You are here:
SEPARATION AND PURIFICATION OF HYDROGEN FROM MIXED GAS STREAMS USING HOLLOW GLASS MICROSPHERES
Impact/Purpose:
In light of current fears surrounding global warming, the appeal of alternative fuel sources continues to grow. A major obstacle standing in the way of the hydrogen economy is the expense and difficulty of the purification and separation of hydrogen from mixed gases. Due to different permeabilities of gases through vitreous silica, it is believed that glass would make an exceptional membrane for gas separation, which has been proposed as far back as Barrer. Hollow glass microspheres have extremely thin walls which makes them ideal membranes for this application. The other exceptional benefit of using hollow glass microspheres is the ease at which they can be produced, using recycled glass frit from beer bottles. Other inorganic membrane materials used for this application often rely upon the fabrication of continuous palladium/palladium alloy films which selectively allow hydrogen to pass. As could be imagined, processing a membrane as such is no simple task, and if any pinhole is present in the film the membrane is rendered useless.
Description:
Due to this study, HGMS have been shown to be a viable means of separating hydrogen from various mixed gases.5 Mass transport through the glass walls of the microspheres is determined through the use of Boyle’s Law:
where P1 is the fill pressure, V1 is the initial volume of the container used to fill the microspheres, P2 is the outgassing pressure, and V2 is the volume inside the microspheres. The amount of hydrogen and helium gas released from the microspheres used in this study always increases linearly with fill pressure which indicates mass transport of these species is occurring at 400 °C whether using pure or mixed gases containing these species. The amount of gas released after filling the microspheres with pure gases of nitrogen, argon, and carbon dioxide is not dependent on the fill pressure, which indicates that these gases are not permeating into the microspheres in detectable quantities. The same is true for these species present in the mixed gases, which indicates that they are only adsorbed on the surfaces of the microspheres.
When outgassed, a sample filled with 700 torr of pure hydrogen and then transferred in air to the RGA indicates that carbon dioxide, nitrogen, and argon are present. Only 93% of the resulting gas is hydrogen. Since the fill gas only contains hydrogen, adsorption of atmospheric gases must have occurred. A comparison of the amount of carbon dioxide and nitrogen outgassed from this exposure to air to results for exposures to 100% carbon dioxide or nitrogen atmospheres indicates that these values are similar within the standard deviation of the data. The amount of argon found in the experiment indicates that less argon was adsorbed than expected.
An attempt to increase the purity of the hydrogen gas recovered with the microspheres was made using both furnace and IR light treatments. Samples were filled with 700 torr of pure hydrogen. One sample was subjected to a preheating treatment via a furnace set to 150 °C while monitoring the gases released. The sample yielded, in decreasing quantities, hydrogen, nitrogen, and carbon dioxide. Argon was not detected above background in this experiment. More hydrogen was released than any other gas being monitored, which is attributed to adsorbed water and hydrogen outgassing. The hydrogen signal did not return to background during the heat treatment, which could indicate that hydrogen is also diffusing out of the microspheres in very small amounts. The sample was then outgassed normally at ~500 °C. The purity of the hydrogen with respect to the gases monitored is ~97%. This increase in purity from 93% is excellent, and is an indicator that higher purities can be achieved via tailoring of the heat treatment time and temperature. Another treatment to remove the adsorbed gases was carried out using IR light. The temperature reading during this treatment reached ~200 °C. The IR treatment yielded, in decreasing quantities, hydrogen, carbon dioxide, and nitrogen. Again, the argon signal was not found to increase above background throughout this treatment. As before, more hydrogen outgassed than any other gas being monitored. The sample was then outgassed normally at ~500 °C. The purity of the hydrogen with respect to the gases monitored is ~98%. This increase in purity from 93% to 98% is excellent, with minimal signals from adsorbed gases.
Behavior of GL-0179 solid spheres was examined to determine the role of adsorption. Hollow spheres did outgas more argon than the blank and the solid spheres, while the solid spheres and the blank (sample tube with no HGMS) containing fiberglass outgas more carbon dioxide than the hollow spheres. A blank without fiberglass released ~15% as much carbon dioxide as a blank with fiberglass. The solid spheres outgas more nitrogen than either the blank or the hollow spheres. As the experiments were conducted under the same pressures and temperatures, the gases adsorbed should only be a function of the surface state of the microspheres and the gas molecules or atoms in question. Carbon dioxide and nitrogen are known to exhibit quadrupole moments, which increase adsorption on a polarizable surface such as the hydroxyl rich surface of a glass. This effect is seen by comparing relative amounts of argon and nitrogen released, or argon and carbon dioxide (even though the sensitivity of the RGA to carbon dioxide may be different). The argon signal is only slightly above background for these microspheres, which is attributed to the inert nature of the gas and the lack of a quadrupole moment. The GL-0179 solid microspheres release more carbon dioxide and nitrogen than the hollow spheres, which may be due to different surface chemistry resulting in differences in the polarizability of the surface.
