High-speed GC Analysis of VOCs: Tunable Selectivitv and Column Selection J
VFiE.
Part 2 of a Two-Part Article
I niques begin with the
olatile or-
widely used in the manufacturing and service industries as well as at government facilities. Even at very low concentrations, many of these materials are toxic or carcinogenic and thus represent a direct health risk to workers. Because virtually all VOCs adversely affect overall air quality, the Clean Air Act Amendments of 1990 require increased monitoring of major emission sonrces. State and local regulations regarding air quality and emission abatement as well as standards for clean-up operationsat chemical waste and spill remediation sites are changing and will be in a state of flux for the foreseeable future. For all of these reasons, there needs to be increased monitoring of VOCs. Many environmental VOC monitoring problems focus on a small number of target compounds. This is particularly m e for field and screening applications and for industrial hygiene monitoring. For these applications, high-speed gas chromatography [GC) may be extremely useful in reducing analysis time and increasing sample throughput. Instrumentation requirements and inlet systems for high-speed GC were discussed last month in this journal. Although these techniques promise analysis times of a few seconds to a few tens of seconds, the probability of coincidental peak overlap is quite large ( 1 ) . Therefore, careful attention
must be paid to column selectivity and special selectivity adjustment techniques to take full advantage of the capabilities of these high-speed GC instruments. Using tandem or serially linked columns with different stationary phases, it is possible to tune the selectivity of a separation by changing either the relative lengths of the columns (2,3) or the pressure at the junction points between the columns (4, 5). This provides a straightforward way to minimize overlapping peaks and completely separate more complex mixtures. Recently, high-speed column switching has been used to further improve the selectivity of high-speed GC analysis for VOCs. These tech-
428 A Envimn. S i . Technol.. Voi. 28, No. 9, 1994
RICHARD SACKS MICHAEL AKARD University of Michigan Ann Arbor, MI48104
rapid, partial separation of the analytes on a nonpolar capillary separation column. The partially separated compounds are automatically transferred in groups to either a polar or a nonpolar capillary column to complete the separation (6). A separate detector monitors each of the two columns used for the final separation. The combination of high-speed column witching and presure-adjustable selectivity yields a complete separation of mixtures containing as many as 20-30 compounds in less than one minute. In this report, these techniques will be discussed and applied to mixtures of VOCs of environmental interest. Stationary phases for VOC analysis High-speed GC separations require high-speed inlet systems with relatively short capillary columns. The price paid for reduced analysis time and increased sample throughput is a significant loss in chromatographic resolution, reducing the number of components that can be completely separated and increasing the probability of peak overlap. This makes the choice of stationary phase more important in highspeed GC than with conventional GC. Most VOC analyses are performed using the stationary phases described in Table 1. Many capillary GC separations of VOCs use 0.25 or 0.32 mm i.d. bonded-phase capillary columns. Microbore capil-
0013-936X/94/0927-428A$04.50/0C 3 1994 American Chemical Society
lary columns (< 0.1 mm i.d.) offer greater resolution and zone capacity (i.e., the number of peaks, if perfectly spaced, that will fit in the chromatogram at a specified resolution) for comparable analysis times, but they are more difficult to use and require smaller samples than do capillary columns. Microbore columns are only beginning to be widely applied to VOC analysis of environmental samples.
rier gas velocity in the nonpolar column and, thus, increases retention times in that section. At the same time, gas flow velocity in the polar column increases, leading to shorter retention times on that column. The result is a decrease in the influence of the polar column on the overall
separation. A pressure reduction at the column midpoint has the opposite effect.
