Analysis of complex samples by coupled-column chromatography

matographic method for the analysis of complex samples. This method Incorporates three chromatographic steps, the first two of which serve to separate...
0 downloads 0 Views 619KB Size
Anal. Chem. 1982, 5 4 , 169-173

169

Analysis of Complex Samples by Coupled-Column Chromakography Kenneth Ogan+ and Elena Katz The Perkin-Elmer Corporation, Main Avenue, Norwalk, Connecticut 06856

Reversed-phase chromatography Is coupled with slze-exciu$ion chromatography to generate a multldlmensional chromatographlc method for the anaiysls of complex samples. Thls method Incorporates three chromatographlc steps, the flrst two of which serve to separate the sample into fractions which are more readlly handled by the final hlgh-resoiutlon analytlcal step. The first fractlonatlon step is a low-resolution version of the flnal step and selects compounds with retentlon times comparabile to those of the compounds of analytlcal Interest. The second fractionation step uses size-excluslon chromatography to select compounds with molecular slzes comparable to those of the compounds of analytical Interest. These flrst two lfractlonatlon steps are coupled on-llne. Individual COmpOMndS are ldentlfled In a final hlgh-resolutlon, reversed-phase (chromatographic step. Thls muitldimenslonal chromatography system is appiled to the determlnatlon of polycyclic aromatlc hydrocarbons In coal llqulds and 011s as an example. Wkth thls system,such samples are fractionated and ready for the final analytlcal step In less than 20 min.

Modern chromatographic methods can readily separate a fair number of compounds in complex mixtures. Nevertheless, there are still samples that are too complex to be analyzed in a single chromatographic step, petroleum and coal liquefaction products being examples. Such samples require a preliminary fractionation step in order to divide the sample into portions that are more amenable to the determination of individual compounds. The high selectivity of chromatographic techniques has resulted in their extensive use in both steps of such analytical methods. The final, high-resolution separation step, in particular, has been dramatically improved by modern technology. The preliminary fractionation step, which in the pmt has utilized classical column chromatography, can also be significantly improved by the incorporation of modern chromatographic technology. In the preliminary fractionation step, the sample is divided into groups of compounds sharing some characteristic defined by the fractionation method. Chromatographic fractionation selects groups of compounds which have similar retention times, reflecting comparable type and strength of interaction with the stationary phase. The choice of a chromatographic system which utilizes some characteristic unique to the compounds of analytical interest yields a very effective fractionation method. By this means, the number of compounds applied to the analytical column is reduced, as compared to the direct injection of the whole sample. This results in fewer overlapping peaks and better resolution in the analytical chromatogram. Many chromatographic fractionation methods have utilized gel filtration or fdze-exclusion chromatography (SEC). A few recent reports have described the application of microparticulate SEC columns to sample preparation (1-7) and have demonstrated the improvements in speed and efficiency that can be achieved with these columns. Fractionation by size-exclusion chromatography selects only those compounds comparable in molecular size to the com0003-2700/82/0354-0169$01.25/0

