Narrow bore open tubular columns for improvement of gas

and quantitation of tetrachlorodibenzo-p-dioxins by HRGC and HRGC/MS ... Narrow-bore wall-coated open-tubular columns for fast high-resolution gas...
0 downloads 0 Views 762KB Size
Anal. Chem. 1983, 55, 2115-2120

21115

emission from the indium dimer becomes depressed. The most interesting interference was due to phosphorus. At mass flows as small as 200 pg of P/s the background emission was depressed resulting in a “negative” peak. Unlike the inverted peaks from hydrocarbons which are symmetrical, the inverted peaks due to phosphorus had a substantial amount of tailing. Similar results were reported for phosphorus by Versino and Rossi ( 4 ) . A similar effect was not observed with burners (U) or (C). The resolution of the monochromator used in this study made it impossible to identify the existence of residual indium chloride emission in the presence of the indium dimer background. If a portion of the background were due to the chloride emission, the “negative” peak could have been caused by the formation of indium phosphide (rnp 1070 “C)on the catalytic surface. A temporary poisoning of the surface by the phosphorus would thus reduce the emission from the background indium chloride.

drogen embrittlement and literally falls apart. Stainless steel and nickel are also prone to this problem, to a lesser extent. Thus, long-term changes in response may be due to more than just the covering of catalytic sites. The mechanism for the formation of InCl in a flame is not known. If the first step requires the formation of HCl, as has been suggested by previous investigators (1-5), then the catalytic action must occur after this step since no enhancement of the response is observed with burner (B). Further clarification of the mechanism will require the identification of some of the intermediate species in the reaction. Registry No. Stainless steel, 12597-68-1;tantalum, 7440-25-7; nickel, 7440-02-0;quartz, 14808-60-7;platinum, 7440-06-4; graphite, 7782-42-5; BHC, 58-89-9; lindane, 58-89-9; heptachlor, 76-44-8;aldrin, 309-00-2;heptachlor epoxide, 1024-57-3;p,p’-DDE, 72-55-9; dieldrin, 60-57-1; o,p’-DDD, 53-19-0; endrin, 72-20-8; o,p’-DDT, 789-02-6;p,p’-DDT, 50-29-3;indium, 7440-74-6; indium chloride, 13465-10-6.

CONCLUSION

LITERATURE CITED

The variations in sensitivity that are observed when different flame tip materials are used appear to be caused by surface catalysis. An enhanced response requires the simultaneous presence of indium, the catalytic surface, and the decomposition products of the sample from the hydrogen-rich lower flame. The catalytic surface is sulbject to short-term poisoning by phoriphorus compounds and long-term poisoning by indium oxide. The mechanism of poisoning appears to be the covering of surface sites by the nonvolatile indium phosphide and oxide. The degreee to which the poisoning iEi reversible is complicated by the physical changes that occur to many of the metals used in this study, when they are exposed to hydrogen and high temperatures. Tantalum is subject to severe hy-

Gilbert, P. T. Anal. Chem. 1968, 38,1921. Overfield, C. W.; Winefordner, J. D. J . Chromatogr. Sci. 1970, 8 , 233. Bowman, M. C.;Beroza, M. J . Chromatogr. Scl. 1071, 9 , 44. Versino, B.; Rossi, G. Chromatographla 1971, 4 , 331. Gutsche, B.; Hudiger, K.; Herrmann, R. 2. Anal. Chem. 1077, 285, 103. (6) Gutsche, B.; tierrmann, R. 2.Anal. Chem. 1970, 2 4 9 , 168. (7) Gutsche, B.; Herrmann, R. 2.Anal. Chem. 1971, 253,257. (8) Folestad, S.;Josefsson, B. J . Chromatogr. 1981, 203,173. (9) Patterson. P. L.: Howe, R. L.; Abu-Shumavs, A. Anal. Chem. 1078, 50,339. (10) Pearse, R. W. B.; Gaydon, A. G. “Identification of Molecular Spectra”, 3rd ed; Chapman and Hall: London, 1965. (11) Patterson, P. L. Anal. Chem. 1078, 50,345. (12) Mavrodlneanu, R.; Ed. “Analytical Flame Spectroscopy”; MacMlllan: London, 1970; pp 286-295.

