1112
Anal. Chem. 1989, 67. 1112-1117 0,512
1 2 3 4 5 6
Hexanesulfonate Octanesultanate DI-Isopropylamine Triethylamine Hexylamine Trlpropylamlne
Figure 3. Test of the equation for relative response (eq 22). A is naphthalene-2-sulfonate: the analytes used are numbered in the figure. The data are taken from ref 15.
for site contributes to about 10% of the relative response. It is interesting to note that the relative response for uncharged analytes amounts to about 10% of the response for charged analytes (15). More elaborate studies are needed to quantitatively test all predictions made in the presented theory.
LITERATURE CITED (1) De Vault, D. J , Am. Chem. SOC. 1943, 65, 532. (2) Helfferich, F.; Klein, G. Multicomponent Chromatography; Marcel Dekker: New York, 1970. (3) Riedo, F.; Kovats, E. J . Chromatogr. 1982, 239, 1 (4) Helfferich, F. J . Chromatogr. 1986, 373, 45. (5) Mangelsdorf, P. C., Jr. Anal. Chem. 1968, 38. 1540. (6) Melander, W. R.; Erard, J. F.; Horvath, Cs. J . Chromatcgr. 1963, 282, 211. (7) Melander. W.R.: Erard, J. F.: Horvath, Cs. J . Chromatogr. 1983, 282, 229. (8) McCormick, R. M.; Karger, E. L. J . Chromatogr. 1980, 199, 259. (9) Bidlingmeyer, B. A,; Warren, F. V., Jr. Anal. Chem. 1982, 5 4 , 2351. (10) Stranahan, J. J.; Deming. S. N. Anal. Chem. 1982, 54, 1540. (11) Vigh, G.; Leitold, A. J . Chromatogr. 1984. 312, 345. (12) Levin, S.; Grushka. E. Anal. Chem. 1988, 58, 1602. (13) Levin, S.;Grushka, E. Anal. Chem. 1987, 59, 1157. (14) Takeuchi, T.; Watanabe, S.; Murase, K.; Ishii, D. Chromatographia 1988, 2 5 , 107. (15) Cromrnen, J.; Schill. G.; Westerlund, D.; Hackzell, L. Chromatographia 1987, 2 4 , 252. (16) Denkert, M.; Hackzell, L.; Schill, G.; Sjogren, E. J . Chromatogr. 1981, 218, 31. (17) Hackzell. L.; Schill, G. Chromatographia 1982, 15, 437. (18) Hern6, P.; Rensen, M.; Crommen, J. Chromatographia 1984, 79, 274. (19) Perkin, J. E. J . Chromatogr. 1984, 287, 457. ( 2 0 ) Stihlberg, J. J . Chromatogr. 1988, 356, 231. (21) Stlhlberg, J.; Furangen, A. Chromatographia 1987, 2 4 , 783. (22) Stihlberg, J.; Bartha, A. J . Chromatogr. 1988, 456, 253. (23) Stahlberg, J.; Hagglund, I. Anal. Chem. 1988, 6 0 , 1958. (24) Stihlberg, J. Chromatographia 1987, 2 4 , 820.
RECEIVED for review October 19, 1988. Accepted January 31, 1989.
Calcium Chabazite Adsorbent for the Gas Chromatographic Separation of Trace Argon-Oxygen Mixtures Peter J. Maroulis* and Charles G. Coe Air Products and Chemicals, Inc., Allentown, Pennsylvania 18195
Calcium chabazite used as a gas chromatographic packing provides a practical means for analyzlng trace levels of a variety of permanent gases Including oxygen and argon. An outstandlng feature of properly activated calcium chabazlte is its ablllty to resolve argon and oxygen at temperatures above 343 K. The level of dehydration achieved for thls adsorbent has a direct Influence on its efficlency and ability to resolve Ar and 02. I n addition, we found that treating the chabazlte at elevated temperature in an oxidizing atmosphere improves the resolution and lowers the detectability limits for determlnlng oxygen. Part-per-bllllon levels of Ne, Ar, O,, N,, CH4, and CO could be measured by udng a calcium chabazlte column, in comblnatlon with a helium ionization detector. The column efflclency and resolution of this adsorbent were measured as a function of corrected flow rate and column temperature for AI-0, mixtures. Trace Ar and 0, were base-llne-resolved in under 2 min by using a 6-ft column of this adsorbent at 343 K. Thus, calclum chabazite provides a practical adsorbent for analyzing a wide variety of gases wlth conventional Instrumentation.
