Report

Plasticization-Enhanced Hydrogen Purification Using Polymeric Membranes

See allHide authors and affiliations

Science  03 Feb 2006:
Vol. 311, Issue 5761, pp. 639-642
DOI: 10.1126/science.1118079

Abstract

Polymer membranes are attractive for molecular-scale separations such as hydrogen purification because of inherently low energy requirements. However, membrane materials with outstanding hydrogen separation performance in feed streams containing high-pressure carbon dioxide and impurities such as hydrogen sulfide and water are not available. We report highly permeable, reverse-selective membrane materials for hydrogen purification, as exemplified by molecularly engineered, highly branched, cross-linked poly(ethylene oxide). In contrast to the performance of conventional materials, we demonstrate that plasticization can be harnessed to improve separation performance.

Hydrogen is produced primarily by steam reforming of hydrocarbons followed by the water-gas shift reaction, which yields a hydrogen product containing impurities such as CO2, H2S, and H2O (1). The hydrogen must be purified for further use, and based on the high volumes currently produced and the likelihood for this production to increase, even a small improvement in H2 purification efficiency could substantially reduce the costs. Membrane technology is attractive for molecular-scale separations because of inherent advantages such as high energy efficiency, excellent reliability, and a small footprint (25). The potential applicability of membrane technology relies strongly on the ability of membrane materials to exhibit high separation performance at practical feed conditions (e.g., with feed streams that contain high-pressure CO2 and impurities such as H2S and H2O).

Highly permeable and highly selective membrane materials are desired for CO2/H2 separation. Gas permeability P, which is the steady-state, pressure- and thickness-normalized gas flux through a membrane, is usually expressed as P = S × D, the product of gas solubility S and gas diffusivity D in the polymer membrane (6). Selectivity αA/B, which characterizes the ability of a membrane to separate gases A and B, is given by Math(1) where SA/SB is the solubility selectivity and DA/DB is the diffusivity selectivity (6). The selectivity of CO2 over H2, αCO2/H2, reflects the tradeoff between favorable solubility selectivity (CO2 is more condensable than H2 and, therefore, SCO2/SH2 > 1) and unfavorable diffusivity selectivity (CO2 is larger than H2, so DCO2/DH2 < 1) (7). In conventional polymeric membrane materials (8) and those based on carbon (4) and silica (9, 10), overall gas selectivity is dominated by diffusivity selectivity and, therefore, these materials are typically more permeable to H2 than to CO2. Consequently, the H2 product is produced in the permeate at low pressure, even though further downstream utilization requires H2 at high pressure. Expensive recompression of the H2 product hence diminishes the advantage of membrane technology relative to that of conventional separation technologies, such as pressure swing adsorption, that produce H2 at or near feed pressure (1, 2, 6). To minimize or avoid H2 recompression, optimal membrane materials should be reverse selective (i.e., more permeable to larger molecules, such as CO2, than to smaller molecules, such as H2). Here, we propose that to achieve very high CO2/H2 selectivity, a membrane must exhibit favorable interactions with CO2 to enhance solubility selectivity and have very weak size-sieving ability to bring DCO2/DH2 as close to 1 as possible. Guided by these material design principles, we prepared and characterized a family of highly branched polymers based on poly(ethylene oxide) (PEO) and found that these polymers display excellent CO2/H2 separation performance. Counterintuitively, the CO2/H2 selectivity and CO2 permeability improve as CO2 partial pressure increases (i.e., as CO2 concentration sorbed in the polymer increases). This is in contrast to the behavior of conventional, strongly size-selective materials, for which raising CO2 partial pressure typically decreases selectivity (11).