RECEIVED for review June 6, 1983. Accepted August 4, 1988.

Narrow Bore Open Tubular Columns for Improvement of Gas Chromatographic Analysis Time Marie France Gonnord and Georges Guiochon* Ecole Polytechnique, Lnboratoire de Chiimie Analytique Physique, Route de Saclay, 91128 Palaiseau Cedex, France Frank I. Onuska National Water Research Institute, Analytical Methods Division, Burlington, Ontario, Canada L7R4A6

Large improvementsin speed of analysis or column efflclency can be achleved by using narrow bore open tubular columns wlth most comnnerclai equipment, provided some minor rnodificatlons are made, Le., allowing the use of higher inlet pressures (up to 20 atm)),llmitlng Instrumental dead volume and electronic t h e constants, and posslbly permltting the use of hydrogen as carrler gas. Experlmentaii results as well as theoretical considerations demonstrate that new possibilities exlst, offered by the use of 80-100 pm 1.d. capillary columns.

Increasing interest in high-performance gas chromatography for the analysis of complex mixtures has led chromatographers to investigate the use of narrow bore open tubular columns with inner diameters below 150 ym (1).Most analytical work using capillary columns is still carried out with the same

diameter originally used by Golay (2),Desty ( 3 ) ,and Scott ( 4 ) ,i.e., 250 ym. Considerable advances made in the technologies of sample handling, injection, and detection, however, permit the use of narrower diameters as demonstrated i.n previous publications (5-8). Although the first study demonstrating the interest of narrow bore capillary columns for high-speed gas chromatography dates back to 1962 (9) and the fluidic sampling systems to 1973 (IO),only Simon and Szepesy (11) seem to have carried out some experimental work in this field before the late 1970’s. This is probably because the somewhiit acrobatic experimental conditions of the earlier work (9)together with some doubts regarding the detection limits limited its use by analysts. The use of fluidic sampling systems (5-7,101 permits the fast injection of narrow sample bands necessary for falat analysis when short, very narrow open tubular columns are

0003-2700/83/0355-2115$01.5O/0 0 1983 American Chemical Soclety

2116

ANALYTICAL CHEMISTRY, VOL. 55, NO. 13, NOVEMBER 1983

used. A specially designed instrument is requested. Thus it is interesting to investigate the potential of using such columns with conventional, commercial instruments without modifying them or with minimum, simple modifications, for example, to the pneumatic system. I t will not be possible to achieve very fast analysis but analysis requiring a very large separation power still would be possible in a reasonably short time. We present here some theoretical considerations and experimental results demonstrating the possibility of increasing markedly the speed of analysis and/or the resolution power. This can be achieved with commercial gas chromatographs by using long, 100 pm i.d. open tubular columns. These columns have proven to be as easy to manufacture and handle and as stable and reliable as the more conventional 250 pm i.d. columns. THEORETICAL SECTION There are two ways to increase the resolution power of a chromatographic column: to make it longer or to make it more efficient, i.e., with a shorter height equivalent to a theoretical plate. In the first case the analysis time is markedly increased, especially in gas chromatography, a t high resolution power. I t has been shown (1, 12) that in this case the analysis time is proportional to the 3/2 power of the column length, because of the compressibility of the mobile gas phase. In practice this method is clearly limited as the analysis time cannot exceed a few hours for the method to be useful. For over 20 years, owing to the pioneering work by Golay (2),Desty (3),Scott ( 4 ) ,and Halasz (13) and the recent development of reliable coating technologies, due especially to Grob and Grob (14) and a few others, interest was focused on the systematic reduction of the contributions of resistance to mass transfer in the stationary phase. This was achieved by improving the quality of the film of stationary phase on the wall surface, its stability, and the inertness of the wall itself. This work on coating technologies has permitted the easy and reliable preparation of high-efficiency, stable columns capable of providing routine analysis for months and of handling all kinds of polar compounds. Since the current performances are very close those predicted by the Golay equation (2) which casts the ultimate limit, little further development is expected now, except maybe in the case of very polar compounds or of compounds with extremely low vapor pressure (14). The only possibility of improving resolution power without increasing analysis time is the reduction in column diameter. This has not yet been carefully explored because of the necessity of using specially designed instruments and the more difficult design of equipment satisfying the specifications for a small enough contribution to band broadening. H , the equivalent to a theoretical plate of an open tubular column, is given by the Golay equation (2)