INTRODUCTION The chromatographic performance of a zeolitic adsorbent depends on many properties, including the particular structure 0003-2700/89/0361-1112$01.50/0
and composition of the zeolite. The number, type, position, and hydration state of the charge-compensating cation present in a zeolite strongly influence its selective adsorption and separation properties ( I ) . As a direct result of fundamental studies on zeolitic adsorbents, we developed a superior chromatographic material based on certain ion forms of chabazite. The chabazite-based chromatographic packing reported here is a versatile adsorbent and has the unique ability to quantitatively separate argon from oxygen a t ambient temperatures. We also found that treating the chabazite in an oxidizing atmosphere a t elevated temperatures improved the resolution of oxygen and argon. This lowers the oxygen detection limit to that of the detector. In addition to practical methods for determining low-part-per-billion by volume (ppbv) levels of oxygen, we found the calcium chabazite useful for determining similar levels of Ne, Ar, N2, CHI, and CO. There are scattered reports of the utility of chabazite for gas separations. Other researchers disclose that calcium chabazite gives the highest nitrogen/oxygen selectivities of any known zeolite but do not mention or discuss its use in separation of argon from oxygen ( 2 ) . The data and experimental procedures they present suggest that the calcium chabazite was not thoroughly dehydrated and therefore would not separate argon from oxygen. Vaughan, using gas chromatography to study the influence of the cation on the adsorption properties of chabazite, showed a useful separation t 1989 American Chemical Society
ANALYTICAL CHEMISTRY. VOL. 61. NO. 10. MAY 15. 1989
of oxygen from nitrogen hut did not mention oxygen-argon separation (3). In both the above cases, the chahazites were only dehydrated a t 598-623 K for 1 h. This is insufficient to dehydrate the adsorbent thoroughly. The use of synthetic zeolites for the gas chromatographic analyses of permanent gas mixtures is very old. The first papers preceed 1960. However, there are few early reports of the use of zeolites (5A and 13X) for the specific separation of Ar and 0,under ambient conditions. The two most relevant are described. Walker published a brief note describing the effect of heat treatment at 673 K or greater on the argon-oxygen separation (4). Only retention times were given with no mention of resolution. In addition he stated that Ar/O, resolution deteriorated after 5 days (5). A more thorough study of this same phenomenon clearly demonstrated the ability of heat-treated 5A and 13X to separate Ar from 0,.Karlsson showed that these columns had life in excess of 6 months if argon was used as a purge gas during the thermal activation and high-purity gases were employed for the analyses. In our hands we could not base-line-resolve Ar from 0,at 303 K over 5A even after heat-treating i t for 8 h a t 673 K, whereas the chahazites used in this study (vide infra) consistently separated Ar from O2 under these conditions. At present, the analysis of argon in the presence of oxygen is usually done with very long, cumhenome columns,cryogenic temperatures, preliminary removal of oxygen by catalysts, and/or combinations of these methods. Partial resolution of argon from oxygen has been demonstrated by using comhinations of NaA and either CaX or CaA, hut these separations were not base-line-resolved and could only he carried out at 273 K or lower (6). A complex multistep procedure requiring constant regeneration of a pretreatment column to scavenge the oxygen has been reported for analyzing argon and nitrogen in oxygen (7). Custom-made porous polymers have been used as chromatographic packing hut give poor Ar/O, separation (8). A recent development for the analysis of argon in the presence of oxygen is the use of a capillary column coated with a thin film of zeolitic adsorbent (9,101. This capillary analysis requires a small sample loading, which is generally not applicable for trace gas analyses. Sievers showed that porous polymers containing metal complexes would separate 0,from AI, hut the AI co-elutes with Nz and cannot he used to analyze argon in the presence of both O2 and N, (11). In contrast, the work reported here describes the preparation and utility of calcium chahazite as a chromatographic adsorbent for these separations as well as others. EXPERIMENTAL SECTION Preparation of Calcium Chabazite. A 500-g