Russian Doll Type Cryogenic Traps - American Chemical Society

design in Figure 2), incorporating a heat exchange section which improves cooling, and minimizes problems of warming up the exiting gas stream, which ...
1 downloads 0 Views 100KB Size
Anal. Chem. 1996, 68, 3050-3053

Russian Doll Type Cryogenic Traps: Improved Design and Isotope Separation Effects C. A. M. Brenninkmeijer* and T. Ro 1 ckmann

Division of Atmospheric Chemistry, Max Planck Institute for Chemistry, Mainz, Germany

An improved cryogenic trap for removing micromole quantities of condensables from gas mixtures is described. It is based on the so-called Russian doll design, with which extraordinary trapping efficiencies at flow rates of up to 10 L/min are obtained. The active element consists of one or more nested glass fiber thimbles. Despite the large fiber area, quantitative retrieval of condensed CO2 is obtained. The new design is demountable and incorporates a heat exchange section; trapping efficiency depends little on the level of coolant, as has been tested for CO2. When fractionation is induced by incomplete trapping, a small isotopic enrichment in 13C occurs. It also has been discovered that when He is used as carrier gas, a gas chromatographic effect for CO occurs in Russian doll traps, accompanied by a large isotope separation, in which the elution sequence corresponds to the vapor pressure ratios of the isotopomers. Cryogenic separation of condensable components from gas mixtures is a widespread laboratory practice and various devices are used.1-3 Their design is usually optimized for attaining quantitative trapping, by increasing the length of cooled tubing or using multiple loops.2 These basic designs have the advantage that recovery of the trapped fraction is without problems. Such is of particular importance when a separated compound is to be subjected to isotopic analysis, where fractionation due to incomplete trapping, or incomplete recovery, can change the original isotopic composition. At high flow rates and low concentrations of the condensables, the simple designs are not optimal,4 and traps incorporating a filter3 and/or a support are used.1 A new type of trap incorporating borosilicate glass fiber filters with a high surface area has shown extreme efficiencies for trapping CO2 and N2O from air, even at high flow rates.5 Such traps have become known as Russian doll traps because the active part consists of an assembly of glass fiber thimbles in a nested concentric arrangement, like the traditional wooden Russian dolls. The first generation of Russian doll traps was used exclusively for quantitative (>99.95%) removal of CO2 and N2O from air, at flow rates of up to 10 L/min, and consisted of rugged, versatile, cleanup devices made of stainless steel and suspended using Cajon metal bellows tubing.6 These traps have appeared to be virtually indispensable (1) Smith, R. D.; Mach, R. H.; Morton, T. E.; Dembowski, B. S.; Ehrenkaufer, R. L. Int. J. Appl. Radiat. Isot. 1992, 43, 466-468. (2) Lowe, D. C.; Brenninkmeijer, C. A. M.; Tyler, S. C.; Dlugokencky, E. J. J. Geophys. Res. 1991, 96, 15455-15467. (3) Horibe,Y; Shigehara, K.; Takakuwa, Y. J. Geophys. Res. 1973, 78, 26252629. (4) Graf, W., Ph.D. Thesis, Technical University of Mu ¨ nich, Mu ¨ nich, Germany, 1979. (5) Brenninkmeijer, C. A. M. Anal. Chem. 1991, 63, 1182-1184.