H = -20g UO

9(P: - PO4)(P?- Po2)

8PO3(Pi3 - Po3)

and the last one by (14)

. 3 Pi2 - P o 2 I=-P O 2 Pi3 - Po3

(3)

where Pi and Po are the column inlet and outlet pressures, respectively. At high values of the inlet to outlet pressure ratio, f is equal to 9/g, and will not be taken into account in the following, while j is equivalent to 3P0/2Pi, so the contribution of the last term becomes small or even negligible. On the other hand, an equipment contribution must be added (3,given by

He =

ff.,2uo2

L(l

(4)

+ kq2

where L is the column length and ue2the variance of the equipment contribution to the band variance. This variance contribution is the sum of different contributions, arising in the sampling system, the connecting tubes, the detector, and its ancillary electronics and data system (16) = CXp? (5) Xi is 1for a Gaussian contribution like the detector response time, 1/12 for a rectangular contribution like the one of the injection profile achieved with a fluidic sampling system (5-7), and around 1/4 for a typical syringe injection (7). The columns considered here have a large separation power, they are relatively long, and, as their inner diameter is small, their pneumatic resistance is large. Furthermore they have to be operated at large outlet velocities. Therefore, the inlet pressure is large. Thus Po can be neglected compared to the inlet pressure (5-7,12), and the outlet velocity uois given by ffe2

d:Pi2

u, =

647LP0

(6)

where 7 is the viscosity of the carrier gas. In eq 1we can neglect the third term, first, because we know how to prepare columns with small enough film thickness, or rather small resistance to mass transfer in the stationary phase (14),and second, because the decompression correction factor is very small (15). Equation 1 can thus be written H = B / u , cu, (la)

+

with

B = 20,

6k’+ 1 1 k r 2 d: k’ df2 + 1 +96(1 - uo + -u0 + k q 2 0, 6(1 + kq2 Di

(1) where uois the outlet carrier gas velocity, D, and DI are the diffusion coefficients of the solute in the carrier gas and stationary phase, respectively, k ’is the column capacity factor, d, is the column inner diameter, and df is the average thickness of the film of stationary phase on the wall. Equation 1 is strictly valid for a straight cylindrical tube with a negligible pressure drop that is used with equipment whose contribution to band broadening is very small. In practice, with the coil diameters (10-15 cm) used, there is no correction in the velocity range typically achieved (less than 1-2 m/s with hydrogen). If the pressure drop is not negligible, the first two terms of the plate height must be multiplied by (14)

f=

K a = 1 + 6k’+ 11kr2 3(1 + k q 2 If we can neglect the equipment contributions, the minimum value of the plate height is given by the equation

Hmin= 2

m =

1

+ 6k’+ l l k ’ 2 3(1 + k q 2

dc

= -K’ 2

(7)

and the corresponding velocity is

The column length is related to the plate number and plate height by the conventional equation

ANALYTICAL CHEMISTRY, VOL. 55, NO. 13, NOVEMBER 1983

L=NH Elimination of H and L between eq la, 6, and 10 and solving for u, gives

Combining eq Ilb, 8, and 10 gives the pressure a t optimum flow velocity

in agreement with eq 27 of ref 1. The analysis time is the retention time of the last component of interest of the analyzed mixture. I t is given by t A := ( L / a ) ( l k(I (12)

