From modeling to synthesis and characterization of nitrogen-selective metallic membranes
- Nitrogen-selective membranes are rare since nitrogen has low adsorption or dissolution affinity toward typical membrane materials. Efficient nitrogen separation at high temperature using membranes would be beneficial if applied to post-combustion capture of CO2, in which nitrogen removal from the combustion mixture leads to indirect CO2 capture. This membrane may also be used for natural gas purification to remove nitrogen or applied to enhanced ammonia synthesis for fertilizer production. Using nitrogen-selective membranes, extreme conditions (400--500˚C and over 100 bar) required for ammonia synthesis in the traditional Habor-Bosch process might be alleviated. In this study, metallic membranes (MMs) are proposed for high-temperature nitrogen separation owing to their advantages of high selectivity and enhanced thermal stability compared to polymeric membranes. In nitrogen-selective MMs, nitrogen molecules catalytically dissociate into atomic nitrogen on the membrane surface, and atomic nitrogen diffuses through bulk of the MM. Metallic membranes based on palladium and its alloys are well known for hydrogen separation but with respect to MMs for nitrogen separation, this dissertation work is the first study of its kind. In this study, research progress on developing nitrogen-selective MMs is presented including membrane synthesis, membrane flux measurements and materials characterization, combined with a theoretical approach for understanding nitrogen--metal interactions in potential membrane materials. In the first part of my dissertation, metals that can be used as membrane materials were investigated using a theoretical model based on density functional theory. To establish a thermodynamic equilibrium model to calculate gas solubility in metallic membranes, hydrogen--metal systems, which have been well established in the literature, were first studied to confirm the validity of our equilibrium model. Without experimental data, but only based on first principles and thermodynamics, the hydrogen solubility in eight transition metals were successfully estimated, and the cases with deviations in the estimation were explained by the lack of accuracy in either binding energy or the vibrational frequency calculations of hydrogen in metals. The validated model was then transferred to nitrogen--metal systems to calculate nitrogen solubility in metals. As a representative metal for nitrogen binding, vanadium (V) was first chosen; however, the solubility model significantly deviated from the experimentally measured nitrogen solubility. The deviation was due to the dilute-solution assumption, which was violated due to strong binding of nitrogen in V. This assumption is critical for the nitrogen solubility estimation for an α-phase solid solution, in which the nitrogen atom is mobile. In molybdenum (Mo), where nitrogen is weakly binding, the nitrogen solubility could be closely estimated with some corrections in the calculated vibrational frequencies. The solubility of nitrogen in Mo, however, is too low to render Mo to be practically viable for MM application. There might be an ideal binding strength of nitrogen for enhanced membrane transport, which gives maximum α-phase solubility but is below the level inducing the phase transformation to nitrides. The binding energy was tuned by alloying V with Mo. A Mo-rich V alloy (Mo3V) showed a binding energy of nitrogen close to that of hydrogen in palladium, which may be an ideal binding strength for facilitated membrane transport. The second part of my dissertation is experimental measurements of nitrogen permeation through metallic foils (30- to 40-µm thick) with the body-centered cubic (bcc) structure, including V, niobium (Nb), tantalum (Ta) and iron (Fe). An experimental set-up and a protocol using a pressurized membrane module at high temperature with continuous gas flow were established. For carbon capture applications, nitrogen and CO2 gas permeabilities were measured in V, Nb, and Ta foils under 400 ˚C and 3--6 bar, simulating compressed post-combustion flue gas conditions. For the application to natural gas cleanup, nitrogen, CH4 and their mixtures were tested using V and Fe foils under 400--600 ˚C and 30--60 bar, considering natural gas well-head conditions. In all gas permeation experiments, the ideal selectivity of nitrogen to CO2 and CH4 was infinity, indicating that only nitrogen exclusively permeated through the membranes. The permeability of nitrogen through those metals was surprisingly higher by 5--6 orders of magnitude than that expected from the literature, potentially due to the enhanced nitrogen diffusion through grain boundaries in metals; however, the nitrogen permeability was still low, by 5 orders of magnitude, compared to hydrogen permeation through palladium. Moreover, the surfaces of the MMs became oxidized after the nitrogen permeation experiments, which might prohibit long-term operation of the membranes. The third part of my dissertation presents the experimental results for developing thin composite membranes, in which an enhanced membrane flux was expected owing to the reduced membrane thickness and greater grain boundary density in metals. The thicknesses of these membranes were between 0.3 and 0.9 µm. The thin MMs were synthesized by creating a dense metal coating on a porous substrate. For the coated thin membranes to not have pinhole defects the structure of the porous substrate was very important; specifically, having a smooth surface and small pores in addition to sufficient porosity was necessary. Such criteria were met by porous α-alumina substrates synthesized via colloidal deposition followed by partial sintering. On these porous alumina substrates, thin membranes were deposited via physical vapor deposition using V and Mo. Subsequently, in permeation experiments, nitrogen and Ar gases were tested to confirm nitrogen selectivity. Thin V membranes suffered from surface oxidation above 300 ˚C, which caused stresses on the entire membrane and eventually cracked the membrane. The membrane did not crack under a slow heating rate, and showed 600 times higher flux than that through the foil membrane; however, the increased flux was largely attributed to the flux via Knudsen diffusion through pinholes. As a result, the nitrogen selectivity was 1.20, which is the Knudsen diffusion ratio of nitrogen to Ar. Molybdenum membranes had high resistance to oxidation with surface reactivity toward nitrogen below 500 ˚C; however, metal particles in the vapor-deposited membrane went through sintering at high temperature, developing a columnar structure associated with pinholes. Consequently, the maximum selectivity of nitrogen was limited by the Knudsen diffusion ratio of nitrogen to Ar. In summary, this study represents pioneering work for the development of nitrogen-selective MMs, providing a theoretical understanding of nitrogen absorption in solid solutions of bcc metals, and showing experimental evidence of nitrogen permeation through these metals. Grain boundary diffusion in the metals was suggested as a main transport mechanism of atomic nitrogen. Thin MMs supported on porous alumina exhibited a flux increase by 2--3 orders of magnitude compared to thick foils, but were more prone to defect formation at high temperature due to oxidation or metal sintering, which led to their low selectivity to nitrogen. The lessons learned from this study may guide future development of nitrogen-selective MMs as follows. 1) A theoretical study may investigate nitrogen diffusion through metal defects to provide quantitative predictions of experimental diffusivity. 2) Metal oxidation at high temperature seems ubiquitous even under an ultrahigh-purity anoxic gas environment; the oxidation may be avoided by choosing appropriate operating temperature and gas conditions. Stable surface oxide formation may prevent the membrane from severe oxidation but could impede fast subsurface diffusion of nitrogen. Overall, understanding the effect of oxygen on MMs will be critical in developing new membranes. 3) Thin MM structures need to be engineered to have a more dense structure with a high grain boundary density to facilitate the solid-state atomic nitrogen diffusion. Increasing the substrate temperature during sputtering deposition may be helpful in creating such structures. In addition to resolving current challenges, demonstrating the membrane performance in the mixtures of N2/CO2 and N2/CH4 with simultaneous ammonia synthesis will be crucial for practical applications.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Stanford University, Department of Energy Resources Engineering.
|Wilcox, Jennifer, 1976-
|Wilcox, Jennifer, 1976-
|Statement of responsibility
|Submitted to the Department of Energy Resources Engineering.
|Thesis (Ph.D.)--Stanford University, 2016.
- © 2016 by Kyoungjin Lee
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...