3050 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

in the determination of the isotopic composition of atmospheric CO (20-2000 ppbv)6,7 using the method developed by Stevens and Krout8 who used fragile glass capillary traps. Later, it was established that CO2 could be recovered completely from Russian doll type traps despite the large surface area of the glass fiber thimbles, and a glass version was used for trapping and recovering microliter quantities of CO2.6 Since then, the Russian doll trap has undergone substantial improvements. We also communicate here that the Russian doll traps have a gas chromatographic property for CO, when He is used as carrier gas. The new trap (Figure 1) is of concentric design (cf. the old design in Figure 2), incorporating a heat exchange section which improves cooling, and minimizes problems of warming up the exiting gas stream, which is achieved with a sheathed thermocouple as heater element.9 The central section is provided with undulations to increase surface area and to introduce some turbulence. The lower section has a thin stainless steel cylinder (0.2 mm wall), which forces the incoming gas to the lowest part of the trap body through the outer narrow annulus. Compared to the previous version Russian doll traps,5,6 the available cold surface area is much larger while at the same time the flow speed is reduced giving time for reaching thermal equilibrium. A convenient method for increasing the number of thimbles is to cut a lengthwise slot in a thimble of the same size as used for the outer one and to reduce its length somewhat, so that it exactly fits inside the outer one. Up to three thimbles can be nested this way, using only one size of thimble. The new design has been tested under extreme conditions in a system (Figure 2) for routine separation of small amounts (10200 µL) of CO from air (about 20-2000 ppbv) after its oxidation to CO2.6 The first two old model Russian doll traps5 scavenge H2O, CO2, and N2O to levels well below 0.2 ppbv. Subsequently the CO fraction is oxidized to CO2 using Schu¨tzes reagent (I2O5 with H2SO4 on a silica gel support) and is collected with the new Russian doll trap. A single filter with a diameter of 25 mm and a mass of 1.7 g is adequate for trapping over 99% of the CO2 at the ppbv level and high flow rates. Although by nesting more filters still higher efficiencies can be obtained, the present application focuses on an optimum between quantitative trapping and rapid recovery. After processing typically 340 L of air at 5 L/min at 50-100 hPa, the inlet valve of the trap (Figure 1) is closed and the air pumped away. Next, the exit valve is also shut and the trap (6) Brenninkmeijer, C. A. M. J. Geophys. Res. 1993, 98, 10959-10614. (7) Brenninkmeijer, C. A. M.; Lowe, D. C.; Manning, M. R.; Sparks, R. J.; van Velthoven, P. F. J. J. Geophys. Res. 1995, 100, 26163-26172. (8) Stevens, C. M.; Krout, L. Int. J. Mass Spectrom. Ion Phys. 1972, 8, 265295. (9) Brenninkmeijer, C. A. M. J. Phys. E. Sci. Instrum. 1988, 21, 502-503. S0003-2700(96)00208-9 CCC: $12.00

© 1996 American Chemical Society

Table 1. Recovery of CO2 and the Corresponding Isotopic Ratios (Relative to V-PDB), Using the Trap in Figure 1, and an Air Flow of 5 L/min (STP)a run

sample size (µL)

yield (%)

δ13C (‰)

δ18O (‰)

coolant type

level (cm)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

37.8 39.3 39.4 38.0 39.0 36.3 36.9 28.1 36.0 37.0 36.0 17.0 33.0 6.0

100.0 99.7 100.9 100.1 100.5 99.0 97.3 73.0 36.0 97.5 90.0 46.0 45.0 7.0

-43.93 -43.79 -43.85 -43.99 -43.63 -43.80 -44.21 -45.17 -46.04 -43.98 -44.29 -45.69 -45.78 -51.25

-16.05 -16.12 -16.19 -16.33 -16.09 -16.07 -15.89 -15.0 -18.64 -15.5 -13.92 -9.70 -10.68 -16.24

N2 N2 N2 N2 N2 N2 N2 N2 N2 Ar Ar Ar Ar Ar

24 24 24 24 19 14.5 12 7 3 24 24 12 12 7

a The nominal CO concentration was 100 ppbv and 200 ppbv for 2 runs 9, 13, and 14.

Figure 1. Improved glass Russian doll trap for separating CO2 from air at the ppbv level at flow rates up to 10 L/min. Total length from glass joint G downwards is 33 cm. The diameter is 4.4 cm. The borosilicate glass thimble (Schleicher and Schuell) inner diameter is 2.5 cm, and the thimble mass is 1.7 g. In a standard run, the section submerged in liquid nitrogen is 24 cm. (H) Sheathed thermocouple used as heater, (L) transfer loop, (S) stainless steel sheath, and (T) glass fiber thimble attached using windings of metal wire.

Figure 2. Basics of the preparation system used for the isotopic analysis of atmospheric CO. After complete stripping of H2O, CO2, and N2O with the two stainless steel Russian doll traps, the CO fraction is oxidized using Schu¨ tzes reagent, upon which the formed CO2 is trapped with the improved Russian doll trap.

warmed to room temperature while the small glass loop at the exit is cooled with liquid nitrogen, upon which transfer of CO2 takes place. Subsequently the exit valve is opened to the pump. At this stage a considerable amount of gas, which had become trapped on the filter but which is not condensable at liquid nitrogen temperature without the large surface area of the glass thimbles being at liquid nitrogen temperature, emerges and is pumped away. Now the CO2 fraction can be rapidly transferred and analyzed with an isotope ratio mass spectrometer. The 18O/ 16O ratio of the original CO is calculated by correcting for the contribution of 18O from the Schu¨tze reagent. The isotope ratios are expressed in the usual δ notation as per mil relative deviation from a standard. The effect of coolant level on trapping efficiency and isotopic fractionation for CO2 has been tested (Table 1). Runs 1-4 show the standard performance using a liquid nitrogen level of 24 cm. Reproducible quantitative yields are obtained with consistent isotopic ratios. No significant change was detected at a level of