sample of 0.0625-in. pellets of AW-500 (Linde Division of Union Carbide Corp., Danbury, CT) was exchanged four times at reflux for 2 h, using 1 L of 1 M CaC1, for each exchange. As received, this chahazite contained significant amounts of calcium, magnesium, sodium, and potassium along with small amounts of strontium and iron. The commercial AW-500 also contained some erionite as a contaminant phase. After it was washed exhaustively with deionized water, the calcium chahazite was placed in a flat pan and air-dried at ambient temperature. Elemental analysis of the exchanged adsorbent indicated that 90% of the exchangeable cations were in the calcium form. The zeolite material was ground to a 60180 mesh and placed in 0.125 in. 0.d. X 0.085 in, i.d. 304 stainless steel columns 1.5 and 6.0 f t long. Further details of these exchange procedures are reported elsewhere (2.2). Column Activation. The columns were activated under flowing helium with the use of specific temperature ramps, final temperature, and length of activation. The following procedure is typical. The column containing calcium chahazite was gradually heated at 3 K/min from 313 K to 673 K. Once a t 673 K the material was heat-soaked for an additional 24 h. The material was then cwled to ambient temperature with helium psssing over
1113
MOLECULAR *!EYE
VILCO ~ - P o R T P U R O E O - H O U S I W ( ~ ~ ~ ~ ~ ~YILVEIGSYI MPL~NC
Figure 1. Experimental setup for measuring effiiency and resolution of Ar and 0, separations as a function of flow rate and column temperature. the material. The helium flow rates measured a t ambient temperature were 185 mL/min at a column temperature of 313 K and 67 mL/min at a column temperature of 673 K, inlet helium pressure was 75 psig. In some cases the dehydrated chahazite was oxidized in a flow of C0,free synthetic air (a 79/21 vol 90mixture of nitrogen and oxygen, containing ppmv (part-per-million by volume) levels of less than 1 for CO,, 2 for H,O, and 0.5 for total hydrocarhon). The following oxidation conditions were typical. The column of chabazite was gradually heated at 10 K/min from 313 K to 523 K. Once at 523 K, the adsorbent was oxidized for another 1.5 h. The air flow rate measured at ambient temperature was 130 mL/min at a column temperature of 313 K. In certain cases the dehydration was carried out hy using the synthetic air instead of an inert purge, allowing the chahazite to he both dehydrated and oxidized in a single activation step. Column Efficiency and Resolution. Column efficiency and resolution were determined as a function of both corrected carrier gas flow rate (FJ and column temperature (T,) on a 6-ft column. Flow rate experiments were conducted hy varying the rate from 7.9 to 35.8 mL/min at a constant T. of 323 K. This experiment was then repeated a t 343 K, using almost identical flow rates. These two temperatures were selected because they resulted in a high degree of resolution in a short period of time, as indicated hy 0, capacity factors of 5.28 at 323 K and 3.34 at 343 K. The effect of T, on the resolution of Ar and 0,was examined by varying T, from 313 K to 363 K a t a constant Fc of 10.9 i 0.14 m L / min . Apparatus and Procedures. Chromatographicdata needed to determine the efficiency, resolving ability, and heats of adsorption of the calcium chahazite were collected by using a Hewlett-Packard 5890 gas chromatograph (Hewlett-PackardCo., Valley Forge, PA) with a thermal conductivity detector (TCD) and a Valco eight-port gas sampling valve (Valco Instrument Company, Inc., Houston, TX). Figure 1schematically illustrates the experimental setup. Initially, pressure gauges were placed at the column's inlet and exit. However, because there was no observed pressure drop between the column exit and TCD under a wide range of flow rates and column temperatures, the exit pressure gauge was removed. The sample volume for all these experiments was 0.5 cm3. A Hewlett-Packard 3393A integrator collected retention time (t,) and peak width at half-height ( W0d data. Interfaced to the same HP5890 gas Chromatograph was a Tracor Model 1% ultrasonic detector (USD)(Tracor Instruments, Inc., Austin, TX) to study dilute gas mixtures, A Spectra-Physics 4270 integrator (Piscataway, NJ) was used to record the USD detector signal Ultratrace-gas analyses (pphv levels) of permanent gases were performed with a Valco Model 1000 trace-gas analyzer (TGA). This is a gas chromatography equipped with two column ovens, a He-purged pneumatic housing, four-port and eight-port gas sampling valves, and a helium ionization detector (HID). The purged housing minimizes ambient air diffusion into the sample and carrier gas lines. The HID has a detector volume of 0.180 mL and uses a 1-Ci titrated scandium foil as a 6 source. Sample loop volumes of 1.0 and 0.50 cm3were used, A SP4270 integrator monitored the HID output signal. Selected gas blends were prepared to characterize the Arlo, separating ability of calcium chahazite and to perform application