In a recent review of the influence of primary chemical structure on CO2/H2 separation properties of polymers, ethylene oxide (EO) units were identified as the best chemical groups for such membranes because the polar ether oxygens in EO units interact favorably with CO2, resulting in high solubility selectivity (12). Polymers containing EO can be highly flexible, leading to weak size-sieving behavior and high diffusion coefficients, two factors which contribute directly to high CO2 permeability and high CO2/H2 selectivity (12, 13). However, pure PEO exhibits very low CO2 permeability [approximately 12 Barrers (14) at 35°C and infinite dilution] as a result of high crystallinity levels (7). Additionally, the presence of crystalline regions in pure PEO reduces polymer chain mobility in the amorphous phase and increases size-sieving ability, thereby decreasing CO2/H2 selectivity (12). To circumvent this limitation and effectively frustrate crystallization, short non-PEO segments are introduced into the polymer backbone to interrupt the EO repeat units. Chain branches containing short, noncrystallizable segments of EO are also introduced randomly into the chain backbone to further inhibit crystallinity. This leads to amorphous materials with higher gas permeability and higher CO2/H2 selectivity than semicrystalline PEO. Plasticization further improves their CO2/H2 separation properties, in contrast to the view that plasticization always reduces polymer membrane separation performance, as it does in the case of CO2/CH4 separation in natural gas purification (15). Moreover, all polymers are more permeable to CO2 than to CH4 because CO2 has higher diffusivity (because of its smaller molecular size) and higher solubility (because of its greater tendency to condense) than CH4. In contrast, polymers that are more permeable to CO2 than to H2 are much rarer because the smaller size of H2 favors its permeation over that of the larger CO2.

Our family of amorphous, high-molecular-weight, cross-linked, network copolymers was synthesized by photopolymerizing different composition ratios of poly(ethylene glycol) diacrylate [PEGDA: CH2=CHCOO(CH2CH2O)14OCCH= CH2] and poly(ethylene glycol) methyl ether acrylate [PEGMEA: CH2=CHCO(OCH2CH2)8OCH3] (16, 17). The resulting copolymer network has the general chemical structure shown in Fig. 1. PEGDA contains EO units in its backbone, and PEGMEA has pendant EO units. Cross-linked copolymer samples with 0 to 99 weight percent (wt %) PEGMEA and the balance PEGDA were prepared and characterized. Independent of the concentration of PEGDA and PEGMEA, the copolymers contain about 82 wt % EO. Increasing PEGMEA content increases fractional free volume and, in turn, CO2 permeability and pure-gas CO2/H2 selectivity. However, at low temperatures (≤0°C), materials with very high PEGMEA content (e.g., 91 wt %) crystallize, resulting in a substantial permeability decrease below 0°C. We extensively investigated the CO2/H2 permeation properties of the 70 wt % PEGMEA/30 wt % PEGDA copolymer at –20°, 10°, and 35°C with pure gases and three binary CO2/H2 mixtures of different compositions (16). We chose this temperature range not only because this copolymer does not crystallize over this range but also because this range is consistent with the operating temperature of some industrial processes currently used for H2 purification (1). Additionally, to determine the effect of H2O and H2S impurities on CO2/H2 separation properties, the 91 wt % PEGMEA/9 wt % PEGDA copolymer was characterized at 22°C with a moisture-laden CO2/H2 mixture and a four-component, H2S-containing gas mixture that mimicked the composition of process synthesis gas. In both pure- and mixed-gas studies, the PEGMEA/PEGDA films were physically stable at transmembrane pressure differences up to 21 atm (the maximum value explored in our study), and their gas permeability coefficients were independent of previous thermal and gas exposure history, as expected for rubbery polymers. Permeability coefficients were independent of film thickness, which varied from 70 to 500 μm.

Fig. 1.

Schematic representation of PEGDA/PEGMEA copolymer network. Italicized and bolded parts of the network derive from the cross-linker. R1 is CO(OCH2CH2)8OCH3 from PEGMEA; R2 is COO(CH2CH2O)14OC from PEGDA.

Figure 2 shows the dependence of CO2 and H2 permeability coefficients on CO2 partial pressure in the feed at different temperatures. As CO2 partial pressure increases, CO2 and H2 permeabilities increase at all temperatures; the permeability rise is greater at lower temperatures because of higher CO2 thermodynamic activity (at a fixed partial pressure) in the polymer film at cooler temperatures. Both pure- and mixedgas permeabilities also follow the same trends, suggesting that the permeability increase is in large measure a result of increasing CO2 partial pressure. As indicated by an increase in CO2 diffusivity with increasing CO2 concentration in the film (18), CO2 sorbed in the polymer plasticizes the polymer chains, leading to an increase in fractional free volume and, in turn, gas permeability. With decreasing temperature, CO2 sorption and, hence, plasticization increase so that the effect of CO2 partial pressure on permeability becomes even stronger.