The temperature at which nitrogen, carbon dioxide, and argon outgas from the microspheres is another indicator of an adsorption process. The signals for the nitrogen and carbon dioxide begin to occur at ~320 s i.e. only 20 s after the onset of heating, and return to background after ~7 minutes. The signal for argon initiates at ~320 s and returns to background at ~380 s. The first sign of gas release occurs when the sample reaches ~50 °C. This temperature is very low for activated diffusion; it is unlikely that the large gas molecules are diffusing through the microsphere walls. The signals returned to background before the maximum outgassing temperature was reached. Gas evolution at low temperatures is commonly associated with adsorbed gases outgassing from the surface. As these bonds are merely electrostatic, they are not strong enough to hold the molecule in place with an increase in the system’s energy.
Unexpectedly, the hydrogen signal was found to exhibit a small initial peak at ~320 s, which corresponds to a temperature of ~50 °C. It is unlikely that this peak is due to permeation of hydrogen from the HGMS at this temperature. It is possible that hydrogen or water adsorbed to the surface of the glass spheres could contribute to this peak. The RGA ionizes molecules into their byproduct atoms and molecules. Water vapor thus exhibits a peak for molecular hydrogen due to molecular decomposition. This peak is not present in every curve, which suggests that it may be atmospherically controlled as the water vapor in the atmosphere varies from day to day. The peak was minimized when a window air conditioner was used which effectively removes water from the local atmosphere. It is suggested that this initial hump at ~320 s is due to adsorbed water and hydrogen.
The carbon dioxide signal also consistently exhibits two individual peaks. The first peak is centered at ~320 s and the second at ~370 s, which corresponds to ~50 °C and ~200 °C, respectively. There are a number of possibilities to explain this phenomenon. The carbon dioxide molecules could be both physically and chemically adsorbed to the surface of the microspheres due to filling at elevated temperatures, which would yield two peaks as the energy to liberate the molecules would be different for the different bonds. It is also possible that this phenomenon is related to the glass composition. Since this glass is a soda lime borosilicate, there are a number of different energies associated with the different atoms present at the surface, which could electrostatically bond to the carbon dioxide molecules differently requiring dissimilar energies to liberate the molecules. It is also possible for gas molecules to adsorb in multiple layers. It is possible that the farthest molecules from the surface of the microspheres would require less energy to liberate than molecules electrostatically bonded to the surface of the glass. Nitrogen data occasionally exhibit two peaks, but this behavior was not a consistent occurrence and may be similar to that of carbon dioxide.
The RGA is only useful at a qualitative level without standards since the sensitivity of the RGA is different for different gases, and the sensitivity varies as a function of gas composition. The sensitivity of the RGA to helium appears to be considerably lower than the sensitivity to hydrogen, while the sensitivity to carbon dioxide appears to be higher than the sensitivity to hydrogen. The sensitivity of the RGA to argon and nitrogen is very similar to the sensitivity of the RGA to hydrogen, which is reflected in the gas analysis data.
The RGA and PVT measurements both indicate that the amount of gas released is a function of the density, wall thickness, and diameter of the microspheres. When the PVT data and the RGA data are normalized for the weight of the microspheres used, the amount of hydrogen and helium released always decreases in the order GL-0237, K25, K35, K46. The GL-0237 microspheres had the thinnest walls, the largest diameters, the lowest density, and thus the largest internal volume which results in these microspheres outgassing the largest amount of gas of any of the spheres on a per gram basis. The K46 microspheres had the thickest walls, the highest density, and thus the smallest internal volume which corresponds to these microspheres outgassing the smallest amount of gas of any of the spheres on a per gram basis. The GL-0237 hollow microspheres would be best for applications requiring low density, fast diffusion, or large volumes of gas. The 3MTM microspheres would be useful for applications requiring large pressures due to thick walls which increase strength.