1114
ANALYTICAL CHEMISTRY, VOL. 61, NO. 10, MAY 15, 1989
studies. A gas standard containing 211 ppmv Ne, 199 ppmv Ar, and 216 ppmv O2 in He was used to characterize the separating ability of calcium chabazite. In addition, standard mixtures of 5.1 H,, 10 Ar, 5.3 02,5 N2,4.0 CO, and 3.9 ppmv CH, in He and a second mixture containing 5 ppm Ar in O2 were used in application studies. Standards were prepared at the ppbv level by standard dilution techniques. Research grade He was used as the carrier gas with the TCD, USD, and HID. In addition, for most analyses with the HID, the He carrier gas contained 25-30 ppmv Ne as a dopant (13). All gases and gas mixtures were supplied by Air Products and Chemicals, Inc. (Allentown, PA). In all experiments, the carrier gas was passed through a 5-A molecular sieve immersed in liquid nitrogen. To determine the efficiency and separating ability of calcium chabazite, the number of theoretical plates (n),height equivalent per theoretical plate (HETP), capacity factor ( k ) ,resolution ( R ) , and selectivity ( a )were calculated from GC-TCD data. These factors were determined as a function of both flow rate and temperature. In addition, heats of adsorption (AH) were calculated for Ar and 02.These relationships and physical quantities were calculated according to standard procedures that are well accepted and similar to those used by several workers (e.g. ref 14-16)
RESULTS AND DISCUSSION Resolution and Efficiency. A series of experiments was and resolving ( R ) conducted to determine the efficiency (amm) ability of calcium chabazite to separate Ar and O2 in He. When T, was maintained a t 323 K, maximum Ar and 0, column efficiencies occurred a t F, values of 20.6 and 24.0 mL/min, respectively. This is indicated for Ar and O2 by of 2030 and maximum numbers of theoretical plates (amm) 2098, respectively. The capacity factors ( k ) at this temperature were 4.02 f 0.043 for Ar and 5.23 f 0.058 for 02.At 323 K, Ar and O2 were base-line-resolved ( R 2 1.5) over the entire flow rate regime used in this study. Over this range, R increased from 1.76 to a maximum of 2.44 at an F, value of 20.1 mL/min and then decreased to 2.16. We repeated the efficiency and resolution experiments for Ar and O2 a t a T , of 343 K. The experiments showed the optimum column efficiency for Ar and 0, was achieved a t an F, of 25.6 mL/min. The nmaxcalculated a t these conditions for Ar was 1962, and for O2 was 2029. The k values for Ar and O2 in the F, range studied were 2.72 f 0.024 and 3.33 f 0.028, respectively. Under all flow conditions studied, except a t the slowest Fc, Ar and O2 were base-line-resolved. The lowest R value (1.24) was obtained a t the slowest flow rate (F, = 8.35 mL/min), but it still indicates adequate resolution (R 2 1.0)for quantitation of the two components. The highest degree of resolution determined in our experiment was 1.68. a t an F, value of 24.2 mL/min. The flow rate required for optimum column efficiency is not necessarily the best practical operating condition. Lower column efficiency can also result in base-line resolution of the two components in a much shorter time. A series of experiments showed that an optimum resolution of 2.4 was achieved with an elution time of 1.5 min for Ar and 2.0 for 0,. In addition, under all conditions tested a resolution of 1.5 or greater was obtained. For example, a t 343 K it takes 1.27 min to elute O2 a t the optimum H E T P flow rate, but O2 can be base-line-resolved from Ar in 0.83 min at an F, of 38.1 mL/min. Hence, the time required to analyze a sample for Ar and 0, can be reduced by as much as a third by using a faster flow rate than the one yielding optimum column efficiency. The effect of temperature in A r / 0 2 resolution was determined for temperatures from 313 K to 363 K a t a constant F, of 10.9 f 0.14 mL/min. Data obtained from this set of experiments were used to determine a , k , and R values for Ar and 0,. From the data in Table I one can determine that k, a , and R decreased nonlinearly as T, increased.