pounds of interest. For some sample types, this is all that is needed to achieve sufficient resolution in the final analytical separation. For more complex samples, however, even a narrow molecular size fraction will contain a great many compounds with a wide variety of functional groups. These compounds would elute over a very wide retention range in the final analytical chromatographic step. Strongly retained compounds are particularly troublesome, as they cause lengthy analyses and can even alter the chromatographic behavior of the column. In order to eliminate these strongly retained compounds, an additional chromatographic fractionation step was added. This new fractionation step selected only those compounds which eluted with the compounds of interest on the final high-resolution column. The advantages of high-efficiency columns and modern instrumentation technology were utilized to couple this additional fractionation step directly to the SEC fractionation step, bypassing the need for any additional sample handling. Thus, this was a true two-dimensional chromatography system, with two different chromatographic systems coupled together on-line. Two other groups have reported the use of a C18bondedphase column directly coupled to an SEC column for analytical purposes (2,3). In both of these cases, the CISbonded-phase column was used for an analytical separation after fractionation of the sample on the SEC column. The analytical system described in this paper uses three steps: first a fractionation step on a CI8 column, next a fractionation step on an SEC column, and finally a high-resolution separation on a C18 column. EXPERIMENTAL SECTION Reagents. Acetonitrile (Wgrade, Fisher Scientific, Fair Lawn, NJ, and tetrahydrofuran, THF (UV grade, Burdick & Jackson Labs, Muskegon, MI, or OmniSolv, MCB Cincinnati, OH), were used as received. Water was filtered, deionized, and purified over a carbon bed system (Continental Water Conditioning Corp., El Paso, TX). A variety of solvent-refined coal samples were supplied to us from an industry source. Information as to site or type of liquefaction process, or as to what stage of the process was being sampled, was not available. These samples varied in color, odor, and viscosity. Apparatus. All equipment and columns were from PerkinElmer. The coupled-column chromatography equipment (see Figure 1) included a Series 2/2 liquid chromatograph, a Model 65T column oven and W detector, and a Model 3000 fluorescence detector. Injection valve no. 1 was a Rheodyne Model 7125. Injection valve no. 2, a Rheodyne Model 7001, was part of the Model 420 autosampler. A Sigma 10 chromatography data system controlled the injection timing of the second step and collected retention time and peak area data. The columns were an HC-ODS column (CISbonded phase), 0.26 by 25 cm, a Shodex SEC guard column, and a Shodex A 802-S SEC column (Styrene-divinylbenzene support). The columns were mounted in the Model 65-T column oven, which was set to 50 "C. One pump of the Series 212 liquid chromatograph delivered 70% THF in acetonitrile to the HC-ODS column through injection valve no. 1 (see Figure l),at a flow rate of 0.5 mL/min. The second pump delivered pure THF to the SEC column through injection valve no. 2 at a flow rate of 1.0 0 1982 American Chemical Society

170

ANALYTICAL CHEMISTRY, VOL. 54, NO. 2, FEBRUARY 1982 CHROMATOGRAPHY

AUTOSAMPLER

/I INJECTION W L

TfON VALVE XZ

Table I. PAH Peak Identification 1 indene 17 2 naphthalene 3 1-methylnaphthalene 18 4 2-methylnaphthalene 19 5 acenaphthene 20 6 fluorene 21 7 1,4-dimethylnaphthalene 22 8 phenanthrene 23 9 anthracene 24 1 0 fluoranthene 25 11 pyrene 1 2 9,10-dimethylanthracene 26 1 3 2-methylanthracene 27 14 benzo[a] fluorene 28 1 5 benz[a]anthracene 29 16 chrysene 30 31

7,12-dimethylbenz [a ] anthracene benzo[e] pyrene benzo[ b] fluoranthene dibenz[a,c]anthracene benzo [ h ] fluoranthene benzo [ a ]pyrene dibenz[a,h] anthracene benzo[ghi]perylene indeno[l,2,3-cd]pyrene dibenzo[a,e]pyrene benzo[ b] chrysene dibenzo[b,h]pyrene picene pquarterphenyl coronene

COLLECT FRACTIONS

Flgure 1. Block diagram of the arrangement of the pumps, columns, and valves used for the coupled-column chromatography system.