+

and can be derived by the combination of eq 7,9,11, and 12

The column length is proportional to cl, (cf. eq 7 and 9) and thus the average velocity for columns giving the same efficiency is independent of the column diameter, in agreement with earlier results (1,8). These results offer a framework to compare performances obtained from open tubular columns made using glass tubes with different diameters. For example a t constant efficiency or resolution power constant ( N in eq 13) we see that the analysis time is proportional to the column diameter. Compared to the current 0.25 mm i.d. columns, narrow bore open tubular columns offer the possibility to reduce by a factor 2 to 8 the analysis time ( 1 , 7,8,11). At constant analysis time, a reduction in column diameter offers the possibility of increasing the efficiency in proportion to d;2/3 and the resolution power, as well ais the separation number in comparable conditions, in proportion to dd1l3. Again cornpared to the current 0.25 mm i.d. columns, narrow bore open tubular columns offer the possibility of increasing the efficiency by a factor of 1.6-4 a t constant analysis time and the resolution by a factor of up to 2. Finally it has been suggested (3,9) to compare columns in terms of production of effective theoretical plates. The effective theoretical plate is the plate number calculated by using the corrected retention time or time spent in the stationary phase by the solute

2'117

heating block with temperature control. The injector port was modified to accept a double septum for safety reasons (as the inlet pressure is very high). The injector includes a septum purge and a glass insert. A telescopic column holder was built, permitting easy column handling. The pressure controller of the equipment is replaced by a high-pressure controller (Tescom Corp. Minneapolis MN), Model 26-1024-24, which allows control of the hydrogen pressure up to 35 atm. Hydrogen is preferred because it has the lowest viscosity and the largest diffusion coefficient (cf. eq 13). Analysis time with hydrogen will be about half that with helium for the same column efficiency. We have been using hydrogen on a routine basis in thi>laboratory for 25 years without any incident. The only safety measure required is to check that the column is tightly fastened to the chromatograph and measure the flow rate before switching on the oven. Care is taken to position the column end at the very exit of the flame ionization detector jet, just at the flame basis, to minimize dead volumes. Specially adapted capillary jet and column insert are useful. The electrometers of the chromatographs were found satisfactory (time constant below 200 ms). At the electrometer output, the signal was amplified 100 times by the mean of a laboratory made amplifier with negligible time constant. Data for the measurement of efficiency were collected with a HP 3497A programmable multimeter (Hewlett-Packard, Palo Alto, CA) and a CBM3032 home computer with a floppy disk unit and a graphic plotter. When recorder time constant was not a problem, a Philips (Paris, France) P8202 electronic recorder was used. The time necessary to carry out an injection with a syringe was estimated to be 50 ms, thus contributing (eq 5) by 210 ms2 to the band variance. The detector volume is a few microliters. It is swept by a flow of hydrogen makeup, in addition to the carrier gas of 30 mL/min. The corresponding time constant is around 5 ms, with a variance contribution of 25 ms2. The major contribution is thus the one of the time constant of the detector electronics, which has been estimated at 175 ms from measurements with an oscilloscope. Faster amplifiers are readily available if needed (6). Column Preparation. Glass tubes were drawn on a laboratory-made drawing machine from 6 mm 0.d. Corning Pyrex tubes; 1.0 and 0.7 mm 0.d. tubes with 86 to 290 pm i.d. were prepared for this study (cf. Table I). The columns were prepared by using the technique described by Grob (17) and Lee (18)and coated by the static method with an OV-1 (Supelco, Bellefonte, PA) solution in glass-distilled pentane (Burdick and Jackson, Muskegon, MI). Solution concentrations between 0.2 and 0.6% were used. The film thickness is derived from this concentration by using the conventional relationship

with C in (v/v) % and df and d, in the same unit. Long, narrow columns (d, < 100 pm, L > 30 m) were coated by the dynamic mercury plug technique (19). Hence the theoretical plate production is obtained by combining eq 13 arid 14

which shows that plate production increases with decreasing column diameter and number of plates produced (12). It is maximum for a rather low v a h e of k' (0.91). In eq 15, N is the maximum number of plates the column can generate, as predicted by the Golay equation (eq In). E'XPERIMENTAL SECTION Equipment. Commercial gas chromatographs were used in this study, with minor modifications: a Carlo Erba (Milano,Italy) Fractovap, series 2150 GC, a Perkin-Elmer (Norwalk,CT) Sigma 3, and a Varian (Walnut Creek, CA) 3700. The standard sampling port was replaced by a split-splitless injector from Cairlo Erba (SL 490), inserted in a laboratory-made