Table I. Capacity Factor, Separation Factor, and Resolution Data Determined As a Function of Temperature for Ar and O2 temp, K
Fc, mL/min
k(Ar)
MOP)
a(Ar/OJ
R(Ar/Oz)
313 323 333 343 353 363
10.7 10.8 10.9 10.8 11.1 10.9
5.08 4.08 3.32 2.75 2.28 1.95
6.84 5.31 4.19 3.37 2.74 2.28
1.35 1.30 1.26 1.23 1.20 1.17
2.37 2.03 1.70 1.39 1.10 0.893
Heats of Adsorption. We also determined heats of adsorption for Ar and O2 on calcium chabazite. The specific retention volume (V,) for each component was calculated from data obtained over a T, range of 313-363 K. From this data, the A H values for Ar and O2 were found to be -5.1 and -5.7 kcal/mol, respectively. The isosteric heats for O2 in the Henry law region for typical zeolites such as 4A, 5A, and 13X are all less than 5 kcal/mol. As stated previously, the increased interaction with the surface as evidenced by the higher leads to an O2 separation. Dehydration Effects on Selective Adsorption. When both molecules are small enough to enter the pore volume of a zeolite, their separation is affected by the energetics of competitive adsorption. If the entropy contribution is about the same, then selectivity for one adsorbate over another a t equilibrium is a direct result of the differences in their heats of adsorption. The heat of adsorption for a given adsorbate is strongly influenced by the nature of the charge-compensating cation that is interacting with the adsorbate. Any water, hydroxyl group, or other ligand interacting with these cations lowers its effective charge density and therefore decreases the heat of adsorption (17). For weakly interacting adsorbates, this drop in AH is generally large enough to reduce or in some cases totally eliminate the separation between components in a chromatographic analysis. Highly exchanged, thoroughly dehydrated polyvalent forms of zeolitic adsorbents, especially chabazite, have much higher heats of adsorption for weakly interacting adsorbates than the corresponding monovalent forms. These higher heats lead to a significant increase in their ability to separate weakly adsorbing gases, particularly in the Henry law region where gas chromatography is generally carried out. Barrer has suggested that chabazite in its calcium form is perhaps the most energetically sorbing of all zeolites for both polar and nonpolar gases and therefore should have special advantages as an adsorbent (18). However, the use of chabazite as a versatile chromatographic adsorbent was largely overlooked. One reason may be the degree of thermal dehydration required to adequately dehydrate a calcium chabazite. Previously, we reported the effect of thermal activation procedures on the adsorption properties of highly exchanged polyvalent forms of X zeolite (19). Compared to the sodium form, polyvalent forms of zeolites need longer times and/or higher temperatures to fully dehydrate. This is particularly true for chabazites. Previous workers in the field of gas adsorption failed to recognize that the conditions required to achieve this level of dehydration vary even for a given zeolite structure, depending upon the nature of the charge-compensating cations. I t is likely that the chabazites studied previously in chromatographic applications were not thoroughly dehydrated; therefore their intrinsic adsorption properties were not observed. When completely dehydrated and a t least 50% calciumexchanged, chabazite exhibits unusually high limiting heats of adsorption for weakly interacting adsorbates. In chromatographic applications this property allows oxygen to be
ANALYTICAL CHEMISTRY, VOL. 61, NO. 10, MAY 15, 1989 A
1115
0
n
s
m
A
Flgure 2. Separation of trace Ar and 0, in He at 343 K using a 6-ft column linked to an HP TCD detector.