19 cm (run 5), and even with only 14.5 cm of liquid nitrogen (run 6) no change in efficiency and isotopic ratios occurs. First, at 12 cm liquid nitrogen (run 7), efficiency drops, however, without a marked change in isotopic composition. With very little liquid nitrogen (7 cm, run 8) the efficiency falls to 73% and as previously,6 a depletion in 13C occurs. Unexpectedly a small simultaneous enrichment in 18O occurs, in contrast to earlier tests showing an 18O depletion at 78% yield.6 Using only 3 cm of liquid nitrogen (run 9), the efficiency eventually drops to 36%, with a further depletion in 13C. Surprisingly, now there appears to be a clear depletion in 18O. We have ascertained that the changes in 18O content are not caused by sporadic exchange with oxygen from traces of water. There are at least two effects that may alter the isotopic ratios when efficiency drops. One is that not all CO2 molecules may have had sufficient time to diffuse to the cold surfaces.4 This would imply that only a lighter fraction is trapped. Obviously this effect is inherently small in filter-type traps. The other effect is that the temperature may not be sufficiently low to retain CO2 quantitatively due to the residual vapor pressure at the achieved temperature becoming comparable to the partial vapor pressure. To distinguish these two effects, tests with liquid argon (-185.7 °C) as coolant were performed (runs 10-14). Due to the higher boiling point, some CO2 escapes, and reproducibility is poor. At 12 cm liquid argon (runs 12 and 13) the efficiency drops to about 50%, and again a depletion in 13C takes place. However, a large enrichment in 18O occurs. Using 7 cm liquid argon (run 14), the efficiency drops to only 7%. A large enrichment in 13C is apparent, while a depletion brings the 18O value back to the original value. The important finding is that the efficiency of the new trap is excellent over a wide range of liquid nitrogen levels. Incomplete yields give a systematic isotopic depletion in 13C. For 18O enrichment occurs at intermediate yields, and depletion overrules this at very low yields. If diffusion only limits the efficiency, a mass dependent isotope shift should occur, which for 18O is contradicted by the results. Measurement of yield and δ13C allows calculation of the kinetic fractionation factor involved in the incomplete trapping of CO2 (Figure 3), assuming a Rayleigh-type (10) Craig, H. J. Geol. 1954, 62, 115-118.

Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

3051

Figure 3. Natural logarithm of the change in δ13C versus the corresponding change in yield. A linear fit gives a fractionation factor of 1.0029.

Figure 4. 13C and 18O isotopic composition inferred for the CO eluting from the two Russian doll traps, over the entire range of yields. The δ values are relative to those of the CO gas used. Low yields correspond to collecting only the first fraction of the eluting CO while the reminder was left behind. The yield of over 100% is due to a residue of a previous sample which was not fully eluted. Table 2. Sequential Recovery of Five Fractions of CO2a

condensation process. A fractionation factor of 1.0029 is calculated. This can be compared with a value of 1.004 based on a diffusion controlled fractionation factor as derived by Craig10 for the case of CO2 in air. Although this may indicate that diffusion plays a role for 13C, for 18O the situation is more complex. To investigate the anomalous behavior in 18O fractionation, we conducted tests using helium as carrier gas and injecting CO2 directly upstream of the glass collection trap (point B in Figure 2). The reason for not using CO at point A in these tests is explained later. The rationale for using He is that because the diffusion speeds of C18O16O and C16O16O are more similar in helium, fractionation due to diffusion speed differences can be largely eliminated. However with He, the efficiency of the trap, even using as little as 5-7 cm of coolant, and even with this coolant being liquid argon, remained well above 95%, with little fractionation. Apparently due to the higher thermal conductivity of helium, its low heat capacity, increased diffusion speed, and/ or physical change at the glass fiber surface, the efficiency of the new glass Russian doll trap remained remarkably high. Further attempts to determine the mechanism of the fractionation effects were abandoned. The experiments using He instead of air as carrier gas, but injecting CO, as in our standard procedure6) (at point A, Fig 2), gave totally unexpected results, which lead to further tests exposing a previously unnoticed, interesting property of the Russian doll traps. Some test runs with He produced zero yields (below 1%), and generally, recovery was erratic. The experiments described above prove, however, that this definitely is not due to a loss of efficiency of the CO2 collection trap itself. Because replacement of air with He cannot simply affect the oxidation of CO to CO2 by Schu¨tzes reagent, the tests show the loss or strong delay of CO in the two Russian doll cleanup traps used upstream of the Schu¨tze reactor. No loss or delay occurs with air or pure N2 as carrier. By changing the helium flow speed, and by varying the timing over which the CO2 (from CO) emerging from the Schu¨tze was collected, results with a wide range of yields and fractionations were obtained (Figure 4). The effect is best explained as a cryogenic isotope gas chromatographic effect in the two Russian doll traps. The leading fraction of the eluting CO2 peak (from CO) is strongly depleted in 13C, whereas the trailing fractions show as expected a compensating enrichment (Table 2). Very large isotopic enrichments (δ13C ) 61‰) and depletions (δ13C ) -84‰) are obtained, which cancel each other 3052 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