mL/min. The eluent line from the HC-ODS column was connected to the sample loading port of the second injection valve. Two instrument interfaces to the Sigma 10 were used, one to time the operation of the injection valve in the Model 420 autosampler and one to collect retention time and area data from the fluorescence detector located after the SEC column. Injection of the second injection valve was initiated by the closure of a relay on this first interface, under control of the external control section of a Sigma 10 method. Rotation of injection valve no. 2 back to the INJECT position initiated a run on the second interface, which collected data from the Model 3000 fluorescence detector. The final, high-resolution separation step (8,9) was carried out with a Series 3B liquid chromatograph, a Model 650-10s fluorescence detector, and a PAH/10 column (0.26 x 25 cm). (The PAH/ 10 column uses a CISbonded-phase material specifically tested to separate the 16 PAHs on the E.P.A. priority pollutant list (8,10-12). The HC-ODS and PAH/10 stationary phases are virtually identical, the PAH/10 material being culled from HC-ODS material on the basis of improved selectivity for particular PAHs.) Peak area and retention data were collected and calculated by the Sigma 10 chromatography data station. This separation used a two-segment gradient: first, a 20-min linear segment from 40% to 100% acetonitrile in water and then a 40-min segment at 100% acetonitrile. The flow rate was 0.5 mL/min. Detection wavelengths were Ex = 280 nm and Em = 340 nm initially, with a change to Ex = 305 nm and Em = 430 nm at 17.7 min into the run. Procedure. The sample was diluted 1 5 or 1:lO in 70% tetrahydrofuran in acetonitrile, the mobile phase used for the HC-ODS column. Injection of this sample with the first injection valve initiated the f i i t Sigma 10 interface. Twenty seconds before the unretained peak from the HC-ODS column reached the second injection valve, this first interface directed the autosampler controller to begin its injection cycle. Injection valve no. 2 was rotated to the FILL position and, 20 s later, when the unretained peak had filled the sample loop of this valve, it was rotated back to the INJECT position, sending the contents of the sample loop to the SEC column. Experiments with standards served to identify the time interval for collection of the SEC eluent fraction, this interval being 9.8-11.3 min for the PAH analysis used as an example. This collected fraction was then analyzed by using an analytical LC method for PAHs (8, 9).

RESULTS AND DISCUSSION Figure 2 is the chromatogram of 31 PAH standards on a Perkin-Elmer PAH/10 column. The direct injection of petroleum or coal liquefaction samples gave chromatograms with a great many more peaks, including many eluting much later

0

5

10

I5

20

25

30

35

40

45

50

55

60

MINUTES

Flgure 2. Chromatogram of PAH standards on a Perkin-Elmer PAH/lO column. Condltlons are given In the text. For peak identification, see Table I. Coronene is peak no. 31.

than the standards in Figure 2. Fractionation of these samples by SEC was attempted first. The samples were injected onto an SEC column, and the fraction of the eluent corresponding to the elution volume of the PAH standards was collected. Small portions (2 wL) of these fractions were then injected onto the analytical PAHI10 column. The resulting analytical chromatograms exhibited improved resolution (relative to injection of the untreated sample), but with peaks eluting long after those in Figure 2. However, the analytical retention times in runs subsequent to these injections were significantly shortened. This problem was accentuated by repeated injections of the SEC fractions. The original retention times could be recovered by purging the column with THF and chloroform, indicating that the problem was due to very nonpolar compounds that remained on the column and altered the chromatography in subsequent runs. While the SEC fractionation removed many of the overlapping compounds by selection of a narrow molecular size range, the compounds in the SEC fraction still encompassed a wide polarity range, In order to eliminate the compounds that were very strongly retained on the analytical column, a highly condensed version of the separation on the final analytical column was added. The stationary phase of the column used in this additional fractionation step had to be identical with that of the final analytical column; hence, an HC-ODS column was used for this step. This new fractionation step preceded the SEC fractionation step. The complete method incorporated two fractionation steps and a final, high-resolution step. The first step was a lowresolution separation process that functioned to limit the range of retention times which had to be handled in the final analytical step. The second fractionation step served to reduce

ANALYTICAL CHEMISTRY, VOL. 54, NO. 2, FEBRUARY 1982 SAMPLE

h

171

COUPLED -COLUMN CHROMATOGRAPY: COLUMN 1' UV: X - 3 4 0 n m

Cis COLUMN RETENTION RANGE SELECTION

L 41.t

I

WmKKlT ALm)-INJECTION

WITH AUTO-INJECTION

I WALENE

I@

ISEC COLUMN MOLECULAR SIZE RANGE SELECTION

WASTE

WRONENE I

CB COLUMN

~JR'ROMA;?OORAPHV

HI H RE OLUTION

DETECTOR

\

Figure 3. Pictorial presentatlon of the multidimensional analytical procedure.