switchins The instrument shown in Figure 1 also allows high-speed column switching after the initial separa-
I :olumn
Mixed phases and tandem columns For relatively simple mixtures, a single stationary phase may be adequate for high-speed GC analysis. However, in many cases, significant time can be saved by using mixed stationary phases or tandem columns of different selectivity. In this scheme, the phase mixture or column length ratios can be tuned for the optimal separation of the target compounds. Several mixed-phase capillary columns for analysis of specific sets of VOCs, as required by U S . regulatory agencies, are available commercially. Continuously adjustable selectivity can be achieved with a pressure-tunable tandem column ensemble. Figure 1 shows a diagram of a pressure-tunable system for highspeed GC.A high-speed cryofocusing inlet system (described in more detail in Reference 7 ) introduces a narrow sample vapor plug into the column ensemble. A nonpolar capillary column C, (blue) is followed by a highly polar column C, (red). Varying the pressure of carrier gas supply G, adjusts the selectivity of the tandem column ensemble. If this pressure increases, the pressure difference along the nonpolar column decreases and along the polar column increases. This reduces car-
TABLE 1
Properties of stationary phases frequently used for VOC analysis Oirnathyl polysiloxane (DW
Phenyl methyl polyoiloxsne
Trifluompmpyl methyl I siloxane
FFh ~~
Nonpolar High
LOW
Intermediate Intermediate
Envimn. Sci. Techmi., Vol. 28. No.9.1094 419 A
tion on the nonpolar column. Column switching increases the zone capacity and enhances the tuning characteristics of the nonpolar/ polar tandem pair. For the instrument described here, components or groups of components that elute from column C, can he switched to another nonpolar column C, (blue) using carrier gas source G, to divert these components to the second nonpolar column. Because columns C, and C, have the same selectivity, temperature, and stationary phase film thickness, no tuning is possible for this tandem pair. For the sample chromatogram shown from column C,, the wellseparated peaks 1,2,5,6, and 9 were switched to column C,. These compounds were then detected on FID, (for flame ionization detection) as shown in the figure. Component pairs 3 and 4, and 7 and 8 cannot be completely separated on C, and were instead selectively switched to column C,. The component pairs are passed through the tandem combination of C, and C,, which was previously tuned to completely separate these components, and the components were detected by FID,. The pressure at G, was earlier optimized for separating these components.
k = (tR- t,)/t,
= k,
+
(tm/&)(k, - kJ (2) Plots of the overall capacity factor k versus t,& are straight lines with slopes equal to k& and intercepts equal to k,. Thus, the slopes of such plots provide a useful comparison of the polarity of the polar column C, relative to the nonpolar column C,. Optimization strategy A wide range of selectivities can be obtained by coupling columns with different stationary phases in series. In many cases it is possible to tune the selectivity for a particular mixture and drastically reduce analysis time. The DMP and PEG columns show significantly greater efficiency (smaller plate height) than the BCP
Selectivity tuning with tandem column ensembles The instrument shown in Figure 1 allows for the direct measurement of the carrier gas hold-up times, t,.,>and k2, h m the individual columns C, and C,, respectively. [Hold-up time is the carrier gas transport time through the column. It is determined by measuring the retention time for a compound, which is completely unretained on the column and thus travels at the same velocity as the carrier gas. Methane is usually chosen when detection is by FILl.1 These key parameters aid in the optimization of the tandem columns. The overall hold-up time t,,,, the sum oft,,,, and tmZ,is related to the overall retention time t, and an overall capacity factor k as describkd in Equation l. [Capacity factor is the mass of a sample component in the stationary phase ratioed to the mass in the moving phase (carrier gas).]
The overall capacity factor can be described in terms of the capacity factors k, and k, for columns C, and C,, respectively, and the time haction k),,, as shown in Equation 2. 430 A Environ. Sci. Technoi.. Voi. 28, No. 9, 1994
and TFP columns. This feature, coupled with the very large selectivity range available for many environmentally important compounds, makes the DMP-PEG tandem ensemble particularly useful. Optimization refers to the selection of the column hold-up time fraction t,,,,lt, that will provide the best separation of the worst-case component pair (critical pair). Optimization is based on Equation 2 and a relative resolution function R, as given in Equation 3. R, = (k,k,)/(k,+l)