RESULTS A N D DISCUSSION Plots of the height equivalent to a theoretical plate vs. flow velocity for different columns are shown in Figure 1. The minimum plate heights are given in Table I together with the values derived from eq 7. The coating efficiency is the ratio of the theoretical plate height to the experimental one, where the theoretical plate height is given by eq la, thus neglecting the resistance to mass transfer in the stationary phase. A coating efficiency larger than 0.80 is deemed acceptable; if it is larger than 0.90, the column can be considered as very good. Thus the 108 pm i,d. column is very good; the 136 pm i.d. column is fair. There is obviously a problem with the 86 pm i.d. column, which can be explained by the Contribution of the instrument to band broadening as discussed later. Also given in Table I are the inlet pressures corresponding to optimum conditions as calculated from eq 11. For this calculation the diffusion coefficient is derived from the Fuller,

2118

ANALYTICAL CHEMISTRY, VOL. 55, NO. 13, NOVEMBER 1983 H

(mm) 30 25

' I : P

20

25

20

15

d,

15

10

10

05

200pm

Ak

31

Ok' 4 7

05 15

30

45

60(cm

9)

15

H (mm

b

30

1

8

45

60

,

75

( c m 8)

!;F 25 .

h

E

Y

cd

3

d,

20

d

Yc: t-Io

*o?

I

05

0 k'4.7

I

l5 .10 .

15

30

45

60 1cm:s)

15

30

45

00

.05

1-

290Um

A k'3 1 0 k'4.7

0 . .

4 a

c)

a

$

v

an

.s A4

75

(cm's)

1

Figure 1. Van Deemter plots for columns with different internal dlameters: (a) d , = 86 pm: (b) d , = 108 pm; (c) d , = 136 pm; (d) d o = 290 pm.

Table 11. Variation of the Average Flow Velocity with the Column Length for a 108 wm i.d. capillary columna 10

P= u, t,= L , m~ P~IP, cmis L I U , s N 42 66 296 000 27.8 7 140 489 000 9 33 46 22.6 440 1064000 13.2 100 Column a D, = 0.36 cm2/s,k' = 7,u o = 200 cm/s. length.

Schettler, and Giddings equation (20) and the viscosity of hydrogen is interpolated at 170 "C from handbook data (21). The agreement with the measured data is striking: carrier gas pressure corresponding to the optimum flow velocity can be correctly predicted using this equation. The agreement between measured and predicted values of the outlet gas velocity was not as good, most probably because the direct determination of the outlet velocity, carried out through flow rate measurements is quite imprecise in the range of flow rate considered. It has been shown that optimum performance, Le., the minimum analysis time for a given efficiency is obtained for a gas velocity and a pressure slightly larger than those corresponding to the maximum efficiency ( I ) . In any case the data in Table I show that the use of narrow-bore columns to achieve very large efficiencies does not necessarily require the use of very large pressures, a t least in the range covered in this study. Table I1 shows the pressures that would be required to operate longer columns and the corresponding values of the average velocity, hold-up time, and efficiencies for a compound with k' = 7. For such a compound an efficiency of 1 X lo6 plates could be achieved with a 100 m long column, with an analysis time of about 1 h, which is quite attractive for the analysis of complex mixtures. If helium is used as a carrier gas, the results are markedly deteriorated because of its much larger viscosity and slightly smaller diffusion coefficient. In order to study the effect of the instrument contribution to band broadening, we have determined the variation of the