- -o

MINUTES

-

0

MINUTES

the number of interfering compounds, thereby simplifying the body of the final analytical chromatogram. These steps are summarized in Figure 3, The details of the arrangement of pumps, valves, columns, and detectors are shown in Figure 1. The analysis of PAHs was used as an example of the application of this system. The appropriate molbile phase for the retention-time fractionation step was determined with model compounds. Coronene, peak 31 in Figure 2, was arbitrarily selected as the largest PAH to be included in the analyses. Ovalene was selected as the model compound for tlhe larger PAHs that were!to be excluded from the analyses. With a mobile phase of 70% THF in acetonitrile, coronene was unretained, but ovalene was slightly retained, on the HC-OD8 column, as shown in the left-hand chromatogram in Figure 4. With this mobile phase, all compounds less polar than eoronene would also be unretained, while the larger, more nonpolar IPAHs would be retained. The left-hand chromatogram in Figure 4 was obtained with the Model 65-'I? detector inserted in the eluent line after injection valve no. 2 (at the asterisk in Figure 1). This chromatogram was obtained without the second injection valve operating. The solid lines in the chromatograms on the right in Figure 4 weire obtained with the second injection valve operating. In this specific system, the unretained peak from the HC-ODS column appeared in the sample loop of the second injection valve 2.34 min after injection at the first valve. Rotation of the second valve from the FILL to the INJECT position at this time caused the coronene peak (which is unretained) to disappear, as is demonstrated in the upper right chromatogram. An additional delay of 0.31 min (the difference in retention time between these two peaks) resulted in ovalene being removed hom the eluent stream, us is demonstrated in the bottom right chromatogram. In this case, it was the ovalene that was injected onto the SEC column. These two chromatograms demonstrated the nearly exclusive selection of one or the other of these two compounds. Additional confirmation of this selectivity was provided by the SEC chromatograms obtained simultaneously with the chromatograms in Figure 4. The 2.01-min delay time (selection of the unretained peak) was used in all subsequent work reported here. An important aspect of this method was the automated transfer of esserntidy the complete peak from the fist column to the second column. This permitted the nearly complete recovery of the compounds of interest. This high recovery was achieved by matching the volume of the unretained peak from the CIS bonded-phase column to the maximum allowable injection volume for the SEC column. Selection of one of these columns set limits on the physical dimensions of the other column. It was determined that a maximum of 100 p L could

Figure 4. Chromatogram of coronene and ovalene on the HC-ODS column, obtained with a UV detector located after the second Injection valve, at the asterlsk In Figure 1.

be injected onto the SEC column before peak broadening occurred. Hence, the volume of the unretained peak from the C18 bonded-phase column had to be about 100 pL. With a smaller peak volume, some of the retained peaks could be in the sample loop with the unretained peak. A larger unretained peak volume would mean less than complete recovery of this peak. From the plate theory, the peak volume is inversely related to the square root of the efficiency, N , of the column. The chromatographic peak is usually modeled by a Gaussian peak with width parameter u. For a pure Gaussian peak, 6a includes 99.9% of the peak volume, while 4u includes 95% of the peak volume. For the unretained peak, the retention volume is just the geometrical volume of the empty column times the porosity, E, of the packing material. (The porosity, e, is the fractional volume not occupied by the solid, impenetrable part of the support material. A typical value for c is 0.75.)

The combination of these two relationships gives

r,2L = uN1/2/at

(1)

where L and r, are the length and inner radius of the column. For a given column length (and required resolution), the peak volume is determined by the column radius. For u = 17 pL (6a = 102 ML),a column length of 250 mm, and an efficiency of 5000 plates, this equation gives re = 1.4 mm. On this basis, a commercially available 2.6 mm diameter column was used for this first column. The solid line in Figure 5 is the SEC chromatogram of four PAH standards injected directly onto the SEC column (de. tection at Ex = 305 nm, Em = 430 nm). The dotted line is the SEC chromatogram obtained by using the coupled-column system to fractionate a solvent-refiied coal sample. The PAH standards correspond to the later part of the SRC chromatogram, indicating the presence of many compounds of larger molecular size (weight) in this sample. The use of the low-resolution separation on a CISbondedphase column prior to fractionation on the SEC column also solved another problem associated with the sole use of the SEC column. In Figure 5, the first three PAH standards eluted according to their size, but coronene eluted much later than expected. Similar anomalously long retention times were observed for other highly fused compounds (for example, benzo[ghi]perylene). For these compounds, evidently adsorption as well as size exclusion was occurring on the Shodex 8025 column. The adsorption of PAHS on styrene-divinyl-