separated from argon. We believe the oxygen/argon separation is a direct result of the large electric field density present in the zeolitic cavity. The small quadrupole of O2interacting with the strong field gives rise to a higher heat of adsorption compared to that of argon and thus has a selectivity for 0, in the presence of argon. Oxygen has a smaller quadrupole moment than nitrogen and concomitantly a smaller degree of interaction with a given electrostatic energy field. Thus O2 is much less sensitive to the charge density of the cation present; and, O2requires the presence of very high electric fields to be selectively adsorbed in the presence of molecules having similar energetics but no quadrupole moment (such as argon). Applications. Our work shows that calcium chabazite is an improved gas chromatographic packing for separating permanent gases, including H2, Ar, O,, N,,CH,, and CO, in various gas mixtures. It is particularly useful for separating Ar and O2at superambient column temperatures over a wide concentration range. Some of the more significant applications of this material for trace-gas analysis are discussed below. Separation of O 2and Ar in He. The separating ability of calcium chabazite was studied with two different gas blends. At the higher concentration (-200 ppmv), an R value of 2.70 was observed at 313 K. A typical chromatogram for this standard is shown in Figure 2. For the low-concentration standard (-3 ppm), the resolution obtained at 313 K was 2.94. Additional dilution of a low-concentration standard produced a blend containing 60 ppbv H2,37 ppbv Ar, and 28 ppbv 0, in helium, which was used to study this separation further. These concentrations combined with the sample volume correspond to actual gas masses of 5.6, 69, and 42 pg for H,, Ar, and 02, respectively. The optimum operating parameters for separating this very dilute mixture were determined, and a chromatogram from this study is shown in Figure 3. Even at these low-mass loadings, a significant signal-to-noise ratio was observed. Prior to oxidation of the calcium chabazite, these low concentrations ( 5 3 ppmv) of oxygen could not be observed. Oxidation of the material also eliminated the O2 peak tailing observed at higher concentrations.
Figure 3. Two chromatograms obtained with the Valco HID dualcolumn unit. The first chromatogram (A) shows the separation of trace N, CH, and CO in He using a 1 . 5 4 length of calcium chabazite at 378 K. The second (B)shows the separation of trace H, Ar. and O2 In He using a 6-ft column at 358 K.
.
The Ar/Oz R value calculated from the chromatogram in Figure 3 was 1.41. This chromatogram shows both the ability of calcium chabazite to separate low concentrations of permanent gases and that of the HID to detect them. Separation of N,, CH,, and CO in He. The separation of Nz, CH,, and CO is another important application where calcium chabazite shows improved performance. At a column temperature of 378 K and column length of 1.5 ft, the three gases are well resolved from one another and can be quantified in under 4 min (see Figure 3). Although the leading side of the N2 peak does elute very close to the combined H,/Ar/02 peak, it does not present a problem for quantitating the material present. Such early elution times, especially for N2, are an important factor in obtaining low-ppbv detection limits. The chromatogram (Figure 3) obtained with this material demonstrates the separation of concentrations as low as 90 ppbv N,(117 pg), 40 ppbv CH, (30 pg), and 58 ppbv CO (75 pg) in He. Furthermore, even lower concentrations of these gases could be separated on this material as indicated by the signal-to-noise ratio of the HID. In fact, the detection limits will probably be limited by the detection system, not the chabazite. Molecular sieve 5A can also separate N2,CH,, and CO at above-ambient temperatures, but the elution times are longer. Also, porous polymers are reported to separate Ar, Oz, N,, and CO, but only with 36-ft columns maintained at room temperature. In addition, the reported analysis was not performed with trace-level gases (8),possibly because porous polymers have been observed to exhibit a continual elution of low-concentrationcontaminations, which currently prevents them from being used to analyze ppbv levels of certain gases. Separation of Ne, Ar, 02,and N2in He. Separation time is an important consideration when one is assaying gases for impurities. Separation of Ne, Ar, 02,and N2 can typically
1116
ANALYTICAL CHEMISTRY, VOL. 61, NO. 10, MAY 15, 1989
2 0 3
a
0
Figure 4. Chromatogram for separation of trace Ne, Ar, 0,, and N, in He, performed at 343 K with a 6-ft column linked to the Tracor USD.
take over 40 min with a 33-ft molecular sieve 5A column a t room temperature (20). In contrast, with calcium chabazite the time required for analysis of low-ppm levels has been reduced to approximately 11 min with the use of a 6-ft column maintained a t 343 K as shown in Figure 4. Also, in comparison to operating a t ambient temperature, it is easier to maintain constant column temperature, resulting in a greater degree of reproducibility in retentioatime and peak width. Separation of Trace Gases in Other Bulk Gases. Calcium chabazite can also separate low-ppbv concentrations of permanent gases in bulk gas matrices other than He, including H,, Ar, and N2. The separation of low-ppbv concentrations of 0, from bulk H, is especially useful. This separation has not been reported in the open literature at these mass loadings with either molecular sieve 5A or porous polymers. However, 25 ppbv O2 was separated from bulk H, with a 6-ft column packed with calcium chabazite a t 303 K. The standard mixture was analyzed as soon as it was generated, since 0, reacts in bulk H,. Figure 5 shows the high degree of resolution obtained for these two gases. Based on the retention of Ar in this system, if it were present in the standard, it would also be resolved from the bulk H2and trace
Figure 5. Chromatogram for separation of 25 ppbv 0, in H, using a 6-ft column at 303 K with the HID unit.
a U
z
PI-
Y
7
z
02.