ANALYTICAL CHEMISTRY, VOL. 55, NO. 13, NOVEMBER 1983 2119 Table 111. Calculated and Measured HETP of Various Alkanes on a 108 pm i.d. Capillary Column c,, s D,,cm2/s heor, wm hmeasd, I.tm A I.tm eluent k' 88 110 24 1.55 x 10-5 0.418 nC14 1.37 92 7 2.36 x 10-5 nCl6 3.25 0.390 93 98 5 nc,, 5.00 2.73 x 10-5 0.378 96 98 104 2 3.05 x 10-5 0.367 nC1* 7.68 0.357 99 94 1 3.31 X nc,, 11.29 a A , HETP contribution resulting from extracolumn band broadening, calculated by using eq 4 with U , = 175 ms, contribution of the response time of the amplifier.

L 50

140

J h----

Jw

-- . p -

I ,

40

120

30

2b

lb

T I m e ,mln

100 Tnmperature

3

80

60

40

O C

Flgure 2. Chrormatographlc analysis of a distillation cut of a crude oil (C,< C2,,) on an OV-1 glass open tubular column: d , = 108 pm; L = 33 m; carrier gas, HP; temperature program, 40-150 O C , at 2

OC/mln.

HETP of a series of alhanes as a function of k'on the 108 pm i.d. open tubular column. The results are reported in Table 111. The data were acquired with a CBM 3032 microcomputer so the major contribution originated in the amplifier response time. This contribution accounts fairly well for the difference between the predicted and measured values of the plate height. The same contribution becomes very large in the case of the 86 pm column which explains thle rather poor results obtained. Replacing the equipment amplifier by a faster one with 10 ms response time, effectively results again in performances close to those predicted but unfortunately also with a marked increase in the noise level. Improvements in the detector design and its connection to the electrometer appear necessary. The detection limit with a 108 pm i.d. 33 m long column is less than 1 pg for n-hexadecane (dmldt = miV/2/tR(2r)1/2 =5 X g/s. Noise ca. 1 X A). Figure 2 shows a chromatogram obtained with a 100-ng sample of crude oil distillate. It should be emphasized that this chromatogram exhibits almost no hump in the middle region, in contrast to what is obtainecl, this chromatogram indicates a significant improvement in the separation of that kind of mixture, showing a considerable decrease of interferences between the elution profiles of minor components.

CONCLUSION This work demonstrates that it is possible to use open tubular columns having much smaller inner diameters than usual with conimercial equipment after only minor modifications, the main modification being the replacement of the pressure controller. The time constant of the electronics then becomes the limiting factor, the weakest link in the chain. This reduction in column diameter permits 'a proportional reduction in analysis time for difficult or very difficult separations. The peak width must be larger than 10-12 times the electronics response time to limit the loss of efficiency to an acceptable level (7)so the choice of the smaller diameter

to use with the equipment also depends on the difficulty of the separation and the length of column necessary. For example significant loss of performance for compounds with moderate retention is already observed with the 33 m long, 100 pm i.d. column, which nevertheless offers almost 400000 plates for compounds with k'around 3 (cf. Table 111). On the other hand it can be calculated that a 100 m long, 85 pm i.d. diameter column could be used to achieve well over 1 X lo6 plates for the same compounds, using the same instrument. The increase in analysis time due to the increased column length results in an increase in bandwidth which compensates for the reduction in that bandwidth due to the better efficiency. Further improvement in performances would be obtained with faster electrometers. We cannot accept, however, a decrease in detector and amplifier response time at the expense of an increase in the noise level. On the contrary a faster detector to be used with narrow bore open tubular columns should be less noisy than the current ones for the following reason. The sample capacity of the column is proportional to the square of the column diameter. Only very small amounts of sample can be injected on a narrow bore column and sensitive detection becomes a serious problem, all the more because the analysis of more complex mixtures, made possible by these columns because of their higher separation power, tends to require lower detection limits: it is useless to be able to separate many compounds if a number of them cannot be detected. On the other hand the detection limits also depend on the analysis time, since the shorter the residence time, the smaller the extent of dilution of the compound vapor in the carrier gas and the lower the detection limits in terms of actual muss of component in the sample. Accordingly at constant efficiency, the analysis time being proportional to column diameter (eq 13),the detection limits, in terms of component concentration in the sample, increase in proportion to l/dc which is not good, while at constant analysis time, the dilution remains about the same, the detection limits increase in proportion to l/d,2, which is worse, and the plate number increases as dc-2/3 (eq 13). In both cases we would like to reduce the detector noise. Finally it should be underlined that the best GC performances so far have been obtained with narrow bore columns i.d. 34 pm (9) 65 pm (7), or 50 pm (22). In the first case a plate production of 2500 plgtes/s has been achieved a few times in difficult experimental conditions while in the second case about 2000 plates/s were produced on a routine basis. More recently plate productions well in excess of 1000 plates/s were obtained by Schutjes et al. (22) on an 8 m long 50 pm i.d. column (170000 plates in 75 s for k' = 2).

