3.1. Experimental apparatus

The test apparatus for dynamic adsorption experiments is shown schematically in Fig. 1. N₂ and Xe were used as the carrier gas and adsorbate, respectively, while the impurity gas included CO₂, CO and water vapor. The pressure sensor, dew point meter and flow meter were used to measure inlet and outlet pressures, dew point and flow rate, respectively. A gas chromatograph (GC) was used to measure the concentrations of Xe, CO₂ and CO.

Fig. 1Full-screen View

The test apparatus for dynamic adsorption experiments

The experiment procedures are as follows. The N₂ and Xe gases were pressurized initially at 0.4 MPa using their respective regulators. The mass flow controllers were adjusted to obtain desired Xe concentration and gas flow rates so that the gases could be mixed adequately and homogeneously. The gas pressure was then adjusted using the adjustable valves at the bed inlet, after which Xe concentration of the inlet gas was determined using the GC (the gas circuit denoted by the dotted line in Fig. 1), employing a 13XHP molecular sieve column to remove water vapor and CO₂ before these species condensed inside the packed bed at low temperature. The pre-cooling column and the fixed bed adsorber were placed in a thermostat filled with LN₂ at 77 K. Various fixed bed adsorbers were used, which were made of cylindrical copper tubes and filled with different adsorbents (hereinafter referred to as the adsorption columns). After the effluent gas from the buffer tank achieved a pressure sufficiently stable for the GC analysis, the Xe concentration in the effluent was monitored at 5-min intervals. At the end of gas circuit, the flow rate was determined with the soap-film flow meter, and dew point was monitored by dew point meters near Valve 4 and the buffer tank.

The effects of gaseous impurities on Xe adsorption were determined by switching on the pressurized containers of CO₂ and CO or the water vapor generator. This was accomplished by closing Valves 3, 4 and 6 and opening Valve 5.

3.2. Adsorbents

The adsorbents included carbon molecular sieves and granular coconut shell-activated carbon. The 01-CMS, 501CMS and 601CMS were produced by Shanghai Institute of Fine Chemicals, China. TJ-CMS was manufactured by Tianjin Chemical Reagent Co., Ltd., China. NM-GAC was produced by Zhongsen Activated Carbon Co., Ltd., Inner Mongolia, China. HN-GAC was procured from Shuangxinlong Industrial and Trading Co., Ltd., Hainan Province, China. Properties of the adsorption columns and adsorbents are summarized in Table 1, where S_BET is the BET specific surface area; V_t is the total pore volume; S_micro and V_micro are the specific surface area and the pore volume for micropores calculated with the t-Plot method, respectively; L_micro is the diameter of micropores calculated with the D-R model; and E₀ is the characteristic energy of adsorption.

3.3. Gas chromatography

A HP6890 GC equipped with a thermal conductivity detector (TCD) was used to analyze the gas concentrations. The operational conditions are presented in Table 2.

4.1. Dynamic adsorption of Xe on different adsorbents

Characteristics of dynamic Xe adsorption on different adsorbents were determined using the gas circuit in Fig. 1. The results are given in Table 3. Based on the specific Xe desorption volume of 8 for activated carbon or a carbon molecular sieve [15], and purities of Xe adsorbed by different adsorbents, the overall adsorption rate constant (k_v) could be calculated by Eq. (6).

Typical breakthrough curve of TJ-CMS at 77 K is shown in Fig. 2(a). The curve is S-shaped, and multilayer adsorption can be seen clearly. The Xe boiling point is 165 K, while LN₂is at 77 K, xenon may be liquefied. The kinetic energy of xenon molecules then decreases, and the inner pressure of the adsorber increases. This increases the contact time between the adsorbent pore surfaces and xenon molecules, and xenon in liquid state is captured by carbon adsorbents. Meanwhile, because of the large specific surface area of carbon adsorbents, more adsorption sites can be provided for xenon. These factors cause advantageous conditions for larger adsorption capacity.

Fig. 2Full-screen View

(Color online) Breakthrough curve (a) and adsorption isotherm (b) of TJ-CMS at 77 K.

From Table 3, k_dB and k_d values of the carbon molecular sieves are higher than those of the activated carbon. For carbon molecular sieves, the maximum k_dB and k_d values for the TJ-CMS are 385.5 and 1417 L/g, respectively, and the purity of Xe is 82%, which is 2.1×10⁴ times the inlet concentration of Xe. For activated carbons, the k_dB of the NM-GAC is greater than that of the HN-GAC, though its k_d is lower. The LUB of 01-CMS, TJ-CMS, and 501CMS are approximately 14 cm. The k_d of NM- GAC is 547.4 L/g at 77 K, which is over 200 times the k_d at 276 K [9]. It is apparent that multilayer adsorption occurs on the CMS and activated carbons at 77 K, leading to the much higher adsorption capacities of the materials. From Table 1, TJ-CMS has a larger BET area and micropores surface area, with narrower micropore diameter than those of other adsorbents, which means larger adsorption capacity and better adsorption selectivity for Xe. The k_dB values are important, and they are commonly used to design the first stage of an adsorption column to ensure high recovery of Xe.

