MXene molecular sieving membranes for highly efficient gas separation

Molecular sieving membranes with sufficient and uniform nanochannels that break the permeability-selectivity trade-off are desirable for energy-efficient gas separation, and the arising two-dimensional (2D) materials provide new routes for membrane development. However, for 2D lamellar membranes, disordered interlayer nanochannels for mass transport are usually formed between randomly stacked neighboring nanosheets, which is obstructive for highly efficient separation. Therefore, manufacturing lamellar membranes with highly ordered nanochannel structures for fast and precise molecular sieving is still challenging. Here, we report on lamellar stacked MXene membranes with aligned and regular subnanometer channels, taking advantage of the abundant surface-terminating groups on the MXene nanosheets, which exhibit excellent gas separation performance with H2 permeability >2200 Barrer and H2/CO2 selectivity >160, superior to the state-of-the-art membranes. The results of molecular dynamics simulations quantitatively support the experiments, confirming the subnanometer interlayer spacing between the neighboring MXene nanosheets as molecular sieving channels for gas separation.


Supplementary Note 2:
The concentration of the colloidal solution was estimated by the gravimetric method. In detail, a blank AAO substrate was first weighed (W0), then a certain amount of the solution (V) was filtered on it to form a supported membrane. Then the dried sample of the MXene membrane supported by AAO substrate was also weighed (W). Finally, the concentration of the colloidal solution (C) was be calculated by the formula as follows: Supplementary Figure 11. FTIR spectrum of the MXene membrane. The stretching vibration at 3457 cm -1 represents -OH, which demonstrates the -OH terminal groups on the MXene surface or the H2O molecules absorbed on it. The absorption band at 1641 cm -1 corresponded to C=O from carbonyl or conjugated carbonyl groups, which might have resulted from the terminating groups connected to the edge of MXene nanosheets 2 .
Supplementary Figure 12. Thermogravimetric analysis result of the MXene membrane. The curve shows a relatively good thermal stability of MXene membrane from room temperature to 1000 °C in nitrogen atmosphere with heating rate of 10 o C min -1 , however there is a slight weight loss of the MXene membrane during heating process. The H2O content around 2.4 % could be deduced from the weight loss occurred at ~120 °C 9 , which is also in accordance with the XPS and FTIR results discussed before.  10 . b, In the Ti 2p region, the majority of the species are Ti atoms (Ti, Ti 2+ , Ti 3+ ) that each belong to a mixture moieties of I, II, and/or IV, and the presence of Ti-F bond belongs to C-Ti-FX (viz. moiety III). The weak Ti-O bond shows the very low oxidation state of the membrane. The presence of the Ti-C and Ti-O bonds is evident from both spectra, indicating the formation of Ti3C2TX with oxygen-containing terminations after treatment. c, In the C1s region, the spectrum is fitted by three peaks. The largest peak is located at 282.0 eV binding energy corresponding to C-Ti-TX (moieties I, II, III, and/or IV). The other two peaks are assigned to graphitic C-C and C-O bounds. In addition, the weak peak at a high binding energy of 288.9 eV is assigned to the O-C=O and/or C-F groups [10][11][12] . d, In the O1s region, the fitted peaks are mainly assigned to C-Ti-OX (moiety I), C-Ti-(OH)X (moiety II) and H2Oads (moiety IV). The H2Oads component (moiety IV) reflects the presence of H2O in the MXene membrane. e, In the F1s region, two fitted peaks correspond to Ti-F and C-F bonds. Note that the presence of C-F is probably derived from the adsorption of exposed C after etching, and the O-C=O species are surface contaminations, similar to the result from the exposure of the high-surface area material to the ambient atmosphere 10-12 . The as-synthesized MXene membranes were characterized by XRD at low angles for dspacing calculation. Four membranes were measured for increased accuracy. The table below shows the detailed parameters. The average d-spacing of the as-synthesized MXene membranes is 13.53 Å based on Bragg's Law with the standard deviation of 0.063, which shows reliable repeatability.

Supplementary Note 3:
The crystalline structures of the MXene membranes were characterized by XRD; the d-spacing was calculated using Bragg's law: where n is an integer (1, 2, 3…), λ is the wavelength of the X-ray, θ is the incident angle and d is the spacing between the diffraction planes 13 .  Table 5). The CO2 curve (red) in main text Fig. 2c fluctuates because CO2 adsorption occurred on the MXene membrane, as indicated by the experimental isothermal adsorption ( Supplementary Fig. 23). When the permeated CO2 molecule re-adsorbed on the membrane, the number of CO2 molecules in the permeate chamber decreased. When CO2 desorbed from the membrane to the permeate chamber, the number of permeated CO2 molecules increased to the former value. Therefore, the fluctuating CO2 curve indicates the balance between the CO2 adsorption and desorption on the MXene membrane. By contrast, the N2 curve (orange) in main text Fig. 2c does not fluctuate, which agrees with the experimental finding that N2 did not adsorb on the MXene. This result also agrees with the simulation results suggesting that N2 had much weaker interactions with MXene (-97.5 kJ mol -1 ) compared to CO2 (-175.1 kJ mol -1 ). In addition, using the H2 flux (0.9) of the long-box simulation (feed chamber length = 12.6 nm, permeate chamber length = 60 nm, see the end of the MD simulation method section), the calculated selectivity of H2/CO2 and H2/N2 increased to 240 and 144, respectively.