ACKNOWLEDGMENT The technical skills of Guy Pr6au were extremely useful and are highly appreciated.

LITERATURE CITED (1) Guiochon, G. Anal. Chem. 1978, 50, 1812. (2) Golay, M. J. E. "Gas Chromatography 1958"; Desty D. H., Ed.; Butterworths: London 1958; p 36. (3) Desty, D. H. Adv,; Chfomatogr. 1965, 1 , 199. (4) Scott, R. 1'. \N. Gas Chromatography 1960"; Scott, R. P. W., Ed.; ' Butterworths: London 1960; p 144.

2120

Anal. Chem. 1983, 55,2120-2126

(5) Gaspar, G.; Arpino, P.; Gulochon, G. J . Chromatogr. Sci. 1977, 15, 256.

(6) Gaspar, G.; Annino, R.; VidaCMadjar, C.; Guiochon, G. Anal. Chem. 1976, 50, 1512. (7) Gaspar, G.; VidaCMadjar, C.; Guiochon, G. Chromatographla 1982, 75, 125.

(8) Schutjes, C. P. M.; Vermeer, E. A.; Rijks, J. A.: Cramers, C. A. J . Chromatogr. 1982, 253, 1. (9) Desty, D. H.; Goidup, A.; Swanton, W. T. "Gas Chromatography"; Brenner, N. B., Callen, J. E., Weiss, M. D., Eds.; Academic Press: New York, 1962; p 105. (IO) Bowen, B. E.; Cram, S. P.; Leitner, J. E.; Wade, R. L. Anal. Chem. 1973, 45, 2165. (11) Simon, J.; Szepesy, L. J . Chromatogr. 1978, 779, 495. (12) Guiochon, G. Adv. Chromatogr. 1989, 8, 179. (13) Halasz, I.; Hartmann, K.; Helne, E. "Gas Chromatography 1964"; Goldup, A., Ed.; The Institute of Petroleum: London, 1965; p 38. (14) Grob, K.; Grob, G.; Blum, W.; Walther, W. J . Chromatogr. 1982, 244, 179.

(15) Giddlngs, J. C.; Seager, S. L.; Stucki, L. R., Stewart, G. H. Anal. Chem. 1960, 32,867. (16) Sternberg, J. C. Adv. Chromatogr. 1966, 2, 205. (17) Grob, K.; Grob, G. J . Chromatogr. 1976, 725,471. (18) Lee, M. L.; Wrigth, 8. W. J . Chromatogr. 1980, 784, 235. (19) Schomburg, G.; Husmann, H. Chromatographia 1975, 8 , 517. (20) Reid, R. C.; Prauznitz, J. M.; Scherwood, T. K. "The Properties of Gases and Llquids", 3rd ed.; McGraw-Hill: New York, 1977; Chapter 7, p 544. (21) Weast, R. C., Ed. "Handbook of Chemistry and Physics", 60th ed.; CRC Press: Cleveland, OH, 1980. (22) Schutjes, C. P. M.; Cramers, C. A.; VldaCMadjar, C.; Guiochon, G. Capillary Chromatography, Proceedings 5th International Symposium, Riva del Garda, Italy, April 26-28, 1983: p 304. J . Chromatogr., in press.