The results indicate that the minimum k_v is 5010 min⁻¹ (TJ-CMS), and the maximum k_v is 9034 min⁻¹ (NM-GAC). The k_v varies with the gas diffusion rate, which depends on the volume proportion of macropores and medium pores.

For confirming the multilayer adsorption of Xe on carbon adsorbents, the adsorption equilibrium isotherm of TJ-CMS was measured at 77 K with nitrogen as the analysis gas by autosorb iQ Station (Quantachrome Instruments, USA). The isotherm of TJ-CMS is shown in Fig. 2(b). According to the classification of IUPAC [16], the adsorption isotherm of TJ-CMS at 77 K is Ⅰ-type, which agrees well with the multisite Langmuir model. Micropore filling occurs at smaller P/P₀ ratios. Due to the existence of macropores and medium pores, multilayer adsorption takes place as the relative pressure P/P₀ increases; while as P/P₀ increases, the phenomenon of capillary condensation appears. Both multilayer adsorption and capillary condensation are useful for enhancing the adsorption. However, when the degree of capillary condensation becomes too large, the adsorbed gas may be frozen, and it may even jam the fixed bed adsorber.

4.2. The Xe concentration effects

The Xe concentration affects the dynamic adsorption performance. The coefficients of Xe dynamic adsorption on TJ-CMS tested at different Xe concentrations are shown in Fig.3(a). The k_dB and k_d decrease with increasing Xe concentration until k_d=117×10⁻⁶ V/V, where they begin to increase. Polynomial fittings were performed between lnk_dB or lnk_d and lnC₀, and the results are shown in Eqs.(7) and (8).

Fig. 3Full-screen View

(Color online) Effects of Xe concentration on k_dB and k_d of TJ-CMS (a) and the fitted line (b) between Xe concentration (C₀) and adsorbed amount (q) of Xe on TJ- CMS, at 492– 519 mL/min of low rate and 19.9×10⁻⁶–235×10⁻⁶ V/V of Xe concentration.

In Fig. 3(b), the adsorbed quantity increases linearly with the Xe concentration. As the concentration increases, the gases in the adsorption column tend to liquefy at 77 K, and capillary condensation occurs. Also, because of the large specific area of TJ-CMS, sufficient adsorption sites can be provided, and the liquefaction of Xe at 77 K can be captured easily by micropores of the TJ-CMS adsorbent. According to the gas-solid adsorption theory, the equilibrium adsorption amount of gas adsorbent increases as the partial pressure of the key gas component increases. The increasing Xe concentration is equivalent to the increasing partial pressure of Xe. Therefore, the above factors result in the decreased adsorption coefficient and the increased adsorption amount.

The k_v values listed in the Table 4 indicate that k_v varied slightly with the Xe concentration.

4.3. The gas flow rate effects

Using the TJ-CMS carbon molecular sieves, the effects of gas flow rate on the dynamic adsorption of Xe at 77 K were studied under flow rate of 509–1348 mL/min and Xe concentrations of 37×10⁻⁶– 45×10⁻⁶ V/V. As shown in Fig.4, the k_dB and k_d decrease with increasing flow rate, from k_dB=380 L/g and k_d=1500 L/g at the gas flow rate of 509 mL/min, to k_dB=150 L/g and k_d=450 L/g at the gas flow rate of 1348 mL/min. Because the linear velocity of gas increases with the flow rate, the gas flowing through the adsorption column may be not adequately cooled, hence the increase in thermal motion of the gas molecules. Under these conditions, flow rate makes a great difference in the dynamic adsorption performance.

Fig. 4Full-screen View

(Color online) k_dB and k_d(a), and k_V (b), as function of the flow rate, at 509–1348 mL/min of flow rate and 37×10⁻⁶– 45×10⁻⁶ V/V of Xe concentration.

As shown in Fig. 4(b), k_v increases with the gas flow rate. This agrees with the results in Ref. [15,17]. As the Xe diffusion rate increases with the gas flow rate, the contact between Xe molecules and the adsorption interfaces of adsorbents becomes easier, and contact time shortens. Therefore, k_v increases with the flow rate, but the adsorption capacity decreases.