(3)
Here, k, and k, are the overall capacity factors for components a and b on the tandem column ensemble, and k., is their average value. This function is a reliable measure of chromatographic resolution for all capacity factor values. Figure 2(a)
FIGURE 2
Plots of overall capacity factor kversus time fraction tm9/tm( and relative reteniion window diagram (b) for seven chlorinaled hydrocarbon compounds using a DMP-PEG tandem column ensemble (a)
shows plots of overall capacity fac- I tor versus $,t, for a mixture of seven chlorinated hydrocarbon compounds using a DMP-PEG tandem combination. Table 2 lists the compounds and their boiling points. Figure 2(b) shows plots o &e relative resolution (Equation 3 of the critical pair using the samc t m?It, axis. Different componen paus constitute the critical pair foi different t,Jt, values. For example, for tmA/t, value. near 0.1, components E and F I (1,1,2-trichloroethane and tetrachloroethylene) are the critical pair; for t,,, It, values near 0.45, components b and B (I-chloropentane anc FIGURE 3 1,2-dichloroethane) form the criti Chromatograms of the seven-component chlorinatedhydrocarbon cal oair. In Fieure 2. the critical mixture using the DMP-PEG tandem column ensemble with tm: pair: in (a) a n i the corresponding values of 0.26 (a), 0.37 (b), 0.40 (c), and 0.48 (d) regions in the relative resolution plot in (b) are identified by various colors. Note that there are seven points, corresponding to tmZ/t, values, where pairs of plots cross in Figure 2(a). For each of these points, the corresponding compounds co-elute, and the relative retention is zero. Also note that there are several “windows” of large relative resolution of the critical pair. Plots of thr type shown in Figure 2(b) often an called window diagrams ( 8 ) . ThL tm,l& value of 0.58 gives the greatest relative resolution and would allow for the best separation of the worst case pair of components with a specified total column length. A high degree of selectivity control for these chlorinated hydrocarbon compounds is obtained using the DMP-PEG combination. Monochlorinated compounds have negative slopes for the plots in Figurp 2(a), as does the symmetric tetra chloroethylene molecule. Positiw slopes are obtained for all other polychlorinated compounds. A large positive slope is obtained for l,l,Z-trichloroethane. Figure 3 shows chromatograms obtained with t,,lt, values of 0.26 (a), 0.37 (bl, 0.40 (cl, and 0.48 (d). components B and D are not ade- ooiiing points, and tnew concentraThe critical pair in each chromato- quately separated. tions are listed in Table 3. For both gram is indicated by contrasting chromatograms, the column temcolors. For chromatogram (a), com- Tunable selectivity with column perature was 4o oc. ponents A and C are not adequately switching On the 8.0-m DMP column sevseparated; note that plots A and C in For more complex mixtures, even era1 unresolved or poorly resolved Figure 2(a) cross at at,& value of tunable selectivity may not allow a peaks are observed. In particular, about 0.22. For chromatogram (b) in complete separation on a time scale note unresolved pairs C-D,J-K, and Figure 3 all components are well of 30-60 s. Figure 4 shows chro- L-M. However, many of the composeparated: note the peak at t,jt, = matograms of a 21-component mix- nents are separated in about 40 s. 0.37 in Figure 2 b ) . Using this t,,lt, ture using an 8-m-long nonpolar On the tandem column ensemble, value, the overall column length DMP column (a) and the tandem the separation is poorer, and more could be significantly reduced with combination of a 4.0-m-long DMP overlapping components are oba further reduction in analysis time. column, and a 4.0-m-long PEG col- served. For chromatograms (c) and (d), umn (b). The components, their Figure 5(a) shows plots of overall
I
Environ. Sci. Technol., Val. 28, NO. 9,1994 431 A
I
:hromatograms of a 21-component mixture using only a DMP olumn (a) and using the DMP-PEG tandem column ensemble (b
1
blue-line window diagram plot in Figure 5 b ) . If only the eight components represented by red lines in Figure 5(a) are switched from the nonpolar C, column to the polar C, column and the other 13 components are allowed to continue on the nonpolar C, column, the window diagram shown by the red plot in Figure 5(b) is obtained. Here, there are a variety of hold-up time fractions that result in an adequate separation. Figure 6 shows chromatograms obtained simultaneously from the two FIDs when the 8 components indicated by the red plots in Figufe 5(a) are switched from column C, to column C, and the pressure set to give a t,It, value of about 0.50.Note that a1121 components are adequately separated in about 40 s. The use of column switching and two parallel measurement channels increases available zone capacity and more efficiently uses available zone capacity. Conclusions Tunable selectivity is necessary for the high-speed separation of complex mixtures. Even for relatively simple mixtures the probability of peak overlap is large, and tnnable selectivity often can simificantlv reduce reauired