RECEIVED for review December 17, 1982. Resubmitted April 18, 1983. Accepted July 5, 1983.

Determination of Long-chain Alkylbenzenes in Environmental Samples by Argentation Thin-Layer Chromatography/High-Resolution Gas Chromatography and Gas Chromatography/Mass Spectrometry Robert P. Eganhouse,*' Edward C. Ruth, and Isaac R. Kaplan Department of Earth and Space Sciences, and Institute of Geophysics and Planetary Physics, Los Angeles, California 90024

The long-chain aikylbenzenes used In the productlon of alkylbenzenesulfonatesurfactants have recently gained attentlon as potential molecular tracers of domestic wastes In the environment. Two methods have been developed for the determlnatlon of both h e a r and branched varieties of these alkylbenzenes in complex envlronmental samples. One relies upon the isolation of a pure aikyibenzene fractlon from total llpld extracts using AgNOS thln-layer chromatography and subsequent analysls by high-resolution gas chromatography. The other approach Involves dlrect analysis of hydrocarbon fractlons by high-resolutlon gas chromatography/electron impact mass spectrometry (HRGCIMS). The AgNO, TLCIGC technlque Is better suited for routine analyses of samples contalnlng only one of the two alkyibenzene types (e.g., wastes, detergents). For the more complex alkylbenrene assemblages sometlmes encountered In waste-affected sedlmentary deposits, GC/MS has the advantage of being able to discriminate between llnear and branched varletles based on differences in their respectlve fragmentation patterns.

Since 1950 alkylbenzenesulfonates have been the dominant surfactants used in commercial detergent formulations (1,2). During the period prior to 1965, synthetic anionic surfactants of this type were produced by sulfonation of complex alkylbenzene mixtures generated by the Friedel-Crafts alkylation of benzene with tetrapropylene. However, the biochemical stability of these tetrapropylene-based surfactants (due to the extensive branching of the alkyl side chains) proved to be environmentally troublesome. This led to the development and introduction in the mid-1960s of a new group of surfacPresent address: Environmental Science Program, University of Massachusetts, Boston, MA 02125.

tants which were simpler in structure, the linear alkylbenzene sulfonates (LAS). The linear alkylbenzenesulfonate surfactants in common use today are synthesized in two steps: (1)a Friedel-Crafts alkylation of benzene using linear internal olefins or chloroparaffins ranging in chain length from Clo to C1, and (2) sulfonation of the benzene ring (para isomers predominate) with H2S04or SO3. The alkylation step produces a mixture of all the possible secondary phenylalkanes (26 isomers for C10-14 side chains) and minor amounts of various cyclic compounds (3-5). Sulfonation of the benzene ring is driven as near to completion as possible; nevertheless, some residual unsulfonated linear alkylbenzenes (LABs) persist and are carried with the sulfonated alkylbenzenes into detergents. Until recently these LAB residues were considered to be of no environmental consequence. Several reports over the last 4 years, however, have shown that the LABs occur in domestic wastes (6),suspended marine particulates near coastal waste discharges (7), and polluted marine and riverine sediments adjacent to urban centers (8, 9). Because of the unique isomer and homologue distributions of synthetic LABs, there is no doubt that those reported to be present in environmental samples arose from contamination by anthropogenic, detergent-bearing wastes. Although it is believed the LABs do not represent a serious environmental threat, they can be exploited as molecular tracers of domestic wastes. In this context, they have a specific advantage over their sulfonated analogues. Because the LABs are hydrocarbons they can be used in efforts to differentiate between waste-derived hydrocarbons deposited in sediments and those contributed by other sources (e.g., combustion products, biogenic residues, oil spills, natural seeps, etc.). In the past, this has been an extremely difficult, if not intractable, problem for the organic geochemist because of the complexity of petroleum assemblages, weathering and diagenetic alterations, and the mul-

0003-2700/83/0355-2120$0 1.50/0 0 1983 American Chemical Society