172

ANALYTICAL CHEMISTRY, VOL. 54, NO. 2, FEBRUARY 1982

-

REVERSED PHASE CHROMATOGRAPHY

B

C

,

MINUTES

W&o.40min. I

,

8:s 1;.2min

The SEC chromatogram of an SRC sample overlaid on the SEC chromatogram of four PAH standards. Peak identification: p-QPh = pquaterphenyl, DiBA = dibenz[a ,h]anthracene, An = anthracene, and Cor = coronene. The 400 pL slices of SEC eluent that were collected for analyses by reversed-phase chromatography are indicated below the chromatograms. Flgure 5.

benzene columns has been noted before (13). (One of the reasons for the use of elevated temperatures in the method was the reduction of these absorptive interactions.) There are many large, highly fused compounds in samples such as the solvent-refined coal samples. Injection of these sample directly onto the SEC column resulted in broad SEC chromatograms continuing well beyond the permeation limit of the column. In some extreme cases, these SEC chromatograms took 1or 2 h. This would have significantly reduced the rate of sample preparation. The SEC chromatograms obtained with the coupled-column system, such as in Figure 5, exhibited little tailing, but returned to the base line at times close to that corresponding to the permeation volume. The strongly retained compounds on the CIS column were evidently just those compounds that were adsorbed on the SEC column. In order to assess the effect of the SEC fractionation step on the final analytical step, we collected narrow 400-pLslices of the SEC eluent after the injection of the SRC sample, as shown in the lower part of Figure 5. Each of these narrow slices were then subjected to the final reversed-phase chromatography step on the PAH/ 10 column. The chromatograms from these analyses are shown in Figure 6, in order of increasing retention volume on the SEC column. The first chromatograms, obtained from volume slices early in the SEC chromatogram, correspond to higher molecular weight molecules, while successive chromatograms correspond to lower molecular weight (size) molecules. The amplitude of the broad, featureless hump of the chromatogram in Figure 6A decreases and individual peaks become apparent in progressive chromatograms. A great many more isomers are possible for higher molecular weight compounds, and this broad hump is due to many overlapping, chromatographically similar, compounds. Many of these compounds would be extracted in a liquid-liquid extraction sample treatment process and would cause a broad background in the analytical chromatogram of the concentrated extract, an artifact familiar to those working in the field. From injection of PAH standards, it was determined that the SEC eluent between 9.8 and 11.3 min contained the PAHs of interest for this application example. This fraction did not contain any of the compounds present in the first two chro-

MINUTES

Analytlcal chromatograms of collected slices of the SEC eluent from Injection of the SRC sample (see Figure 5). The SEC fractlons were collected in sequence: A = 8.8-9.2 mln, B = 9.2-9.6 min, C = 9.6-10.0 min, D = 10.0-10.4 mln, E = 10.4-10.8 min, and F = 10.8-1 1.2 mln. The analytical chromatography conditions are the same as In Figure 2. Figure 6.

n

I

I

0

IO

I 20

I

I

30 40 MINUTES

I 50

I 60

Figure 7. Analytical chromatogram of a solvent-refined coal sample, fractionated with the coupled-column system, as described in the text. The analytical chromatography conditions are the same as in Figure 2.

I

b

E 8

b

8

3

Y

I 0

1

I

IO

20

I

i

i

I

30

'40

50

60

MINUTES

Figure 8. Analytlcal chromatograms of a process oil sample, fractlonated with the coupled-column system, as described in the text. Analytical chromatography condltions are as before.