Componnt
Methylene chloride t-l,Z-Dichlomethylene c-I,2-Dichlomethylene Chloroform 1,Z-Dichlomethane Carbon tetrachloride ErOmOdichlomrnethane Methylcyclohexane c-I ,3-Dichloropropne 1-1,3-Dichloropropene 1,1,2-Trichlomethane 2-Fluomtoluene 1,3-Dichloropropane Cycloheptane Tetrachlorwthylene Methyl pentanoate Chlorobenzene 1-Chlorohexane m-Xylene 4-Heptanc"o-Xylene capacity factor k versus bold-up time fraction t,Jt, for all 2 1 components. The components represented by blue lines are all adequately separated on the 8.0-m DMP column. If all 2 1 components (red and blue lines) are separated on the
tandem combination of the DMP and the PEG columns, no bold-up time fraction will adequately separate the mixture in the short timeframe considered here. This is illustrated by the large number of lowamplitude windows shown in the
4392 A Environ. Sci. Technol.. Voi. 28, No. 9. 1994
tandem combination of a nonpolar column and a highly polar column provides a relatively wide polarity range. The use of a nonpolar dimethyl polysiloxane column in tandem with a polyethyleneglycol column provides high selectivityand a wide tuning range for chlorinated hydrocarbons. The work described here employed two series-coupled columns. Extension of these principles to three or more series-coupled columns is in progress and should provide additional tunability with further reductions in analysis times. High-speed column switching can be extremely useful for the highspeed separation of mixtures involving more than 10 or 15 components. The increased zone capacity provided by a dual-channel detection system, together with selective peak switching, reduces the number of components separated in the tuned portion of the system. The combination of selective peak switching and tandem column tunable selectivity may be extremely useful in increasing sample throughput for many environmental samples. The techniques are relatively straightforward to implement, and the optimization strategies described in this report are readily employed
Plots of overall capacity factor kversus time fraction tm,/tm (a) nd relative retention window diagrams (b) for a 21-component mixture using a DMP-PEG tandem column ensemble
UUY*
r l Y I r x r7"Yy""..,Ly
UY-
YY...
I . . .
samples or library capacity factor values for target compounds on the individual columns. Because each separation is very rapid, methods development is also rapid, and the time and cost savings can be enormous. These techniques could have a major impact on the analysis of VOCs in environmental samples.
Richard Sacks,co-founderand president of Chromatofastand professor of chemistryat the UniversityofMichigan,received his Ph.D. in analyticol chemistryfrom the University of Wisconsin in 1969. He has published more than 100 papers on atomic spectroscapy, high-speed CC,and chemical instrumentation. He has been working in the area of high-speed GC since 1982 ond served as a consultant to seveml chemical companies.
Y 0.2
0.8
06
0.4
1 .o
k,"" The blue-line window diagram is lor the case where all 21 components are separated on the tandem column ensemble. The red-line window diagram IS for the case where Only the eight components indicated by red lines in part (a) are Separated On the tandem column ensemble See Table 3 for mixture components.
r
r w m E6
lichael Akard, a research chemist at hromatofast, received his B.S. degree in chemistry and mathematics from the University of New Mexico in 1990. In 1994, he received his Ph.D. in analytical chemistry from the University of Michigan. His dissertation work focused on selectivity tuning strategies for highspeed gas chromotogmphy.
2
Chromatoarams of the 21-com~onentmixture usina column switching 2nd tunable selectiv/ty: (a), chromatogram obtained from the nomolar DMP column:_(bl. ... chromatoaram obtained from the PEG column
-
I
References (1) (21
Davies, M.: Giddings, J. C. Anal. Chem. 1983.55,418-24. Vililobos, R.;Pearson, R. IST Transac-
tion 1988.25, 5 5 4 0 . (3) Sandra, P. et al. HRCbCC 1985, 8 . 782-97. (4) Deans, D.:Scott. 1. Anal. Chem. 1973, 45,113741. (5) M a w r , T.:Engewald. W.:Steinborn, A. J. Chromatogr. 1990,517,7746. (6) Akard. M.; Sacks, R. I. Chromatog. Sci. 1993,31, 297-304. (7) Klemp, M.; Akard. M.; Sacks, R. Anal. Chem. 1993,65,2516-21. (8) Laub, R.: Purnell. I. Anal. Chem. 1976,48,1720. Environ. Sci. Technol., Vol. 28, No. 9, 1994 433 A