4.4. The inlet pressure effects

The inlet pressure effects on the dynamic adsorption of Xe at 77 K were assessed at the gas flow rates of 565–573 mL/min and Xe concentration of 33×10⁻⁶– 45×10⁻⁶ V/V. As shown in Fig. 5, the k_dB and k_d increase with the inlet pressure, agreeing with reports in Ref.[10]. The inlet pressure is an most important factor for designing adsorption columns and gas circuits. The k_v data in Table 5 indicate that k_v changes slightly with the inlet pressure.

Fig. 5Full-screen View

(Color online) The k_dB and k_d as function of the inlet pressure (P), at flow rate of 565–573 mL/min and Xe concentration of 33×10⁻⁶– 45×10⁻⁶ V/V.

4.5. The inner diameter effects

The adsorption capacity of adsorbents is known to decrease with increasing linear velocity of the gas. A common way of reducing the effects of a high linear velocity is to increase the pre-column pressure and inner diameter of the adsorption column. Therefore, we studied inner diameter effects of the adsorption column on Xe dynamic adsorption over the carbon molecular sieve TJ-CMS at 77 K. The results are summarized in Table 6. The linear velocities in each test were about the same. While the k_dB data obtained at the inner diameters of 1.6 and 4.2 mm were similar (283 and 257 L/g), the k_dB became 1215 L/g at the inner diameter of 6.8 mm, when the inlet pressure increased to 401 kPa from a few hundred kPa.

4.6. The CO₂ concentration effects

The CO₂ concentrations effects on the dynamic adsorption of Xe were examined on TJ-CMS at 77 K, under the gas flow rates of 518–576 mL/min and Xe concentration of 37×10⁻⁶– 40×10⁻⁶ V/V.As shown in Fig. 6(a), the k_dB value on TJ-CMS decreases with the CO₂ concentration and then increases. Because of the competitive adsorption, the existence of CO₂ results in the decrease of xenon adsorption capacity. The sublimation temperatures of CO₂ and Xe are 194 K and 165 K, respectively, so CO₂ can be liquefied easier than xenon, and the extent of gas liquefaction inside the adsorption column increases with the CO₂ concentration. The extent of capillary condensation is very high during the adsorption process of CO₂, and the micropores used to adsorb Xe are occupied by CO₂ molecules so that the micropores volumes which are fit for the adsorption of xenon become smaller. The inlet pressure versus the ventilation duration at 2722×10⁻⁶V/V of CO₂ is plotted in Fig. 6(b). It shows that the inlet pressure would increase to some degree due to CO₂ liquefaction. As the CO₂ concentration increases, the flow rate reduces owing to CO₂ condensation. So, CO₂ must be removed prior to Xe adsorption.

Fig. 6Full-screen View

The dynamic adsorption of Xe as a function of CO₂ concentration (a) and the pre-column pressure vs ventilation duration (b) at CO₂ concentration of 2722×10⁻⁶V/V, flow rate of 518– 576 mL/min and Xe concentration of 37×10⁻⁶– 40×10⁻⁶V/V.

4.7. The CO concentration effects

The CO concentration effects on the dynamic adsorption of Xe were assessed with TJ-CMS at 77 K, under the gas flow rate of 509–554 mL/min, Xe concentration of 35×10⁻⁶–37×10⁻⁶V/V and inlet pressure of 114–129 kPa. The data are summarized in Table 7. The k_dB and k_d decrease with increasing CO concentration. Therefore, CO must also be removed before Xe adsorption.

4.8. The water vapor concentration effects

According to Ref.[1], water vapor tends to reduce the adsorption of Xe, and it may also form ice plugs within the adsorption column. The water vapor effects on the dynamic adsorption of Xe were therefore analyzed using the carbon molecular sieve TJ-CMS at 77 K. The k_dB values are 415.8, 256.5 and 245.0 L/g at moisture contents of 1×10⁻⁶, 76×10⁻⁶ and 553×10⁻⁶ V/V, respectively, at Xe concentrations of 36.6–38.5×10⁻⁶V/V and flow rates of 533–575 mL/min. The k_dB values over TJ-CMS decrease with increasing water vapor concentration. Moreover, at higher concentrations, the water vapor readily condenses and freezes in the adsorption column, creating a blockage. Then, water vapor must be removed before Xe adsorption.

4.9. Uncertainty estimation

The sources of uncertainties of the dynamic adsorption coefficients of Xe include the uncertainty of the concentration of standard gases (Xe, CO₂, and CO; <2%), the uncertainty of the values of flow rates with the flow rate controller (<2%), the uncertainty of the test values of the pressure sensor (<2%), and the uncertainty of the peak area of GC (<3%). The above entries are independent from each other, and the combined standard uncertainty of the dynamic adsorption coefficient is less than 10% (k=2).