Supplementary Note 6:
The terminal groups on the surface of 2D membrane may affect the separation performance to some extent. We have compared the separation performance of Ti3C2O2 with Ti3C2F2, but the difference is not significant (the ideal selectivity of H2/CO2 were 200 and 171 for Ti3C2O2 and Ti3C2F2, respectively), indicating that the effect of the terminating groups on the gas permeation performance is not significant. The insignificant difference may be attributed to the -O to -F transition not bringing sharp change to the MXene-gas molecule interactions. The effect of the channel orderliness on the gas permeation has also been investigated by MD simulations. Taking into account that less orderliness would disturb the interlayer distance, and the uniform interlayer distance would be changed to non-uniform distance with a distribution, i.e. some interlayer distances are larger than 0.35 nm and some are smaller than 0.35 nm. Therefore, the confined diffusion coefficients of gas in the MXene nanochannels with different distances are simulated, demonstrating that gas molecules prefer to diffuse via wider channels with lower mass transfer resistance, which would dominate the gas separation performance. As a result, the membrane with wider interlayer distance gives much lower selectivity of H2/CO2 (e.g. membrane with interlayer distance of 0.45 nm gives a H2/CO2 selectivity of only ~70, which is much lower than the selectivity (> 200) with interlayer distance of 0.35 nm). In other words, deviations from 0.35 nm as little as 0.1 nm would deteriorate the gas selectivity significantly. Therefore, it can be indicated that our MXene membrane structure is highly-ordered, because slight disorder with only 0.1 nm disturbance to the interlayer distance would lead to a poor gas selectivity. Not to mention that disorder usually brings wide slits or flaws between MXene nanosheets, which would even ruin the gas selectivity.  Fig. 3a). The gas permeability decreases with increasing membrane thickness due to the prolonged gas diffusion pathway, while the selectivity increases. Figure 31. Effect of H2 concentration in the feed gas on the separation performance of a 2-μm-thick MXene membrane with mixed H2/CO2 gas feeding. The H2 permeability increases with the increasing H2 concentration in the feed gas due to the enhanced driving force for gas permeation through the MXene membrane, while CO2 permeability decreases. As a result, the H2/CO2 selectivity jumps up to ~600 when feeding with gas mixture of (90% H2 + 10% CO2).

Supplementary
where P is the gas permeance, A is the pre-exponential factor, Eact is the apparent activation energy, R is the ideal gas constant (8.314 J mol -1 K -1 ) and T is the absolute Kelvin temperature (K). ln(P) versus 1/T displays a straight line, whose slope is used to calculate Eact. As shown in Supplementary Fig. 33, the Eact,H2 is about 0.5 kJ mol -1 , and Eact,CO2 is about 3.4 kJ mol -1 .
The apparent activation energy is an association of diffusion activation energy and heat of adsorption, considering much weaker adsorption of H2 on MXene than CO2, heat of adsorption of CO2 on MXene is also expected to be higher than H2. Therefore, diffusion activation energy of CO2 through the MXene membrane is at least 3.4 kJ mol -1 higher than that of H2, which indicates much more activated diffusion of CO2 through MXene membranes or much tighter fit of CO2 with MXene flakes. It can also explain why the H2/CO2 separation factor decreases with increasing temperature (Main text Fig. 3b). Because CO2 permeability rises faster than that of H2, there is more activated CO2 diffusion than H2 in the MXene membrane.

Supplementary Note 8:
For the mechanical testing, the MXene membranes were cut into strips (30 mm×10 mm). The tensile tests were performed at a loading rate of 1 mm min -1 at room temperature.

MXene membrane
Elemental EDX analysis (atomic ratio) For the unknown membrane thickness, permeance is converted to permeability assuming the thickness of 0.1 μm 25,32 . Table 9 and main text Fig. 3d summarized the latest various membranes in separation of H2/CO2, including zeolites, MOFs, ZIFs, GO and MoS2. In similar conditions, MXene membranes display outstanding separation performance in terms of the permeability and selectivity.