ANALYTICAL CHEMISTRY, VOL. 54, NO. 2, FEBRUARY 1982

Table 11. PAH cConcentrations arocreosote process matic oil, solvent, oil, PAH

benz[a] anthracene benzo[ blfluoranthene benzo[k] fluoranthene benzo[a]pyrene dibenz[a,h :I anthracene benzotghi ] perylene

MJg/mL d m L 2650 600 250 9 20

1100

1420

200 60

430 570 6 20

MJghL

266 60 25 92

110

matograms in Figure 6. Figures 7 and 8 are the final analytical chromatograms that were obtained in the analysis of two representative rgamples, an SRC sample and a process oil. While there are still very many peaks ]present, these chromatograms have much less of the broad hump found in the chromatograms from these samples when prepared by more conventional means. This reduction of the number of overlapping peaks greatly improves the accuracy of quantitative measurements. The combined method, coupled-column fractionation and high-resolution ]reversed-phaseseparation, was applied to the analysis for PAHs in several samples. For the SRC sample, a portion of the collected SEC fraction was diluted 1 : l O with acetonitrile and then 10 p L was injected on the analytical PAH/10 system. The analytical chromatogram is shown in Figure 7 . (The chromatographic conditions are the same as for Figure 2.) The corresponding analytical chromatogram for the process oil sample is shown in Figure 8. The SEC fraction from this sample was diluted 1:50 with acetonitrile before injection on the analytical column. There are no peaks eluting beyond 50 min in the analytical chromatograms in Figures 7 and 8. This correlates well with the 45-min retention time of coronene and demonstrates that the coupled-column system was successful in defining the retention range of the compounds reaching the final analytical column. This has important advantages in achieving the maximum analysis rate. The maximum retention time is defined (and adjustable), and there are no strongly retained compounds to be purged from the column, so the column can be more quickly reequilibrated. The retention time reproducibility for these and other analytical chromatograms was well within 0.5% (relative standard deviation). This result, in contrast to the shifting retention times observed with direct injection or with SEC fractionation alone, clearly demonstrates the success of the coupled-column preparation method. This degree of precision

173

is required in order to utilize the benefib of a computer-based system for data collection and reduction from complex chromatograms such as those shown in Figures 7 and 8. Table I1 gives the results for selected important PAHs in three samples. This particular application example covered a wide range of PAHs, running from 2 to 6 fused rings. A simpler analytical step could be used if the analytical fraction were less complex. The coupled-columnsystem described could be easily adjusted to be more selective in terms of retention range and molecular size range. Since such fractions can be generated very quickly, an alternative approach might be to use a simpler final analytical step on several sequential fractions from the coupled-column system. Although only its application to PAH analyses has been discussed here, this is a general method with potential utility for many other analyses of complex mixtures. Analytical fractions can be generated from very complex samples in less than 20 min. Retention times and peak areas in the analytical chromatograms are very reproducible. Also, while CIS reversed-phase columns are used in this example, the concept is just as applicable to analyses based on a normal-phase silica column. In this latter case, a silica column in the first stage of the fractionation method would prevent very polar compounds from reaching the analytical silica column and reducing its activity.

ACKNOWLEDGMENT We wish to thank J. G. Atwood for helpful comments and encouragement.

LITERATURE CITED Popl, M.; Stejaskal, M.; Mostecky, J. Anal. Chem. 1975, 4 7 , 1947-1950. Ernl, F.; Frel, R. W. J . Chromatogr. 1978, 149, 561-569. Johnson, E. L.; Gloor, R.; Majors, R. E. J . Chromatogr. 1978, 149, 571-585. MaJors,R. E.; Johnson, E. L. J . Chromatogr. 1978, 167, 17-30. Dark, W. A. J . Chromatogr. S d . 1978, 16, 289-293. Hausler, D. W.; Hellgeth, J. W.; McNair, H. M.; Taylor, L. T. J . Chromatogr. Scl. 1979, 17, 617-623. Apffel, J. A.; Alfredson, T. V.; Majors, R. E. J . Chromatogr. 1981, 206, 43-57. Ogan, K.; Katz, E.; Slavln, W. Anal. Chem. 1979, 50, 1315-1320. Katz, E.; Ogan, K. Chromatogr. News/. 1980, 8 , 18-20. Ogan, K.; Katz, E. J . Chromatogr. 1980, 188, 115-127. Katz, E.; Ogan, K. J . Liq. Chromatogr. 1980, 3 , 1151-1163. Atwood, J. G.; Qoldsteln, J. J . Chromatogr. Sci. 1980, 18, 650-654. Popl, M.; Fahnrich, J.; Stejskal, M. J . Chromatogr. Sci. 1978, 14, 537-540.

RECEIVED for review August 27, 1981. Accepted October 20, 1981.