Fig. 2.

Effect of temperature and CO2 upstream partial pressure on (A)pure- and mixed-gas CO2 permeability and (B) pure- and mixed-gas H2 permeability in 70 wt % PEGMEA/30 wt % PEGDA copolymer (14). Pure-gas permeability data are shown as open circles(◯). Mixed-gas CO2/H2 feed compositions (in mol % CO2:mol % H2) were 10:90 (⚫), 50:50 (Δ), and 80:20 (▾). Uncertainty in the permeability data was ±10% or less. The lines are provided to guide the eye.

Decreasing temperature typically decreases gas permeability as a result of a reduction in polymer chain mobility and, therefore, diffusivity at lower temperatures (6, 19). However, Fig. 2A demonstrates that decreasing temperature does not necessarily decrease CO2 permeability, especially at high CO2 partial pressure. For example, at a CO2 feed partial pressure of 17 atm, the CO2 permeability at –20°C is 410 Barrers, which is higher than that at 10°C (300 Barrers) and very similar to that at 35°C (440 Barrers). Thus, the permeability decrease that would normally accompany a temperature reduction is essentially offset by the increase in CO2 solubility and the increased plasticization of the polymer by CO2, which increases diffusivity at lower temperatures.

Decreasing temperature considerably increases mixed-gas CO2/H2 selectivity. As shown in Fig. 3, at a CO2 partial pressure of 17 atm, mixed-gas selectivity increases from 9.4 to 31 as temperature decreases from 35° to –20°C. Furthermore, the selectivity of 31 is accompanied by a CO2 permeability of 410 Barrers, which is orders of magnitude higher than that observed in conventional polymer membranes used for CO2 separations (6). Facilitated transport membranes can exhibit high CO2/light gas selectivity at low CO2 partial pressure (∼1 atm or less); however, the selectivity in these materials decreases strongly as CO2 pressure increases (2022). Such materials are often studied for low CO2 partial pressure applications (e.g., removal of CO2 from breathing gas aboard spaceships) (20). However, they cannot operate at the high pressures required for hydrogen applications because they are often based on liquids supported in porous media, and the liquids typically cannot be maintained in the porous support when a large pressure difference is applied across the membrane (23, 24).

Fig. 3.

Effect of CO2 partial pressure and temperature on mixed-gas CO2/H2 selectivity (αCO2/H2 in 70 wt % PEGMEA/30 wt % PEGDA copolymer. Mixed-gas CO2/H2 feed compositions (in mol % CO2:mol % H2) were 10:90 (⚫), 50:50 (Δ), and 80:20 (▾). The lines are provided to guide the eye.

Mixed-gas selectivity is essentially independent of CO2 partial pressure at 35° and 10°C (Fig. 3). However, as CO2 partial pressure increases at –20°C, mixed-gas CO2/H2 selectivity increases by 35%. In conventional size-sieving polymers, plasticization of the polymer by CO2 or other condensable components results in mixed-gas selectivity values that decrease, often markedly, as CO2 partial pressure increases (11) or when other strongly sorbing impurities such as higher hydrocarbons are present (25). For example, when the CO2 partial pressure increased from 1 to 10 atm, the CO2/CH4 selectivity of a rigid, glassy, strongly size-sieving, aromatic polyimide decreased from 40 to about 4 (11).

We ascribe the mixed-gas CO2/H2 selectivity results for the PEGMEA/PEGDA copolymers to the inherent transport properties of these reverse-selective materials (3, 26). At high CO2 partial pressures, these materials sorb considerable amounts of CO2, leading to swelling and an increase in free volume, which presumably decreases the glass transition temperature (Tg) (2729) and weakens the size-sieving ability of the membrane (26).