fraction

yield (µL)

yield (%)

δ13C

δ18O

ACO-435 ACO-436 ACO-437 ACO-438 ACO-439

68.50 67.30 20.70 7.76 8.14

38.40 37.60 11.50 4.30 4.60

-84.16 -41.35 3.11 24.34 61.60

-28.29 -15.15 -2.11 4.01 14.80

172.40

96.40

-43.63

-15.95

total

a A total of 179 mL of CO was injected with δ13C ) -43.60‰ and δ18O ) -16.00‰.

Figure 5. Correlation between δ13C and δ18O for all eluted fractions. The change in δ18O is only 60% of the corresponding change in δ13C.

when the combined recovery is close to 100%. The trailing CO fraction could only be fully recovered within a reasonable time by interrupting the helium flow and flushing the system with CO free air. Furthermore, Table 2 and Figure 4 show that for C18O an identical but smaller effect occurs than for 13CO. This implies an elution order according to mass 28, 30, followed by 29 (13C16O), i.e., according to vapor pressures (12C16O > 12C18O > 13C16O) that do not correspond to the respective molecular masses.11 Figure 5 reveals a compact linear correlation between 13C and 18O, showing that the effect for 18O is only about 60% of that for 13C. Although the effects observed are systematic, interpretation is difficult. Isotope chromatography effects depend on the stationary phase and its interaction with the solute isotopomers. (11) Bigeleisen, J. J. Chem. Phys. 1961, 34, 1485-1493.

For CO2, the elution order of 12C16O18O (mass 46) was opposite to that of 13C16O16O (mass 45) when a switch was made from Poropack column fillings to silica gel.12 In the case of silica gel, the interaction may be influenced by silanol groups, and the same applies to our glass fiber thimbles. Van Hook13 gave results for the chromatographic separation of CH4 and CD4 in which the fractionation changed sign when a tubular glass column with wetted wall was used. For hydrocarbons on a silicone column it is observed that the first fraction to elute is enriched in 13C.14 The reason that the effect we have observed here occurs with He as carrier gas, and not air, is probably due a change in the surface properties of the glass fibers by a low coverage with the very (12) Gunter, B. D.; Gleason, J. D. J. Chromatogr. Sci. 1971, 9, 191-192. (13) van Hook, W. A. Isotope separation by gas chromatography. In Isotope Effects in Chemical Processes; Gould, R. F., Ed.; Advances in Chemistry 89; American Chemical Society: Washington, DC, 1969. (14) Merrit, D. A. Continuous and highly precise analysis of stable isotopes of carbon and nitrogen in gas chromatography effluents. Ph.D. Thesis, Indiana University, Bloomington, IN, 1993. (15) Song, H.; Parcher, J. F. Anal. Chem. 1990, 62, 2616-2619.

volatile He. However, although the surface properties do play a role, it seems that the vapor pressure differences determine the separation. A chromatographic effect due to a stationary liquid film15 can be excluded. We conclude that the improved Russian doll trap is very efficient in trapping and releasing minute amounts of CO2 extracted at the ppbv level from an air flow of 5 L/min and more. The level of liquid nitrogen is not critical. The new design can be easily constructed and should be of benefit in many instances in which quantitative removal of condensables is required. Furthermore, it was found that the Russian doll traps have a gas chromatographic property, which may have useful applications.

Received for review March 1, 1996. Accepted May 17, 1996.X AC960208W X

Abstract published in Advance ACS Abstracts, July 15, 1996.

Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

3053