We also determined trace concentrations of Ar in O2 by using calcium chabazite. A chromatogram obtained for the separation of 5 ppmv Ar from bulk O2 is shown in Figure 6. This separation used a 10-ft column maintained at 298 K. To perform this analysis with other chromatographic materials, a deoxygenating material is required to remove the bulk O2 from the sample before reaching the detector, or an 0, carrier gas is used to eliminate detector response to 02.In contrast, calcium chabazite allows direct determination of Ar in 0, using a standard gas chromatograph and inert carrier gas. Conclusion. Our studies have shown that when chabazite is highly calcium-exchanged, thoroughly dehydrated, and exposed to an oxidizing atmosphere a t high temperature, it can quickly and easily separate percent to ppbv concentrations of Ar and 0, a t above ambient temperatures. This material also has the ability of performing other unique separations such as trace O2 in a H, matrix and trace Ar in an O2 matrix.
Figure 6. Separation of 5 ppmv Ar in bulk 0, using a 10-ft column at 298 K with the Valco HID system.
ACKNOWLEDGMENT We wish to thank L. Schoonmaker and L. Yoder for performing a large portion of the experimental work described in this paper. LITERATURE CITED ( 1 ) Breck, D. W. Zeolite Molecular Sieves; John Wiley (L Sons: New York, 1974; p 666. (2) Ciambelli, Nato diPaolo; DeSimone. Vincenzo et al. Rend. Accad. Sci. Fis. Mat ., Naples 1983, 50, 227-233. (3) Vaughan. D. E. W. GB 1443197, 1976. (4) Walker, D. A. J. Nature 1966, 209, 197. (5) Karlsson, B. M. Anal. Chern. 1968, 38, 668-669. (6) Melikashvili Institutes of Physical and Organic Chemistry, Academy of Sciences of the Georgian SSR. Translated from Zavod. Lab. 1975, 47(4), 398-401. (7) Verzele. M.; Verstappe. M.; Sandra, P. J. Chromatogr. 1981, 209, 455-457.
Anal. Chem. 1909, 6 1 , 1117-1128 (8) Pollack, G. E.; O'Hara, D.; Hollis, 0. L. J . Chromatogr. Sci. 1984, 2 2 , (9) (10) (11) (12) (13) (14) (15) (16) (17)
343. McNair, H. M.; Ogden, N. W.; Hensley. J. L. Am. Lab. 1985, 34. deZeeun, J.; deNijs, R. C. M. Chrompack Top. 1985, 12, 1. Sievers. R. E.; Giiiis. J. N. Anal. Chem. 1985, 5 7 , 1572-1577. Maroulis, P. J.; Coe, C. G.; Kuznicki, S. M.; Clark, P. J.; Roberts, D. A. US 4,713,362, 1987. Andrawes, F. F.; Gibson, E. K. Anal. Chem. 1980, 5 2 , 846-851. Ettre, L. S.Basic Relationships of Gas Chromatography;Perkin-Elmer Corp.: Norwalk, CT, 1979. Purnell, H. Gas Chromatography; John Wiley & Sons, Inc.: New York, 1962. Littlewood, A. B. Gas Chromatography;Academic Press: New York, 1970. Coe, C. G.; Kuznicki. S. M. US 4,544,378, 1985.
1117
(18) Barrer, R. M.; Mainwaring, D. E. J. Chem. Soc., Dalton Trans. 1972, 1254. (19) Coe, C. G.; Parris, G. E.; Srinivasan, R.; Auvil, S. R . I n New Developments in Zeolite Science and Techno/ogy-~rocee~ings of the 7th International Zeolite Conference, Tokyo; Iijima, A., Ward, J. W., Murakami, Y., Eds.; Elsevier: New York, 1986; p 1033. (20) Brettell, T. A.: Grob, R. L. Am. Lab. 1985, 17, 19.