Figure 4 presents a permeability/selectivity map for CO2/H2 separation. The upper bound line in the figure gives an estimate of the highest pure-gas selectivity possible for a given permeability in polymer-based materials at 25°C (8). Unlike separation based on strong size-sieving ability [where there is a strong tradeoff between permeability and selectivity (8, 30)], the positive slope of the upper bound indicates that high CO2 permeability and high CO2/H2 selectivity may be achieved simultaneously. The PEGMEA/PEGDA copolymers we explored exhibit excellent separation performance for CO2/H2 mixtures, and decreasing temperature actually moves the CO2/H2 separation performance above the upper bound line. However, because this line is commonly established by permeation properties of polymers at or near 25°C (8), the upper bound may shift as temperature deviates from 25°C. Currently, there is no model to predict the temperature dependence of the upper bound, and there are not yet enough systematic experimental data available to provide clear evidence of its temperature dependence.

Fig. 4.

Permeability/selectivity map for CO2/H2 separation. Mixedgas separation performance data of the 70 wt % PEGMEA/30 wt % PEGDA copolymer at 35°C (⚫), 10°C (Δ) and –20°C (▾) are included for comparison. The various symbols at each temperature represent data points measured at different feed pressures and binary CO2/H2 mixture compositions. Each open circle on the graph represents the separation characteristics of a different material from the literature. Data at lower CO2 permeability correspond to lower CO2 partial pressures in the feed gas and vice versa. The upper bound is drawn according to a model prediction of this phenomenon (30) with the adjustable parameter f set to 0. The parameter f characterizes the interchain spacing at equilibrium. Rubbery polymers such as those of interest in this work do not exhibit the nonequilibrium excess volume that is associated with nonzero values of f in glassy polymers.

The effect of impurities such as H2O and H2S on CO2/H2 separation properties was investigated at 22°C with the 91 wt % PEGMEA/9 wt % PEGDA copolymer. Although this material crystallizes at about 0°C, it is wholly amorphous at the experimental temperature studied. Because it contains more PEGMEA, this copolymer is more permeable than the 70 wt % PEGMEA/30 wt % PEGDA material presented in Figs. 2, 3, 4. At 25°C, the infinite-dilution CO2 permeability is 380 Barrers for the 91 wt % PEGMEA copolymer and 280 Barrers for the 70 wt % PEGMEA material. The addition of 0.33 mole percent (mol %) H2O vapor (i.e., 100% relative humidity) to a feed gas containing a 1:3 mixture of CO2:H2 at 22°C and a feed pressure of 8.0 atm increased CO2 permeability of the 91 wt % PEGMEA copolymer from 360 to 515 Barrers and its mixed-gas CO2/H2 selectivity from 7.8 to 12. We ascribe this change to plasticization by H2O, which tends to improve CO2/H2 separation performance by a mechanism probably very similar to that by which higher CO2 partial pressure increases CO2/H2 selectivity.

The 91 wt % PEGMEA copolymer was also evaluated with a four-component gas mixture composition representative (on a dry basis) of a synthesis gas stream produced by a commercial General Electric (formerly, Texaco) quench gasifier (31). This mixture contained 1.0% H2S, 12.5% CO2, and 35.7% H2 in CO. The separation of H2S from H2, similar to that of CO2 from H2, requires reverse-selective membrane materials with weak size-sieving ability and a polar nature to interact more favorably with this acid gas. At a feed pressure of 7.8 atm, the copolymer had an H2S permeability of 2500 Barrers and a mixed-gas H2S/H2 selectivity of 50. H2S is considerably more soluble than CO2 in polar polymers (32, 33), and this effect contributes to the much higher H2S permeability than CO2 permeability in the material. Additionally, CO2 permeability and CO2/H2 selectivity remain unchanged in the presence of H2S, indicating the robustness of this series of polymers for CO2/H2 separation. If the H2S partial pressure were higher, it might sorb to high enough levels to alter the gas transport properties of the polymer.

In summary, a family of reverse-selective membrane materials based on highly branched, cross-linked PEO exhibits outstanding separation performance for H2 purification by removing acid gases such as CO2 and H2S from feed streams of practical interest. Moreover, the presence of moisture and high-pressure CO2 in the feed actually improves permeability and selectivity, in contrast to the detrimental behavior associated with plasticizing agents in conventional membrane materials. In addition to hydrogen purification applications, these molecularly engineered copolymers may also be used to remove CO2 and H2S from natural gas as well as SO2 and NH3 from nonpolar gases.

Supporting Online Material

www.sciencemag.org/cgi/content/full/311/5761/639/DC1

Materials and Methods

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article