a silicon oil bath at 125 ̊C. The solution was allowed to cool and the dispersion was vacuum filtered through 0.45-μm PTFE filter (Millipore FH). The product was then dried in a vacuum oven at 80 ̊C for 1.5 hours to remove any residual silane and toluene. XPS analyses were performed using a Physical Electronics (PHI) 5000Versa Probe Scanning XPS system at the Stanford Nanocharacterization Laboratory at Stanford University. Spectra were collected under high vacuum (10 Torr) at ambient temperature using monochromatic Al Kα radiation at 1486 eV. All spectra were calibrated to the C1s peak located at 284.6 eV. Samples were prepared by dispersing the functionalized MWNT using bath sonication in deionized water and then filtering the dispersion through 0.4-mm polycarbonate filters. Scanning Electron Microscopy (SEM) was conducted using a FEI XL30 Sirion SEM at 5 KV and using a FEI Magellan 400 XHR at 5 KV. The SEM samples were prepared by dispersing the functionalized MWNT using bath sonication in deionized water. A drop of this dispersion was placed on a heated (120 ̊C) bare silicon wafer plate. Micro‐Raman spectroscopy was performed in a LabRamAramis confocal Raman spectrophotometer (Horiba). Measurements were carried out at 633 nm (1.96 eV) excitation at 100x magnification and 1-mm spot size, and 1800 grating. Samples for Raman spectroscopy were made in a similar manner as the SEM samples. Data was normalized to the G-band. FTIR was performed in a Thermo Nicolet Nexus 470 ESP FTIR spectrophotometer. FTIR pellet samples were made by grinding CNT samples with KBr and then pressing in a pelletizer. The porosity and surface area analysis was performed using a Quantachrome Autosorb iQ gas sorption analyzer. Each sample was outgassed at 0.03 torr with a 2 ̊C/min ramp to 110 ̊C, followed by a 5 ̊C/min ramp to 140 ̊C, where the sample was held for 3 hours and tested for continuing outgassing. If the sample was still outgassing, it was further held at 140 ̊C for up to 3 additional hours, with a test performed every 15 minutes. The sample was then held at vacuum (0.03 torr) until the analysis was run. Pore analysis was performed using N2 at 77 K (P/P0 range of 1×10 to 0.995) and Ar at 87 K (P/P0 range of 1×10 to 0.995). Mesopore distributions were estimated using N2 and Ar desorption isotherms in combination with the Barrett-Joyner-Halenda (BJH) method, while micropore distributions were estimated using N2 and Ar adsorption isotherms in combination with the Dubinin-Astakhov (DA) method. Breakthrough experiments were performed in a quartz reactor with packed beds ranging from 100 mg to 200 mg of CNT. The experimental schematic can be seen in Figure 1. Nitrogen is passed through a bubbler containing H2O (the bubbler is bypassed in dry experiments) at 20 mL/min and passed through the packed bed. At a designated time, 2 mL/min of CO2 is passed into the flow stream. Breakthrough is measured downstream of the packed bed using an Extrel 300Max-LG quadrupole MS. Both CO2 and H2O concentrations are monitored, and CO2 is not sent to the bed until the H2O concentration has stabilized. Typical experimental parameters are 9.09% CO2 and 2.1% H2O, with a balance of N2. Residence times across the bed are on the order of several seconds. The bed is held at a constant temperature of 40 ̊C for adsorption experiments. The reactor may be rapidly heated to 120 ̊C for desorption measurements. The capacity of the sorbent is determined by measuring the difference between CO2 breakthrough curves through an empty reactor and through the packed bed. Experiments were performed to determine the effect of pressure drop across the bed using a 500 mg bed of Sigma Aldrich 50-70 mesh quartz sand. Changes due to pressure drop were determined to be minimal, and within the experimental error. Figure 1: Schematic for CO2 breakthrough experiments. MWNTs were functionalized with primary (APTES) and tertiary (DMAPS) aminosilanes and the material properties of the functionalized MWNTs were investigated. The results of this characterization analysis are described in the following sections. Scanning Electron Microscopy Morphology was investigated by scanning electron microscopy (SEM) by drop casting dispersions of MWNTs in deionized water onto clean and heated silicon/silicon oxide wafers. There was no significant difference in appearance of the OH-MWNTs before or after heat treatments prior to functionalization (not shown). Figure 2 shows highresolution images of large aggregates of MWNTs with and without functionalization. The SEM images show MWNTs with a large available surface area and little order in aggregation. In addition to the large surface area outside the SWNTs, this loose architecture should allow some access to the interior pore space inside the MWNT. However, some of the higher contrast particles observed in the micrographs could be metal catalyst, which could block this inner pore space. It should be noted that the silane molecules are not expected to be observable by SEM unless there is extensive polymerization of the silane resulting in large aggregates. Toluene was also used as a dispersion media and no significant differences in MWNT morphology were observed. However, the “coffee ring effect” due to material deposited during solvent evaporation was observed in the low-resolution images on the SiO2 surface of the APTES-MWNT samples (not shown). It is likely that these are due to residual APTES dissolved in