RECEIVED for review August 30, 1988. Revised manuscript submitted January 27,1989. Accepted February 14,1989. The authors wish to thank Air Products and Chemicals, Inc., for permission to publish this work.
Improved Resolution of Glycoproteins by Chromatography with Concanavalin A Immobilized on Microparticulate Silica via Temperature-Programmed Elution Alan F. Bergold' and Peter W. Carr* Department of Chemistry and Institute f o r Advanced S t u d y in Biological Process Technology, University of Minnesota, Minneapolis, Minnesota 55455
The ability of the column temperature to control elution in the affinity chromatography of glycoproteins (e.g., Ovalbumin and horseradish peroxidase) on silica immobilized concanavalin A has been studied. Column temperature programs can be achieved by placing a small HPLC column within a commercial mobile phase preheater assembly. It is shown that elution of adsorbed proteins can be initiated by changing the column temperature without altering the chemical composition of the mobile phase. Further, due to the enhancement in the rate of dissociation of the sample from the ligand, the peaks are narrowed. The resolution can be controlled by changing the initial temperature, dwell time at the initial temperature, and the rate of change of the temperature program. Addition of a competitive binding agent to the mobile phase decreases the temperature needed to elute strongly retained proteins. The effect of heating the column through many thermal cycles is assessed by periodically measuring the retention of a small monosaccharide that binds to the immobilized concanavalin A. The effect of two different immobilization procedures (glutaraldehyde and carbonyidiimldazoie), as well as the effect of including a monosaccharide In the mobile phase, on the stability of the column is easily monitored by thermal elution chromatography. The effect of column temperature on the above glycoproteins has been assessed through studies of enzyme activities and anion exchange and isoelectric focusing patterns before and subsequent to temperature-programmed elution affinity chromatography.
INTRODUCTION In the decade that has passed since Mosbach (1) introduced the technique of high-performance affinity chromatography (HPAC), our understanding of the factors governing the success or failure of this technique has increased enormously (2). Despite this increased understanding, it is fair to say that Present address: P r o t e i n Structure Facility, University of Iowa, I o w a C i t y , IA 52242. 0003-2700/89/0361-1117$01.50/0
HPAC has not mirrored the success of either conventional high-performance liquid chromatography or classical affinity chromatography. Slow desorption kinetics of the surface adsorbed eluite from the immobilized affinity ligand has been identified by many groups (3-5) as the principal rate-limiting factor in HPAC and, therefore, as a controlling factor in achieving narrow zones. In this work we attempt to exploit the column temperature as a means of enhancing the rate of desorption. The elution method used in affinity chromatography depends on the goals of the separation as well as the thermodynamics and kinetics of the biospecific interactions involved. Desorption is typically achieved by changes in pH (6) or ionic strength (7), the use of an inhibitor (8),or a combination of these methods (9). The most common elution scheme for the separation of monosaccharides and glycoproteins when bound to concanavalin A (Con A) is biospecific elution with a competitive inhibitor. The reason for this is 2-fold. First, competitive inhibitors such as a-methyl-D-mannopyranoside (MDM) are readily available and inexpensive. Second, elution with an inhibitor is very mild and selective. Nonbiospecific elution by changes in pH or ionic strength is less commonly used because of the susceptibility of Con A to loss of Ca2+and Mn2+ at low pH (10) and the relative insensitivity of the interaction between Con A and polysaccharides to ionic strength (11). A major disadvantage of inhibitor-based elution schemes is the above mentioned dominant role of surface desorption kinetics on the efficiency of Con A affinity columns (12). We and others have previously pointed out the extreme sluggishness of this process for small monovalent eluities (3-5, 12-14). The desorption of such species from immobilized Con A is very slow (approximately 1 s-l) compared to the rates observed for the desorption of small solutes from a reversed-phase column (>lo0 s-l) (15). As a consequence of the slow kinetics, traditional gradient elution methods are relatively ineffective in sharpening broad, kinetically tailed peaks. Muller has studied the effect of kinetics on the elution of glycoproteins from HPAC columns (12). Under conditions of a linear gradient of 0.1 M MDM, using ovalbumin as the solute, no protein other than the 0 1969 American Chemical Society
